Comparisons of Videolaryngoscopes for Intubation Undergoing General Anesthesia: Systematic Review and Network Meta-Analysis of Randomized Controlled Trials

Background: The efficacy and safety of videolaryngoscopes (VLs) for tracheal intubation is still conflicting and changeable according to airway circumstances. This study aimed to compare the efficacy and safety of several VLs in patients undergoing general anesthesia. Methods: Medline, EMBASE, and the Cochrane Library were searched until 13 January 2020. The following VLs were evaluated compared to the Macintosh laryngoscope (MCL) by network meta-analysis for randomized controlled trials (RCTs): Airtraq, Airwayscope, C-MAC, C-MAC D-blade (CMD), GlideScope, King Vision, and McGrath. Outcome measures were the success and time (speed) of intubation, glottic view, and sore throat (safety). Results: A total of 9315 patients in 96 RCTs were included. The highest-ranked VLs for first-pass intubation success were CMD (90.6 % in all airway; 92.7% in difficult airway) and King Vision (92% in normal airway). In the rank analysis for secondary outcomes, the following VLs showed the highest efficacy or safety: Airtraq (safety), Airwayscope (speed and view), C-MAC (speed), CMD (safety), and McGrath (view). These VLs, except McGrath, were more effective or safer than MCL in moderate evidence level, whereas there was low certainty of evidence in the intercomparisons of VLs. Conclusions: CMD and King Vision could be relatively successful than MCL and other VLs for tracheal intubation under general anesthesia. The comparisons of intubation success between VLs and MCL showed moderate certainty of evidence level, whereas the intercomparisons of VLs showed low certainty evidence.


Introduction
The tracheal intubation during general anesthesia can be often unsuccessful. Although the intubation is successful, it can cause several complications. These included respiratory (sore throat, airway trauma), hemodynamic (bradycardia, tachycardia, hypotension) or mechanical complications (mucosal bleeding, dental injury) [1,2]. Difficult intubation

Data Identification and Extraction
The study data were collected and extracted using a standardized form. Two investigators (W.K. and J.L.) independently screened articles by title, abstract, and full texts according to the prespecified inclusion criteria. A full-text review was subsequently performed for potentially relevant articles. Any discrepancies were resolved by consensus after consulting a third investigator (Y.C.). The inclusion criteria were as follows: (1) adult patients who underwent tracheal intubation by experienced anesthetists during general anesthesia, and (2) RCT studies published in English for VL or MCL. The exclusion criteria were (1) non-relevant intervention, (2) studies that failed to acquire outcomes of interest, (3) non-adult studies, and (4) non-RCTs such as review letters, before and after studies, observational studies, case-control studies, case reports, pre-prints, and conference abstracts.

Outcome Measures
The primary outcome of efficacy was the success rate for first-attempt intubation. The success of intubation was defined by capnography confirmation. Other success outcomes measured by chest rise or visual confirmation using VL were excluded. The secondary outcomes were intubation time, glottic view on the first attempt of intubation, and the incidence of sore throat within 24 h. The intubation time was defined as the time from picking up the laryngoscope to confirmation by capnography. The glottic view was assessed using the Cormack-Lehane grade (CLG, I-IV) or modified CLG. The good-glottic view was also defined as CLG I-II or modified CLG I and IIa [20]. Better efficacy or safety means a higher success rate, shorter intubation time, better glottic view, and lower incidence of sore throat. These four outcomes were evaluated by three categories of airway status (all vs normal vs difficult airway). All airways were defined as the normal airway mixed with a difficult airway. Normal airway was additionally defined as airway circumstance that did not predict a difficult airway. Difficult airways were predicted using the following definitions: morbidly obese participants (body mass index > 35 kg/m 2 ); patients with immobilized cervical spines; Mallampati classification 4; retrognathia; more than one

Data Identification and Extraction
The study data were collected and extracted using a standardized form. Two investigators (W.K. and J.L.) independently screened articles by title, abstract, and full texts according to the prespecified inclusion criteria. A full-text review was subsequently performed for potentially relevant articles. Any discrepancies were resolved by consensus after consulting a third investigator (Y.C.). The inclusion criteria were as follows: (1) adult patients who underwent tracheal intubation by experienced anesthetists during general anesthesia, and (2) RCT studies published in English for VL or MCL. The exclusion criteria were (1) nonrelevant intervention, (2) studies that failed to acquire outcomes of interest, (3) non-adult studies, and (4) non-RCTs such as review letters, before and after studies, observational studies, case-control studies, case reports, pre-prints, and conference abstracts.

