Videolaryngoscopy versus direct laryngoscopy for adult patients requiring tracheal intubation

Sharon R Lewis; Andrew R Butler; Joshua Parker; Tim M Cook; Andrew F Smith

doi:10.1002/14651858.CD011136.pub2

Videolaringoscopia versus laringoscopia directa para pacientes adultos que requieren intubación traqueal

Authors' declarations of interest

Version published: 15 November 2016 Version history

https://doi.org/10.1002/14651858.CD011136.pub2

Collapse all Expand all

Resumen

available in

Antecedentes

La intubación traqueal exitosa durante la anestesia general tradicionalmente requiere una línea de visión a la laringe lograda mediante el posicionamiento de la cabeza y el cuello y el uso de un laringoscopio para retraer la lengua y las partes blandas del suelo de la boca. Es común que surjan dificultades con la intubación, y se han diseñado laringoscopios alternativos que emplean tecnología digital o de fibras ópticas para mejorar la visibilidad cuando se predicen o se encuentran dificultades en las vías respiratorias. Entre estos dispositivos, un videolaringoscopio (VLS) rígido utiliza una hoja para retraer las partes blandas y transmite una imagen de video iluminada a una pantalla.

Objetivos

El objetivo primario fue evaluar si el uso de videolaringoscopia para la intubación traqueal en adultos que requerían anestesia general reduce los riesgos de complicaciones y el fracaso en comparación con la laringoscopia directa. El objetivo secundario fue evaluar los beneficios y los riesgos de estos dispositivos en grupos de población seleccionados, como los adultos con obesidad y los que presentan dificultades conocidas o previstas en las vías respiratorias.

Métodos de búsqueda

Se hicieron búsquedas en el Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials) (CENTRAL), MEDLINE y en Embase el 10 febrero 2015. Los términos de búsqueda fueron relevantes para la pregunta de la revisión y no se limitaron según los resultados. Se realizaron búsquedas en los registros de ensayos clínicos y el seguimiento de citas hacia adelante y hacia atrás. Se volvió a realizar la búsqueda el 12 de enero de 2016; a esa búsqueda del 2016 se añadieron nuevos estudios potenciales de interés a una lista de "Estudios en espera de clasificación", y se incorporarán estos estudios en la revisión formal durante la actualización de la revisión.

Criterios de selección

Se consideraron todos los ensayos controlados aleatorizados y los estudios cuasialeatorizados con pacientes adultos sometidos a la laringoscopia realizada con un VLS o un laringoscopio Macintosh en contextos clínicos, de emergencia o extrahospitalarios. Se incluyeron diseños de estudios de grupos paralelos y cruzados (cross‐over).

Obtención y análisis de los datos

Dos autores de la revisión evaluaron de forma independiente la calidad de los ensayos, extrajeron los datos y consultaron con un tercer autor para resolver los desacuerdos. Se utilizaron procedimientos metodológicos Cochrane estándar, incluyendo la evaluación del riesgo de sesgo.

Resultados principales

Se incluyeron 64 estudios identificados durante la búsqueda de 2015 que incluían a 7044 participantes adultos y comparaban un VLS de uno o más diseños con un laringoscopio Macintosh. Se identificaron 38 estudios en espera de clasificación y siete estudios en curso. De los 64 estudios incluidos, 61 incluían a pacientes sometidos a cirugía electiva, y tres se realizaron en un contexto de emergencia. Entre 48 estudios que incluían a participantes sin dificultades previstas en las vías respiratorias, 15 usaron técnicas para simular una vía respiratoria difícil. Siete incluyeron a participantes con dificultades conocidas o previstas en las vías respiratorias, y los estudios restantes no especificaron ni incluyeron las vías respiratorias con dificultades previstas y no previstas. Sólo dos estudios reclutaron específicamente a los participantes obesos. No fue posible cegar al intubador al dispositivo, y se observó un nivel alto de heterogeneidad inevitable, debido al número grande de estudios.

Estadísticamente, se informaron significativamente menos intubaciones fallidas cuando se utilizó un VLS (odds‐ratio (OR) Mantel‐Haenszel (M‐H), efectos aleatorios 0,35; intervalo de confianza (IC) del 95%: 0,19 a 0,65; 38 estudios; 4127 participantes) y menos intubaciones fallidas cuando se utilizó un VLS en participantes con dificultades previstas en las vías respiratorias (OR M‐H, efectos aleatorios 0,28; IC del 95%: 0,15 a 0,55; seis estudios; 830 participantes). La calidad de esta evidencia se consideró moderada sobre la base del sistema GRADE. Las intubaciones fallidas fueron menores cuando se utilizó un VLS en participantes con una simulada vía respiratoria difícil (OR M‐H, efectos aleatorios 0,18; IC del 95%: 0,04 a 0,77; nueve estudios; 810 participantes), pero los grupos sin previsión de tener vías respiratorias difíciles no proporcionaron resultados significativos (OR M‐H, efectos aleatorios 0,61; IC del 95%: 0,22 a 1,67; 19 estudios; 1 743 participantes).

Ocho estudios informaron sobre la hipoxia y sólo tres de ellos describieron algún evento; los resultados no mostraron diferencias entre los dispositivos para este resultado (OR M‐H, efectos aleatorios 0,39; IC del 95%: 0,10 a 1,44; 1319 participantes). De igual manera, pocos estudios informaron sobre la mortalidad y no observaron diferencias entre los dispositivos (OR de M‐H, efecto fijo 1,09; IC del 95%: 0,65 a 1,82; dos estudios; 663 participantes) y sólo un estudio informó sobre la aparición de complicaciones respiratorias (78 participantes); estos tres resultados se calificaron como de muy baja calidad debido a la falta de datos. No se encontraron diferencias estadísticamente significativas entre los dispositivos en la proporción de primeros intentos exitosos (OR M‐H, efectos aleatorios 1,27; IC del 95%: 0,77 a 2,09; 36 estudios; 4731 participantes) ni en aquellos que necesitaban más de un intento. La calidad de esta evidencia se calificó como moderada. Los estudios no informaron diferencias estadísticamente significativas en la incidencia de dolor de garganta en la unidad de cuidados posanestésicos (UPA) (OR M‐H, efectos aleatorios 1,00 (IC del 95%: 0,73 a 1,38); 10 estudios; 1548 participantes) ni a las 24 horas postoperatorias (OR M‐H, efectos aleatorios 0,54; IC del 95%: 0,27 a 1,07; ocho estudios; 844 participantes); se calificó la calidad de esta evidencia como moderada. Los datos combinados para incluir estudios de diseño cruzado (cross‐over) revelaron significativamente menos traumatismos laríngeos o de las vías respiratorias (OR M‐H, efectos aleatorios 0,68; IC del 95%: 0,48 a 0,96; 29 estudios; 3110 participantes) y menos incidencias de ronquera postoperatoria (OR M‐H, efecto fijo 0,57; IC del 95%: 0,36 a 0,88; seis estudios; 527 participantes) cuando se usó un VLS. Un mayor número de laringoscopias realizadas con un VLS lograron una visión de la mayor parte de la glotis (OR M‐H, efectos aleatorios 6,77; IC del 95%: 4,17 a 10,98; 22 estudios; 2240 participantes), menos laringoscopias realizadas con un VLS no lograron una visión de la glotis (OR M‐H, efectos aleatorios 0,18; IC del 95%: 0,13 a 0,27; 22 estudios; 2240 participantes) y un VLS fue más fácil de usar (OR M‐H, efectos aleatorios 7,13; IC del 95%: 3,12 a 16,31; siete estudios; 568 participantes).

