Videolaryngoscopy versus direct laryngoscopy for adults undergoing tracheal intubation: a Cochrane systematic review and meta-analysis update

Abstract

Background

Tracheal intubation is a commonly performed procedure that can be associated with complications and result in patient harm. Videolaryngoscopy (VL) may decrease this risk as compared with Macintosh direct laryngoscopy (DL). This review evaluates the risk and benefit profile of VL compared with DL in adults.

Methods

We searched MEDLINE, Embase, CENTRAL, and Web of Science on February 27, 2021. We included RCTs comparing VL with DL in patients undergoing tracheal intubation in any setting. We separately compared outcomes according to VL design: Macintosh-style, hyperangulated, and channelled.

Results

A total of 222 RCTs (with 26 149 participants) were included. Most studies had unclear risk of bias in at least one domain, and all were at high risk of performance and detection bias. We found that videolaryngoscopes of any design likely reduce rates of failed intubation (Macintosh-style: risk ratio [RR]=0.41; 95% confidence interval [CI], 0.26–0.65; hyperangulated: RR=0.51; 95% CI, 0.34–0.76; channelled: RR=0.43, 95% CI, 0.30–0.61; moderate-certainty evidence) with increased rates of successful intubation on first attempt and better glottic views across patient groups and settings. Hyperangulated designs are likely favourable in terms of reducing the rate of oesophageal intubation, and result in improved rates of successful intubation in individuals presenting with difficult airway features (P=0.03). We also present other patient-oriented outcomes.

Conclusions

In this systematic review and meta-analysis of trials of adults undergoing tracheal intubation, VL was associated with fewer failed attempts and complications such as hypoxaemia, whereas glottic views were improved.

Systematic review registration

This article is based on a Cochrane Review published in the Cochrane Database of Systematic Reviews (CDSR) 2022, Issue 4, DOI: 10.1002/14651858.CD011136.pub3 (see www.cochranelibrary.com for information). Cochrane Reviews are regularly updated as new evidence emerges and in response to feedback, and the CDSR should be consulted for the most recent version of the review.

Editor's key points.

• •Videolaryngoscopes are used with increasing frequency in adults undergoing tracheal intubation. There is mounting evidence of improved safety compared with direct laryngoscopy.
• •This Cochrane systematic review and meta-analysis, the largest and most complete to date, provides new summative evidence on the benefits of videolaryngoscopy in various settings.
• •Videolaryngoscopes of all designs likely result in fewer failed intubations and hypoxaemic events, and improve glottic views.

Tracheal intubation is a common procedure, performed for almost 40% of general anaesthetics in the UK and as an emergency procedure in the pre-hospital, emergency department (ED), and ICU setting.¹ A Macintosh direct laryngoscope (DL) is classically used to facilitate intubation. Failed or difficult intubation can have catastrophic consequences for individuals and may result in complications, including hypoxaemia, pulmonary aspiration, arrhythmias, cardiac arrest, and death.^2,3 Various factors influence the probability of failed intubation and complications arising from intubation. These include the location in which a person undergoes intubation, factors such as obesity or difficult airway predictors, and the experience of the intubator.^2,4, 5, 6, 7, 8, 9, 10

Videolaryngoscopes (VLs) rely on video technology to transmit an image from the distal portion of the laryngoscope to an eyepiece or monitor where it is viewed by the person performing tracheal intubation. These devices may be flexible or rigid in design for the purpose of assisting in difficult intubations and reducing failure, trauma and other complications. For this review we are interested in rigid VLs, which use a blade to retract the soft tissues and transmit an image to an eyepiece, a screen attached to the handle, or to a standalone monitor. This design enables glottic visualisation without requiring a direct line of sight. Some hospitals and anaesthetic departments have adopted the universal use of VLs for all intubations, and many have a range of different devices routinely available in various areas where intubation is likely to be performed.^11,12

A range of VLs are available in the marketplace with considerable heterogeneity in blade design. Some designs are broadly based on the original Macintosh blade shape, others have a more acutely curved blade or tip, allowing better visualisation of anterior structures at laryngoscopy. Another subset of VLs have a guiding channel incorporated into the blade. In this review, we refer to these three categories as Macintosh-style, hyperangulated, and channelled blade designs, respectively. The technique required for intubating with a Macintosh-style VL is broadly similar to that when using a Macintosh DL. In contrast, the technique when using a hyperangulated device de-emphasises displacement of the tongue and soft tissues, using the shape of the blade to clearly see the anteriorly situated glottis, normally utilising a preformed stylet. Channelled devices guide the tracheal tube but do not permit manipulation of the tip in space, requiring an optimally positioned blade to ensure successful intubation.

