A Systematic Review and Meta-Analysis of Randomized Controlled Trials on Supine vs. Nonsupine Endotracheal Intubation

Background This systematic review and meta-analysis of randomized controlled trials (RCTs) was performed to compare the safety and efficacy of supine vs. nonsupine positions during intubation. Methods Based on the literature from inception to October 2020, 13 studies with nonemergent intubation in supine and nonsupine positions were chosen using PRISMA and MOOSE protocols. Pooled estimates were calculated using random-effects models with 95% confidence interval (CI). The primary outcome was a successful intubation, attempt, and duration of intubation. The secondary outcome was adverse events (trauma and hypoxia). Bias was evaluated qualitatively, by visual analysis, and quantitatively through the Egger test. Results The final analysis included 13 clinical trials with 1,916 patients. The pooled success rates in the supine vs. lateral positions were 99.21% and 98.82%. The supine vs. semierect positions were 99.21% and 98.82%. The 1st attempt success rate in the supine vs. lateral position was 85.35% and 88.56% compared to 91.38% and 90.76% for the supine vs. semierect position. The rate of total adverse events in the supine position was 3.73% vs. 6.74% in the lateral position, and the rate of total adverse events in the supine position was 0.44% vs. 0.93% in semierect position. Low to substantial heterogeneity was noted in our analysis. Discussion. There is no significant difference between total successful intubations and success from 1st intubation attempt between supine and nonsupine positions. However, there are slightly higher rates of adverse events in nonsupine position. Addition of more recent studies on supine vs. nonsupine intubations would improve this study. Given these findings, it is important to develop more studies regarding different intubation positions and techniques with the aim of improving efficacy and decreasing adverse outcomes. Other. This review is not registered in a public database. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.


Introduction
Endotracheal intubation is a procedure performed to secure the airway for various elective and emergent indications, including surgeries, respiratory failure, and altered mental status [1]. Te frst reported endotracheal intubation was performed in 1878 by William Hack to remove vocal cord polyps [2]. Multiple enhancements have been applied to this procedure to improve outcomes and reduce adverse events [3]. Success in the frst attempt and hypoxia are welldetermined factors for adverse events related to endotracheal intubation.
Tere are currently no guidelines regarding optimum positioning during endotracheal intubation. Te supine position is the most common position used for endotracheal intubation [4]. With the advent of medicine, other positions, including lateral and semierect, are being used for various indications [5,6]. Experience in intubation in nonsupine positions is vital as inadequate airway management can be catastrophic, leading to hypoxia, brain injury, and even death [7]. Intubation in a nonsupine position, especially in the lateral position, could be challenging for many reasons, such as unfamiliarity and lack of experience between providers and distortion of the upper airway anatomy.
Multiple studies have suggested that intubation in the semierect position can improve the upper airway's alignment by providing a better glottic view and reducing the time of successful intubation [8][9][10]. It is hypothesized that intubation in the semierect position can lead to increased functional residual capacity and duration of nonhypoxic apnea compared to the supine position [11][12][13][14]. Te lateral position during intubation is needed in many situations: patients with back pain/lesions, increased aspiration risk, dislodgement of the endotracheal tube during surgery in the lateral position, amongst various other indications [15][16][17].
Tis study was intended to do a systematic review and meta-analysis of randomized controlled trials (RCTs) to compare the safety and efcacy of semierect vs. supine and lateral vs. supine positioning during endotracheal intubation.

Search Strategy.
A comprehensive search of several databases and conference proceedings, including PubMed, Google Scholar, Embase, CINAHL, and Cochrane (earliest inception to October 2020), was carried out to identify RCTs comparing endotracheal intubation in the supine vs. semierect position and supine vs. lateral position. We performed the search according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines and Meta-Analysis of Observational Studies in Epidemiology (MOOSE) protocol ( Figure 1).
We searched the abovementioned databases using a combination of keywords "intubation," "supine," "lateral," "ramped-up," "semi-erect," and "complications." Two authors reviewed study titles and abstracts and excluded studies not related to the research question according to semidetermined inclusion and exclusion criteria. Te remaining articles were read in detail to fnd studies with relevant information. Conficts in the fnal inclusion of a study were resolved by discussion with a third author. Te authors reviewed references of the selected articles and other relevant articles, and the appropriate studies were added.

