Despite increasing evidence supporting the use of the new generation of supraglottic devices,1 endotracheal intubation (ETI) remains a cornerstone of the practice of anaesthesia, but complications, with critical incident is estimated to be between 0.5 and 7%,2 continue to occur with social and economic costs. In a survey of the British National Health Service, airway and respiratory claims accounted for 12% of anaesthesia-related claims and more than half of deaths (53%); 8.6% of the total claims involved airway management, mainly due to failed or difficult ETI.3
In recent years, the interest in implementing strategies to predict difficult ETI has grown considerably. The Mallampati score4 was the first attempt to provide a scale for difficult ETI and its modern version is still used today. Thereafter an impressive body of work sought to introduce and validate different bedside indices of predicting difficult ETI.5 Notable examples are the thyromental distance,6 the upper-lip bite test7 and the multivariate el-Ganzouri index.8 However, the reliability of these indices has been questioned over the years and a recent systematic review failed to confirm a significant predictive role for any of them.9 Despite this many are still incorporated in current guidelines.10
We hypothesise that transnasal flexible endoscopic laryngoscopy (TFEL) might provide additional information to anaesthesiologists in their prediction of difficult ETI. TFEL is a minimally invasive procedure performed by ear, nose and throat (ENT) specialists.11 An endoscope is inserted through the nasal cavity into the nasopharynx to give direct visualisation of the oropharynx, hypopharynx and larynx. The examination is quick to perform, well tolerated with topical anaesthesia, and complications are extremely rare.12 TFEL has already been proposed as a possible tool for pre-operative evaluation of expected difficult ETI, but its role in specifically predicting difficult tracheal ETI has never been assessed.13 It was found to provide reliable and reproducible hypopharyngeal airway examination in obstructive sleep apnoea patients, comparable with the Cormack–Lehane and the modified Cormack–Lehane (MCL) scoring systems.14
Although it is clearly unadvisable to suggest that TFEL be performed on any patient scheduled to undergo general anaesthesia, it seems appropriate to take advantage of this examination when it is performed as routine pre-operative evaluation on other grounds. It could be a useful additional tool in planning airway management when the degree of tracheal intubation difficulty is unclear.
The current study aims at evaluating if TFEL, when performed by an ENT specialist, improves the predictability of difficult ETI obtained with routine bedside pre-operative evaluation by the anaesthesiologist.
Study design and population
The current prospective, single-centre, cohort study was approved by the Ethical Committee of Vita-Salute University, Milan, Italy on 9 December 2015 with the number CE/132/INT2015, and was carried out according to the Helsinki Principles. It was registered in the international registry www.clinicaltrials.gov with the number NCT02671877.
Between January and June 2016, following written consent, adults (>18 years) scheduled for elective ENT surgery under general anaesthesia were enrolled in this study. Exclusion criteria were suspected or confirmed neoplastic disease of the nose, oral cavity, pharynx or larynx; indwelling tracheostomy; bleeding lesions; elective fibreoptic awake intubation scheduled because of expected difficult laryngoscopy; previous radiation therapy to the head and neck region.
Bedside pre-operative evaluation by an anaesthesiologist
The sex, age, BMI and American Society of Anesthesiologists (ASA) status of all patients were recorded.
During routine bedside pre-operative evaluation, the modified Mallampati scale was recorded. This score is based on the finding that visualisation of the glottis is impaired when the base of the tongue is disproportionally large. Patients are evaluated while sitting upright with fully opened mouth and protruded tongue. The score includes the following four categories:
- (1) Class I: faucial pillars, soft palate and uvula are visible.
- (2) Class II: faucial pillars and soft palate may be seen but the uvula is masked by the base of the tongue.
- (3) Class III: only soft palate is visible.
- (4) Class IV: soft palate is not visible.
Moreover, the following data were recorded as dichotomous (N/Y) variables: prognathism, macroglossia, head extension impairment (cervical spine mobility impairment, defined as inability to touch the chest with the chin or to extend the neck freely), mouth opening impairment (interdental distance <40 mm with fully opened mouth), dentures, facial malformations, previous record of difficult intubation, reduced thyromental distance (distance from the lower border of the mandible to the thyroid notch with the neck fully extended <60 mm).
