Network meta-analysis of comparative efficacy and safety of intubation devices in children

To evaluate the comparative efficacy and safety of different intubation devices on intubation outcomes in pediatric intubation. We identified relevant studies from previous meta-analyses and literature retrieval in PubMed, EMBASE, and Cochrane Library. The primary outcome was the first-pass success (FPS), and the secondary outcome included the time to intubation (TTI) and the risk of local complications (LC). Network meta-analysis was performed using STATA 14.0. Twenty-three randomized comparative trials (RCTs) including 12 devices were included. Compared with Macintosh, Airtraq (odds ratio [OR] = 13.05, 95% confidence interval [CI] = 4.68 to 36.38), Miller (OR = 4.77, 95%CI = 1.32 to 17.22), Glidescope (OR = 2.76, 95%CrI = 1.60 to 4.75) and McGrath (OR = 4.61, 95%CI = 1.18 to 17.99) obtained higher PFS. Meanwhile, Airtraq was superior to Glidescope (OR = 0.21, 95%CI = 0.07 to 0.65) for PFS. For TTI, Canada was superior to other intubation devices, as well as CMAC was superior to TruViewEVO2, Glidescope, and StorzDCI. Airtraq lowered the risk of LC compared with Macintosh and Pentax but there was no statistical difference between Airtraq and KingVision. Airtraq may be the optimal option for FPS, Canada for TTI, and KingVision for LC in pediatric intubation.


Data extraction
Two independent authors extracted the following data: first author, origin, publication year, American Society of Anaesthesiologists (ASA) physical status, sample size, the proportion of male patients, age, weight, and device used for videolaryngoscopy and direct laryngoscopy, and outcomes.Moreover, detailed information on the risk of bias was extracted as requested by the Cochrane risk of bias tool 18 .Any disagreements were resolved by consulting a third author for consensus.

Data items
In this network meta-analysis, we defined FPS as the primary outcome and TTI and LC as secondary outcomes.LC was defined to include dental or lip trauma, minor bleeding, or mucosal injury 19 .For the continuous variable (i.e., TTI), we used the recommended formula to estimate the sample mean and standard deviation where data was not available in this format 20 .When the underlying data were not available, we contacted the leading author for more information.

Geometry of the network
We generated a conventional network graph to explore the configuration of the network for a single outcome.A node represented an intubation device, and a line between the two nodes indicated that the two intubation devices were directly comparable.Furthermore, the size of a node represented the number of patients, and the thickness of a line represented the number of direct comparisons.

Risk of bias within the study
Two independent authors assessed the methodological quality of included studies using the Cochrane Collaboration tool, Risk of Bias 2 tool (RoB2) (The Cochrane Collaboration, Oxford, UK) 18 .Individual studies were assessed on the following seven items: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other sources.Each item was classified as low, unclear, or high based on how well the actual information matched the assessment criteria.Any disagreements between the authors were resolved by discussion or consultation with a third reviewer.

Risk of bias across studies
We used funnel plots to test if reporting bias, heterogeneity, methodological quality, or chance interfered with the final results 21 .Plots were generated for the primary and secondary outcomes.

Statistical analysis
Random-effect frequentist network meta-analyses were performed using STATA software (version 14; Stata-Corp LP, College Station, Texas, USA) with the "network" command.Odds ratios (ORs) with 95% confidence intervals in (CIs) were calculated as effect sizes.Meanwhile, we used graphical tools developed by Chaimani and colleagues 22 to visually represent the results.
We first appraised the plausibility of the transitivity hypothesis based on the design characteristics and methods of the studies included in this network meta-analysis 23 .We examined overall consistency using the design-bytreatment interaction model 24,25 , and we also used the method described by Lu and Ades 26 to assess loop inconsistency.We used the side-splitting model to check the inconsistency between the direct and indirect evidence 27 .
We calculated the surface under the cumulative ranking (SUCRA) line to determine how different intubation devices rank in terms of individual outcomes.The higher the SUCRA value, the greater the probability t of a higher rank 28 .Comparison-adjusted funnel plots for individual outcomes were generated 29 .

Basic characteristics of included studies
Table 1 documented the basic characteristics of the included studies.Among 23 studies, 1873 children were enrolled in our network meta-analysis.Six studies 14,32,33,36,44,45 reported details of the ASA status of eligible patients.A total of 12 intubation devices were identified, including Macintosh, Miller, GlideScope (hyperangulated), Pentax (channeled), CMAC (Macintosh-style), TruviewEVO2 (hyperangulated), Airtraq (hyperangulated), KingVision (hyperangulated), StorzDCI, McGrath MAC (Macintosh-style), and Berci-Kaplan.Figure 1 shows the structure of the evidence for all results categorised by specific device name, and Fig. S5 shows the netplot categorised by device type.