Study Selection.
In this meta-analysis, we included RCTs comparing the safety and technical success of endotracheal intubation in the supine vs. semierect position and supine vs. lateral position.
We applied the following exclusion criteria in our study: (1) all studies other than randomized clinical trials, (2) studies with fewer than 50 patients, (3) studies with patients undergoing emergent endotracheal intubation, (4) studies performed in pediatric populations (age <18 years), and (5) studies not published in the English language.
In the case of cohort overlap or multiple publications from a single cohort, data from the most recent and most relevant articles were included.  independently performed quality scoring. We used the Oxford JADAD score for evaluation of the quality of our study. Tis quality score consisted of 3 aspects, as shown in Table 1.

Outcomes Assessed
Primary outcomes are as follows: (1) Pooled rate of total successful intubations: supine vs. lateral position and supine vs. semierect position (2) Pooled rate of success from 1st, 2nd, and 3rd intubation attempts: supine vs. lateral position and supine vs. semierect position (3) Pooled rate of time for successful intubation: supine vs. lateral position and supine vs. semierect position Secondary outcomes are as follows: (1) Pooled rate of total and individual adverse events: supine vs. lateral and supine vs. semierect positions (2) Correlation of Mallampati score to successful intubation

Assessment Methodology and Defnitions.
We matched the data collected between all the study groups (supine position, lateral position, and semierect position) before sending it for statistical analysis. Te semierect position was defned as raising the back by 25°from horizontal plane in 2 studies, the elevation of the upper body until an imaginary horizontal line can be drawn between the sternal notch space and the external ear in one study, and as raising the table's back-end to the patient's comfort level [5,21,24,29].

Statistical Analysis.
To measure the individual pooled estimates, we utilized meta-analysis techniques, as suggested by DerSimonian and Laird, using the random-efects model. If an incidence of an outcome was zero, we performed statistical analysis after adding a continuity correction of 0.5 to the number of incident cases. We performed a heterogeneity assessment of study-specifc estimates by utilizing the Cochran Q statistical test for heterogeneity, 95% prediction interval (PI), which addresses the efects' dispersion, and the I 2 statistics. Tis test reported heterogeneity as low, moderate, substantial, and considerable based on values of <30%, 30-60%, 61-75%, and >75%, respectively. We evaluated publication bias both qualitatively, by visual analysis of funnel plot, and quantitatively, by the Egger test. In the case of publication bias, we performed additional analysis using the fail-safe N test and Duval and Tweedie's "trim and fll" test to determine the impact of the bias. Te degree of publication bias impact was classifed into three levels based on the similarity between the reported results and estimated results if no bias. Te impact was classifed as minimal if both results were estimated to be the same, modest if changed efect size substantially but the conclusion remained the same, and severe if bias compromised the fnal fndings of the analysis. We performed all analyses using comprehensive metaanalysis software, version 3 (BioStat, Englewood, NJ).

Search Results and Population Characteristics.
Te search yielded 13 studies comparing intubation in the supine position vs. other nonsupine positions. Two studies with cohort overlap were found, and one was excluded based on the relevance of the trial and information available within the manuscript. Overall, 9 studies provided data on supine vs. lateral positioning and 4 studies provided data on the supine vs. semierect position.
We followed PRISMA guidelines for study selection as illustrated in Figure 1.
Population characteristics were similar between the study groups, as illustrated in Tables 2 and 3.

