Difficult airway management algorithms: a directed review
Summary
The primary aim of this study was to identify, describe and compare the content of existing difficult airway management algorithms. Secondly, we aimed to describe the literature reporting the implementation of these algorithms. A directed search across three databases (MEDLINE, Embase and Scopus) was performed. All articles were screened for relevance to the research aims and according to pre-determined exclusion criteria. We identified 38 published airway management algorithms. Our results show that most facemask employ a four-step process as represented by a flow chart, with progression from tracheal intubation, facemask ventilation and supraglottic airway device use, to a rescue emergency surgical airway. The identified algorithms are overwhelmingly similar, yet many use differing terminology. The frequency of algorithm publication has increased recently, yet adherence and implementation outcome data remain limited. Our results highlight the lack of a single algorithm that is universally endorsed, recognised and applicable to all difficult airway management situations.
Introduction
Maintaining oxygenation is the most critical task performed by airway professionals 1. Difficult airway management requires a rapid and co-ordinated team response to prevent patient harm 2. This response can be challenging due to cognitive overload in a highly stressful environment 3. Difficult airway management algorithms, along with cognitive aids, may reduce cognitive overload and provide a framework for appropriate decision-making 4. For the purposes of this review, definitions for difficult airway and difficult airway algorithm were adopted from the American Society of Anesthesiologists (ASA) 5. Before the advent of published algorithms, difficult airway management depended solely on the experience of a skilled individual. However, systematically developed airway management recommendations allow for streamlined and standardised patient care, which aims to improve patient outcomes 1.
There are several nationally or internationally recognised airway societies, such as the ASA, the Difficult Airway Society UK (DAS) and the All India Difficult Airway Association (AIDAA). These societies allow for collaboration to establish guidelines and accompanying algorithms for difficult airway management. The first recognised international airway management guidelines were developed by the ASA in 1993 6, which were quickly followed by guidelines from Canada 7, Italy 8, Germany 9 and the UK 10.
In 2011, the 4th National Audit Project (NAP4) in the UK made a strong recommendation for the adoption of airway algorithms, proposing that the utilisation of these airway algorithms might lead to an improvement in patient outcomes 2. However, there remains a lack of published evidence as to whether the implementation of airway algorithms results in modification of clinical practice, or a change in patient outcomes. Moreover, there is a lack of consensus among practitioners as to which airway management approaches should be endorsed. Thus, the Project for the Universal Management of Airways (PUMA) was established in 2016 with the aim of promoting collaboration, garnering consensus and developing a united approach to airway management that is not context-specific 11. However, an accurate description of the current landscape of difficult airway management algorithms does not exist to further support the development of projects such as PUMA.
We, therefore, aimed to identify, describe and compare all difficult airway management algorithms published in the last 20 years. We also aimed to describe the literature examining the implementation of these algorithms.
Methods
This directed review adhered to the preferred reporting items for systematic reviews and meta-analyses recommendations (PRISMA) 12. A literature search for published algorithms of difficult airway management and articles reporting the implementation of algorithms was performed across three databases (MEDLINE, Embase and Scopus), seeking papers published between 6 December 1998 and 6 December 2018. Keywords searched were “difficult airway”, “front of neck access”, “CICO”, “emergency airway”, “complete ventilation failure”, “emergency cricothyrotomy” or “can't intubate can't ventilate”, in conjunction with “guideline”, “algorithm”, “cognitive aid”, “clinical consensus”, “consensus”, “checklist” or “implementation tool”. Furthermore, “airway management” was searched for as a title term rather than a keyword. Articles were limited to those published in English within the last 20 years which focused on adult populations.
All manuscripts underwent title and abstract screening for relevance to our primary aims. Relevant manuscripts proceeded to full-text screening. Authors then retrieved full-text manuscripts and assessed for eligibility. Both primary and secondary research articles were included, which consisted of published difficult airway management algorithms, surveys, cohort studies and randomised controlled trials, as well as reviews and editorials. Pre-defined exclusion criteria were non-full-text articles, non-English papers or manuscripts that were not directly related to airway management algorithms. Conference abstracts and commentary critiquing the specific content of algorithms were also excluded. Reference lists of included studies were then searched for further relevant papers. The quality of all included papers was assessed with the Medical Education Research Study Quality Instrument (MERSQI) 13 or the Critical Appraisal Skills Programme (CASP) 14 tools for quantitative and qualitative papers, respectively.