The original version of this review with 64 studies and 7044 participants provided evidence that VLs collectively improve glottic visualisation and may reduce the incidence of failed intubation compared with direct laryngoscopy.¹³ Since then many further RCTs have been published, adding to the evidence available in this area. Our review updates and builds on the original, by differentiating outcomes for the three blade designs vs direct laryngoscopy with a Macintosh blade.¹³

Methods

Protocol

This is an abridged version of a Cochrane systematic review, itself an update of previous published Cochrane review.¹³ The original protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO; CRD42015019127) and was published in the Cochrane Library.¹⁴ We prepared this manuscript according to Cochrane guidelines,¹⁵ the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement for systematic reviews and meta-analysis.¹⁶ Changes between protocol and final manuscript are specified in the original Cochrane publication.¹⁷

Eligibility criteria

We included RCTs of both parallel and cross-over design comparing the use of any model of VL with a Macintosh DL in adults aged ≥16 yr undergoing tracheal intubation in any setting. See Supplementary material, Appendix S1, for full criteria.

Information sources

On February 27, 2021, we searched Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE and Embase via Ovid, and Web of Science. Because this was an update of a previous Cochrane review, we limited the searches from January 2015 to 27 February 2021. We applied the Cochrane highly sensitive filter for RCTs in MEDLINE and Embase, using both medical subject headings (MeSH) and free text.¹⁸ We applied no restrictions on language or publication status. We searched www.clinicaltrials.gov and the WHO International Clinical Trials Registry Portal (ICTRP) for ongoing trials on June 10, 2021. We undertook forward citation tracking of eligible studies, and backward citation tracking of key articles, and reviewed studies of AirTraq laryngoscopes excluded in the previous review.¹³ See Supplementary material, Appendix S2, for search strategies.

Data collection and analysis

After removal of duplicates, two review authors (JH and AR) used Covidence to screen titles and abstracts.¹⁹ We reviewed the full texts of potentially relevant titles to identify eligible studies; consensus was reached through discussion. Three review authors piloted the data extraction form on 10 studies. We extracted data for the following outcomes.

Critical outcomes

• 1.Failed intubation: defined as more than three attempts, or change of device or intubator required
• 2.Hypoxaemia: defined as oxygen saturation <94% between start of induction and recovery from anaesthesia
• 3.Successful first attempt at tracheal intubation
• 4.Oesophageal intubation

Important outcomes

• 1.Dental trauma
• 2.The Cormack–Lehane (CL) grade²⁰
• 3.Time for tracheal intubation
• 4.Patient-reported sore throat: within the first 24 h of anaesthesia or closest to 6 h postoperatively
• 5.Number of attempts at tracheal intubation
• 6.Intubation Difficulty Scale (IDS)²¹
• 7.Percentage of Glottic Opening (POGO)²²
• 8.Mortality within 30 days of intubation

Study risk of bias assessment

We assessed the risk of bias using the Cochrane risk of bias tool for the following domains: sequence generation, allocation concealment, blinding of participants, personnel and outcome assessors, incomplete outcome data, selective reporting.²³ We considered bias arising from intubator expertise. It was not possible for the intubator to be blinded to the intervention for this research question and, similarly, it was difficult for assessors of outcomes during intubation to be unaware of the allocation of the participant.

Summary measures and synthesis methods

Data were analysed using the calculator in RevMan Web.²⁴ We calculated risk ratios (RRs) for dichotomous data outcomes with 95% confidence intervals (CIs). We expressed treatment effects for continuous data outcomes as mean differences (MDs) with 95% CI. We did not combine outcomes measured using different scales. We extracted data for each type of VL into separate analyses, whereby we categorised them as Macintosh-style, hyperangulated, or channelled. See Supplementary material, Appendix S3, for devices assigned to each category.

We conducted meta-analyses only when studies were clinically and methodologically comparable. To account for anticipated complexity of this intervention in a broad population, we used random-effects models. We used the I² statistic to quantify the possible degree of statistical heterogeneity and assumed moderate heterogeneity at I² 30–60%, substantial heterogeneity at 50% and 90%, and considerable heterogeneity at 75% and 100%.