Aunque un gran número de estudios informó el tiempo necesario para la intubación traqueal (55 estudios; 6249 participantes), no se presentó una estimación de los efectos para este resultado debido al nivel extremadamente alto de heterogeneidad estadística (I² = 96%).

Conclusiones de los autores

Los videolaringoscopios podrían reducir el número de fracasos en la intubación, en particular entre los pacientes que se presentan con dificultades en las vías respiratorias. Mejoran la visión glótica y podrían reducir el traumatismo laríngeo/de las vías respiratorias. En la actualidad, no hay evidencia para indicar que el uso de un VLS reduce el número de intentos de intubación o la incidencia de hipoxia o complicaciones respiratorias, ni evidencia para indicar que el uso de un VLS afecta el tiempo requerido para la intubación.

PICOs

Population

Intervention

Comparison

Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Resumen en términos sencillos

available in

Videolaringoscopios para guiar la inserción de tubos respiratorios en pacientes quirúrgicos adultos

Antecedentes

Los pacientes que requieren anestesia general necesitan asistencia respiratoria durante la cirugía. Para brindar dicha asistencia, el anestesista podría insertar un tubo a través de la boca o la nariz, bajar por la tráquea y llegar a los pulmones. Para realizar este procedimiento, conocido como intubación traqueal, el anestesista por lo general usa un instrumento metálico llamado laringoscopio para correr la lengua y los partes blandas de la boca y poder ver las cuerdas vocales directamente antes de la intubación. Sin embargo, podría ser difícil ver las cuerdas vocales, p.ej. cuando el paciente presenta restricciones en el movimiento del cuello, y cualquier dificultad en la intubación podría dar lugar a complicaciones para el paciente. Otros laringoscopios, llamados videolaringoscopios, emplean tecnología de video y podrían mejorar la visión del anestesista antes de la intubación. Esta tecnología permite al anestesista ver realmente la posición del tubo en una pantalla de video mientras lo está insertando. Esta revisión procuró evaluar si los videolaringoscopios reducen los riesgos de complicaciones y el fracaso en la intubación.

Características de los estudios

La evidencia está actualizada hasta el 10 de febrero de 2015. Se encontraron 64 estudios con 6895 participantes. Los estudios compararon a anestesistas que utilizaron diferentes tipos de videolaringoscopios con anestesistas que utilizaron un laringoscopio estándar Macintosh sin la característica de video. Se repitió la búsqueda el 12 de enero de 2016 y se considerarán los nuevos estudios de interés cuando se actualice la revisión.

Resultados clave

Se combinaron los resultados de los estudios mediante pruebas estadísticas y se encontraron menos fracasos en las intubaciones que requirieron intubación con el dispositivo alternativo cuando se usó un videolaringoscopio en los pacientes, incluidos los que presentaban dificultades en las vías respiratorias, en comparación con el uso del laringoscopio estándar. Los participantes también presentaron menos probabilidades de tener lesiones menores en la boca/garganta o de experimentar ronquera después de la intervención quirúrgica. Los anestesistas tuvieron una mejor visión antes de la intubación y evaluaron el videolaringoscopio como más fácil de usar que un laringoscopio estándar. Los investigadores no informaron diferencias en el número de participantes adultos con dolor de garganta ni diferencias en el número de primeros intentos exitosos o el número general de intentos. No fue posible combinar los datos para comparar los estudios estadísticamente para el tiempo que tomó el uso de un videolaringoscopio debido al número de diferencias en los puntos temporales medidos. Se identificaron 38 estudios para su posible inclusión y se evaluarán estos estudios durante la actualización de la revisión.

Calidad de la evidencia

Aunque se observaron buenos métodos en algunos de los estudios, no fue posible para los investigadores ocultar al anestesista al tipo de laringoscopio utilizado, y se cree que este hecho podría haber comprometido la calidad de la evidencia a favor de cualquiera de los tipos de laringoscopio.

Conclusiones

La evidencia indica que los videolaringoscopios podrían mejorar el éxito de la intubación traqueal, en particular cuando el paciente presenta dificultades en las vías respiratorias.

Authors' conclusions

Implications for practice

Our evidence suggests that videolaryngoscopes may aid intubation, particularly in patients presenting with a predicted or known difficult airway. Their use is likely to improve the glottic view and reduce the number of laryngoscopies in which the glottis cannot be seen, irrespective of predicted or known difficulty, and may reduce the incidence of laryngeal/airway trauma. We found no evidence to indicate that use of a VLS would result in fewer attempts to intubate. We were not able to establish whether intubation is likely to take less or more time with a VLS, nor whether this would result in fewer incidences of hypoxia or respiratory complications. However, we are aware of relevant ongoing studies that compare different videolaryngoscopes with direct laryngoscopy, and a large number of studies were identified in searches run in January 2016, along with completed studies identified from clinical trials registers. This demonstrates continued research interest in this field, and incorporation of data from these studies may lead to changes in the results of this review.

Implications for research

This review has not sufficiently explored the use of VLS devices in particular clinical scenarios, for example, VLS intubation in the emergency setting during anaesthesia, and in the intensive care unit and emergency department and outside hospitals. Further research is needed on the effect of intubator experience on potential benefits of VLS. We would recommend that studies incorporate useful data on respiratory complications, hypoxia and time to intubate. Finally, we were not able to usefully distinguish performance differences between different types of VLS, but it is unlikely that devices of differing designs would perform equally; research is needed to elucidate the differential effects of different types of VLS.