Discrete airway management algorithms were identified and data were tabulated on to a Microsoft Word 2016 for Mac (Microsoft Inc., Redmond, WA, USA) document. Data extracted included the year of publication, algorithm endorsement, intended patient population and included research evidence. Descriptive analysis was performed to identify common and divergent recommendations for clinical practice, such as airway techniques, patient position, equipment choice, pre-oxygenation, neuromuscular blocking drug use and human factors. Superseded algorithms were not included in the final analysis. For manuscripts reporting the implementation of difficult airway management algorithms, the primary study aims were identified for further descriptive analysis. These included algorithm adherence, implementation and knowledge retention assessments.
Results
After removing duplicates, 1066 studies underwent screening for relevance to the aims of the review. Of these, 189 progressed to full-text review. After excluding 104 studies, a further three were added from reference screening, yielding a total of 88 included papers (Fig. 1; Appendix S1 and S2). Analysis of the included papers using the MERSQI and CASP tools revealed the overall quality of evidence to be low 13, 14.
From 88 included articles, 38 difficult airway management algorithms were identified (see also Supporting Information, Appendix S1). Fourteen algorithms were generated by recognised airway societies. There has been a steady increase in the frequency of algorithm publications recently, with 58% of algorithms being published in the last 5 years (Fig. 2). Figure 3 demonstrates intended patient populations of all identified algorithms. Twenty-eight (74%) of the airway algorithms specified their role in difficult airway scenarios, with 10 (26%) purposed for all airway management. Of the identified algorithms, 35 (92%) use a flow chart comprised of rectangular boxes connected by arrows to represent a stepwise approach to airway management with progression through different procedures, which include tracheal intubation, face-mask ventilation and insertion of a supraglottic airway device (SAD). Out of the 38 included algorithms, 37 (97%) follow a similar flow chart format. Of the 14 society-produced algorithms, three employ a colour coding system with progression from green to yellow and eventually to red. Twenty-three (61%) algorithms state a specific stepwise order of airway techniques, while 13 (34%) allow for situational or clinician judgement, or level of experience, to alter the order (Table 1). Three society-produced algorithms make suggestions regarding device choice during specific steps of the algorithm.
Thirty-seven (97%) published algorithms recommend an invasive emergency airway rescue procedure to maintain oxygenation if required (Table 1). The terminology used differs across the 14 society-produced algorithms. Of these, seven (50%) explicitly suggest cricothyroidotomy 15-21, two (14%) suggest front-of-neck airway 22, 23, one suggests invasive airway access 5, one suggests surgical airway 24, one suggests a trans-laryngeal/trans-tracheal procedure 25 and two (14%) suggest multiple techniques 26, 27. There is also variation in the terminology used to identify a ‘can't intubate, can't oxygenate’ (CICO) situation. DAS 2015 uses the term CICO 18, whereas ASA 2013 uses the term ‘can't intubate, can't ventilate’ 5 and AIDAA 2016 uses terms ‘complete ventilation failure’ 15.
Ten (71%) out of the 14 society-produced algorithms make mention of human factors (Table 1). The DAS 2015 guidelines for the management of unanticipated difficult intubation in adults (DAS 2015) and DAS 2018 guidelines for the management of tracheal intubation in critically ill adults (DAS 2018) both clearly include the concept of ‘stop and think’ within their accompanying algorithms 18, 26, 28.
In terms of endorsement, only DAS 2018 guidelines detail which organisations and countries approved their recommendations 26. This guideline also specifically endorses the use of the Vortex cognitive aid in conjunction with the algorithm 26, 29. No other algorithms provide details on endorsement.
Out of the 14 society-produced algorithms, nine (64%) were formed with a combination of expert consensus and evidence-base 5, 15-20, 26, 27, two are solely based on expert consensus 22, 25, one is exclusively evidence-based 24 and two do not state the basis for their recommendations 21, 23. Of the algorithms that include evidence-based recommendations, four clarified the level of evidence for each recommendation made. One used the modified Delphi method to grade evidence 20, two used the Grading of Recommendations, Assessment, Development and Evaluations method 24, 27 and one categorised evidence by study design 23.
There were 55 articles that discussed or investigated the implementation, adherence or knowledge retention of airway algorithms (Appendix S2). Of these, 44 (80%) focused solely on society-produced algorithms 1, 3, 8, 22, 28, 30-69. Five focused on non-society-produced algorithms 70-74 and six included both in their analysis 75-80. Five of these articles also served as publications for a total of seven algorithms 70-74. Algorithms published by DAS were the most commonly included, with DAS 2015 (n = 16, 29%) 3, 22, 28, 32, 35, 37, 40-42, 46, 49, 51, 55, 62, 64, 81 and DAS 2004 (n = 13, 24%) 1, 3, 22, 36, 43, 44, 48, 50, 53, 60, 61, 79, 80 being the most referred to algorithm within the included articles. This was followed by algorithms published by the ASA, with ASA 1993 being referred to in 13 (24%) of the included papers 1, 3, 30, 33, 38, 39, 47, 58, 69, 75-77, 80. Moreover, ASA 2013 was referred to in a further 11 (20%) of the included papers 31, 40, 41, 44, 45, 55, 57, 62, 66, 68, 78. The Vortex was the most commonly referred to non-society algorithm (n = 4, 7%) 35, 63-65. The study methodology in the majority of articles was observational (n = 28, 51%) 30, 33-35, 38, 39, 44-48, 50, 51, 53, 57, 59, 64, 66, 67, 70-75, 77, 78, 80, with 18 (33%) being opinion articles or editorials 3, 22, 28, 31, 32, 36, 37, 40, 41, 49, 54-56, 58, 60, 63, 69, 79. Five papers were reviews of published airway algorithms 1, 43, 61, 62, 65 and four were interventional studies, including three randomised controlled trials 42, 52, 68, 76. Of the observational and interventional papers, the methodological quality of these articles primarily ranged from low to medium according to MERSQI and CASP (Appendix S2) 13, 14. Of the non-algorithm studies, algorithm adherence was the most common outcome measurement (n = 14) 30, 33, 38, 39, 45, 47, 50, 57, 59, 70, 74, 75, 77, 80. Other primary outcomes included the effects of simulated training on algorithm compliance (n = 4) 44, 48, 54, 68 and the effects of algorithms on clinical practice (n = 2) 64, 66.
The rates of adherence for society-produced algorithms, such as ASA 1993, varied among anaesthetists from 19% to 94% 33, 38. In 2006, a survey of 211 anaesthesia departments in Germany found 53% used a department-specific difficult airway management algorithm, compared with 28% using an ASA algorithm and 3% of departments utilised a combination of both. Furthermore, 20% used no difficult airway algorithm. 80. Variable adherence among practitioners was reported, as clinicians with greater clinical experience reported higher rates of adherence to airway management algorithms 47, 57.
Two included studies specifically investigated the change in clinical practice following algorithm implementation. One detailed the steady increase in the use of SADs in the period of 2001–2013, coinciding with the publication of ASA 2003, which recommended their use 66. Comparatively, another found that SAD usage in difficult airway situations between 2008 and 2012 remained steady at 12.4% 64. One paper demonstrated that 97% of anaesthetists had difficulties recalling ASA content 59. Variations in the perceived role of algorithms were evaluated in another study in 2017, with a variable understanding of the algorithm demonstrated by over half of the surveyed anaesthetists 46. Specifically, 14 out of 20 (70%) studies reported using algorithms as a cognitive aid, while algorithms as a ‘rule of law’ was only reported in 3 out of 20 (15%) 46. Four studies investigating the effect of simulation training on algorithm compliance were inconsistent. Two demonstrated that educational training improved algorithm compliance which was maintained for a year, as determined by a reduction in deviations from the ASA 2013 algorithm and The French Society of Anesthesia & Intensive Care Medicine 2008, respectively 44, 68. Another found that the improved compliance, measured by a reduction in deviations from DAS 2004, was maintained for 6 weeks but declined by 6 months 48 and a further study demonstrated that clinical practice in anaesthesia is unaffected by algorithmic training sessions 34.
Discussion
We identified 38 airway management algorithms, with 58% of algorithms being released in the last 5 years, and 14 published by recognised national or international airway societies. Overall, despite some differences in context, geography and target patient population, the identified algorithms are overwhelmingly similar. Many independently generated algorithms appear to be adaptations of society-produced ones. All analysed algorithms outlined a similar stepwise approach to airway management, suggesting the use of tracheal intubation, face-mask ventilation and SADs. Following these steps, all algorithms, except one, suggest ending with invasive airway access if necessary. It is important to note that a majority of the algorithms only address the situation in which a tracheal tube is the intended airway. However, in modern anaesthetic practice, most episodes of airway management begin with the intention of placing a SAD 82.
Despite all algorithms suggesting an emergency surgical airway in a CICO situation, there is a lack of consistent terminology for both terms. Multiple and ambiguous terminologies can potentially impede effective team performance 83. In a situation of high stress and cognitive overload, such as a difficult airway scenario, having inconsistent critical language could potentially add to the cognitive barrier in initiating necessary invasive airway access 83. During an airway crisis, it is paramount to have terminology that is clear and understood by all team members. Adoption of globally recognised consistent language would help in this space. Consistency in language for crisis management exists for cardiorespiratory arrest, with terms such as ‘shockable rhythm’, ‘CPR’ and ‘stand clear’ being universally recognised 83. Furthermore, algorithms such as ‘DRSABCD’ are well known and implemented in both pre-hospital and hospital settings 84. Currently, an attempt is being made to address this via a DAS-lead global initiative, looking at developing a lexicon for airway management that is concise, precise and universally consistent.