Pre-operative evaluation by an ear, nose and throat specialist
Pre-operative TFEL was carried out by two experienced ENT surgeons (SB, DDS). After topical anaesthesia with endonasal lidocaine spray, a video rhinolaryngoscope (model EN-VQ on Olympus Evis Exera II CV-180 Video System Center, Olympus Corporation, Tokyo, Japan) was passed through the more patent nasal cavity and the nasopharynx and stopped just below the velum to assess the morphology of the upper airway.
To standardise the reporting of the examinations, we defined the Endoscore, a five-grade scoring system categorised as follows:
- (1) Grade 1: complete view of the vocal folds, including the anterior commissure.
- (2) Grade 2a: incomplete view of the vocal folds, anterior commissure not visible.
- (3) Grade 2b: incomplete view of the vocal folds, anterior two thirds of the vocal folds not visible, only vocal process of the arytenoid cartilages can be visualised.
- (4) Grade 3: vocal folds not visible, only epiglottis can be visualised.
- (5) Grade 4: larynx not visible, only base of the tongue can be visualised.
The current scoring system was defined by analogy with the familiar MCL scale. Endoscore was recorded in the resting position (ES-RES), with protruded tongue (ES-TP) and hyperextended neck (ES-HYP).
During TFEL examiners also reported any individual characteristic that might contribute to difficult ETI, such as tongue base hypertrophy and epiglottis malformations such as an omega-shaped epiglottis.
Finally, patients were asked to perform a Muller manoeuvre to evaluate the collapse of soft palate, lateral pharyngeal wall and tongue base.15 This negative-pressure manoeuvre was performed in the resting position, both above (MUL-A) and below (MUL-B) the soft palate; and was intended to mimic the muscle relaxation obtained immediately before ETI. Results were reported on a five-point scale (0 to 4).
On the day of surgery, anaesthesia was induced with intravenous fentanyl (1 to 2 mg kg−1) and propofol (2 mg kg−1). Following muscle relaxation obtained with intravenous rocuronium bromide (0.6 mg kg−1), ETI was performed by one of three senior anaesthesiologists experienced in ENT surgery (MG, MRC, AR), who scored difficulty in laryngeal exposure with the MCL scale16 and difficulty in intubating with the intubation difficulty scale (IDS).17 A Macintosh laryngoscope blade was used. The intubating anaesthesiologist was blind with respect to the ENT endoscopic pre-operative evaluation.
The MCL scale evaluates the quality of direct laryngoscopy and includes the following categories:
- (1) Grade 1: full view of glottis.
- (2) Grade 2a: partial view of glottis.
- (3) Grade 2b: only posterior extremity of glottis seen or only arytenoid cartilages.
- (4) Grade 3: only epiglottis seen, none of glottis seen.
- (5) Grade 4: neither glottis nor epiglottis seen.
The IDS scores the intubation difficulty by summing up seven items:
- (1) N1: Number of supplementary attempts. An attempt is defined as an advance of the tube in the direction of the glottis during direct laryngoscopy or a blind intubation trial.
- (2) N2: Number of supplementary operators: the number of additional persons directly attempting (not assisting) intubation.
- (3) N3: Number of alternative techniques used.
- (4) N4: Glottic exposure as defined by the Cormack grade minus one. N4 = 0 for complete visualisation of the vocal cords, N4 = 1 for visualisation of the inferior portion of the glottis, N4 = 2 for visualisation of only the epiglottis, N4 = 3 for epiglottis not seen. Glottic exposure is evaluated during the first attempt by the first operator. If intubation is successful after blind nasotracheal intubation, N4 = 0. If the blind attempt(s) fail, glottic exposure is evaluated during the first subsequent alternative visualised laryngoscopic attempt.