For each comparison group, we performed prespecified subgroup analyses for failed intubation: setting (theatre vs non-theatre); obesity (obese vs non-obese participants); airway difficulty (participants with predicted, known or simulated difficulty vs those without such features); intubator experience (inexperienced vs experienced; defined as more than 20 intubations with each device). We combined all three VL designs in post hoc subgroup analyses of these subgroups. We removed extreme outliers where identified and used sensitivity analysis to explore this decision. We used the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach to assess the certainty of the body of evidence associated with seven critical outcomes in the review.²⁵

Results

Study selection and characteristics

After removal of duplicates from the results of the search conducted from January 1, 2015 to February 27, 2021, we screened 2344 titles and abstracts. We assessed 227 full-text records and identified 158 new studies for inclusion in this review update. We included the 64 studies identified in the original review. We identified 222 trials (with 26 149 participants) to include in the review (Fig. 1). See Supplementary material, Appendix S4 for a full list of references. We identified 46 ongoing studies, and 27 studies awaiting classification. These studies are reported in the full version of the Cochrane review.¹⁷

Fig 1.

PRISMA diagram. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analyses.

Twelve studies reported no outcomes of interest to the review. The remaining studies reported at least one outcome of interest. We did not include data for two cluster RCTs because they did not account for the effect of clustering in the manuscript. Six studies were conducted in the prehospital setting, seven in the ED, and eight in the ICU. The remaining studies were conducted in the operating theatre.

Sixteen studies included only obese or morbidly obese individuals, with a further eight studies including a mix of obese and non-obese participants. Five studies included pregnant participants undergoing Caesarean section. Eight studies included adults in cardiac arrest. Twenty-one studies recruited participants with a known or predicted difficult airway, whereas a difficult airway was simulated in 29 studies.

We present a table with the various VL types by category and the studies that included each device in Supplementary material, Appendix S3. Sixty-three studies with 10 222 participants included a Macintosh-style VL as one of the comparisons. One hundred and two studies with 11 857 participants included a hyperangulated VL as one of the comparisons. Seventy-seven studies with 7385 participants included a channelled VL as one of the comparisons. We included 16 studies that used double-lumen tracheal tubes for intubation. Most studies used a two-arm design, comparing one type of VL with a classic Macintosh laryngoscope. However, 30 studies reported multi-arm comparisons with two or three types of VL compared with a Macintosh blade.

In 12 studies the intubators were described as novices or were trained on manikins, but had no or limited clinical experience. Thirteen studies included intubators who were experienced in the use of DL, but were inexperienced with VL. Eleven studies included a combination of novice and experienced intubators. Twenty-five percent of the studies did not specify intubator experience explicitly, and we were not able to infer it from the manuscript. The remaining studies had experienced intubators performing laryngoscopy and intubation.

Detailed study characteristics are reported in the full version of the Cochrane review.¹⁷

Risk of bias in studies

All studies were described as RCTs, with approximately half describing adequate methods to randomise participants to treatment groups, and we judged these studies to be at low risk of selection bias for sequence generation (Fig. 2). Sufficient detail about methods used to conceal allocation was provided in 40% of the studies, and we judged those studies to be at low risk of bias in this domain. We judged performance and detection bias to be high in all studies as it was not possible to blind intubators and critical outcome assessors. We found low risk of attrition bias in approximately three-quarters of the included studies. A detailed risk of bias for every study is presented graphically in Supplementary material, Appendix S5, with descriptions available in the full version of the Cochrane review.¹⁷

Fig 2.

Risk of bias graph. Review authors' judgements about each risk of bias item presented as percentage across all included studies.

Synthesis of results

We present results for each of the four critical and eight important outcomes for the three comparisons separately. Summary of findings with GRADE assessments for each comparison are presented in Supplementary material, Appendixes S6–S8.

Macintosh-style videolaryngoscopy vs direct laryngoscopy

Failed intubation

Forty-one studies reported the number of failed intubations. Analysis demonstrated fewer failed intubations when a Macintosh-style VL was used (RR=0.41; 95% CI, 0.26–0.65; I²=28%; favours videolaryngoscopy; 41 studies, 4615 participants; moderate-certainty evidence) (Fig. 3). We generated a funnel plot which demonstrated possible asymmetry on visual assessment but found no statistical evidence of small-study effects (Harbord modified test, P=0.202).

Fig 3.