Outcome Measures
The primary outcome of efficacy was the success rate for first-attempt intubation. The success of intubation was defined by capnography confirmation. Other success outcomes measured by chest rise or visual confirmation using VL were excluded. The secondary outcomes were intubation time, glottic view on the first attempt of intubation, and the incidence of sore throat within 24 h. The intubation time was defined as the time from picking up the laryngoscope to confirmation by capnography. The glottic view was assessed using the Cormack-Lehane grade (CLG, I-IV) or modified CLG. The good-glottic view was also defined as CLG I-II or modified CLG I and IIa [20]. Better efficacy or safety means a higher success rate, shorter intubation time, better glottic view, and lower incidence of sore throat. These four outcomes were evaluated by three categories of airway status (all vs. normal vs. difficult airway). All airways were defined as the normal airway mixed with a difficult airway. Normal airway was additionally defined as airway circumstance that did not predict a difficult airway. Difficult airways were predicted using the following definitions: morbidly obese participants (body mass index > 35 kg/m 2 ); patients with immobilized cervical spines; Mallampati classification 4; retrognathia; more than one of the following: Mallampati classification 3, inter-incision distance of 35 mm or less, and a thyromental distance of 65 mm or less [21,22].

Quality Assessment
Quality assessment was also independently performed by the reviewers using the risk of bias tool developed by the Cochrane group [23]. Evaluated biases included: (1) random sequence generation; (2) allocation concealment; (3) blinding of participants and personnel; (4) blinding of outcome assessments; (5) incomplete outcome data; (6) selective reporting; and (7) other bias. The methodological quality of the identified studies was assessed independently by W.K. and J.L. Investigators selected the terms "low risk of bias," "high risk of bias," or "unclear" to define each study. Any disagreements between the investigators were resolved by a third investigator.

Reporting Guidelines and Certainty of Evidence
The modified Grades of Recommendation, Assessment, Development and Evaluation (GRADE) tool for network meta-analysis was used to evaluate the quality of evidence [24]. The quality of the results were classified as follows: (1) high quality-further research is very unlikely to change the confidence in the estimated effect; (2) moderate quality-further research is likely to have an important impact on the confidence in the estimated effect and may change the estimate; (3) low quality-further research is very likely to have an important impact on the confidence in the estimated effect and is likely to change the estimate; and (4) very low quality, where any estimated effect is highly uncertain.

Statistical Analysis
Odds ratio (OR) with 95% confidence interval (CI) was used to calculate the difference for dichotomous outcomes, while the standardized mean difference (SMD) with 95% CI was used for continuous variables. If the studies only reported the median and measure of dispersion, the data were converted to mean and standard deviation assuming a normal distribution, by using two simple formulae.
We performed a frequentist network meta-analysis of aggregate data to obtain network estimates for the aforementioned outcomes of interest. The model framework used random effects to allow for apparent heterogeneity among studies in treatment comparison effects [25]. We conducted a pairwise meta-analysis to generate direct estimates for outcomes using a random-effects model.
Transitivity assumption, the distribution of patient, and study characteristics that modify treatment effects (effect modifiers) across treatment comparisons were explored to assess whether these characteristics were sufficiently similar between comparisons. Additionally, we evaluated the incoherence assumption (the statistical disagreement between direct and indirect evidence in a closed loop), locally using a loop-specific approach, and globally using a design by treatment interaction model.
The surface under the cumulative ranking curve (SUCRA) values and rankograms were used to present the hierarchy of interventions for each outcome [26]. SUCRA values show the percentage of effectiveness of each intervention compared to the hypothetically best intervention, which is always the best without uncertainty. The certainty of evidence was assessed using GRADEpro in the Cochrane group. Publication bias was evaluated using a comparison-adjusted funnel plot for network meta-analysis.
The results were considered statistically significant at a two-sided p-value of less than 0.05. All statistical analyses were performed using STATA 14.0 software (StataCorp, College Station, TX, USA).