Summary of findings

Open in table viewer

Summary of findings 1. Videolaryngoscopy compared with direct laryngoscopy for tracheal intubation

Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	Number of participants (studies)	Quality of the evidence (GRADE)	Comments
Videolaryngoscopy compared with direct laryngoscopy for tracheal intubation
Patient or population: patients requiring tracheal intubation Settings: clinical, emergency or out‐of‐hospital, worldwide Intervention: videolaryngoscopy Comparison: direct laryngoscopy
	Assumed risk	Corresponding risk
	Direct laryngoscopy	Videolaryngoscopy
Failed intubation	Study population		OR 0.35 (0.19 to 0.65)	4127 (38 studies)	⊕⊕⊕⊝ moderate^a	Downgraded by 1 level. See footnote.
	94 per 1000	35 per 1000 (19 to 63)
	Moderate

Hypoxia	Study population		OR 0.39 (0.1 to 1.44)	1319 (8 studies)	⊕⊝⊝⊝ very low^a,b,c	Downgraded by 3 levels. See footnotes.
	58 per 1000	23 per 1000 (6 to 81)
	Moderate

Serious respiratory complications	See comment	See comment	Not estimable	78 (1 study)	⊕⊝⊝⊝ very low^a,d	Insufficient data to complete meta‐analysis. Downgraded by 2 levels. See footnotes.
Mortality	Study population		OR 1.09 (0.65 to 1.82)	663 (2 studies)	⊕⊝⊝⊝ very low^a,e,f,g	Downgraded by 3 levels. See footnotes.
	106 per 1000	114 per 1000 (71 to 177)
	Very low

Proportion of successful first attempts	Study population		OR 0.79 (0.48 to 1.3)	4731 (36 studies)	⊕⊕⊕⊝ moderate^a,h	Downgraded by 1 level. See footnotes.
	831 per 1000	795 per 1000 (702 to 865)
	Moderate

Sore throat	Study population		OR 1.00 (0.73 to 1.38)	1548 (10 studies)	⊕⊕⊕⊝ moderate^a,i	Downgraded by 1 level. See footnotes.
	250 per 1000	289 per 1000 (211 to 385)
	Moderate

Time for tracheal intubation	See comment	See comment	Not estimable	4488 (37 studies)	⊕⊝⊝⊝ very low^a,j	High level of statistical heterogeneity between studies; therefore meta‐analysis not completed. Downgraded by 3 levels. See footnotes.
The basis for the assumed risk* (e.g. median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI = confidence interval; OR = odds ratio.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
^aNot possible to blind intubator to device. Downgraded for study limitations. ^bI² statistic shows high level of heterogeneity at 70%. Downgraded for inconsistency. ^cOnly three studies with event data. Downgraded for imprecision. ^dOnly one study. Downgraded for imprecision. ^eOnly two studies with event data. Downgraded for imprecision. ^fBoth studies include only trauma patients. ^gNo assessment of publication bias made for this outcome. ^hI² statistic shows high level of heterogeneity at 79%. Downgraded for inconsistency. ⁱI² statistic shows moderate level of heterogeneity at 55%. Downgraded for inconsistency. ^jI² statistic shows very high level of heterogeneity at 96%. Downgraded for inconsistency.

Background

Description of the condition

Securing the patient’s airway is a critical step in providing general anaesthesia. Recent data from the Fourth National Audit Project of the Royal College of Anaesthetists and Difficult Airway Society (NAP4) in the UK suggest that tracheal intubation is used for airway management in 38.4% of general anaesthetics, estimated at 1.1 million procedures per year (Woodall 2011). A cuffed tracheal tube, which is considered the most reliable device for securing the airway, is inserted through the mouth and larynx and into the trachea to enable oxygenation and ventilation, and to prevent aspiration, during general anaesthesia.

A clear view may be achieved by flexing the lower cervical spine and extending the upper cervical spine (a 'sniffing the morning air' position), enabling the intubator to create 'line of sight' to the larynx to pass the tracheal tube. Retractor type laryngoscopes, typically a detachable metal blade with handle (e.g. the Macintosh curved blade), are used to retract the tongue and soft tissue in the floor of the mouth during this procedure, which is termed 'direct laryngoscopy'. However, although these laryngoscopes may be adequate for moving soft tissue, the intubator still requires line of sight to the larynx, provided by correct head and neck positioning of the patient.

Failed or difficult intubation is associated with complications, such as increased risk of hypertension, desaturation, unexpected admission to the intensive care unit (ICU) and death (Caplan 1990; King 1990; Rose 1994). Such difficulties during intubation are estimated to occur in 1% to 6% of cases, whereas failed intubation occurs in only 0.1% to 0.3% (Crosby 1998; Shiga 2005).

Airway management difficulties are increased when patients are obese (Juvin 2003; Lundstrom 2009). In the UK, NAP4 showed that obese patients accounted for 42% of patients who experienced a major airway complication during anaesthesia (Cook 2011). Functional residual capacity (FRC), which is the volume of air left in the lungs at the end of normal expiration, is reduced in obese patients; this, along with other factors, reduces respiratory reserve and makes these patients vulnerable to hypoxia if an airway is lost, making airway management more time critical and increasing the risk of postoperative chest infection and other complications (Adams 2000; Malhotra 2008; Marley 2005).

In addition to obesity, intubation may prove difficult for other reasons, for example, restrictions in neck flexion, a narrow jaw opening, an enlarged tongue, poor tissue mobility and cervical instability. Predictive tests, for example, the Mallampati or Wilson index test (Mallampati 1985; Wilson 1988), are used before anaesthesia is given. The Mallampati score, which is based on the view of the soft palate when the patient opens his mouth, is the most widely used predictor of difficult intubation, but this and other prediction tests have been shown to have low positive predictive value for difficult intubation (Shiga 2005).

Patients who are admitted to the ICU and to the emergency department may differ from elective patients scheduled for general anaesthesia. Many patients are admitted to the ICU or the emergency department because they have vulnerable airways, which may be due to major trauma requiring cervical spine protection, airway swelling, direct airway trauma or lung injury, major head and neck surgery or infection. Critical care teams may need to provide airway management in the emergency department at very short notice without the presence of an anaesthetist (Cook 2011).

Description of the intervention and how it might work

Alternative devices, such as a videolaryngoscope (VLS), rely on fibreoptic or digital technology to transmit an image from the tip of the laryngoscope to an eyepiece or monitor, where it is viewed by the intubator. These devices may be flexible or rigid in design for the purpose of assisting in difficult intubations and reducing difficulty, failure, trauma and other complications. For this review, we are interested in the rigid videolaryngoscope, which uses a blade to retract the soft tissues and transmits a video image to a screen attached to the end of the handle or to a monitor. This design enables a lighted view of the larynx without direct 'line of sight' and therefore can assist when difficulty is encountered (or predicted) with direct laryngoscopy.

The Cormack and Lehane classification system describes the intubator’s view of the larynx during laryngoscopy (Cormack 1984), with a score or 4 indicating a poor view and a score of 1 indicating a good view. Studies suggest that the use of videolaryngoscopes improves these visualization scores (e.g. a Storz V‐Mac videolaryngoscope compared with a Macintosh laryngoscope in Kaplan 2006). Videolarngoscopes may therefore provide the possibility of more successful intubation for patients in whom direct laryngoscopy may be difficult. They also may be used after unsuccessful attempts to intubate with direct laryngoscopy.

Why it is important to do this review

Use of a videolaryngoscope may aid visualization, but evidence is required to establish whether this equates with increased success of intubation with reduced complications. Recent non‐Cochrane reviews of VLS models have concentrated on their impact on process measures, such as the view of the larynx, first‐time and overall intubation success rates and intubation time, and have concluded that there is limited evidence to support their use in tracheal intubation in unselected populations and in those with a known or anticipated difficult direct laryngoscopy (Griesdale 2012b; Healy 2012; Niforopoulou 2010). A systematic review and meta‐analysis of 17 studies of the GlideScope reported advantages for non‐expert intubators (Griesdale 2012b).