Forest plot for Macintosh-style videolaryngoscopy (VL) vs direct laryngoscopy (DL) comparison: failed intubation. CI, confidence interval; M-H.

Hypoxaemia

Sixteen studies reported the number of hypoxaemic events. Analysis demonstrated fewer hypoxaemic events when a Macintosh-style VL was used (RR=0.72; 95% CI, 0.52–0.99; I²=26%; favours videolaryngoscopy; 16 studies, 2127 participants; moderate-certainty evidence).

Successful first attempt

Forty-two studies reported rates of successful intubation on the first attempt. Analysis demonstrated higher rates of success on the first attempt at intubation when a Macintosh-style VL was used (RR=1.05; 95% CI, 1.02–1.09; I²=77%; favours videolaryngoscopy; 42 studies, 7311 participants; moderate-certainty evidence).

Oesophageal intubation

Fourteen studies reported rates of oesophageal intubation. We found little or no difference in the rate of oesophageal intubation when a Macintosh-style VL was used (RR=0.51; 95% CI, 0.22–1.21; I²=39%; favours videolaryngoscopy; 14 studies, 2404 participants; low-certainty evidence).

Important outcomes

Eighteen studies reported data for dental trauma. We found little or no difference in the rate of dental trauma when a Macintosh-style VL was used (RR=0.68; 95% CI, 0.16–2.89; I²=0%; favours videolaryngoscopy; 18 studies, 2297 participants; very low-certainty evidence).

Thirty-eight studies reported data for CL grades in a format that we were able to extract. Analysis showed a lower proportion of grade 3 and 4 CL views when a Macintosh-style VL was used (RR=0.38; 95%, CI 0.29–0.48; I²=37%; favours videolaryngoscopy; 38 studies, 4368 participants; moderate-certainty evidence).

Thirty-five studies with 4061 participants reported data on time required for tracheal intubation. When we combined these studies, we noted considerable statistical heterogeneity (I²=96%), which could likely be explained by the variation in the definitions of time for intubation used in the included studies. We did not report a pooled effect estimate.

Seventeen studies reported data for patient-reported sore throat. We found little or no difference in the rate of patient-reported sore throat when a Macintosh-style VL was used (RR=0.85; 95% CI, 0.68–1.07; I²=65%; favours videolaryngoscopy; 1960 participants).

Thirty-one studies reported the number of attempts required for successful intubation. Analysis demonstrated a higher proportion of successful single attempts at intubation when a Macintosh-style VL was used (RR=1.05; 95% CI, 1.01–1.10; I²=74%; favours videolaryngoscopy; 31 studies, 3240 participants). We found no difference between Macintosh-style VL and DL when more than one attempt was required (RR=0.68; 95% CI, 0.46–1.01; I²=63%; favours videolaryngoscopy; 31 studies, 3240 participants).

Four studies reported intubation difficulty scale (IDS) scores in a way to allow extraction. We found little or no difference between Macintosh-style VL and DL (for IDS=0: RR=1.22; 95% CI, 0.87–1.72; I²=0%; for IDS=1–5: RR=1.04; 95% CI, 0.84–1.28; I²=0%; for IDS>5: RR=0.60; 95% CI, 0.25–1.45; I²=21%; favours videolaryngoscopy; four studies, 267 participants). Five studies reported POGO scores for this comparison. We noted considerable statistical heterogeneity (I²=94%) and did not report a pooled effect estimate.

Three studies with 719 participants reported mortality rates. All three were conducted outside the operating theatre environment, either in the ICU or ED. We found little or no difference in mortality rates when a Macintosh-style VL was used (RR=1.01; 95% CI, 0.82–1.24; I²=0%; favours direct laryngoscopy; three studies, 719 participants).

Hyperangulated videolaryngoscopy vs direct laryngoscopy

Failed intubation

Sixty-five studies reported the number of failed intubations. We did not include the study by Trimmel and colleagues²⁶ in the meta-analysis as the number of failed intubations in the VL group was disproportionately large, rendering the study an outlier (see sensitivity analysis below). Analysis demonstrated fewer failed intubations when a hyperangulated VL was used (RR=0.51; 95% CI, 0.34–0.76; I²=41%; favours videolaryngoscopy; 63 studies, 7146 participants; moderate-certainty evidence) (Fig. 4).

Fig 4.

