On occasion I have noticed a face at a distance across a room and had the thought, “That person would be difficult to intubate.” This happens from time to time to other anesthesiologists I know. In the book Blink, Malcolm Gladwell uses the term “thin slicing” to describe unintended snap judgments that occur in a rapid, spontaneous manner.1 He explains that the subconscious mind can recognize patterns in a brief flash of visual information and will derive conclusions from the patterns as a matter of course. Someone with finely honed instincts could be successful in making accurate rapid judgments after a momentary glance. However, first impressions formed in the blink of an eye lack objective reflection and can be swayed by an individual’s recent experience and by multiple biases.2
Considering the potential pitfalls, it is unclear what weight should be given to an impression regarding potential airway difficulty made in a thin slice of time. The question is not purely academic. Thin slicing is part of human nature, and it can occur in professional settings and informal environments. Anesthesiologists may be influenced by thin slicing when they perform preoperative airway examinations. The issue is the extent to which rapid, subjective impressions based on patient appearance affect the decision about potential airway difficulty and whether the result is beneficial. Thin slicing could make a positive contribution to airway evaluation if it were founded on subconscious but insightful pattern recognition, but jumping to an unfounded conclusion could interfere with accurate airway assessment.
In a study reported in this month’s issue, Connor and Segal3 hypothesized that the availability of facial photographs would augment the ability of anesthesiologists to predict intubation difficulty over that achieved with Mallampati score (MP) and thyromental distance (TMD). Anesthesiologist subjects assessed airways based on TMD and MP scores for 80 patients who were known to be either easy (40 patients) or difficult to intubate (40 patients). The 160 anesthesiologists were then given frontal and lateral photographs of the patients along with TMD and MP and asked to repeat the predictions.
The group changed their assessment for some of the patients because of the photographs, most commonly deciding that an individual with a favorable MP score (< 3) or TMD (≥ 3 cm) would present greater difficulty than initially imagined. Occasionally, they downgraded a difficult rating for patients with TMD <3 cm to easy. The bedside airway evaluation tests have notoriously low sensitivity,4 but sensitivity increased when facial photographs were added. It is important to note that the positive predictive value for intubation difficulty improved significantly from 57% (95% confidence interval, 56% to 58%, with TMD and MP alone to 62% [confidence interval, 61%–62.4%]). Thus, seeing patients’ facial characteristics helped the anesthesiologists predict intubation difficulty in this study.
Determining how feature recognition improves prognostic accuracy would be useful in understanding how anesthesiologists make airway predictions and might lead to insights about airway evaluation. The subjects were not asked, so we do not have explicit information about why viewing images changed decisions. Connor and Segal suggest that anesthesiologists develop a subjective opinion about difficulty based on facial patterns. The impression may be subconscious and difficult to put in words. However, the investigators surmised some facial characteristics that anesthesiologists might link with difficulty by analyzing the images that affected airway assessment. Photographs of patients with a thick jaw relative to the height of the face (a high chin-to-nose/nose-to-chin ratio), those with a large chin-neck slope, or those with a large body mass index were likely to trigger a change in airway assessment.3 Thus, these facial features and/or a hefty appearance may have been part of the pattern the subjects perceived as related to intubation difficulty. In a previous study, the authors had used facial analysis software and logistic regression to show that the same facial features, brow-nose-chin ratio and the jaw-neck slope, indeed discriminated between patients who were easy and difficult to intubate.5 (See Fig. 3 and Fig. 5 in that article for illustrations). To my knowledge, neither measurement had been previously recognized to predict difficult intubation. Posterior mandibular depth has been tested as a marker of difficulty (with inconclusive results) but not as a ratio normalized to face height.6–8
Connor and Segal have entered a novel, previously untapped field by studying the role of facial analysis in airway assessment. Additional work will be needed to establish the validity of their findings and to address unanswered questions. The prevalence of intubation difficulty in the current study population was 50%, much higher than the usual rate of around 5% in most surgical populations.4,9 It would be worthwhile to test whether facial images add appreciably to airway prognostication when the prevalence of difficulty is lower. On another topic, various bedside tests, such as sternomental distance, interincisor gap, head extension, mandible length, and upper lip bite test are in use for predicting difficult intubation.9 Future investigators could conduct a more stringent test of whether examining facial images carries an advantage by augmenting the list of airway parameters beyond MP score and TMD. One might expect that the images would still be beneficial because the combination of MP score and TMD perform as well or better than other sets of airway parameters.4 A final suggestion for future research would be exploration of whether anesthesiologists consciously recognize specific facial features as markers for prognostication or whether they modify their opinion based on an overall pattern that is difficult to explain.