Characteristics and Quality of Included
Studies. All 13 studies were RCTs. Eight studies were single-center and fve were multicenter. Tere were no population-based studies. All studies commented sufciently on clinical outcomes and variables under the study, including the intubation method, side of lateral position, and degree of semierect position, as summarized in Tables 4 and 5. Te studies included were not double blinded due to the nature of the procedure under investigation, as it is difcult to blind the investigator from the patient's position during intubation. Based on the JADAD score provided in Table 1, 8 studies were considered high quality and fve studies were low quality.
In studies comparing supine vs. semerect positions, rates of success from the frst, second, and third attempts in the supine position were 91.38% (95% CI 78.89-98.93, . P values were P ≤ 0.001 for the frst attempt, P � 0.54 for the second attempt, and P ≤ 0.001 for the third attempt ( Figure 5).
(2) Correlation of Mallampati Score with Adverse Events. In the supine vs. lateral position, there was a positive correlation between increased Mallampati score and rate of adverse events in both positions. However, it is not possible to analyze the correlation with the supine vs. semierect position because the majority of the studies did not investigate adverse events.

Sensitivity Analysis.
To determine if any individual study had an infuential efect on the meta-analysis, we sequentially excluded a study each time and analyzed the impact on the primary outcomes. Te outcomes and heterogeneity of the meta-analysis were not predominantly afected by a single study.

Heterogeneity.
We evaluated the dispersion of analysis outcomes using I 2 percentage values, which determine whether the dispersion is true vs. chance. Te I 2 test reported heterogeneity as low, moderate, substantial, and considerable based on values of <30%, 30-60%, 61-75%, and >75%, respectively.