Risk of bias across studies
We generated comparison-adjusted funnel plots for all outcomes.Symmetric funnel plots indicated no bias is present for FPS and LC.However, an asymmetric funnel plot for TTI suggested a potential bias.See Fig. S2 for a funnel plot categorised by device name and Fig. S6 for a funnel plot categorised by actual type of device.

Classified by device type
As shown in Fig. S7, we provide estimated pairwise pooled effects as well as 95% confidence intervals (CIs) for the results.We chose the consistency model to calculate all results as the consistency model tests for FPS (chi2 = 0.17, P = 0.667), TTI (chi2 = 0.67, P = 0.715), and LC (chi2 = 0.01, P = 0.924) showed consistent results.
According to the network meta-analysis, FPS, TTI, and LC parameters did not vary significantly between devices.Figure 3 is a forest plot of the meta-analysis as a whole, showing the relative impact of several intubation device types.Figure S8 shows the SUCRA plots for each intubation device at different outcomes.

Inconsistency examination
Node-splitting method was used to examine loop inconsistency in our network meta-analysis.The direct effects were found to be inconsistent with indirect effects in terms of FPS and LC.However, we found that the direct effect was inconsistent with the indirect effect in terms of TTI.All results of inconsistency examinations are shown in Table S2.

Discussion
There is no consensus on the optimal device for pediatric endotracheal intubation due to the anatomical and physiological differences between the airways of pediatric and adult patients 2 .A previous meta-analysis 1 showed that, compared with direct laryngoscopy intubation, intubation with videolaryngoscopy devices did not significantly reduce first attempt failure and lower TTI.However, many intubation devices are available for videolaryngoscopy, it is unclear which devices should be preferred in the specific setting of pediatric patients.We therefore performed this network meta-analysis to compare the relative performance of different intubation devices for intubation in children.
Our network meta-analysis reported several important findings as follows: (1) Airtraq, GlideScope, and McGrath significantly improved the FPS in pediatic intubation, and Airtraq is better than GlideScope, CMAC, and StorzDCI in the improvement of FPS; (2) Canada is associated with significant reduction of TTI compared with other intubation devices, and CMAC also significantly reduce the TTI compared with TruViewEVO2, Gli-deScope, and StorzDCI; (3) Airtraq significantly is associated with the reduced risk of LC compared with Macintosh and Pentax; (4) Airtraq ranks at first place for the improvement of FPS, followed by Miller and McGrath; (5) Canada ranks first for the reduction of TTI, followed by CMAC and Miller; and (6) KingVision ranks first for the reduction of LC, followed by Airtraq and Miller.
The Airtraq is a type of indirect laryngoscope with an exaggerated curative of the blade that is anatomically shaped 50 .It has mirrors and prisms that provide a wide-angle view of the airway during intubation, and the image is transmitted to a viewfinder 50 .Compared to the conventional laryngoscope, Airtraq does not need any alignment of oral, pharyngeal, and laryngeal axes for intubation and also has a quicker learning curve 30,31 .Studies in the adult population have demonstrated that the Airtraq laryngoscope facilitates faster and more accurate intubations 51 .Children have higher metabolisms and oxygen consumption than adults, and oxygen desaturation www.nature.com/scientificreports/can more rapidly during intubation 35 ; therefore, the FPS and TTI of intubation are especially important in pediatric patients.This meta-analysis has shown that Airtraq was superior to other video and conventional laryngoscopes in improving FPS, and the unique blade design and ease of learning might be contributing to the superiority.However, Airtraq involves a single-use, disposable blade, and its impact on the environment should be considered.
In included studies comparing other types of video laryngoscope with a conventional direct laryngoscope, authors found that proficiency is a main barrier to the FPS and TTI of video laryngoscope.Inal et al. stated that the primary reasons for the increased duration of TruviewEVO2 included less experience and lack of eye-hand coordination and practice 32 .Pangasa et al. also had similar findings 39 .Kim et al. also found that experience is needed to be skillful in the GlideScope to have comparable TTI with conventional intubation 34 .An earlier study suggested that a learning experience of 10-30 cases was required for proficiency in the use of GlideScope 52 .Sufficient training and practice of video laryngoscope should be implemented and evaluated when new devices are introduced to the clinical setting.Furthermore, the size and design of the blade should be taken into consideration when evaluating individual patients.
Yoo et al. and Kim et al. compared intubation methods in pediatric nasotracheal intubation 34,48 .Yoo et al. found that compared to the McGrath video laryngoscope and Pentax Airway Scope, the Macintosh laryngoscope has shorter TTI 48 , while Kim et al. demonstrated that GlideScope had similar performance as direct laryngoscope 34 .Further studies are needed to investigate the application of video laryngoscopes in pediatric nasotracheal intubation.Vadi et al. performed the comparison in manual in-line stabilization 43 .While the outcomes were similar among groups, the authors highlighted the importance of additional technical skills when performing video laryngoscopy 43 .