Forest plot for hyperangulated videolaryngoscopy (VL) vs direct laryngoscopy (DL) comparison: failed intubation. CI, confidence interval; M-H, Mantel-Haenszel.

We performed a sensitivity analysis including the above excluded study, and this did not alter our interpretation of the effect for this outcome (RR=0.57; 95% CI, 0.34–0.95; I²=69%; favours videolaryngoscopy; 64 studies, 7472 participants).

Hypoxaemia

Fifteen studies reported the number of hypoxaemic events. We found little or no difference in the rate of hypoxaemic events when a hyperangulated VL was used (RR=0.49; 95% CI, 0.22–1.11; I²=39%; favours videolaryngoscopy; 15 studies, 1691 participants; low-certainty evidence).

Successful first attempt

Sixty-eight studies reported rates of successful intubation on the first attempt. Analysis demonstrated higher rates of success on the first attempt at intubation when a hyperangulated VL was used (RR=1.03; 95% CI, 1.00–1.05; I²=76%; favours videolaryngoscopy; 66 studies, 8086 participants; low-certainty evidence).

Oesophageal intubation

Fifteen studies reported rates of oesophageal intubation. Analysis demonstrated fewer oesophageal intubations when a hyperangulated VL was used (RR=0.39; 95% CI, 0.18–0.81; I²=0%; favours videolaryngoscopy; 14 studies, 1968 participants; moderate-certainty evidence).

Important outcomes

Thirty-one studies reported data for dental trauma. We found little or no difference in the rate of dental trauma when a hyperangulated VL was used (RR=0.51; 95% CI, 0.16–1.59; I²=0%; favours videolaryngoscopy; 30 studies, 3497 participants; very low-certainty evidence).

Fifty-five studies reported data for CL grades in a format that we were able to extract. Analysis showed a lower proportion of grade 3 and 4 CL views when a hyperangulated VL was used (RR=0.15; 95% CI, 0.10–0.24; I²=60%; favours videolaryngoscopy; 54 studies, 6058 participants; moderate-certainty evidence).

A total of 59 studies with 6644 participants reported data on time required for tracheal intubation when a hyperangulated VL was used. On combining the above studies, we again noted considerable statistical heterogeneity (I²=99%) and therefore we did not report a pooled effect estimate.

Thirty-one studies reported data for patient-reported sore throat. Analysis showed a lower incidence of patient-reported sore throat when a hyperangulated VL was used (RR=0.81; 95% CI, 0.66–1.00; I²=60%; favours videolaryngoscopy; 31 studies, 3725 participants).

Fifty-one studies reported the number of attempts required for successful intubation. Analysis demonstrated a higher proportion of successful single attempts at intubation when a hyperangulated VL was used (RR=1.02; 95% CI, 1.00–1.05; I²=60%; favour videolaryngoscopy; 50 studies, 5502 participants). Again, we found little or no difference between hyperangulated VL and DL when more than one attempt was required (RR=0.84; 95% CI, 0.66–1.08; I²=48%; favours videolaryngoscopy; 50 studies, 5502 participants).

Ten studies reported IDS scores, and 14 studies reported POGO scores. We noted considerable statistical heterogeneity (I²=93% and 95%, respectively) and did therefore not report a pooled effect estimate.

Three studies with 826 participants reported mortality rates when a hyperangulated VL was used. All three were conducted outside the operating theatre environment, either in the ICU or ED. We found little or no difference in mortality rates when a hyperangulated VL was used (RR=1.15; 95% CI, 0.73–1.79; I²=0%; favours direct laryngoscopy; three studies, 826 participants).

Channelled videolaryngoscopy vs direct laryngoscopy

Failed intubation

Fifty-three studies reported the number of failed intubations. We did not include the studies by Arima and Trimmel in the meta-analysis as the number of failed intubations in the VL groups were disproportionately large, rendering these studies outliers (see sensitivity analysis below). Of note, both studies were conducted in the prehospital setting. Analysis demonstrated fewer failed intubations when a channelled VL was used (RR=0.43; 95% CI, 0.30–0.61; I²=0%; favours videolaryngoscopy; 53 studies, 5367 participants; moderate-certainty evidence) (Fig. 5).

Fig 5.

Forest plot for channelled videolaryngoscopy vs direct laryngoscopy comparison: failed intubation. CI, confidence interval; M-H, Mantel-Haenszel.