Quality Assessment of the Included Studies
The results of the quality assessments of the included studies are presented in Figure S1. All studies showed a high risk of bias in the two domains for blinding of participants and personnel (performance bias) or outcome assessors (detection bias). Most studies showed low or unclear risk of bias in four domains: random sequence generation, allocation concealment, incomplete outcome data, and selective reporting. Only two studies showed a high risk of bias among these four domains. The study by Sarkilar in 2015 was identified as a high risk of bias in random sequence generation due to a time interval of more than one week in the randomization between the MCL and C-MAC groups. Additionally, the study by Cavus in 2011 showed a high risk of bias for incomplete outcome data in the C-MAC group.

Intubation Success Rate at First Attempt (Success)
In the rank analysis using SUCRA, CMD was the overall most successful VL (SUCRA 77.7) and in the context of difficult airway (SUCRA 85.2) status. The pooled success rates of CMD were 90.6% (358/395 patients) in the all airway status category (the range of success rate = 81-100% in eight included studies) [13,29,61,78,89,92,93,103] and 92.7% (178/192 patients) in the difficult airway status category (the range of success rate = 92-95% in three included studies) [29,61,89]. In the normal airway status category, King Vision was the most successful (SUCRA 72.7), and the pooled success rate of King Vision was 92% (169/183 patients; the range of success rate = 68-100% in five included studies) [30,47,48,78,88] (Figures 3-5). The success rates in all airway were ranked as follows

Intubation Time to Confirmation by Capnometry (Speed)
In the rank analysis using SUCRA, C-MAC was the fastest VL in the context of all (SUCRA 84.9) and normal (SUCRA 84.7) airway status. The intubation time of C-MAC ranged from 25 to 32 s (25 ± 7 s vs. 27 ± 7 s vs. 32 ± 6 s; resulted from 144 patients in three included studies) [28,41,116] in all airway status and 27 s (27 ± 7 s; resulting from 39 patients in one included study) [42] in normal airway status category. In the difficult airway condition, the Airwayscope was the fastest (SUCRA 88.7). The intubation time of the Airwayscope was 34 s (34 ± 25 s; resulting from 35 patients in one included study) [67] (Figures 3-5

Sore Throat within 24 h after Extubation (Safety)
In the rank analysis using SUCRA, CMD showed the lowest incidence of sore throat within 24 h after extubation in all (SUCRA 99.9) and normal (SUCRA 99.3) airway status categories. The pooled incidence of sore throat for CMD was 0% (0/65 patients resulted from one included study) in both airway status categories [103]. In the difficult airway status category, Airtraq also showed a low incidence of sore throat (SUCRA 83.5). The pooled incidence rate of sore throat in Airtraq was 12.6% (26/206 patients; the range of incidence = 0-17.5% in four included studies) [39,60,76,102] (Figures 3-5). When sorting by rank, safety ranking in all airway based on SUCRA values were as follows; CMD 99.

Quality Evidence in GRADE Assessment
The evidence level of each comparison between VL and MCL or intercomparison of the VLs is fully described in Table S1. We extracted and summarized the comparisons of all moderate certainty of evidence assessed by the GRADE tool in Table 2.
For intubation success, five VLs (Airtraq, CMD, GlideScope, King Vision, and McGrath) were more successful than MCL. Moderate evidence did not exist in the intercomparison of VLs. For intubation time, four VLs (Airwayscope, C-MAC, GlideScope, and McGrath) were faster than MCL. In the intercomparison of VLs, the Airtraq was faster than other VLs (CMD, GlideScope, King Vision in difficult airway, and McGrath). The Airwayscope was also faster than GlideScope. However, GlideScope was faster than McGrath and King Vision in the normal airway category, whereas it was slower than McGrath in the difficult airway category. For the glottic view, Airtraq and Airwayscope showed better glottic views than MCL. No moderate evidence existed in the intercomparison of VLs for the glottic view. For safety after extubation, three VLs (Airtraq, C-MAC, and CMD) showed a lower incidence of sore throat than MCL. Moreover, two VLs (Airtraq and McGrath) showed a lower incidence of sore throat than GlideScope in the intercomparisons of VLs.

Publication Bias
In the comparison-adjusted funnel plot, most funnel plots showed symmetry for the success, speed, view, and safety in three airway status categories (all vs. normal vs. difficult). Asymmetry was only observed in sore throat (safety) in all, normal, and difficult airway categories, which suggested the presence of small-study effects (Figures S2-S4).