- (5) N5: Lifting force applied during laryngoscopy. N5 = 0 if little effort is necessary, and N5 = 1 if increased lifting force is necessary. This notion is based on the operator's impression that an abnormal amount of force was used compared with routine practice.
- (6) N6: Need for applied external laryngeal pressure for optimising glottic exposure. N6 = 0 if no external pressure is applied. N6 = 1 if external laryngeal pressure is necessary.
- (7) N7: Position of the vocal cords. N7 = 0 if vocal cords are in abduction or not visualised, N7 = 1 if the vocal cords are in adduction, presenting an impediment to tube passage.
We planned to study 200 patients, the number who could be conceivably enrolled in a 6-month time span in our hospital.
Continuous variables are reported as mean ± SD. Discrete variables are reported as number (percentage). Odds ratios are reported as OR [95% confidence interval (CI)].
Ordinal logistic regression models with correlated data were built using the proportional odds (or cumulative logit) model and taking MCL and IDS as the dependent variables. Potentially predictive variables were tested separately for variable selection in univariate analysis. Variables exhibiting P value less than 0.25 were considered candidates for the multivariable models, and their interactions were checked. In building the final models, P value less than 0.05 was considered statistically significant. Assuming the observations did not correlate, multicollinearity of variables included in the final model was excluded with the variance inflation factor evaluation. To verify the proportional-odds (or parallel regression) assumption, an approximate likelihood-ratio test of whether coefficients are equal across categories.
Model comparison was performed with the Akaike Information Criterion (AIC) and the Likelihood-ratio test.
After recoding MCL and IDS into dichotomous variables, the relevant logistic regression models were built and the corresponding areas under the receiver operating characteristics curve (AUROC) were compared with a nonparametric approach.
No available former study addressed the issue of pre-operative endoscopic evaluation of tracheal intubation difficulty. Although no formal sample size evaluation was performed, we calculated the power of the comparison between the AUROCs of the two dichotomous MCL logistic regression models to provide some clues about the strength of our result. Calculations were performed with MedCalc Statistical Software version 19.0.7 (MedCalc Software bvba, Ostend, Belgium; https://www.medcalc.org; 2019). Taking an alpha error of 0.05 and a power of 80%, 157 cases (102 negative and 55 positive) were required to observe an AUROC difference between a 0.62 and 0.75 with a negative/positive ratio of 1.85 (65% controls and 35% cases). Data were analysed using dedicated Stata 13.0 software (Copyright 1985 to 2013 StataCorp. LP, StataCorp, College Station, Texas, USA).
We enrolled 200 patients and excluded 31; six because surgery was suspended and the remaining 25 because ETI was not carried out by one of the designated investigators due to organisational reasons, leaving 169 for analysis.
Mean age was 47.6 ± 13.7 years, 80 (47.6%) were males, their BMI was 24.7 ± 4.2 kg m−2, and their ASA status was I in 70 (41.4%), II in 84 (49.7%), III in 15 (8.9%) and there were none with ASA IV.
Bedside pre-operative evaluation by an anaesthesiologist revealed reduced interdental distance in four (2.4%) patients, prognathism in two (1.2%), macroglossia in two (1.2%), impaired head extension in 11 (6.5%), reduced thyromental distance in two (1.2%), dentures in 30 (17.8%) and no facial malformations.
Modified Mallampati scores are reported in Table 1.
TFEL was performed in all patients without complications. Tongue base hypertrophy was reported in 11 (6.5%), it was always symmetrical, and never suggested a neoplastic cause. Omega-shaped epiglottis was apparent in two (1.2%).
MUL-A was 0 in 34 (20.1%), 1 in 53 (31.4%), 2 in 34 (20.1%), 3 in 31 (18.3%) and 4 in 17 (10.1%). MUL-B was scored: 0 in 50 (29.6%), 1 in 58 (34.3%), 2 in 30 (17.8%), 3 in 25 (14.8%) and 4 in 6 (3.5%).
Endoscore scores for both resting (ES-RES), tongue protruded (ES-TP) and hyperextended neck (ES-HYP) are reported in Table 2.