Discussion
Our study is the frst meta-analysis of RCTs comparing supine, lateral, and semierect positions in endotracheal intubation. Data analysis showed no evidence of favorable outcomes for intubation in the supine position vs. lateral and semierect positions regarding total success rate; success from the frst, second, and third attempts; and time required to complete intubation.
In comparing randomized-controlled trials for the supine vs. semierect position, the majority of the studies showed a comparable total success rate. However, the success rate from the frst attempt in the supine vs. semierect position was noticeably lower in one study [18]. From reviewing the diferent studies available that compared the supine vs. semierect position, experience was not a factor as anesthesiologists who participated in Chang's study were noted to be "two experienced anesthesiologists." Tough Chang et al. did not mention the number of years to quantify experience, other studies, such as the study by Reddy et al. involved anesthesiologists with varying years of experience who had slightly more successful frst attempt intubations with the semierect position (92.4 vs 92.9%) in a larger sample size. In addition, the use of the Mallampati score vs. the Cormack-Lehane grade did not show the study was skewed on difcult airways. Despite this information, there are several notable diferences that could contribute to the diference in this fnding.    6 Critical Care Research and Practice Te frst noticeable diference is the region where the study was conducted. Te study by Chang et al. was performed in Korea, whereas studies by Collins, Gupta, and Reddy were carried out in western countries. Te diference in the bone or cranial structures between eastern and western counterparts [30] could have contributed to a lower success rate in the frst attempt of intubation. Given the diferences in body metrics, it can also be considered that the BMI and the Mallampati score may not be adjusted appropriately as these studies were usually performed on western body types and could have underestimated the difculty of intubation [31]. It is also important to note the diferences in BMI among all four studies. Chang's study has the lowest BMI (23.85) compared to Reddy's (28), Gupta' (46.8), and Collins' (49.9). As Reddy's study has mentioned, the semierect position can improve the view on patients who will likely be more difcult to intubate. Terefore, the angle of 25°may not be the most optimal angle for patients in Chang's study given the diferences mentioned.
Aside from the structure of the patient, there are also diferences in the position, intubation method, and medication used that could account for the lower success rate on the frst attempt of intubation. Te degree of semierect positions in the studies was variable. Some semierect positions were measured in 25°, 30°, and 35°, while others were dependent on patient preference and the angle the EAM is horizontal to the sternal notch. Te variability of the degrees of nonsupine positions may have afected the outcomes. It is possible that studies may not have optimized the view of intubation due to these diferences. Compared to the other studies, Chang used fberoptic bronchoscopy instead of laryngoscopy with a GlideScope to perform the procedures. For sedation, sevofurane was used instead of fentanyl. Compared to the laryngoscope, the bronchoscope is a more        Critical Care Research and Practice method itself of intubating through a bronchoscope could also account for the increased attempts given the criteria of 60 seconds or desaturation of less than 90% as the bronchoscope is longer and more fexible than the laryngoscope which the other studies used. Lastly, the type of sedation can also be considered as to help with the relaxation to improve airway clearance. Another diference worth noticing is that the rate of hypoxia in semierect patients' intubations was statistically signifcant. Tough some studies hypothesized that the nonerect position can increase functional residual capacity and duration of nonhypoxic apnea events, the analysis shows that there was actually less rates of hypoxia in supine positions. A possible explanation is that the studies that had these fndings were carried out with angles of 20°, 25°, or 30° [11][12][13][14]. One explanation could be that two studies with the highest BMIs, Collins' and Gupta°s, did not follow these recommendations. In addition, lack of familiarity with intubating in the supine positions could have contributed to this diference.
In RCTs comparing supine vs. lateral positions, the total success rate was comparable in all studies. Two studies [25,26] showed a signifcantly lower success rate from the frst attempt in the supine position. Similarities between the studies by Li et al. and Komatsu et al. include outcomes (more intubation attempts required in supine positioning), investigator characteristics (a small number of experienced anesthesiologists), and patient population characteristics (predominantly female in their early to mid-50s with average BMI 23). Both papers theorized that with supine positioning, gravity causes physiologic changes within oropharyngeal tissues that may hamper intubation [25,26]. Under anesthesia, the tongue relaxes and obstructs the trachea, causing resistance to bronchoscopic advancement. In addition, gravity causes the diameter between the epiglottis and posterior pharyngeal wall to decrease, causing the laryngoscopic blade to advance into the vallecula instead of the glottis. Tese obstructions may require the anesthesiologist or an assistant to reposition the mandible, increasing risk of hemodynamic instability in vulnerable patients.