Discussion
This systematic review and network meta-analysis demonstrated that the CMD was relatively successful compared with other VLs and MCL for the tracheal intubation undergoing general anesthesia in all airway circumstances. Additionally, while the KV was more successful in normal airway circumstances, the CMD was more successful in difficult airway circumstances compared with other VLs and MCL. Other VLs such as MG, GVL, AWS, and CM were top three ranked VLs for intubation success in all, normal and difficult airway circumstances. The comparisons of intubation success between VLs and MCL showed moderate certainty of evidence level, whereas the intercomparisons of VLs showed low certainty evidence.
Previous meta-analyses have provided limited evidence for the usefulness of VLs compared with MCL. De jong et al. only identified the usefulness of VLs compared with DL in critical care settings through a 2014 meta-analysis [117]. In both emergency and critical care settings, the meta-analysis for seven RCTs reported that the first-pass intubation success was not significantly improved by VLs compared with DL [118]. In the 2016 Cochrane systematic review by Lewis, the 64 included studies were composed of 61 elective surgery patients and three patients in emergency settings [18]. However, this study found that VLs mostly showed better performance than DL. We hypothesized that more widespread usage and the availability of VLs in training programs will lead to improved VL performance in clinical settings. Although this study demonstrated that VLs might reduce failed intubation in difficult airways, no evidence indicated that the use of VLs affected the time required for intubation. No evidence was provided for the outcomes for intubation in the intercomparison of VLs in these meta-analyses.
To obtain consistent results and minimize the heterogeneity in the comparison of laryngoscopes, we categorized airway circumstances (normal vs. difficult) and only included studies for intubation by experienced anesthetists in elective surgery. In emergent or critical care settings, some concerns might be raised for inaccurate or rough evaluation of airway status. Furthermore, the urgency of the situation might significantly affect the intubation time or success, especially in cases of difficult airway.
The highest-ranked VLs for outcomes were CMD (success, safety), King Vision (success), Airwayscope (speed and view), C-MAC (speed), McGrath (view), and Airtraq (safety). In this study, CMD, C-MAC, and McGrath were categorized as non-channeled VLs, whereas King Vision, Airwayscope, and Airtraq were categorized as channeled VLs. Through these results, we realized that no absolute superiority existed between non-channeled and channeled VLs. These results also inspired that the most appropriate VL should be clinically decided by considering airway circumstances and the characteristics of VLs.
The included VLs have slightly different characteristics in the blade or video screen, to aid intubation. Most channeled VLs such as Airtraq, Airwayscope, and King Vision have angulated disposable blades and direct screens combined with handles [30,68,73]. Because Airtraq has an exaggerated blade curvature (90 • ), a view of the glottis can be provided with minimal need for airway optimisation manoeuvres such as hyperextension [68]. Compared with Airtraq, King Vision has a wider field of view (160 • vs. 80 • , respectively) with a similar blade curvature [30]. Non-channeled VLs such as GlideScope, C-MAC, and CMD require the use of a stylet and, with the exception of McGrath, have indirect screens [34,43,119]. GlideScope and C-MAC have a lesser angulated blade (60 • vs. 80 • , respectively) than Airtraq and King Vision [28,34]. In particular, CMD has an exaggerated curvature of the distal end of the blade, which faces markedly upward [119]. As a result of the curvature of the blade components, anesthetists require less cervical spine movement.
In the analysis of intubation success in a difficult airway, we additionally found that the different experiences for specific VLs might act as a confounding factor even though experienced anesthetists were enrolled. Although the pooling success rate of intubation using King Vision was 92% in the normal airway category (169/183; 68-100% in five included studies) [30,47,48,78,88], the study by Abdulmohsen (2016) [78,88]. However, two other studies by El-Tahan reported unclear or insufficient experience for intubation (median King Vision intubation <12 times), even if the reported success rate was 100% [47,48]. Thus, the experience of intubation for King Vision might play a role as a confounder for the success rate. However, in the analysis for intubation experience in Tables S2-S4, the included studies did not report outcomes by consistent manner enough to perform meta-regression. Therefore, we only summarized the information of intubation experiences and intubators. This information suggested that most investigators have enough experiences for MCL, however, the experience for VLs was relatively insufficient compared with MCL. The difference of intubation experience between MCL and VLs should be considered in the evaluation of intubation performances of VLs.