ETI was successfully accomplished via direct laryngoscopy in all cases.
Modified Cormack–Lehane score analysis
The MCL score distribution is shown in Table 3.
When only bedside evaluation variables were taken into account, the MCL score was predicted by two variables, namely reduced head extension: OR 5.41 (95% CI 1.60 to 18.35, P = 0.007) and the Modified Mallampati Scale OR 2.47 (95% CI 1.39 to 4.38, P = 0.002, whole model P < 0.001, pseudo R2 = 0.069, proportionality of odds across response categories P = 0.712).
For TFEL variables, only ES-TP entered the multivariate predictive model for MCL: reduced head extension OR 9.07 (95% CI 2.54 to 32.32, P = 0.001), Modified Mallampati Scale OR 2.66 (95% CI 1.47 to 4.82, P = 0.001) and ES-TP OR 2.09 (95% CI 1.45 to 3.01, P < 0.001, whole model P < 0.001, pseudo R2 = 0.125, proportionality of odds across response categories P = 0.375). This predictive model had a higher goodness-of-fit than the bedside evaluation model not involving ES-TP (Likelihood-ratio test P < 0.001; AIC 275.60 vs. 290.65).
When the MCL was considered as a dichotomous variable (grade = 1 vs. grade >1), the ensuing logistic regression model considering only the bedside pre-operative evaluation exhibited an AUROC = 0.65 (95% CI 0.58 to 0.74) and when also ES-TP was considered, the model exhibited a significantly (P = 0.005) better AUROC = 0.75 (95% CI 0.67 to 0.83).
Intubation difficulty scale score analysis
IDS score distribution is shown in Table 4. When only bedside evaluation variables were taken into account, the IDS score was predicted by a single variable, the Modified Mallampati Scale OR 3.63 (95% CI 1.97 to 6.70), P = 0.001, whole model P < 0.001, pseudo R2 = 0.053, proportionality of odds across response categories (P = 0.900).
When TFEL variables were taken into account, only ES-TP entered the multivariate predictive model for IDS: Modified Mallampati Scale OR 3.83 (95% CI 2.07 to 7.10, P < 0.001) and ES-TP OR 1.46 (95% CI 1.03 to 2.08), P = 0.034, whole model P < 0.001, pseudo R2 = 0.068, proportionality of odds across response categories (P = 0.413). This predictive model had a higher goodness-of-fit than the bedside evaluation model not involving ES-TP (Likelihood-ratio test P = 0.032; AIC 335.55 vs. 338.16).
When the IDS was considered as a dichotomous variable (grade = 0 vs. grade >0), the ensuing logistic regression model considering only the bedside pre-operative evaluation exhibited an AUROC = 0.66 (95% CI 0.58 to 0.74) and when also ES-TP was considered, the model exhibited a slightly better (P = 0.049) AUROC = 0.70 (95% CI 0.61 to 0.80).
We suggest that pre-operative TFEL may improve the ability of pre-operative bedside evaluation to predict difficult ETI. We propose a scoring system (Endoscore) to standardise reporting of the predictive findings of the ENT evaluation. The Endoscore grading is defined by analogy with the familiar MCL scale, to facilitate understanding and cooperation between ENT and anaesthesia specialists.
Many predictors of difficult ETI have been studied, mainly evaluating anatomical characteristics at the bedside together with the patient history, and each predictor has been tested in its association with others in the attempt to improve sensitivity and specificity.4,7–9,18,19 Recently, a promising role for ultrasound-guided evaluation has been reported.20
A major difficulty in studying difficult ETI is that its incidence is low, ranging between 5 and 8%.9 This increases the number of false positives, as Yentis has pointed out,21 making the prediction less accurate. In addition, the definition of difficult laryngoscopy and ETI varies among authors, as some score the IDS, others the Cormack–Lehane or the MCL scale, while others simply note the number of failed attempts.9 What is worse is that the knowledge of this pivotal topic among professionals proved to be surprisingly low: only 25% of the anaesthesiologists interviewed during a major European anaesthesia congress in 2008 were able to correctly identify all four Cormack–Lehane grades.22
Although the anaesthesiologist's bedside evaluation is the cornerstone of any difficult ETI prediction, it is possible that hints from other specialists could add power to such an evaluation. A notable example of this is the ENT evaluation that is required in some patients.