We performed a sensitivity analysis including the two excluded studies. The effect estimate with these studies is different to the primary analysis as it crosses the line of no effect with wider confidence intervals (RR=0.55; 95% CI, 0.29–1.03; I²=58%; favours videolaryngoscopy; 55 studies, 5685 participants).

Hypoxaemia

Fifteen studies reported the number of hypoxaemic events. Analysis demonstrated fewer hypoxaemic events when a channelled VL was used (RR=0.25; 95% CI, 0.12–0.50; I²=0%; favours videolaryngoscopy; 15 studies, 1966 participants; moderate-certainty evidence).

Successful first attempt

Forty-seven studies reported rates of successful intubation on the first attempt. Analysis demonstrated higher rates of success on the first attempt at intubation when a channelled VL was used (RR=1.10; 95% CI, 1.05–1.15; I²=84%; favours videolaryngoscopy; 47 studies, 5210 participants; very low-certainty evidence).

Oesophageal intubation

Sixteen studies reported rates of oesophageal intubation. We found little or no difference in the rate of oesophageal intubation when a channelled VL was used (RR=0.54; 95% CI, 0.17–1.75; I²=39%; favours videolaryngoscopy; 16 studies, 1756 participants; low-certainty evidence).

Important outcomes

Twenty-nine studies reported data for dental trauma. We found little or no difference in the rate of dental trauma when a channelled VL was used (RR=0.52; 95% CI, 0.13–2.12; I²=0%; favours videolaryngoscopy; 29 studies, 2375 participants; very low-certainty evidence).

Forty studies reported data for CL grades in a format that we were able to extract. Analysis showed a lower proportion of grade 3 and 4 CL views when a channelled VL was used (RR=0.14; 95% CI, 0.09–0.21; I²=14%; favours videolaryngoscopy; 40 studies, 3955 participants; moderate-certainty evidence).

A total of 57 studies with 5676 participants reported data on time required for tracheal intubation when a channelled VL was used. When we combined data for the above studies, we noted considerable statistical heterogeneity (I²=98%) and therefore did not report a pooled effect estimate.

Eighteen studies reported data for patient-reported sore throat. We found little or no difference in the rate of patient-reported sore throat when a channelled VL was used (RR=0.91; 95% CI, 0.73–1.14; I²=40%; favours videolaryngoscopy; 18 studies, 1666 participants).

Thirty-eight studies reported the number of attempts required for successful intubation. Analysis demonstrated a higher proportion of successful single attempts at intubation when a channelled VL was used (RR=1.09; 95% CI, 1.04–1.14; I²=81%; favours videolaryngoscopy; 38 studies, 4157 participants). We noted considerable heterogeneity. When more than one attempt was required, this was less commonly observed for channelled VL as compared with DL (RR=0.47; 95% CI, 0.33–0.68; I²=56%; favours videolaryngoscopy; 38 studies, 4157 participants).

Sixteen studies reported IDS scores. Analysis demonstrated a higher proportion of easy intubations (IDS=0) and lower proportion of difficult intubations when channelled VLs were used (for IDS=0: RR=3.34; 95% CI, 2.43–4.60; I²=66%; for IDS=1–5: RR=0.38; 95% CI, 0.27–0.53; I²=73%; for IDS >5: RR=0.21; 95% CI, 0.12–0.37; I²=0%; favours videolaryngoscopy; 16 studies, 1004 participants). Five studies reported POGO scores for this comparison. We noted considerable statistical heterogeneity (I²=99%) and did therefore not report a pooled effect estimate.

No studies reported data for mortality rates in this comparison.

Additional analyses

Subgroup analyses

Out of the planned prespecified subgroup analyses for the outcome of failed intubation in each comparison, we were only able to perform meaningful analyses for the difficult airway subgroup because of a small number of studies reporting data for other subgroups. We present the data in Table 1.

Table 1.

Prespecified subgroup analyses. ∗Denotes statistical significance. CI, confidence interval; RR, risk ratio; VL, videolaryngoscopy.

