Anesthesiologists perform endotracheal intubation (ETI) in the operating room under controlled circumstances, and the procedure carries a low risk of complications.1 Although laryngoscopy is difficult in 6-10% of intubations,2 - 4 difficult or failed intubations are much less frequent, occurring in 1.8-5.8% and 0.13-0.30%, respectively.2 , 5 - 8 Unfortunately, physical findings on examination of the airway discriminate poorly between potentially easy and difficult intubations.9 Thus, anesthesiologists need to be prepared for the unanticipated difficult airway, as many of these patients have had a “reassuring” airway physical examination. In addition to the unanticipated difficult airway, there are circumstances that lend themselves to a high risk of difficult laryngoscopy and tracheal intubation. In particular, emergent ETI outside of the operating theatre is associated with a much higher risk of difficult laryngoscopy and intubation.10 - 13 As such, techniques that may improve successful intubation may be especially beneficial in these emergent environments. Laryngoscopy with the Glidescope® video-laryngoscope (Verathon Medical, Bothell, WA, USA) appears to be associated with improved glottic visualization.14 , 15 Whether the improved visualization translates into increased success at ETI, when compared to direct laryngoscopy, remains unclear.14 , 16 Given this uncertainty, our goal was to perform a systematic review and meta-analysis of randomized and quasi-randomized trials comparing Glidescope® video-laryngoscopy to direct laryngoscopy regarding glottic visualization, successful first-attempt intubation, and time to intubation. In addition, we explored the heterogeneity in these outcomes based on operator expertise and according to the difficulty of the intubation.

Methods

This article reports our meta-analysis of controlled trials of Glidescope® video-laryngoscopy compared to direct laryngoscopy in accordance with the Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA) statement.17 A review protocol was not published for this study.

Search strategy

We systematically searched MEDLINE (1966 to June 13, 2011), EMBASE (1977 to June 13, 2011), and The Cochrane Central Register of Controlled Trials (CENTRAL) (1948 to June13, 2011) for randomized and quasi-randomized trials comparing Glidescope® video-laryngoscopy to direct laryngoscopy regarding the glottic view, successful first-attempt intubation, and time to intubation. We included non-English publications. We hand-searched abstracts of selected conferences from 2000 to 2010, including those of the American Society of Anesthesiologists, the Canadian Anesthesiologists’ Society, and the International Anesthesia Research Society. We also hand-searched bibliographies of all relevant trials and review articles.

For the bibliographic review, we constructed search filters for the concepts “Glidescope video-laryngoscope” and “clinical trials” using a combination of exploded Medical Subject Heading (MeSH) terms and text words all combined with the Boolean operator “OR.” The Glidescope® video-laryngoscope filter contained the text words glidescope and video-laryngoscope. The clinical trials filter included the MeSH terms clinical trials [publication type], clinical trials as topic, placebos with text words trial*, random* or placebo. A similar search strategy was used for both EMBASE and CENTRAL.

Selection criteria, data abstraction, and methodological quality

In duplicate and independently, two authors (D.G., D.L.) screened all articles and abstracts, which were included if they 1) were randomized or quasi-randomized controlled trials, 2) compared direct laryngoscopy to Glidescope® video-laryngoscopy, 3) addressed adult patients, and 4) contained any outcome of interest (Cormack-Lehane view,18 successful first-attempt intubation, time to intubation).

The same two authors abstracted the data and assessed the study quality in duplicate and independently. Disagreement was resolved by discussion and arbitrated if necessary by a third author (P.C.). We abstracted the year of publication, sample size, country of origin, operator training and experience, physical examination of the airway, anticipated or history of difficult intubation, application of manual in-line stabilization, Cormack-Lehane grade, successful first attempt at intubation, and time required to intubate. We contacted investigators for missing data as necessary.