No reviews have considered the use of VLS specifically in obese patients. The prevalence of obesity is increasing in both developed and developing countries (current figures: http://www.oecd.org/), as is the number of obese patients requiring anaesthesia. It is important to establish whether videolaryngoscopy is a more effective technique for this patient group, as well as for other selected and unselected groups.

We wish to update the non‐Cochrane reviews above by focusing only on evidence derived from randomized controlled trials (RCTs) and by considering, when possible, patient relevant outcomes such as complications. We aimed to consider studies in both unselected and selected populations, and to include studies of obese participants. This review will continue the work of the current review authors in published reviews such as "Supraglottic airway devices versus tracheal intubation for airway management during general anaesthesia in obese patients" (Nicholson 2013a) and "Tracheal intubation with a flexible intubation scope for obese patients requiring general anaesthesia" (Nicholson 2013b). This review does not focus on videolaryngoscopy in children, as this topic is the focus of another Cochrane review (Abdelgadir 2014).

Objectives

Our primary objective was to assess whether use of videolaryngoscopy for tracheal intubation in adults requiring general anaesthesia reduces risks of complications and failure compared with direct laryngoscopy. Our secondary aim was to assess the benefits and risks of these devices in selected population groups, such as adults with obesity and those with a known or predicted difficult airway.

Methods

Criteria for considering studies for this review

Types of studies

We included randomized controlled trials (RCTs) of both parallel and cross‐over design. We did not include simulation or mannequin studies.

Types of participants

We included trials of participants aged 16 years and older who required tracheal intubation during general anaesthesia. We included participants scheduled for surgery, as well as participants requiring tracheal intubation in the emergency department or the ICU under general anaesthesia. We included trials with unselected patient populations, those restricted to participants with known or predicted difficult laryngoscopy (e.g. Mallampati score III or IV (Mallampati 1985) or previous Cormack and Lehane score III or IV (Cormack 1984) with direct laryngoscopy) and those restricted to participants with a body mass index (BMI) > 30 kg/m².

Types of interventions

We included studies that compared the use of a videolaryngoscope of any model versus direct laryngoscopy with a Macintosh blade.

We provide a list of example models and manufacturers in Appendix 1. We excluded optical stylets.

Types of outcome measures

Our primary outcomes were serious complications that may arise from difficulties with intubation. We included failed intubation with the first choice of device as a primary outcome. This is an important indicator of the success of an intubation technique. Failed intubation with the first device may not always result in an adverse consequence for the patient, but it increases the risk of serious complications, especially in obese patients (Cook 2012). The other primary outcome was hypoxia. Our secondary outcomes included mortality and serious airway complications, as well as surrogate process markers for airway problems, such as the number of attempts at intubation. We also assessed the impact of sore throat or hoarseness after surgery on patient‐reported measures as surrogate measures of airway trauma.

We did not include outcomes as part of the study eligibility assessment. We included studies that reported on any of the relevant outcomes even if they were not primary study outcomes.

Primary outcomes

Failed intubation or change of device required
Hypoxia between start of intubation and recovery from anaesthesia, with dichotomous data (episodes of arterial oxygen saturation < 90%) or continuous data (lowest or mean arterial oxygen saturation)

Secondary outcomes

Mortality within 30 days of anaesthesia
Serious respiratory complications (including aspiration) within 30 days of anaesthesia
Laryngeal or airway trauma – including any one of damage to vocal cords, bleeding or dental injury
Patient‐reported sore throat or hoarseness ‐ both early (within two hours of anaesthesia) and late (within 48 hours of anaesthesia)
Proportion of successful first attempts at tracheal intubation
Number of attempts at tracheal intubation
Total time for tracheal intubation and commencement of ventilation
Difficulty of tracheal intubation ‐ assessed by intubator or observer, using a locally derived or validated difficulty scale
Improved visualization of the larynx as measured on a validated scale (such as the Cormack and Lehane classification system (Cormack 1984); the POGO (percentage of glottic opening) score (Levitan 1998); or classification system by (Cook 2000).

Search methods for identification of studies

Electronic searches

We searched for eligible trials on 10 February 2015 in the following databases: Cochrane Central Register of Controlled Trials (CENTRAL; 2015, Issue 2) in the Cochrane Library (searched 10 February 2015), MEDLINE via Ovid (1970 to 10 February 2015) and Embase via Ovid (1980 to 10 February 2015). We applied the Cochrane highly sensitive filter for randomized controlled trials in MEDLINE and Embase. We searched the trial register www.clinicaltrials.gov for ongoing trials. We have presented our search strategies for MEDLINE, Embase and CENTRAL in Appendix 2, Appendix 3 and Appendix 4. We searched using both medical subject headings (MeSH) (or equivalent structured vocabulary in other databases) and free text.

We included publications that reported study data, including abstracts. We applied no restrictions on language of publication.

We reran the searches in the databases above (CENTRAL, MEDLINE and Embase) on 12 January 2016. We have added potential new studies identified during the 2016 search to Characteristics of studies awaiting classification and will incorporate these into the formal review during the review update.

Searching other resources

We undertook forward and backward citation tracking for key review articles and eligible articles identified through the electronic resources.

Data collection and analysis

Selection of studies

We collated results of the searches and removed duplicates.

Two review authors (Sharon Lewis (SL) and Andrew Butler (AB)) screened all titles and abstracts to remove studies that were ineligible. If no abstract was available but the title was possibly relevant, we obtained the full text of the article.

We (SL and AB) reviewed the full texts of potentially relevant titles. Each review author used software (www.covidence.org) to record decisions and reach consensus at each stage. We reported in a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta‐analyses) flow chart the numbers of full‐text papers assessed and exclusions at each stage, along with reasons for those reviewed in full text.

Data extraction and management

Two of three review authors (SL, AB and Joshua Parker (JP)) extracted data from eligible studies using Covidence software (Covidence).

We successfully contacted the authors of Ahmad 2013, Cordovani 2013 and Suzuki 2008 for additional information. We resolved disagreements by discussion and, if necessary, by consultation with Tim Cook (TC) or Andrew Smith (AS).

Assessment of risk of bias in included studies

We used the Cochrane risk of bias tool to assess the quality of study design and the extent of potential bias (Higgins 2011) by considering the following domains.

Sequence generation.
Allocation concealment.
Blinding of participants, personnel and outcomes assessors.
Incomplete outcome data.
Selective outcomes reporting.

It was not possible for the anaesthetist or the intubator to be blinded to the intervention for this research question and, similarly, it was difficult for assessors of outcomes during intubation to be unaware of the allocation of the participant. Outcomes assessed during or after the operation, such as airway trauma or respiratory complications, could be assessed by staff other than the intubator who were unaware of the laryngoscopy device. It is feasible that the asleep participant may not know the device used, which may be important for patient‐reported outcomes, such as sore throat.

Other sources of bias

We paid particular attention to sources of funding and the role of manufacturers and also considered the potential for selective reporting bias. We reviewed the original protocol of the trial, if this was available, to identify any changes to procedure or missing outcome data that may indicate bias.

We considered baseline characteristics of participants as well as the expertise of the anaesthetist, which has the potential to be an important confounder in this review.

We included cross‐over trials, but we conducted sensitivity analyses to determine whether they had introduced bias into the results.

Measures of treatment effect

The outcomes in this review are mainly dichotomous outcomes (mortality, complications, successful first attempt, failed intubation). For dichotomous outcomes, we entered totals and numbers of events within each randomization group into RevMan 5.3 and calculated odds ratios with 95% confidence intervals. For continuous measures (e.g. time for intubation), we calculated mean differences. We recorded some outcomes on short ordinal scales (e.g. number of attempts, intubation difficulty scores, scales of improved visualization). We converted these to dichotomous data when appropriate.

Unit of analysis issues

As well as including studies of cross‐over design, we included studies that reported more than one comparison, for example, groups allocated to two designs of videolaryngoscopes compared with a direct laryngoscopy group. We compared an amalgamated comparison group (combining each type of videolaryngoscope) with the control group, initially at least, to create a single pair‐wise comparison (Section 16.5.4 of Higgins 2011). In subgroup analyses, we presented the data for each device separately. When it was not possible to amalgamate data without unit of analysis error, we chose to include data from the VLS group that would be closest to a result of 'no effect' ‐ we then addressed these decisions in a sensitivity analysis.

Dealing with missing data

We attempted to contact study authors to request missing data and included results only when study authors confirmed data. We did not include results reported in abstracts in which denominator figures were not explicitly stated and for which we were unable to reach study authors.

Assessment of heterogeneity

We expected that the findings for any given outcome may differ between the studies included in the review. This heterogeneity may be due to:

BMI > 30 kg/m² and degree of obesity;
anticipated degree of difficulty of airway, with measures such as Mallampati score;
expertise of intubator, VLS device used (e.g. GlideScope or Pentax);
urgency of intubation (emergency vs elective); or
site of intubation (operating theatre, emergency department, ICU).

We assessed heterogeneity by using Chi² and I² statistics. We investigated important heterogeneity (Chi² test with P < 0.1 or I² > 50%) by performing subgroup analyses.

Assessment of reporting biases

We examined a funnel plot to assess the potential for publication bias for our primary outcome.

Data synthesis

We carried out meta‐analysis for outcomes for which we had comparable effect measures from more than one study, and when measures of heterogeneity indicated that pooling of results was appropriate. An I² statistical value > 80% would argue against presentation of an overall estimate. Our choice of a fixed‐effect or random‐effects statistical model for any meta‐analysis was influenced by study characteristics, in particular, the extent of methodological or clinical differences between studies. We used Mantel‐Haenszel models for all dichotomous outcomes. For our continuous outcome (i.e. time for tracheal intubation) we used the inverse variance method.

We initially combined all designs of VLS and all population types, when appropriate, before dividing data by VLS design and by unselected and selected participant groups.

Subgroup analysis and investigation of heterogeneity

We considered whether the results of meta‐analysis for the outcome of failed intubation differed for:

different designs of VLS;
obese and non‐obese participants;
anticipated or known difficult laryngoscopy;
different sites of intubation (operating theatre, emergency department, ICU); and
experienced and inexperienced intubator.

We defined experienced intubators as those who had equivalent experience in the clinical setting of at least 20 uses with each device, and inexperienced intubators as those with fewer than 20 uses of a VLS.

Sensitivity analysis

We undertook sensitivity analyses to explore the potential impact of missing data in our risk of bias assessment. We also considered the potential impact of data analysis decisions on the results.

Summary of findings

We used the principles of the GRADE system to give an overall assessment of evidence related to each of the following outcomes (Guyatt 2008).

Failed intubation or change of laryngoscopy device required.
Hypoxia between start of intubation and recovery from anaesthesia.
Mortality within 30 days of anaesthesia.
Serious respiratory complications (including pulmonary aspiration of gastric contents and lower respiratory tract infection) within 30 days of anaesthesia.
Sore throat.
Proportion of successful first attempts.
Total time for tracheal intubation and commencement of ventilation.

The GRADE approach incorporates risk of bias, directness of evidence, heterogeneity of data, precision of effect estimates and risk of publication bias to give an overall measure of how confident we can be that our estimate of effect is correct. SL used GRADEpro software to create a 'Summary of findings' table for each outcome and discussed discrepancies with AS.

Results

Description of studies

Results of the search

We screened 3412 titles and abstracts, of which we identified 406 through forward and backward citation searching. We also screened titles from clinical trials register searches. We assessed 283 full texts for eligibility. See Figure 1.

Figure 1

Study flow diagram.

We reran the search in January 2016 and screened an additional 424 titles and abstracts, following removal of duplicates. See Characteristics of studies awaiting classification.

Included studies

From the search in February 2015, we identified 64 studies that we included in the review (Abdallah 2011; Ahmad 2013; Andersen 2011; Aoi 2010; Arici 2014; Arima 2014; Aziz 2012; Bensghir 2010; Bensghir 2013; Bilehjani 2009; Carassiti 2013; Cavus 2011; Choi 2011; Cordovani 2013; Dashti 2014; Enomoto 2008; Frohlich 2011; Griesdale 2012; Gupta 2013; Hirabayashi 2007a; Hirabayashi 2009; Hindman 2014; Hsu 2012; Ilyas 2014; Ithnin 2009; Jungbauer 2009; Kanchi 2011; Kill 2013; Kim 2013; Komatsu 2010; Lee 2009; Lee 2012; Lee 2013; Lim 2005; Lin 2012; Maassen 2012; Malik 2008; Malik 2009a; Malik 2009b; Maruyama 2008a; Maruyama 2008b; McElwain 2011; Najafi 2014; Nishikawa 2009; Peck 2009; Pournajafian 2014; Robitaille 2008; Russell 2012; Russell 2013; Sandhu 2014; Serocki 2010; Serocki 2013; Shippey 2013; Siddiqui 2009; Sun 2005; Suzuki 2008; Takenaka 2011; Taylor 2013; Teoh 2010; Turkstra 2005; Walker 2009; Woo 2012; Xue 2007; Yeatts 2013). All identified studies were RCTs. We identified no quasi‐randomized studies and no cluster trials. We have summarized details of the individual studies, including countries in which studies were conducted, in the Characteristics of included studies section. See Characteristics of studies awaiting classification for potentially relevant studies identified in the search conducted in January 2016.

A total of 7044 participants were included in the 64 studies. One study took place in the intensive care unit (Griesdale 2012), one at a trauma centre (Yeatts 2013) and one in an out‐of‐hospital setting (Arima 2014), all with participants requiring emergency treatment. The remaining 61 studies took place in the hospital theatre setting with elective surgical participants. Two studies specified inclusion of only obese participants (Abdallah 2011; Andersen 2011), one study included only obstetrical participants (Arici 2014), one study only participants with untreated hypertension (Dashti 2014) and one study only participants from the burns unit (Woo 2012).

We identified 17 studies conducted by a cross‐over design (Carassiti 2013; Cavus 2011; Cordovani 2013; Enomoto 2008; Hindman 2014; Ilyas 2014; Lee 2009; Lee 2012; Maassen 2012; Maruyama 2008a; Peck 2009; Robitaille 2008; Russell 2012; Serocki 2010; Serocki 2013; Taylor 2013; Turkstra 2005) and 47 studies with a parallel design. Those studies described by study authors as cross‐over designs used one type of laryngoscope initially to assess glottic view, followed by the other type of laryngoscope to assess glottic view and perform intubation. The exception to this was Hindman 2014, which intubated participants after laryngoscopy with each device. Participants in both cross‐over designs were randomized by different orders of laryngoscope.

We included nine different types of VLS in our analysis; data showed comparisons with GlideScope (29 studies: Ahmad 2013; Andersen 2011; Bilehjani 2009; Carassiti 2013; Choi 2011; Cordovani 2013; Dashti 2014; Griesdale 2012; Hsu 2012; Ithnin 2009; Kill 2013; Lee 2012; Lim 2005; Malik 2008; Malik 2009b; Najafi 2014; Pournajafian 2014; Robitaille 2008; Russell 2012; Russell 2013; Sandhu 2014; Serocki 2010; Serocki 2013; Siddiqui 2009; Sun 2005; Teoh 2010; Turkstra 2005; Xue 2007; Yeatts 2013), Pentax AWS (20 studies: Abdallah 2011; Aoi 2010; Arima 2014; Enomoto 2008; Hirabayashi 2007a; Hirabayashi 2009; Kanchi 2011; Kim 2013; Komatsu 2010; Lee 2013; Malik 2008; Malik 2009a; Malik 2009b; Maruyama 2008a; Maruyama 2008b; Nishikawa 2009; Suzuki 2008; Takenaka 2011; Teoh 2010; Woo 2012), C‐MAC (C‐MAC ‐ nine studies: Aziz 2012; Cavus 2011; Gupta 2013; Jungbauer 2009; Lee 2009; Lee 2012; Maassen 2012; McElwain 2011; Teoh 2010; DCI ‐ one study: Serocki 2010) and McGrath (McGrath Series 5 ‐ six studies: Arici 2014; Frohlich 2011; Ilyas 2014; Lee 2012; Taylor 2013; Walker 2009; McGrath with unspecified design ‐ two studies: Peck 2009; Shippey 2013). The remaining VLS comparisons included X‐lite for only two studies (Bensghir 2010; Bensghir 2013) or individual studies; C‐MAC D‐blade (Serocki 2013); Airtraq (with video) (Hindman 2014; McElwain 2011); Truview EVO2 (Malik 2008); and CEL‐100 (Lin 2012). Most studies used a two‐arm design, comparing one type of VLS with a Macintosh blade. However, eight studies (Cavus 2011; Lee 2012; Malik 2008; Malik 2009b; McElwain 2011; Serocki 2010; Serocki 2013; Teoh 2010) conducted multi‐arm comparisons with two or three types of VLS versus a Macintosh blade. Gupta 2013 used a multi‐arm design but compared a C‐MAC blade and a Macintosh blade, both with and without the use of a stylet, to aid intubation. We have provided further details of included VLS designs in Appendix 5.

Four of the multi‐arm studies (Cavus 2011; Lee 2012; Serocki 2010; Serocki 2013) used a cross‐over design.

We included three studies that used a double‐lumen tracheal tube for intubation (Bensghir 2010; Cordovani 2013; Russell 2013). All remaining studies used a single‐lumen tube.

Forty‐eight studies recruited patients without predicted difficult airways (Andersen 2011; Aoi 2010; Arici 2014; Bensghir 2010; Bensghir 2013; Bilehjani 2009; Carassiti 2013; Choi 2011; Dashti 2014; Enomoto 2008; Griesdale 2012; Gupta 2013; Hirabayashi 2007a; Hirabayashi 2009; Hindman 2014; Hsu 2012; Ilyas 2014; Ithnin 2009; Kanchi 2011; Kill 2013; Kim 2013; Komatsu 2010; Lee 2012; Lee 2013; Lim 2005; Lin 2012; Maassen 2012; Malik 2008; Malik 2009a; Maruyama 2008a; Maruyama 2008b; McElwain 2011; Najafi 2014; Nishikawa 2009; Peck 2009; Pournajafian 2014; Robitaille 2008; Russell 2012; Russell 2013; Shippey 2013; Siddiqui 2009; Sun 2005; Takenaka 2011; Taylor 2013; Turkstra 2005; Walker 2009; Woo 2012; Xue 2007). Six studies recruited patients with a known or predicted difficult airway (Aziz 2012; Cordovani 2013; Jungbauer 2009; Malik 2009b; Serocki 2010; Serocki 2013); of these, two studies specified inclusion of patients with restricted cervical mobility (Aziz 2012; Serocki 2013). Two studies specified recruitment of participants both with and without predicted airway difficulties (Cavus 2011; Teoh 2010). Eight did not specify airway difficulties in the inclusion or exclusion criteria (Abdallah 2011; Ahmad 2013; Arima 2014; Frohlich 2011; Lee 2009; Sandhu 2014; Suzuki 2008; Yeatts 2013). For those participants recruited without predicted difficult airways, 15 studies used techniques (such as manual in‐line stabilization) to simulate a difficult airway (Aoi 2010; Enomoto 2008; Gupta 2013; Ilyas 2014; Komatsu 2010; Lim 2005; Malik 2008; Malik 2009a; Maruyama 2008a; McElwain 2011; Peck 2009; Robitaille 2008; Shippey 2013; Taylor 2013; Turkstra 2005).

Most studies specified the use of experienced anaesthetists to perform laryngoscopies (47 studies: Abdallah 2011; Ahmad 2013; Andersen 2011; Aoi 2010; Arici 2014; Arima 2014; Aziz 2012; Bensghir 2013; Bilehjani 2009; Carassiti 2013; Cavus 2011; Choi 2011; Cordovani 2013; Dashti 2014; Frohlich 2011; Gupta 2013; Hindman 2014; Hsu 2012; Ilyas 2014; Ithnin 2009; Jungbauer 2009; Kanchi 2011; Kim 2013; Komatsu 2010; Lee 2009; Lee 2012; Lee 2013; Lin 2012; Malik 2008; Malik 2009a; Malik 2009b; Maruyama 2008a; McElwain 2011; Najafi 2014; Nishikawa 2009; Pournajafian 2014; Robitaille 2008; Russell 2013; Serocki 2010; Serocki 2013; Siddiqui 2009; Sun 2005; Takenaka 2011; Teoh 2010; Turkstra 2005; Woo 2012; Xue 2007). Five studies used anaesthetists who were described as novices or who were trained with mannequins but had no patient experience (Griesdale 2012; Hirabayashi 2007a; Hirabayashi 2009; Taylor 2013; Walker 2009). Five studies used both novice and experienced anaesthetists (Bensghir 2010; Kill 2013; Lim 2005; Russell 2012; Yeatts 2013). Seven studies did not specify the experience of anaesthetists (Enomoto 2008; Maassen 2012; Maruyama 2008b; Peck 2009; Sandhu 2014; Shippey 2013; Suzuki 2008).

Ten study authors declared that they had received one or more of the intervention devices from the manufacturers for the purpose of the study (Abdallah 2011; Frohlich 2011; Komatsu 2010; Malik 2008; Malik 2009a; Malik 2009b; Maruyama 2008a; Maruyama 2008b; McElwain 2011; Serocki 2010). Five study authors declared that one of their study team had an interest in the company that manufactured the intervention devices (Storz manufacturers: Aziz 2012; Cavus 2011; Serocki 2013. Pentax AWS manufacturers: Enomoto 2008. McGrath manufacturers: Taylor 2013). Other studies reported department or government grant sources or did not report on this.

Excluded studies

We excluded 211 studies at the full text review stage; we have listed 70 of these under Characteristics of excluded studies. A large number of studies had used an Airtraq laryngoscope, which can be used with or without a video camera attachment. We excluded those studies in which it was unclear whether the laryngoscope had been used with the camera device. We also excluded studies of other devices in which it was not clear whether a video camera had been used. Thus we excluded from this review 30 studies comparing an Airtraq scope with a Macintosh blade (Ali 2012; Amor 2013; Chalkeidis 2010; Corso 2010; DiMarco 2011; Erden 2010; Ferrando 2011; Gaszynski 2009; Hayes 2011; Hayes 2012; Hirabayashi 2008a; Koh 2010; Maharaj 2006; Maharaj 2007; Maharaj 2008; Marco 2011; Ndoko 2008a; Park 2010; Ranieri 2012; Ranieri 2014; Sansone 2012; Saxena 2013; Stumpner 2011; Terradillos 2009; Tolon 2012; Trimmel 2011; Turkstra 2009a; Turkstra 2009b; Wang 2009; Wasem 2013) and eight studies that used other devices (Bullard, Truview, WuScope and Optiscope) (Araki 2002; Arora 2013; Barak 2007; Carlino 2009; Hastings 1995; Smith 1999; Watts 1997; Yang 2013). We excluded other studies because they lacked comparison with a Macintosh blade, used nasotracheal intubation, included patients not undergoing general anaesthesia, provided abstracts with insufficient details, did not report relevant outcomes or used the wrong study design. See Characteristics of excluded studies.

Ongoing studies

We identified seven studies through a clinical trials register search (NCT01914523; NCT01914601; NCT02088801; NCT02167477; NCT02292901; NCT02297113; NCT02305667). All studies were potentially eligible and were listed as at the stage of recruiting participants. See Characteristics of ongoing studies.

Studies awaiting classification

We identified a total of 38 studies that required further assessment for inclusion and have listed these under Characteristics of studies awaiting classification.

We identified eight studies through a clinical trials register search (NCT00178555; NCT00602979; NCT00664612; NCT01029756; NCT01114945; NCT01488695; NCT01516164; NCT02190201). All were potentially eligible and were listed as complete. However, study results were not published on the register, and we were unable to establish whether these studies had been published.

We found five additional studies that were reported in abstract form only, with insufficient detail, and we were unable to contact study authors (Ahmadi 2014; Eto 2014; Gharehbaghi 2012; Ishida 2011; Morello 2009). We will await the publication of full texts for these studies. We identified three studies that are awaiting translation before they can be assessed for inclusion (Kita 2014; Liu 2010; Wang 2008).

We identified 21 new studies for potential inclusion through screening of titles and abstracts during the search conducted in January 2016 (Ahmad 2015; Ahmadi 2015; Akbar 2015; Amini 2015; Bakshi 2015; Bhandari 2013; Bhat 2015; Colak 2015; Hamp 2015; Janz 2015; Kido 2015; Laosuwan 2015; Nakayama 2010; Pieters 2015; Postaci 2015; Rovsing 2010; Silverberg 2015; Wallace 2015; Yao 2015; Yousef 2012; Zhao 2014) and one study during the peer review process that we had excluded at an earlier stage (Cattano 2013). We will incorporate these studies into the formal review during the review update.

Risk of bias in included studies

We have included a summary of risk of bias assessments in Figure 2 and Figure 3.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

All studies were described as randomized, and 36 studies provided sufficient details on the method of randomization (Abdallah 2011; Andersen 2011; Arici 2014; Aziz 2012; Bensghir 2010; Bensghir 2013; Bilehjani 2009; Carassiti 2013; Cavus 2011; Cordovani 2013; Dashti 2014; Enomoto 2008; Griesdale 2012; Gupta 2013; Hirabayashi 2007a; Hindman 2014; Ilyas 2014; Ithnin 2009; Jungbauer 2009; Kanchi 2011; Komatsu 2010; Lin 2012; Malik 2008; Malik 2009a; Malik 2009b; McElwain 2011; Najafi 2014; Nishikawa 2009; Pournajafian 2014; Robitaille 2008; Russell 2012; Russell 2013; Siddiqui 2009; Sun 2005; Teoh 2010; Turkstra 2005). Other studies failed to provide details, or review authors determined that it was unclear if the method described would be adequate to reveal whether bias had been introduced. We judged only one study (Woo 2012) to be at high risk of selection bias regarding methods of randomization.

Only nine studies provided sufficient detail about methods used to conceal allocation from personnel (Abdallah 2011; Andersen 2011; Griesdale 2012; Hindman 2014; Komatsu 2010; Lin 2012; Pournajafian 2014; Teoh 2010; Walker 2009), and we were unable to make judgements other than 'unclear' for all remaining studies.

Blinding

We judged all studies to be at high risk of performance bias, as it was not possible to blind the anaesthetist from the type of scope used.

Similarly, it was not possible for outcome assessors of the primary outcomes of failed intubation and hypoxia to be blinded, and so again we judged all studies to be at high risk of detection bias. However, seven studies reported that researchers had made attempts to blind assessors to particular outcomes such as assessment of sore throat (Abdallah 2011; Kill 2013; Lee 2013; Lin 2012; Najafi 2014; Nishikawa 2009; Siddiqui 2009). In all, 15 studies described observers as 'independent' for some outcomes (Aoi 2010; Bensghir 2013; Enomoto 2008; Gupta 2013; Hirabayashi 2007a; Hsu 2012; Kanchi 2011; Kim 2013; Lee 2012; Lim 2005; Malik 2008; Malik 2009a; Malik 2009b; McElwain 2011; Teoh 2010); although this does not equate to being blinded to group allocation, these study authors made attempts to reduce detection bias in their studies.

Incomplete outcome data

Most studies reported no participant losses during the trial or only a small number of losses that were unlikely to affect results. We obtained insufficient data for some studies reported in abstract format only (Ahmad 2013; Sandhu 2014; Shippey 2013; Suzuki 2008), and so we were unable to make judgements of bias for these. We judged seven studies as having high risk of bias (Arima 2014; Cavus 2011; Ithnin 2009; Lee 2009; Maruyama 2008b; Woo 2012; Yeatts 2013) because they reported large numbers of losses, used exclusion criteria that introduced bias to the results or made changes to the protocol during the trial.