Comparison	Subgroup analysis	Number of studies	Participants	Effect estimate	Subgroup difference
Macintosh-style VL	Setting	4	969	Not pooled	N/A
	Obesity	4	392	Not pooled	N/A
	Difficult airway	12	1393	RR=0.37; 95% CI, 0.19–0.74 (favours VL)	P=0.85
	Intubator experience	1	60	Not pooled	N/A
Hyperangulated VL	Setting	7	837	Not pooled	N/A
	Obesity	7	477	Not pooled	N/A
	Difficult airway	15	1520	RR=0.29; 95% CI, 0.17–0.48 (favours VL)	P=0.03∗
	Intubator experience	9	927	Not pooled	N/A
Channelled VL	Setting	1	120	Not pooled	N/A
	Obesity	4	359	Not pooled	N/A
	Difficult airway	20	1433	RR=0.22; 95% CI, 0.1–0.49 (favours VL)	P=0.07
	Intubator experience	7	1152	Not pooled	N/A

For the hyperangulated VL comparison, 15 studies reported data for failed intubation in 1520 participants with either a known, predicted or simulated difficult airway. Analysis demonstrated fewer failed intubations when a hyperangulated VL was used for participants with predicted, known or simulated difficult airway (RR=0.29; 95% CI, 0.17–0.48; I²=27%; favours videolaryngoscopy; 15 studies, 1520 participants). Subgroup analysis demonstrated a significant difference in favour of hyperangulated VLs when features of difficult airway were present (P=0.03, I²=78.2%).

Sensitivity analyses

We performed sensitivity analyses on the critical outcomes of all three device types combined. These sensitivity analyses were consistent with our primary analyses. We present the forest plots for the analyses in Supplementary material, Appendices S9–S12.

When all device types were included in the same analysis, we found that VL use led to fewer failed intubations (RR=0.44; 95% CI, 0.35–0.56; I²=22%; favours videolaryngoscopy; 139 studies, 16 228 participants), fewer hypoxaemic events (RR=0.61; 95% CI, 0.44–0.85; I²=34%; favours videolaryngoscopy; 41 studies, 5434 participants), and increased rates of success on the first attempt (RR=1.05; 95% CI, 1.03–1.07; I²=81%; favours videolaryngoscopy; 138 studies, 19 797 participants). We also found a lower rate of oesophageal intubation with all VL designs (RR=0.47; 95% CI, 0.29–0.77; I²=16%; favours videolaryngoscopy; 40 studies, 5768 participants).

Discussion

Summary of main results

We found 222 studies comparing videolaryngoscopy with direct laryngoscopy in adults requiring tracheal intubation. We found all three VL designs likely decreased rates of failed intubation, improved success on the first attempt, and were associated with improved glottic views. Oesophageal intubation rates were lower when hyperangulated VLs were used, but this effect did not translate to statistical significance for the other two designs. Rates of hypoxaemia were reduced for Macintosh-style and channelled VLs. We found no difference in rates of dental injury for either design, but our certainty in this evidence is very low. We were not able to reliably pool data for time for tracheal intubation because of considerable statistical heterogeneity.

We found little or no difference in patient-reported rates of sore throat when Macintosh-style or channelled VLs were used, but there was a lower incidence with hyperangulated VLs. Analysis demonstrated a higher proportion of single first attempts at intubation with all VL designs as compared with DL. In terms of IDS scores, we found little or no difference between Macintosh-style VLs and DL, and increased rates of easy intubations when channelled VLs were compared with DL. POGO scores were associated with considerable heterogeneity, and we therefore decided not to pool the data. Mortality rates were reported for Macintosh-style and hyperangulated VL comparisons, and we found little or no difference between VL and DL in those comparisons.

We were only able to conduct a subgroup analysis for features of a difficult airway, where hyperangulated VLs have been shown to reduce the likelihood of failed intubation. Sensitivity analyses combining all VL designs for the critical outcomes demonstrated lower rates of failure, oesophageal intubation, and hypoxaemia, with increased rates of success on the first attempt. This further strengthens our effect estimates of a highly favourable safety profile of videolaryngoscopy for adults undergoing tracheal intubation.

Limitations

In order to address the differences between blade designs, we also presented results in the review according to three discrete categories of devices: Macintosh-style, hyperangulated, and channelled. We acknowledge that this decision represents an artificial construct. It was at times necessary to apply expert judgement to allocate a given device into a category, potentially introducing bias. This categorisation, however, would in theory improve the power of analysis for dichotomous outcomes, especially ones with low numbers of events. A drawback to our approach, as opposed to performing a network meta-analysis, is that this discards potential indirect evidence on which group of VLs might perform better than others.