Statistical analysis

We used relative risk (RR) as the summary measure for dichotomous outcomes (glottic view and successful first intubation attempt) and the weighted mean difference (WMD), in seconds, as the summary measure for time to intubate. We applied a half-integer continuity correction to all four cells if the event rates were zero. The random effects method of DerSimonian and Laird was used to generate a pooled RR or WMD across studies.19 Random effects analysis yields a more conservative estimate than the fixed-effects model in the presence of between-study heterogeneity. We assessed statistical heterogeneity using Cochran’s Q statistic20 (with P < 0.10 considered significant) and expressed the quantity using the I 2 statistic and 95% confidence interval (CI). The I 2 statistic indicates the percentage of variation in study results that is due to between-study heterogeneity rather than sampling variability.21 We assessed for the following outcomes: Cormack-Lehane view grade 1 vs grade ≥ 2, successful first-attempt intubation, and time to intubate (in seconds).

Sources of potential heterogeneity identified a priori were the experience level of the operator (anesthesia or casualty consultants or house staff vs “other”) and potential difficulty. Intubations were considered difficult in studies that included patients with a known prior difficult intubation, physical examination features suggesting a difficult intubation, or in whom difficult intubation was simulated by providing manual-in-line stabilization. Random-effects meta-regression was used to evaluate the relation between these subgroups on the final pooled estimates.22 We evaluated the presence of publication bias by visual inspection of the funnel plot and by using Egger’s and Begg’s tests, with P < 0.05 considered statistically significant. All analyses were done using Stata 10.0 (2007) (StataCorp LP, College Station, TX, USA).

Results

Literature search

A total of 297 citations were identified during the bibliographic search: 76 from MEDLINE, 150 from EMBASE, and 71 from CENTRAL. We excluded 264 citations on the initial abstract screen (178 duplicate citations, 86 from screening). We identified three published abstracts from conference screening and five citations from reference lists. This resulted in 41 citations for full text review. The exclusion of 24 citations (for reasons listed in Fig. 1) resulted in 17 trials being included in the current analysis.14 - 16 , 23 - 36 We contacted one author, who provided the raw data for the number of attempts required for intubation, which was not included in the published article.25

Fig. 1
figure 1

Study selection flow chart

Full size image

Study characteristics

Table 1 lists the trial characteristics. Of the 17 included trials with a total of 1,998 subjects, three were published abstracts.33 - 35 One trial was published in Japanese.36 Although most of the studies randomized subjects to Glidescope® video-laryngoscopy vs direct laryngoscopy, in four studies subjects underwent both techniques sequentially, with the order of the techniques allocated randomly.24 , 29 , 31 , 33 The operators in most of the studies were anesthesiologists experienced with both techniques. There were two studies in which the primary operators were inexperienced personnel consisting of nonanesthesia house staff36 or trainees consisting of paramedics, nurses, and medical students.14 Although the trial by Jones and colleagues included anesthesia consultants and residents, only 39% were experienced with the Glidescope® (≥ 10 intubations).16 In contrast to all the other studies of elective patients in the operating theatre, the trial by Yeatts et al. examined patients presenting to the casualty department.35

Table 1 Characteristics of randomized and quasi-randomized trials comparing Glidescope® video-laryngoscopy to direct laryngoscopy
Full size table

Most of the studies specifically excluded patients with a known or anticipated difficult airway.14 , 16 , 23 , 25 , 27 , 28 , 30 - 34 In contrast, two studies selected patients with clinical examination features suggesting a difficult intubation.24 , 26 Five studies attempted to increase the difficulty of laryngoscopy by applying manual in-line stabilization.23 , 28 , 29 , 31 , 34 Finally, three studies did not specify any exclusion or inclusion criteria based on prior or anticipated difficulty of laryngoscopy.15 , 35 , 36

Grade 1 glottic view

Twelve studies presented outcomes corresponding to our primary outcome, glottis visualization (Table 2).14 - 16 , 23 , 24 , 26 - 29 , 32 - 34 A forest plot is presented in Fig. 2. The pooled RR across all studies was 2.0 (95% CI 1.5 to 2.5, P < 0.001), indicating improved glottic visualization using the Glidescope® when compared to the direct laryngoscope. There was significant between-study heterogeneity in our primary analysis (Q = 74.8, df = 11, P < 0.001), with a corresponding I 2 statistic of 85% (95% CI 76 to 91). Only one study used inexperienced operators14; thus, we were unable to explore heterogeneity by expertise. We examined for effect modification by anticipated or simulated difficult laryngoscopy (manual in-line stabilization). Meta-regression demonstrated that the benefit to glottic visualization afforded by Glidescope® was even more pronounced in studies that considered patients with anticipated or simulated difficult airways (P = 0.003). The resultant pooled estimates were as follows: for nondifficult intubations (RR 1.5, 95% CI 1.2 to 1.9) and for difficult intubations (RR 3.5, 95% CI 2.3 to 5.5). Visual inspection of the funnel plot revealed an absence of small studies favouring direct laryngoscopy (not shown). This publication bias was confirmed on Begg’s (P = 0.04) and Egger’s (P = 0.07) regression testing.

Table 2 Outcomes of randomized and quasi-randomized trials comparing Glidescope® video-laryngoscopy to direct laryngoscopy
Full size table
Fig. 2
figure 2

Risk ratios (RR) of Cormack-Lehane (CL) grade 1 (vs ≥ grade 2) in clinical trials comparing Glidescope® video-laryngoscopy to direct laryngoscopy stratified by the difficulty of the intubation. Subjects were considered to have difficult intubations in studies that included patients with known prior difficult intubation, physical examination features suggesting difficult intubation, or in which difficult intubation was simulated by providing manual-in-line stabilization. The pooled estimate was derived using the DerSimonian and Laird random effects method with grey squares depicting individual study point estimates of the RR. Larger squares indicate a larger weight of the study when calculating the pooled estimate. Solid horizontal lines display the 95% confidence interval (CI) of the point estimate. Dashed vertical line represents an RR of 1.00, indicating no difference between Glidescope® video-laryngoscopy and direct laryngoscopy. Solid vertical lines represent the pooled estimates. Test for heterogeneity was significant using meta-regression analysis (P = 0.003). DL = direct laryngoscopy; GS = Glidescope®

Full size image

Successful first-attempt intubation

Fourteen studies presented data on intubation success (Table 2).14 - 16 , 23 - 28 , 30 , 32 , 33 , 35 , 36 A forest plot is presented in Fig. 3. The pooled RR across studies was 1.1 (95% CI 0.99 to 1.2, P = 0.09). There was significant between-study heterogeneity (Q = 117.12, df = 13, P < 0.001), with a corresponding I 2 statistic of 89% (95% CI 83 to 93). Two studies presented data on inexperienced operators,14 , 36 and meta-regression demonstrated effect modification by operator expertise (P = 0.001). Compared to the direct laryngoscope, the Glidescope® increased the success of first intubation attempts in studies with nonexpert operators (RR 1.8, 95% CI 1.4 to 2.4) but not amongst airway experts (RR 1.0, 95% CI 0.94 to 1.20). There was no effect measure modification by potential or simulated difficult airways (P = 0.89). There was no evidence of publication bias on this outcome by Begg’s (P = 0.38) or Egger’s (P = 0.86) testing.

Fig. 3
figure 3

Risk ratios (RR) of successful first-attempt intubation in clinical trials comparing Glidescope® video-laryngoscopy to direct laryngoscopy stratified by operator expertise (anesthesia or casualty consultants or house staff vs “other”). The pooled estimate was derived using the DerSimonian and Laird random effects method with grey squares depicting individual study point estimates of the RR. Larger squares indicate a larger weight of the study when calculating the pooled estimate. Solid horizontal lines display the 95% CI of the point estimate. Dashed vertical line represents an RR of 1.00, indicating no difference between Glidescope® video-laryngoscopy and direct laryngoscopy. Solid vertical lines represent the pooled estimates. Test for heterogeneity by operator expertise was significant using meta-regression analysis (P = 0.001). DL = direct laryngoscopy; GS = Glidescope®

Full size image

Time to intubation

The time required to intubate was available in 15 studies (Table 2).14 - 16 , 23 - 28 , 30 - 32 , 34 - 36 A forest plot is presented in Fig. 4. The pooled WMD across studies did not differ between Glidescope® video-laryngoscopy and direct laryngoscopy (WMD 3.8 sec, 95% CI −1.7 to 9.3 sec, P = 0.17). However, there was significant between-study heterogeneity in these results (Q = 675.7, df = 14, P < 0.001) with an I 2 statistic of 98% (95% CI 97 to 98) that was not explained by the difficulty of the intubation on meta-regression (P = 0.85). Meta-regression did demonstrate that operator expertise explained some of the between-study heterogeneity observed (P = 0.004), with the Glidescope® being associated with a shorter time to intubation in the two studies with nonexperts as the primary operators (WMD −43 sec, 95% CI −72 to −14 sec). There was no difference in time to intubation amongst experts (WMD 8 sec, 95% CI −2 to 17 sec). There was no effect measure modification by airway difficulty on meta-regression (P = 0.74). There was no evidence of publication bias on this outcome by Begg’s (P = 0.18) or Egger’s (P = 0.96) testing.

Fig. 4
figure 4

Weighted mean difference (WMD), in seconds, in clinical trials comparing Glidescope® video-laryngoscopy to direct laryngoscopy stratified by operator expertise (anesthesia or casualty consultants or housestaff vs “other”). The pooled estimate was derived using the DerSimonian and Laird random effects method with grey squares depicting an individual study point estimate of the mean difference. Larger squares indicate a larger weight of the study when calculating the pooled estimate. Solid horizontal lines display the 95% CI of the point estimate. Dashed vertical line represents a WMD of 0, indicating no difference between Glidescope® video-laryngoscopy and direct laryngoscopy. Solid vertical lines represent the pooled estimate. Test for heterogeneity by operator expertise was significant using meta-regression analysis (P = 0.004)

Full size image

Discussion

In this meta-analysis of randomized trials comparing Glidescope® video-laryngoscopy to direct laryngoscopy, the former was associated with improved glottic visualization, particularly amongst studies that considered patients with potential or simulated difficult airways. Although there was an improved successful first intubation attempt and faster time to intubation with Glidescope® video-laryngoscopy, it was confined to studies of nonexpert operators. There was no benefit in either of these outcomes in studies with expert operators. Importantly, there was marked between-study heterogeneity in all three outcomes.

Improved glottic visualization (compared to that with direct laryngoscopy) is a consistent finding with nonstandard laryngoscopes, including video-laryngoscopes.37 Building on this, we have demonstrated that improvement in glottic visualization afforded by the Glidescope® is even greater in studies using patients with either simulated (via manual in-line stabilization) or physical examination predictors of difficult laryngoscopy. This is not surprising as the Glidescope® appears to be used often by clinicians in these situations. A large observation cohort study by Aziz and colleagues of 2,004 Glidescope® intubations showed that most were performed in patients with clinical examination predictors of a difficult direct laryngoscopy.38 Thus, clinicians are triaging patients to video-laryngoscopy when difficulty with endotracheal intubation is anticipated.

As in our current review, a prior systematic review demonstrated significant heterogeneity when comparing the Glidescope® results to those achieved with the direct laryngoscope.37 In contrast, we attempted to quantify and evaluate sources of heterogeneity by both operator expertise and potential difficulty of the intubation. Given that most of the studies were performed by airway management experts on patients without predictors of difficult intubation, it is not surprising that the Glidescope® did not result in improved first-attempt success. Aside from one trial with a markedly low rate of 63%, documented by Morello et al.,33 the rest of the studies with experts—and excluding difficult airways—had a first-attempt success rate of > 90%.15 , 16 , 27 , 32 This high rate of success with direct laryngoscopy by anesthesiologists is reflected in other clinical studies.6 Even in the unlikely scenario that Glidescope® video-laryngoscopy would improve the success rate in patients without difficult airways by experts, it would require a large sample of patients to prove it. Thus, potential benefits of Glidescope® video-laryngoscopy may lie with: 1) use in patients with clinical features indicating difficult laryngoscopy; 2) it being used as a rescue method following failed direct laryngoscopy; or 3) it being used by nonexpert providers. Indeed, the observational study by Aziz et al. demonstrated that the Glidescope® was successful in 96% of patients with predictors of difficult direct laryngoscopy and in 94% following failed direct laryngoscopy.38

Although our review did show increased first-attempt success and decreased time to intubation in studies of nonexperts with the Glidescope® compared to direct laryngoscopy, these results must be interpreted with caution given that there were only two studies in this subgroup.14 , 36 Rather, the possible benefit of Glidescope® video-laryngoscopy amongst nonexperts should be viewed as an area that requires further research.

This systematic review and meta-analysis highlights several areas that need to be addressed. How is expertise developed and defined, particularly when a new technology is introduced? What role should nonexperts play in airway management? Studies examining new technology are prone to proficiency bias. Despite this fact, anesthesiologists have incorporated the Glidescope® into their armamentarium with a high rate of success.39 Although it seems reasonable to assume that anesthesia consultants are experts, it remains less clear how, and at what point, this competence develops. When examining trainees, we have previously shown that anesthesia house staff were successful in 85% of their first attempts at intubating critically ill patients.40 This success rate is very respectable given that this is a population with a 6.6-22.0% risk of a difficult intubation.11 , 13 , 41 Furthermore, anesthesia house staff require fewer attempts to perform tracheal intubation compared to their nonanesthesia counterparts. Having an airway management expert at the bedside for each intubation may be advantageous, but there are many situations when this is not feasible. In many environments, there may be limited, if any, access to anesthesiologists, and airway management must be delivered by physicians from different speciality backgrounds. Endotracheal intubation remains a competence objective of the Royal College of Physicians and Surgeons of Canada in training for internal medicine.42 Also, use of an advanced airway (e.g., endotracheal tube) remains a fundamental skill in Advanced Cardiac Life Support according to the 2005 American Heart Association Guidelines.43 Thus, technologies that can improve the success of airway management, particularly in the hands of nonexperts, are desirable and should be studied. An example is Glidescope® use by prehospital paramedics.44

There are several limitations to our review. As previously stated, there was marked heterogeneity in all of our endpoints that was only partially explained by subgroup analysis. We attempted to account for this heterogeneity by performing a random-effects meta-regression, which yields a more conservative pooled estimate when between-study heterogeneity exists.45 In addition, we explored heterogeneity by a priori defined subgroups and presented these results when they were significant. As with all meta-analyses, our review is subject to information bias. We defined expertise and difficulty a priori, but there may be marked differences between studies with respect to subject or operator characteristics that we were unable to evaluate from the available information. Another limitation is the low number of studies that included nonexperts, which markedly limits the ability to evaluate the effect of video-laryngoscopy in this important subgroup. Finally, there was evidence of publication bias in our primary outcome of the glottic view, suggesting that small studies favouring direct laryngoscopy were not being published. However, tests of publication bias are subject to a high risk of a type I error in the presence of significant heterogeneity, limiting their interpretability.46

In conclusion, we have shown in our meta-analysis that, compared to direct laryngoscopy, Glidescope® video-laryngoscopy is associated with improved glottic visualization, particularly in studies that considered patients with potential or simulated difficult airways. In addition, there is marked heterogeneity in all of our outcomes that is partially explained by operator expertise or the difficulty of intubation. There is a need for further evaluation of potential improvements in successful first-attempt intubations or time to intubate among nonexperts.