However, key diferences exist. Li et al. found that lateral positioning resulted in shorter time for intubation while Komatsu et al. found that time to intubation, intubation success, and airway complications were not signifcantly diferent. Tese diferences in outcomes may be attributed to study design dissimilarities. Te degree of the lateral positions was not provided though 7 studies mentioned that they maintained an axial position for the lateral patients. Tough there is less variability compared to the semierect positioning, this diference might have afected the outcomes of the study. In addition, Li used a fexible fberoptic bronchoscope while Komatsu used a novel video laryngoscope called the airway scope (AWS) that allows for improved visualization of the glottic opening. To our knowledge, there are no randomized control trials comparing the efcacy of fexible fberoptic bronchoscopes compared to the AWS. Suzuki has described one case report of the AWS used in a morbidly obese patient with a full stomach in lieu of fexible bronchoscope [32], but this patient does not match the patient population of either studies, as their population's average BMI was approximately 23. Other studies have compared the AWS to C-MAC, GlideScope, and Macintosh laryngoscope and found that all three instruments had comparable frst intubation attempt success rates, but the  Critical Care Research and Practice AWS had the best time to intubation, ease of intubation and grade 1 laryngeal view [33]. Given that the AWS had differences in performance metrics to the C-MAC and Gli-deScope, it is possible that the AWS performs diferently from the fexible fberoptic scope.
Notably, Komatsu et al. chose to have one experienced anesthesiologist assigned to laryngoscopy with a Macintosh #3 blade before intubation while the AWS was assigned to a diferent anesthesiologist. Te investigator that performed laryngoscopy had limited experience, having only performed 5 with the Macintosh laryngoscope. Conversely, the investigator that performed the intubation was very experienced with the AWS, having performed 100 intubations using the AWS. Laterally positioned patients' bodies (but not heads and necks) were also stabilized with an assistant. Having multiple unblinded participants may have led to some indiscriminate bias, limiting true comparison between studies. In addition, the two studies difered in patient population as Komatsu excluded patients with ASA 4 or more and anticipated airway difculties (Mallampati grade 4 and thyromental distance <6 cm) compared to Li's exclusion of patients with ASA 3 or more and any history of poor cardiopulmonary function. Komatsu's sicker population may have rendered intubation outcomes in diferent positions null.
Te rate of total adverse events in both lateral and semierect positions was almost two times that with the supine position, (6.74% vs. 3.73%, P ≤ 0.001) and (0.93% vs. 0.44%, P ≤ 0.001), respectively. Our results suggest that while intubation in supine, lateral, and semierect positions is technically comparable, nonsupine positions might be associated with more adverse events. Tis result could be attributed to various reasons, including unfamiliarity and lack of experience between providers, distortion of the upper airway anatomy as occurs in the lateral position, and other unidentifed variables [34,35]. It is reasonable to think that as these positions are used more frequently, the rate of adverse events will become comparable to the supine position. However, it is important to be aware of the risk factors for adverse events during intubation.
A higher Mallampati score was associated with a higher rate of adverse efects in supine and lateral positions. Tis score is an independent risk factor for difcult intubation which is associated with multiple adverse events including hypoxia and traumatic intubations [36][37][38]. Older age correlated with an increased rate of adverse events in all positions and decreased rate of total success in supine and semierect positions. Tis fnding is likely related to reduce upper airway size and increased Mallampati score and laryngoscopy grade in older population [39,40]. Although male sex is associated with higher Mallampati score [40], upper airway collapsibility [39], and intubation forces [41], our analysis showed higher percentage of males correlated with increased rate of successful intubation and reduced rate of adverse events. Females have diferent upper airway anatomy, physiology, and biomechanics compared to males [39,[41][42][43]. Furthermore, in obese females, intubation is further complicated by impediment of laryngoscopy by patients' breasts [44]. Hasanin conducted a trial that showed modifed-ramped position for females resulted in better outcomes compared to the classical ramped position [45].
Despite the diferent fndings noted, it is important to mention the limitations of this study. One would be intubation techniques, methods, and positions were not compared. Tese diferences could afect the accuracy of the analysis given the diference in performance metrics. Another is the diferent levels of experience anesthesiologists have per study. However, the information on intubations per level of experience was not discussed in all of the studies, so a proper comparison could not be achieved. Finally, the studies had diferent defnitions of failed intubations, specifcally number of attempts or time period. Te defnitions for failed attempts with time periods also vary. Te diferent criteria for failed intubation were not statistically analyzed, which makes it another limitation of this study.
In conclusion, there are no signifcant diferences regarding success rates, attempts, and time for intubation with supine vs. lateral vs. semierect positions. However, there are slightly higher rates of adverse events in the semisupine position as compared to the supine position. A variety of factors, such as ethnicity, age, and sex, can afect the outcomes. It would be interesting to see if more recent data show any changes in primary and secondary outcomes based on the position and risk factors. Furthermore, studies can be carried out to address the limitations of this study. Lastly, more studies can be performed regarding diferent techniques to improve efcacy and decrease adverse outcomes of these variables.

Data Availability
Te data used to support the fndings of this study are available from the authors upon request.

Additional Points
Te review was not registered in a public database. Aside from the PRISMA, there are no additional amendments to the information and the protocol.

Disclosure
Tis research is an extended version of "Supine vs. Nonsupine Intubation: A Systematic Review and Meta-analysis of Randomized Control Trials" that was presented in CHEST 2021.

Conflicts of Interest
Te authors declare that they have no conficts of interest.