While some studies found that video laryngoscopes had comparable outcomes as the conventional ones, authors suggested that the optical system facilitates a clear image of the glottis which can be a useful option for tracheal intubation in patients with anticipated difficult airways 11,32,40,44 .Nonetheless, the included RCTs recruited pediatric patients scheduled for elective surgery, and generally excluded patients with difficult airways.Further studies are needed to evaluate whether different types of video laryngoscopes are suitable for children with difficult airways.
A recent pair-wise meta-analysis published in November 2020 investigated the comparative efficacy and safety of videolaryngoscopy versus direct laryngoscopy for intubation in children and suggested that videolaryngoscopy required a longer time to intubate and no difference was observed in FPS 1 .Unfortunately, this meta-analysis did not investigate the comparative efficacy and safety of individual videolaryngoscopy devices versus direct laryngoscopy devices or each other individually.In contrast to this meta-analysis, our network meta-analysis first compared individual videolaryngoscopy devices to direct laryngoscopy devices and also compared videolaryngoscopy devices to each other.Therefore, our results were informative in many clinical situations, as our network meta-analysis simultaneously determines the relative performance and ranking of different intubation devices.For example, Airtraq should be preferred for intubation if FPS was listed as the primary indicator of performance.Conversely, we should prioritize Canada if we decide to complete the intubation process within a limited time.
Although our network meta-analysis yielded some promising findings, we must also acknowledge that our network meta-analysis has several limitations: (1) By only including pediatric patients, variations of the results may be reduced.However, differences in ASA status, age, and weight may also confound our pooled results, as we did not perform sensitivity or subgroup analysis to eliminate the effect of these factors on the results; (2) We could not incorporate operator experience and patient characteristics into our analysis, as most studies did not report these outcomes.However, these factors are critical to clinical outcomes and may bias the pooled results significantly.Future studies should control for or report in detail on those confounding factors that play a crucial role in success and safety outcomes.(3) Cost-effectiveness is also an important factor in the choice of intubation.It is difficult to summarize the cost of intubation as the price of intubation equipment and procedures varies from hospital to hospital worldwide.(4)The methodological quality of the included studies varied, which also may affect the robustness of our pooled results; (4) small-study effect and publication bias were detected for TTI, which will reduce the reliability of this outcome to a certain extent; (5) direct comparison for some intubation devices is not available, so the pooled results were only calculated based on indirect comparisons, which also impair the robustness of pooled results; (6) some closed loops of TTI produced significant inconsistency, which will negatively influence our findings.Despite these limitations above, our network meta-analysis has several advantages: (1) as far as we know, our network meta-analysis is the first attempt to comprehensively investigate the relative performance of different intubation devices in pediatric intubation; (2) our network meta-analysis ranks different intubation devices according to SUCRA value, which will aid in clinical decision-making; and (3) our network meta-analysis has also other strengths, such as the comprehensive search and use of the Cochrane risk of the bias assessment tool.

Conclusion
For tracheal intubation in children, Airtraq may be the preferred intubation device because it significantly improves FPS, Canada may be the preferred option for the reduction of TTI, and Canada should be preferred for the significant reduction of LC.However, more studies are warranted to further validate our findings, as some intubation devices cannot be directly compared, and not all devices are included in individual outcomes.

Statement
The work has been reported in line with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) and AMSTAR (Assessing the methodological quality of systematic reviews) Guidelines.

Figure 1 .
Figure1.Network plot for first-pass success (A), time to intubation (B), and local complications (C).The size of each node represents the number of patients included in studies featuring that device.The thickness of the lines connecting the nodes is proportional to the number of head-to-head studies in each comparison.

Figure 2 .Figure 3 .
Figure 2. Risk of bias assessments.y, u, and n indicate low, unclear, and high risk of bias, respectively.

Figure 4 .
Figure 4. Forest of relative effects of different devices in terms of first-pass success (A), time to intubation (B), and local complications (C).

Table 1 .
Basic characteristics of the included studies.FPS, first-pass success; TTI, time to intubation; LC, local complications; ASA, American Society of Anesthesiologists; NA, not applicable.