Human factors include leadership, role allocation and communication, while also referring to optimising the interaction between people and their environment 4. It has been widely suggested that considering such human factors during crisis management may be necessary to ensure the best outcomes for patients 32, 81, 85. However, DAS 2015 and DAS 2018 are the only algorithms that specifically mention ‘stop and think’ as an important step in airway management 18. In the often-cited case of Elaine Bromiley, a lack of situational awareness, teamwork and decision-making allowed a potentially salvageable airway emergency to result in a tragic outcome 86. It is concerning that, more than a decade later, most difficult airway algorithms do not specifically highlight a requirement for considering human factors, such as clear, concise and structured communication, which were deficient in the catastrophe that occurred. NAP4 suggested that human factors were contributory in 40% of adverse airway outcomes 2. It is, therefore, appropriate that new society-produced algorithms appear to now be incorporating the need for human factors during difficult airway management 28.
Traditionally, difficult airway trolleys have included a wide array of all airway equipment. There has been increased consensus that a more restricted and deliberate approach to the availability of equipment may be more beneficial in an airway emergency 4. The Australian and New Zealand College of Anaesthetists and DAS have both published recommendations for airway trolley content to focus attention on categories of equipment, rather than specific devices 4. This minimalistic approach is also evident across the society-produced published algorithms, with only four (29%) making reference to specific tracheal intubation techniques or devices 20, 21, 25, 26.
Currently, DAS are finalising guidelines for the management of the predicted difficult airway in adults with awake tracheal intubation 87. These guidelines may be helpful in further defining the role of awake videolaryngoscopy compared with flexible bronchoscopic intubation, and perhaps a future universal algorithm could allow for a choice between these two techniques.
Having multiple algorithms that are solely relevant to a single patient population could perhaps result in confusion as to which algorithm to employ in a difficult airway, especially in the context of high stress and cognitive overload 4. For example, if a trauma patient who was critically ill came into an emergency department, it may be difficult to decipher which currently published airway algorithm would be most suitable if required. Such a confusion may result in decreased or inappropriate algorithm adherence. Having one universal algorithm for could serve to further reduce confusion and redundancy.
Societies need to balance the risk of publishing algorithms with excessive frequency resulting in redundancy, with not updating them frequently enough to keep up with advances in evidence and device technology. With nearly half of medical guidelines becoming outdated by 6 years, four-yearly updates have been suggested 79. However, the ASA 5, 6 and DAS 18, 88 have updated their difficult airway management in adult algorithms in 10- and 11-year intervals, respectively. The relative infrequency of updates from the societies may have sparked the recent surge in independent algorithm publications. One consistent and universally accepted algorithm for airway management has the potential to remedy this problem.
Adherence to difficult airway algorithms varied from 19% to 94% 80. Even when algorithmic content is retained, there is not always adherence 34. Low adherence rates may be due to difficulty in understanding airway algorithms and the lack of a universal algorithm. However, we must acknowledge that available data on adherence were limited. Data establishing a direct causal relationship between the publication of algorithms and improved difficult airway practices and patient outcomes are not widely available in the literature. Therefore, robust, large-scale, multicentre studies are required before a conclusion can be drawn on the overall effect of implementing airway algorithms 89.
The primary limitation of this paper is the lack of quality published research into the use and content of the algorithms. Moreover, there is a lack of evidence on whether algorithms result in a change in clinical practice or subsequent patient outcomes. Of the included studies, most are observational, contain small sample sizes and have a brief follow-up 90. Furthermore, difficulties in conducting high-quality studies on airway emergencies have led to the majority of algorithms being generated primarily on expert consensus, rather than evidence 89. Moreover, of those that claim to be evidence-based, many do not categorise or detail the quality of the evidence sourced. In addition, by utilising a wide and comprehensive search strategy, both national society-approved and editorial or correspondence-based algorithms were included. Ultimately, the differences in the training, location, personnel and patient populations of each study made comparison and interpretation of results problematic.
Overall, our study demonstrates the abundance of similarly presented currently available airway management algorithms, several of which are published by recognised airway societies. They suggest a similar stepwise approach to airway management with emergency surgical airway rescue. The multitude of international algorithms produced in parallel might engender confusion, hamper widespread implementation and ultimately reduce adherence. Universal endorsement of a single airway algorithm that can be adequately adapted to a range of clinical scenarios could be most effective in further improving difficult airway management.
Acknowledgements
DB is on the advisory group to Project for Universal Management of Airways (PUMA). DE and EP contributed equally as co-first authors. No external funding or competing interests declared.