We also modified the definitions of some critical outcomes in this update, such as failed intubation and hypoxaemia. Previously, we adopted a study-centric approach and used the definitions presented by study author teams. In this update, we agreed on a review-level definition of these outcomes and extracted data according to our definitions rather than study authors. Despite this change, we found that generally the extracted outcome data from studies was largely unchanged. The definition of the newly added outcome of oesophageal intubation also varied among studies, and it was not always clear whether it was detected at the time of initial tube advancement, or later, based on capnography and clinical signs.

We found that establishing the experience of intubators was challenging. A number of studies reported intubators as experienced without quantifying the amount of experience. Others reported experience either in terms of years or number of prior uses of a given device. A study by Cortellazzi and colleagues²⁷ showed that the number of intubations with a hyperangulated VL required for attaining a success rate of 90% by novice intubators was 75. We kept the cut-off at 20 intubations before start of enrolment as it allowed observers to differentiate between complete novices and intubators with some expertise. We could not rule out the possibility that our thresholds for defining intubator experience introduced bias into this particular subgroup analysis; therefore, the results of the subgroup analyses presented in this review are applicable only according to our interpretation of experience as defined.

Our review is the largest systematic review of its type to date, providing a comprehensive overview of relevant published RCTs along with an extensive summary of ongoing research in this domain. The findings of our meta-analyses are in keeping with the conclusions of previous reviews and add to the mounting evidence base on the improved safety and efficacy profile of VL as compared with DL across designs, populations, users, and settings. It is worth noting, however, that our review design does not allow for comparisons between various VLs, and it does not represent a head-to-head comparison of the three designs. We only present evidence for separate comparisons of the three designs to DL.

Implications for future research

There is ongoing interest in this topic in the field of airway research, as evidenced by 46 ongoing studies and 27 studies awaiting classification that may contribute data to future reviews. We encourage future investigators to address the limitations in the quality of evidence, focusing on high-quality study design with clear and complete reporting on methods of randomisation, allocation concealment, and quantified intubator experience. Specifically, more evidence on the relative effects of VLs in the ICU, ED, and prehospital setting would be helpful. A set of core outcome measures would ensure appropriate selection of meaningful outcomes and allow for better summation of evidence. We expect that further expansions of this type of review design would not add much more to the certainty of evidence in this domain as most studies will have inherent design issues, disallowing us from reaching anything beyond moderate-certainty evidence for the main outcomes.

Conclusions and implications for practice

We conclude that VLs of any of the three designs likely reduce rates of failed intubation with increased rates of successful intubation on the first attempt and better glottic views across patient groups and settings. Hyperangulated designs are likely favourable in terms of reducing the rate of oesophageal intubation, and result in improved rates of successful intubation in individuals presenting with difficult airway features. We conclude that VLs likely provide a safer risk profile compared with DLs for adults undergoing tracheal intubation.

Authors' contributions

Concept of the review: AFS

Co-ordinating the review update: JH

Manual searches: JH

Screening search results: JH, AMR

Organising retrieval of papers: JH

Screening retrieved papers against inclusion criteria: JH, AMR, SRL

Appraising quality of papers: JH, AMR, SRL

Abstracting data from papers: JH, AMR, SRL

Writing to authors of papers for additional information: JH

Managing data for the review: JH

Entering data into Review Manager: JH, AMR

Analysing Review Manager statistical data: JH

Interpreting data: JH, AMR, SRL, TMC, AFS

Drawing statistical inferences: JH, AMR, SRL, TMC, AFS

Writing the review: JH, AMR, SRL, TMC, AFS

Reading and checking the review before submission: JH

Performing previous work that was the foundation of the present review: SRL, TMC, AFS

Securing funding for the review: AFS

Guarantor for the review: AFS

Handling editor: Jonathan Hardman

Declarations of interest

SRL, JH, and AMR have no conflicts of interest. TMC was paid for lecturing, several years ago (>36 months), by Intavent Orthofix and the LMA Company. This company manufactures and distributes several supraglottic airway devices and one videolaryngoscope: AP Venner. TMC's department has received free or at-cost airway equipment from numerous ‘airway’ companies for evaluation or research. He and his family have no financial investments and no ownership of any such company of which he is aware. TMC has reported no other conflicts of interest. He spoke at a Storz educational meeting in 2015, and the company paid the costs of travel to this meeting and accommodations. He received no financial benefit from the meeting and was not paid to speak. AFS has received funding for market research relating to airway devices, but not for videolaryngoscopes, nor for any company that produces them.

Appendix A. Supplementary data

The following is the Supplementary data to this article: