A Comparison of GlideScope Video Laryngoscopy Versus Direct Laryngoscopy Intubation in the Emergency Department
Presented at the American College of Emergency Physicians Scientific Assembly, Chicago, IL, October 2008.
No funding or support was received for this study, and none of the authors have a relationship with the maker of the device studied.
A related commentary appears on page 908.
Timothy F. Platts-Mills, MD; e-mail: firstname.lastname@example.org. Reprints will not be available.
Objectives: The first-attempt success rate of intubation was compared using GlideScope video laryngoscopy and direct laryngoscopy in an emergency department (ED).
Methods: A prospective observational study was conducted of adult patients undergoing intubation in the ED of a Level 1 trauma center with an emergency medicine residency program. Patients were consecutively enrolled between August 2006 and February 2008. Data collected included indication for intubation, patient characteristics, device used, initial oxygen saturation, and resident postgraduate year. The primary outcome measure was success with first attempt. Secondary outcome measures included time to successful intubation, intubation failure, and lowest oxygen saturation levels. An attempt was defined as the introduction of the laryngoscope into the mouth. Failure was defined as an esophageal intubation, changing to a different device or physician, or inability to place the endotracheal tube after three attempts.
Results: A total of 280 patients were enrolled, of whom video laryngoscopy was used for the initial intubation attempt in 63 (22%) and direct laryngoscopy was used in 217 (78%). Reasons for intubation included altered mental status (64%), respiratory distress (47%), facial trauma (9%), and immobilization for imaging (9%). Overall, 233 (83%) intubations were successful on the first attempt, 26 (9%) failures occurred, and one patient received a cricothyrotomy. The first-attempt success rate was 51 of 63 (81%, 95% confidence interval [CI] = 70% to 89%) for video laryngoscopy versus 182 of 217 (84%, 95% CI = 79% to 88%) for direct laryngoscopy (p = 0.59). Median time to successful intubation was 42 seconds (range, 13 to 350 seconds) for video laryngoscopy versus 30 seconds (range, 11 to 600 seconds) for direct laryngoscopy (p < 0.01).
Conclusions: Rates of successful intubation on first attempt were not significantly different between video and direct laryngoscopy. However, intubation using video laryngoscopy required significantly more time to complete.
Endotracheal intubation is an important emergency department (ED) intervention. Indications include need for airway protection, respiratory failure, prevention of movement for imaging, upper airway obstruction, and anticipated airway compromise. ED intubations may be complicated by cervical spine immobilization, facial trauma, obesity, recent food consumption, and difficulty preoxygenating the patient. A recent multicenter study of ED intubations found a first-attempt success rate of 83%, with the first physician successfully intubating the patient in 90% of cases, and a failure rate of 2.7%.1,2
The GlideScope (Saturn Biomedical Systems, Burnaby, BC, Canada) is a video laryngoscope that uses a plastic reusable blade similar in shape to a Macintosh 3 blade with a more curved blade and a camera located in the middle of the outer curvature of the blade. During intubation with the GlideScope, the blade is inserted along the tongue in the midline, and the tip is positioned in the vallecula as with a Macintosh blade. The glottis is viewed on an external liquid crystal display monitor rather than by direct visualization. The theoretical advantage is that the oral, pharyngeal, and tracheal axes do not need to align to gain a good view.
Several prior studies have examined the performance of the GlideScope in anesthesia settings.3–6 These studies have demonstrated improved glottic view with this device. A study of emergency physicians using video laryngoscopy on manikins has been performed.7 A recent study reports on the initial use of video laryngoscopy in the ED without a comparison population.8 The goal of this study was to determine if ED residents had a higher first-attempt success rate using GlideScope video laryngoscopy compared with direct laryngoscopy.
This was a prospective observational study. Informed consent was not obtained, because this was an observational study with no anticipated impact on patient care. The study received approval from the hospital’s institutional review board.
Study Setting and Population
The study was conducted in the ED of a Level 1 trauma center with an emergency medicine residency program. Adult patients requiring intubation in the ED were continuously enrolled between August 2006 and February 2008. Exclusion criteria included patients who presented to the ED after being intubated by emergency medical services personnel or an outside hospital or patients under age 16 years. The decision as to which patients required intubation and whether to use video or direct laryngoscopy was left to the resident and attending physician caring for the patient. The initial attempt for all intubations was by a resident; an attending physician only performed or attempted to perform the procedure if the resident was unsuccessful. First-year residents did not participate in the study, because ED intubations are not part of their practice.
Prior to patient enrollment, all residents and faculty attended one of two 30-minute orientations to the GlideScope provided by one of the company representatives. This orientation included a video demonstration of the GlideScope. In addition to receiving the 30-minute orientation session, all providers also practiced using the device at least once on an airway manikin. Intubations with the GlideScope used a standard malleable stylet with a 75- to 90-degree endotracheal tube curve consistent with manufacturer recommendations. Both GlideScopes used for intubations had color monitors and reusable solid blue plastic blades. The GlideScopes used in our study were purchased by our department. The study was not sponsored by the manufacturer, and other than providing two orientation sessions, the manufacturer had no input into the study.
A data collection sheet was placed in each intubation kit and was filled out immediately after the intubation by the treating physician. Data collected included the indication for intubation and patient characteristics including age, estimated weight, and whether the patient was a trauma or medical patient. The reason for intubation was also recorded; these included altered mental status, respiratory failure, facial trauma, and the need to immobilize for imaging. Patients with cardiac arrest were designated as having altered mental status as the reason for intubation. When present, multiple reasons for intubation were recorded. The presence of cervical spine immobilization, cardiac arrest, and obesity were also recorded. The postgraduate year of the resident who initially attempted the intubation was recorded. Outcome data were documented, including whether the intubation was successful on the first attempt, total number of attempts, need to switch to a different device or a different physician, occurrence of an esophageal intubation, time to successful intubation, and initial and lowest oxygen saturation levels.
An attempt was defined as an introduction of the laryngoscope into the mouth and its removal regardless of whether an endotracheal tube was inserted. The time to intubation was defined as the time from when the laryngoscope was first placed into the patient’s mouth until the time of first successful forced inspiration using the newly placed endotracheal tube. If an attempt was made, it was unsuccessful, and subsequent attempts were made, the time for intubation included the entire time from the beginning of the first attempt until the time of successful forced inspiration. The definition for time to complete the intubation was provided to all participants by e-mail, during a brief lecture introducing them to the study, and was written on the data collection form. Time to successful intubation was measured by a nurse, nursing assistant, or a resident not performing the intubation; this person was identified prior to each procedure. For study purposes, failure of intubation was defined as 1) esophageal intubation, 2) changing to a different device or physician, or 3) inability to place the endotracheal tube after three attempts. A change to a different physician was classified as a failure regardless of how many attempts were made prior to the change. In most cases, residents made at least two attempts before the attending physician attempted the intubation. However, the decision to switch from resident to attending physician was left to the discretion of the individual attending physician. In patients who were not victims of cardiac arrest, confirmation of endotracheal tube placement included auscultation, qualitative capnometry, pulse oximetry, chest radiograph, and in most cases visualization of the tube passing through the cords. In cardiac arrest patients, direct visualization of the tube going through the cords, breath sounds, and an esophageal detector device were used to confirm endotracheal tube placement.
The primary outcome variable was successful intubation on first attempt. Secondary outcome variables include time to successful intubation, lowest oxygen saturation level, and intubation failure rate.
Data were entered into a Microsoft Excel (Redmond, WA) spreadsheet. Descriptive statistics including proportions, means, and 95% confidence intervals (CIs) were calculated using Excel tools. For patients with cardiac arrest, oxygen saturation levels were not obtainable; patients without oxygen saturation data were excluded from the calculation of mean and lowest oxygen saturation levels. Because time to intubation had a nonnormal distribution, medians were calculated. The Wilcoxon rank sum test of nonpaired samples was used to compare the time to intubation for the two groups. Logistic regression was conducted to identify predictors of successful first-attempt intubation, and the following interactions were included: type of laryngoscope × patient weight, type of laryngoscope × postgraduate year, and type of laryngoscope × presence of facial trauma. When a significant interaction was identified, a new variable was created consisting of all levels of the interacting variables. Additional independent variables included in the logistic regression analysis were patient age; patient weight; whether the patient had medical illness or traumatic injury, facial trauma, or cervical spine immobilization; and resident who performed the initial intubation attempt. Data were entered into the logistic regression model using a forced entry method. The Wilcoxon rank sum test and the logistic regression analysis were performed using SAS version 9.2 (SAS Institute, Inc., Cary, NC).
Prior to initiating the study, a sample size calculation was performed. Assuming a first-attempt success rate of 75%, a sample size of 100 patients in each group had an 80% power to detect a 15% difference in first-attempt success rates between video and direct laryngoscopy intubations (two-tailed α = 0.05). Although a success rate of 83% may have been more accurate,1 the sample size needed to detect a difference between 83% and 98% is smaller (72 patients in each group) than between 75% and 90%, and the decision to use 75% as the first-attempt success rate for sample size calculation decreases the risk of a type II error.
A total of 281 patients were enrolled in the study. One entry was excluded because a flexible fiber-optic scope was the initial device used. The initial attempt for the remaining 280 patients was with video laryngoscopy in 63 patients (23%), and direct laryngoscopy in 217 patients (77%). Of patients for whom the initial attempt was with direct laryngoscopy, 86 (40%) were initially attempted with a Macintosh 3 blade, 121 (56%) with a Macintosh 4 blade, and 10 (4%) with a Miller blade. Characteristics of patients for whom the initial attempt to intubate was with video laryngoscopy were similar to those of patients for whom the initial attempt was with direct laryngoscopy, with the exception of the postgraduate year status of the resident performing the intubation (Table 1). Third- and fourth-year residents performed a greater proportion of the 63 intubations in which video laryngoscopy was the first device used (83%) than the 217 intubations in which direct laryngoscopy was the first device used (60%).
|Mean age, yr (range)||51 (18–80)||50 (18–100)||0.73|
|Mean weight, kg (range)||74 (45–114)||77 (40–220)||0.42|
|Medical or trauma patient|
| Medical patients (%)||46 (73)||137 (63)||0.15|
| Trauma patients (%)||17 (27)||80 (37)|
|Reason for intubation|
| Altered mental status (%)||42 (67)||137 (63)||0.61|
| Respiratory failure (%)||32 (51)||95 (44)||0.32|
| Facial trauma (%)||6 (10)||18 (8)||0.76|
| Immobilize for imaging (%)||7 (11)||17 (8)||0.41|
| Cervical spine precautions (%)||18 (29)||90 (41)||0.06|
| Cardiac arrest (%)||3 (5)||28 (13)||0.07|
| Obesity (%)||2 (3)||13 (6)||0.38|
|Resident postgraduate year†|
| Second year (%)||11 (17)||86 (40)||0.005|
| Third year (%)||33 (52)||83 (38)|
| Fourth year (%)||19 (30)||48 (22)|
Overall, 233 patients (83%, 95% CI = 78% to 87%) were successfully intubated on first attempt. The first-attempt success rates for video laryngoscopy (81%) and direct laryngoscopy (84%) were similar (p = 0.59; Table 2). Prior to logistic regression, an interaction was identified between the postgraduate year of the resident performing the intubation × intubation method. Therefore, new variables were created combining these variables for regression analysis. Regression analysis showed a significantly higher rate of first-attempt success rate in third- and fourth-year residents using direct laryngoscopy, compared with second-year residents using direct laryngoscopy (odds ratio [OR] = 6.2, 95% CI = 2.3 to 19.4; OR = 3.3, 95% CI = 1.1 to 10.3, respectively). The other combinations of direct and video laryngoscopy for each postgraduate year of the intubator were not significantly different when compared with second-year residents using direct laryngoscopy (p > 0.05). Additionally, no patient characteristics were significantly associated with the primary outcome (Table 3). The Hosmer and Lemeshow goodness-of-fit test (χ2 = 7.11; p = 0.52) indicated satisfactory fit of the logistic model.
|First-attempt success rate, No. (%) (95% CI)||51 (81) (70–89)||182 (84) (79–88)|
|Failure rate, No. (%) (95% CI)||9 (14) (7–25)||17 (8) (5–12)|
|Median time to intubation, seconds (95% CI)||42 (35–50)||30 (30–30)|
|Median time to intubation for patients intubated on first attempt, seconds (95% CI)||40 (30–45)||30 (25–30)|
|Mean initial pulse oximetry, % (95% CI)||97 (95–99)||96 (95–97)|
|Mean lowest pulse oximetry, % (95% CI)||89 (84–94)||91 (89–93)|
Predictors of Intubation Success on First Attempt*
|Patient age||0.999 (0.982–1.016)|
|Patient weight||0.988 (0.975–1.003)|
|Traumatic injury, no facial trauma†||1.400 (0.499–3.930)|
|Facial trauma†||1.165 (0.240–5.651)|
|Cervical spine immobilization||0.776 (0.272–2.214)|
|Direct laryngoscopy by third year‡||6.772 (2.338–19.619)|
|Direct laryngoscopy by fourth year‡||3.305 (1.066–10.250)|
|Video laryngoscopy by second year‡||1.797 (0.352–9.172)|
|Video laryngoscopy by third year‡||1.239 (0.488–3.147)|
|Video laryngoscopy by fourth year‡||3.508 (0.733–16.795)|
Failure rates were also similar for video and direct laryngoscopy (14% vs. 8%; p = 0.12). Of the nine video laryngoscopy failures, two were eventually intubated successfully using video laryngoscopy by a second physician, and seven were eventually intubated by the initial physician using direct laryngoscopy. Of the 17 direct laryngoscopy failures, nine were eventually intubated by the same physician with a different direct laryngoscope blade, six were intubated by another physician, and one required cricothyrotomy. After direct laryngoscopy failure, video laryngoscopy was not used as a rescue device in any cases. Median time to successful intubation was 42 seconds for video laryngoscopy versus 30 seconds for direct laryngoscopy (p < 0.01).
There were a total of 38 different laryngoscopists in the study. All had used the GlideScope on a manikin, but for some this was their first time using the GlideScope on a patient. Laryngoscopists had a median total number of four entries into the study, with a median of one video laryngoscopy intubation and three direct laryngoscopy intubations. Two laryngoscopists used the GlideScope 10 or more times. These two laryngoscopists had a first-attempt success rate with the GlideScope of 82%, almost identical to that for the group as a whole. A logistic regression analysis accounting for random effects due to multiple intubations having been performed by the same residents did not appreciably alter the findings.
In our study of 280 patients, no significant difference was observed in first-attempt success rate or failure rate between video and direct laryngoscopy. Prior to logistic regression analysis for the outcome variable of successful first-attempt intubation, an interaction was identified between the postgraduate year of the resident × the type of laryngoscopic method used, that is, the pattern of success by method of intubation differed by year of the resident performing the intubation. In the logistic regression model accounting for this interaction, there was a significant increase in first-attempt success only for intubations performed by third-year residents using the direct method compared with second-year residents using the video method.
The median time to successful intubation was 12 seconds longer for the video laryngoscopy than direct laryngoscopy. A similar increase was also found by Sun and colleagues.4 An additional 12 seconds for intubation is unlikely to be clinically important, particularly in patients whose oxygen saturation is maintained above 90% during the intubation. However, it is possible that a 12-second delay in intubation could worsen hypoxia and could be clinically relevant in patients who are profoundly hypoxic or have preexisting intracranial disease (e.g., head trauma or stroke).
No significant difference was observed in the mean lowest oxygen saturation recorded for patients intubated with video versus direct laryngoscopy. Upper-level residents performed a greater proportion of the video laryngoscopy intubations than the direct laryngoscopy intubations. This difference probably reflects the preference among attendings that second-year residents master direct laryngoscopy before using video laryngoscopy. This would be expected to improve success rates in patients intubated with video laryngoscopy. Although patient characteristics were recorded, we did not ask providers why they chose one device over the other for each intubation. This information may have facilitated our attempts to evaluate for the presence of selection bias and the role of each device as a rescue technique.
Several previous studies have examined the performance of the GlideScope either by anesthesiologists using manikins or in patients requiring elective surgical care.3–7 In a manikin study, anesthesiologists stated that the GlideScope offered a better view than direct laryngoscopy in the presence of pharyngeal obstruction, but no difference was reported for patients with cervical rigidity or tongue edema.3 A study of emergency physicians intubating manikins with either video or direct laryngoscopy found improved views in patients with cervical immobilization.7 In a study of the GlideScope versus direct laryngoscopy intubation by an anesthesiologist for patients requiring surgical care, GlideScope was found to improve the view by at least one grade in 68% of patients with a Cormack and Lehene (C&L) view grade 2 or 3 when examined initially using direct laryngoscopy.4 In the randomized portion of this study, GlideScope intubations took longer than direct laryngoscopy intubations (46 seconds vs. 30 seconds). This study excluded patients with known airway pathology or cervical spine injury or requiring rapid sequence intubation. An observational study by Cooper et al.5 examined 728 GlideScope laryngoscopies in the operating room.5 Use of the GlideScope resulted in C&L grade 1 or 2 views in 99% of patients. They also reported that 92% of patients predicted to have difficult airways (Mallampati class 3) had a C&L grade 1 view with the GlideScope. An observational study of the use of the GlideScope video laryngoscopy in the ED setting reported that 15 of 21 patients (71%) were successfully intubated.8
Although data on visualization with each approach were not recorded in the present study, it was the authors’ experience that the GlideScope often did provide a good view of the vocal cords. However, the sharp angle created by the blade sometimes made the passage of the tube difficult. In some cases, rotating the tube 90° clockwise or removing the stylet was the only way that the tube could be advanced through well-visualized vocal cords. In other cases, vocal cords were well visualized, but the tube could not be passed by the physician, and the patient was intubated with direct laryngoscopy. Although visualization of the vocal cords is a helpful step in a successful intubation, it is neither necessary nor sufficient. Future studies that examine the use of video laryngoscopy should include the first-attempt success rate, failure rate, and total time to intubation.
As with any method of intubation, experience probably leads to improved performance with video laryngoscopy. Because the physicians in this study had more experience with direct laryngoscopy than with video laryngoscopy prior to patient enrollment, the impact of duration of training on the outcome measures used in this study requires further evaluation. However, because most practicing emergency physicians who purchase this equipment will probably have the same exposure to the video laryngoscope than we did, namely a 30-minute orientation session plus practice on a manikin, the current study provides a realistic comparison of the two methods of intubation after introduction of the GlideScope into an ED. Individuals who used the GlideScope 10 or more times had the same first-attempt success rate as the group overall. Although it is possible that the GlideScope would serve as a valuable backup device for a difficult airway in individual cases, during the 1.5-year study period at our institution, we found no cases where the GlideScope was used after a failed attempt with direct laryngoscopy.
As an observational study, choice of device for intubation was left to the provider. Although patient characteristics were similar for the two groups (Table 1), such similarities do not exclude the possibility of selection bias. There are patient characteristics such as degree of agitation and rapidness of deterioration that may make them difficult to intubate that we did not collect. Even with a more extensive description of patient characteristics, it would not be possible to account for all variables that might differ between the two groups. We used logistic regression analysis in an attempt to control for patient characteristics that may have impacted the success of intubation on first attempt. In retrospect, an objective measurement of airway difficulty, such as a Mallampati score, would have been a useful variable to collect for inclusion in the logistic regression analysis and may have improved our ability to control for patient characteristics. Additionally, a randomized clinical trial (comparing one intubation approach with the other) could have been performed to eliminate potential selection biases. This approach would, however, require a waiver of informed consent and a preceding community consultation, a time-intensive process that would have significantly delayed our study.
Although the study was designed for consecutive enrollment of all patients meeting study criteria, we were not able to confirm that all eligible patients were enrolled in the study. Perhaps a patient was missed if the resident was overwhelmed by clinical responsibilities. We think unlikely that failure to enroll eligible patients, if it occurred, resulted in selection bias.
There was preferential use of direct laryngoscopy in cases of cardiac arrest (direct laryngoscopy in 28 cases, GlideScope in three), probably due to the more immediate availability of the direct laryngoscope. The lack of use of the GlideScope as a rescue technique after failed direct laryngoscopy suggests that providers were probably not biased to select the GlideScope for patients with difficult airways. A randomized study may provide additional information about the contribution of video laryngoscopy to ED intubations.
It is also possible that we found no difference in the lowest reported oxygen saturation because hypoxic events went unnoticed or were underreported. While we acknowledge this, our prospective real-time data collection decreases the likelihood that this occurred.
Our study focused on the GlideScope as a primary airway device rather than as a rescue technique. It is possible that the GlideScope has a more important role in the difficult or failed airway. It might offer an advantage over direct laryngoscopy in patients with challenging anatomy, such as an anterior larynx. Our study did not address this issue and cannot comment on this application.
Initial intubations were all performed by emergency medicine residents who had varying levels of experience in airway management. Also, most of the residents had limited experience using video laryngoscopy. By contrast, all had at least 50 intubations with direct laryngoscopy as part of their first-year training prior to participating in the study. The substantial difference between resident experience with direct laryngoscopy and experience with the GlideScope means that this study does not provide an equal comparison of video versus direct laryngoscopy. It does, however, provide a perspective on the effectiveness of video laryngoscopy when introduced into an ED. We feel that the 30-minute orientation and practice on the manikin is probably similar to the amount of experience that most emergency physicians would get prior to using a video laryngoscope on patients if the equipment were introduced into a practice, although the impact of duration of training on the outcome measure may require further study.
The initial attempt for most patients in the study was direct laryngoscopy. The preference for direct laryngoscopy over video laryngoscopy is probably in part due to the lack of experience with the device and in part because there were only two GlideScopes in the department during the study period. ED providers were aware that a study was being conducted. This may have encouraged them to attempt to intubate the patient more rapidly or use video laryngoscopy when they otherwise might not have. Sedation and paralysis are used in the large majority of intubations in our ED, and their use was not specifically recorded for the study. We think it unlikely that there were meaningful differences in the type of sedation or paralysis used between the groups. There are several video laryngoscopes currently on the market; we studied the GlideScope, and our results reflect only the performance of this device. Future changes in the shape of the blade and stylet and the design of the light, video camera, and monitor may change the performance of video laryngoscopy equipment.
In this observational study, the rate of successful intubation on first attempt was found to be similar between video and direct laryngoscopy, but intubation using video laryngoscopy took more time to complete.