Rapid judgments can be influenced by irrelevant factors, so anesthesiologists who develop a thin slice impression of airway difficulty would do well to look for objective criteria supporting their assessment before proceeding. Thin slice conclusions are derived from one’s memory of how circumstances and events have been associated with outcomes in the past. If the thin slicing is based on only a coincidental association rather than a causal relationship, the conclusion is likely to be faulty. For example, suppose an anesthesiologist encounters in a short space of time 2 bald patients with large noses who are difficult to intubate. The next bald patient with a bulbous nose may make the anesthesiologist think of intubation difficulty, even though that combination of features is not a sure predictor of airway outcome. Daniel Kahneman’s book, Thinking Fast and Slow, reviews in-depth various biases that can impair the accuracy of thin slice judgments.2 Biases are not necessarily bad. Connor and Segal3 estimated that anesthesiologists were inclined to overcall intubation difficulty, preparing unnecessarily for difficult intubations 6.5 times for every unexpected instance of difficulty. In this situation, the benefit of avoiding airway surprises seems to be valued highly compared with the cost of unnecessary preparation.
Connor and Segal’s study has practical implications for anesthesiology. Based on the evidence to date, anesthesiologists should not be averse to using impressions based on patient facial appearance in planning airway management in conjunction with standard measurements. On average, thin slice airway assessments appear to improve the diagnostic accuracy for airway evaluation. Some clinicians had better results than others with thin slicing. If it were possible to measure individual talent in facial evaluation, anesthesiologists could calibrate personal skill and know how much weight to put on their subjective impressions of airway difficulty. Predictive ability had a tendency to improve with experience in this study, although more work is necessary on this point. If talent for thin slice airway evaluation increases with experience, it might be possible to design teaching exercises for facial evaluation that would improve performance.
The 2 Connor and Segal studies3,5 demonstrate that facial analysis can be used to identify new markers for airway difficulty. If facial analysis were used in much larger studies, it might be possible to identify additional, novel predictive measurements that would strengthen airway examination. Furthermore, computerized facial analysis could hold hope for establishing sensitive and specific computer-based algorithms for predicting airway difficulty. The interobserver reliability of airway assessment tests is only moderately good under optimal conditions, and frequent discrepancies are possible when conditions are worse.10,11 A computerized procedure might also improve the reproducibility of airway evaluation. Connor and colleagues have developed a phone application to process facial appearance and physical data for airway evaluation, and they reported preliminary work with the app at the 2013 International Anesthesia Research Society meeting.a Such endeavors could eventually support telemedicine practice of anesthesiology in which airway evaluation tests could be performed at a glance, not just across the room but across the world.
Name: Randolph H. Hastings, MD, PhD.
Contribution: This author helped write the manuscript.
Attestation: Randolph H. Hastings approved the final manuscript.
This manuscript was handled by: Franklin Dexter, MD, PhD.
a Connor CW, Tammineedi VS, Sparling J, Djang R, Oh E, Huh S, Segal S. Bedside recruiting, recording, and processing of data on ease or difficulty of intubation with a handheld app. Anesth Analg 2013;116:S388
1. Gladwell M Blink. 2005;Chapter 1 New York Back Bay Books, Little, Brown and Company:38–47
2. Kahneman D Thinking Fast and Slow. 2011 New York Farrar, Strauss and Giroux:79–88–44:234–88–44 Chapter 7 and Chapter 22:
3. Connor CW, Segal S. The importance of subjective facial appearance on the ability of anesthesiologists to predict difficult intubation. Anesth Analg. 2014;118:419–27
4. Shiga T, Wajima Z, Inoue T, Sakamoto A. Predicting difficult intubation in apparently normal patients: a meta-analysis of bedside screening test performance. Anesthesiology. 2005;103:429–37
5. Connor CW, Segal S. Accurate classification of difficult intubation by computerized facial analysis. Anesth Analg. 2011;112:84–93
6. Naguib M, Malabarey T, AlSatli RA, Al Damegh S, Samarkandi AH. Predictive models for difficult laryngoscopy and intubation. A clinical, radiologic and three-dimensional computer imaging study. Can J Anaesth. 1999;46:748–59
7. White A, Kander PL. Anatomical factors in difficult direct laryngoscopy. Br J Anaesth. 1975;47:468–74
8. Wilson ME, Spiegelhalter D, Robertson JA, Lesser P. Predicting difficult intubation. Br J Anaesth. 1988;61:211–6
9. Khan ZH, Mohammadi M, Rasouli MR, Farrokhnia F, Khan RH. The diagnostic value of the upper lip bite test combined with sternomental distance, thyromental distance, and interincisor distance for prediction of easy laryngoscopy and intubation: a prospective study. Anesth Analg. 2009;109:822–4
10. Karkouti K, Rose DK, Ferris LE, Wigglesworth DF, Meisami-Fard T, Lee H. Inter-observer reliability of ten tests used for predicting difficult tracheal intubation. Can J Anaesth. 1996;43:554–9
11. Rosenstock C, Gillesberg I, Gätke MR, Levin D, Kristensen MS, Rasmussen LS. Inter-observer agreement of tests used for prediction of difficult laryngoscopy/tracheal intubation. Acta Anaesthesiol Scand. 2005;49:1057–62