This study has some limitations. First, some important confounders might affect the outcomes of this study. Following factors might affect the intubation performance: the different intubation experience (resident vs. attending level; working history; total intubation attempts of each device), the systemic diseases of patients such as American Society of Anesthesiologists score, the hemodynamic characteristics of patients, the use of sedatives or muscle relaxants. The size and type of ETT (double lumen vs. single lumen) might also be a contributing factor for sore throat. Non-channeled video laryngoscopes (VLs) performed better in intubation than channeled VLs, videostylets, and direct laryngoscopes, according to a prior network meta-analysis by Kim et al. 2020 [120]. We concentrated on obtaining more general findings, regardless of the lumen type or ETT size. ETT's increased size could be linked to an increase in sore throat. However, the size of ETT is an ordinary variable with a narrow range, making it an insufficient variable for meta-regression. As good eye-hand coordination is necessary for intubation using non-channeled VL compared to channeled VL, the use of a stylet in non-channeled VL may be linked to an increased risk of sore throat [121]. By causing distinct hemodynamic changes during intubation, sufficient muscular relaxation has a major impact on the ease of laryngoscopy and intubation success. Depolarizing or non-depolarizing muscle relaxants may impact the results. We summarized this information in Tables S2-S4. Further sensitivity or subgroup analysis for these possible confounders was difficult to perform in the network meta-analysis because this information was not reported by consistent manner to perform further analyses. Second, small-study effects for sore throat safety were suggested in the comparison-adjusted funnel plot. This did not indicate the presence of publication bias. It was needed to interpret carefully not to overestimate the effect of the intervention. Third, certain critical criteria for defining a difficult airway were missed. Patients' histories, complaints, clinical examinations, and lastly investigative findings all fall under this category. As a result, results from meta-analyses that have been pooled for restricted information cannot adequately reflect the clinical setting. Fourth, there is no absolute distinction between channeled (without stylet) and non-channeled (with stylet) VLs. Intubation with the ATQ, for example, may require the use of a gum elastic bougie, but intubation with the GVL, CM, or MG does not. Fifth, some risk factors that were left out of the study could be linked to the occurrence or duration of sore throat. In the literature search and full text review, the majority of the studies just stated "elective surgery" rather than addressing specific surgeries. As a result, we were unable to determine whether upper airway surgery was linked to sore throat. Other details, such as a history of previous difficult intubation or a non-infectious or infective sore throat, were not clearly indicated. Sixth, the number of attempts at intubation and cuff pressure were significant factors in the rise in intubation-related complications. We could not find any data on cuff pressure in any of the research we looked at. According to the complete text review, the number of intubation attempts ranged from one to five times. For meta-analysis, however, it was insufficient and heterogeneous to synthesize. As a result, we only used the first attempt at intubation as a primary outcome because the data was sufficient and homogeneous. Seventh, several intubation-related complications, as well as sore throat, were thoroughly investigated. Other mechanical complications have already been collected, such as mucosal bleeding and dental injury. As the data were insufficient and heterogeneous, we were unable to synthesize it in the network meta-analysis.
In conclusion, CMD and King Vision could be relatively successful VLs for tracheal intubation in the comparisons of MCL and other VLs under general anesthesia. The comparisons of intubation success between VLs and MCL showed moderate certainty of evidence level, whereas the intercomparisons of VLs showed low certainty evidence.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/jpm12030363/s1. Figure S1: Risk of bias assessment. Author's judgments about each risk of bias item for each included study, Figure S2: Comparison-adjusted funnel plot for the network metaanalysis of (A) the success at first intubation attempts, (B) intubation time, (C) glottic view and (D) sore throat in in all circumstances of airway, Figure S3: Comparison-adjusted funnel plot for the network meta-analysis of (A) the success at first intubation attempts, (B) intubation time, (C) glottic view and (D) sore throat in in normal airway, Figure S4: Comparison-adjusted funnel plot for the network metaanalysis of (A) the success at first intubation attempts, (B) intubation time, (C) glottic view and (D) sore throat in difficult airway, Table S1: The evidence level of each comparison between videolarygoscopes and Macintosh laryngoscope or intercomparison of the videolaryngoscopes; Table S2: Analysis for factors attributing outcomes according to airway circumstances in included studies; All airway, Table S3: Analysis for factors attributing outcomes according to airway circumstances in included studies; Normal airway, Table S4: Analysis for factors attributing outcomes according to airway circumstances in included studies; Difficult airway, Document S1: Search strategies by Medline, EMBASE, and the Cochrane Library databases.