Selective reporting

We were able to source published protocols for eight of the studies and could adequately judge these as having low risk of bias because study authors had reported on all outcomes as stated in the protocol (Andersen 2011; Aziz 2012; Cordovani 2013; Hindman 2014; Hsu 2012; Kim 2013; Walker 2009; Yeatts 2013). We did not seek protocols for all other studies and therefore could not adequately judge the risk of bias for this domain.

Other potential sources of bias

Experience of intubator

We considered the experience of the intubator to be a potential source of bias in this review, in particular whether the intubator had equivalent experience with the VLS as with the Macintosh blade. It was often not possible to judge from the information presented by study authors whether bias had been introduced by intubators' experience.

Several studies adequately described anaesthetists as having equivalent experience with both devices, and we judged these to be at low risk of bias. Some studies described experience in terms of the number of intubations performed with each device.

If anaesthetists had carried out more than 20 intubations with the VLS device in the clinical setting, or had spent a considerable length of time using the device and at least this much time with the Macintosh device, we judged these studies to be at low risk of bias (Ahmad 2013; Andersen 2011; Bensghir 2013; Carassiti 2013; Choi 2011; Cordovani 2013; Gupta 2013; Hindman 2014; Hsu 2012; Kanchi 2011; Kim 2013; Lee 2012; Lee 2013; Lin 2012; Malik 2008; Malik 2009a; Malik 2009b; Nishikawa 2009; Pournajafian 2014; Robitaille 2008; Russell 2012; Serocki 2010; Serocki 2013; Siddiqui 2009; Sun 2005; Suzuki 2008;Teoh 2010; Turkstra 2005; Woo 2012). Two studies described personnel as experienced in the use of both devices; we assumed this to be equivalent experience and judged these studies as having low risk of bias (Aoi 2010; Xue 2007). Frohlich 2011 described operators as having used the devices on at least five occasions, but we believed this information was insufficient for us to judge whether bias was introduced here.

If however anaesthetists had carried out fewer than 20 intubations with the VLS device in the clinical setting, we assumed, unless otherwise stated, that the balance of experience would favour the Macintosh group and therefore judged these studies as having high risk of bias (Abdallah 2011; Taylor 2013).

Some studies used novice personnel only, and if it was implied that the level of experience between all personnel was equivalent, we judged these studies as having low risk of bias (Griesdale 2012; Hirabayashi 2007a). Hirabayashi 2009 described personnel as novices with less experience with the videolaryngoscope compared to the Macintosh; we judged this study to be at higher risk of bias.

Some studies used both novice and experienced personnel; if study authors did not adequately explain whether the balance of experience was equivalent between groups, we judged these studies to be at high risk of bias (Aziz 2012; Kill 2013; Lim 2005). Enomoto 2008 and Lee 2009 provided adequate descriptions of equivalent experience between novice and experienced personnel for review authors to judge these studies as having low risk of bias.

In two studies, anaesthetists had equivalent experience with the devices but not with use of a double‐lumen tube; therefore, we determined that a higher level of bias had been introduced (Bensghir 2010; Russell 2013). Similarly, in studies designed to assess devices at ground level and in the lateral position, operators had less experience with devices in the simulated position; it was not clear if this experience was equivalent between devices and whether bias had been introduced (Komatsu 2010 and Takenaka 2011, respectively).

Nineteen studies did not specify the experience of the anaesthetist at all, or described the anaesthetist as experienced but did not state whether the experience was equivalent in both devices; we were unable to judge the risk of bias for these (Arici 2014; Arima 2014; Bilehjani 2009; Cavus 2011; Dashti 2014; Ilyas 2014; Ithnin 2009; Jungbauer 2009; Maassen 2012; Maruyama 2008a; Maruyama 2008b; McElwain 2011; Najafi 2014; Peck 2009; Russell 2012; Sandhu 2014; Shippey 2013; Walker 2009; Yeatts 2013).

Baseline characteristics

Four abstracts did not present sufficient information on baseline characteristics, and we were unable to make a sufficient judgement of the risk of bias for this domain (Ahmad 2013; Peck 2009; Sandhu 2014; Suzuki 2008). One full study report provided no information on baseline characteristics, and we were unable to make a decision on bias for this (Robitaille 2008). Eight of the cross‐over design studies had presented baseline characteristics for the whole group of randomized patients and not by order of scope; therefore, it was not possible to judge bias for these studies (Enomoto 2008; Hindman 2014; Maassen 2012; Maruyama 2008a; Serocki 2010; Serocki 2013; Turkstra 2005; Walker 2009). Sixteen studies had presented baseline characteristics for which we noted some differences between study groups (Hsu 2012; Ilyas 2014; Ithnin 2009; Kim 2013; Komatsu 2010; Lee 2012; Malik 2008; Malik 2009a; McElwain 2011; Najafi 2014; Russell 2012; Siddiqui 2009; Takenaka 2011; Taylor 2013; Teoh 2010; Yeatts 2013). However, it was unclear how these differences may have affected the results. We noted significant differences in the numbers of participants reported throughout Woo 2012, leading to concerns about the randomization process and adequate reporting of baseline characteristics; therefore, we judged this study as having high risk of bias.

Funding

We judged studies reporting that they had received no funding or department funding only as having low risk of bias; when studies did not report any funding source, we assumed that no funding had been received and judged these studies to be at low risk of bias (in total, 48 studies: Ahmad 2013; Andersen 2011; Aoi 2010; Arici 2014; Arima 2014; Bensghir 2010; Bensghir 2013; Bilehjani 2009; Carassiti 2013; Choi 2011; Cordovani 2013; Dashti 2014; Griesdale 2012; Gupta 2013; Hindman 2014; Hirabayashi 2007a; Hirabayashi 2009; Hsu 2012; Ilyas 2014; Ithnin 2009; Jungbauer 2009; Kanchi 2011; Kim 2013; Lee 2009; Lee 2012; Lee 2013; Lim 2005; Lin 2012; Maassen 2012; Najafi 2014; Nishikawa 2009; Peck 2009; Pournajafian 2014; Robitaille 2008; Russell 2012; Russell 2013; Sandhu 2014; Shippey 2013; Siddiqui 2009; Sun 2005; Suzuki 2008; Takenaka 2011; Teoh 2010; Turkstra 2005; Walker 2009; Woo 2012; Xue 2007; Yeatts 2013).

Six study authors declared that one member of their study team had an interest in the manufacturing company of the intervention devices (Storz manufacturers: Aziz 2012; Cavus 2011; Serocki 2013. Pentax AWS manufacturers: Enomoto 2008. McGrath manufactures: Taylor 2013. GlideScope manufacturers: Kill 2013). We believe that this connection would present increased risk of bias towards the study results, and we therefore judged these studies to be at high risk of bias.