A recent review suggested a role for TFEL as a predictor of difficult ETI, but its generalisability is doubtful, since it considered only ENT patients and predictability was limited to the detection of oropharyngeal, hypopharyngeal or laryngeal anomalies.13 Currently, there are little data about a possible role of ENT evaluation in predicting difficult ETI in a wider context.
Yamamoto et al.23 evaluated indirect oral laryngoscopy for the prediction of difficult direct laryngoscopy at the induction of general anaesthesia. They used the same grading of glottis exposure, the unmodified Cormack–Lehane scale, for both laryngoscopies and they observed improved sensitivity and positive predictive value when indirect oral laryngoscopy was added to bedside evaluation.
A 2011 study by Rosenblatt et al.24 on ENT surgery patients showed that in 26% pre-operative endoscopic evaluation changed the anaesthesia plan either from awake ETI to ETI after anaesthesia induction or vice versa. In this article, no specific assessment quantifying the risk for difficult ETI was proposed. Rather it measured the extent to which endoscopic evaluation influenced the subjective decision of the attending anaesthesiologist. Moreover, patients enrolled often presented with airway malignancy.
A recent study25 showed a good correlation between the number of visible subglottic tracheal rings on TFEL and ease of ETI. However, the ability to explore the subglottic region and to count the tracheal rings seems highly dependent on the quality of the examination, the expertise of the examiner, the compliance of the patient and even the dose of topical anaesthetic administered during the examination.
Torre et al.14 demonstrated a correlation between findings at endoscopic laryngoscopy, scored according to a self-defined 5-grade scale, and both the Cormack–Lehane and the MCL scales. The authors performed TFEL only in the resting position, and their newly defined scale was proposed as an alternative to the Cormack–Lehane scale to be adopted by both the ENT specialist and the anaesthesiologist.
In our study, Endoscore predicted MCL and IDS only when evaluated with tongue protrusion (ES-TP) and not in the resting position (ES-RES) or with hyperextended neck (ES-HYP). It is possible that tongue protrusion during TFEL better mimics the changes produced by direct laryngoscopy, in which the tongue is pulled distally and upwards. We also observed that impaired head extension predicts difficult visualisation of the larynx (MCL) but not difficult ETI (IDS). This seems reasonable, since reduced neck mobility may significantly impair glottis visualisation, but imperfect visualisation is not necessarily an impediment to ETI when performed by an experienced anaesthesiologist.
A first limitation of our study is that we considered in the variable selection of our predictive models only a part of the great number of potentially predictive variables reported in the current literature; we did not considered the el-Ganzouri index. This is the result of an arbitrary, albeit necessary, choice.
A second and more important limitation is that we used the same sample for both the definition and the validation of our models. As Yentis pointed out, this could artificially improve model fitness and lead to overoptimistic conclusions. The low incidence of difficult ETI precluded the splitting of our series into a training and a test set.
The incidence of difficult ETI itself may have been lowered with respect to general hospital patients by our choice of involving in the ETI manoeuvre only three senior anaesthesiologists highly experienced in ENT surgery. Although this choice improves standardisation in our results, it could obviously limit their generalisability.
Our results, do not apply to patients with neoplastic lesions of the upper airway. The low incidence of anatomical abnormalities observed on TFEL precludes any definite conclusion about their predictability of difficult ETI in our patients. Hence, our models should be regarded as useful in patients with ‘normal’ anatomy of the upper airway.
We assessed intubation difficulty on the basis of direct laryngoscopy, which warrants comparability with previous studies on this topic. Our results are not generalisable to tracheal intubation performed with videolaryngoscopy,26 an increasingly available tool in many centres providing better visualisation of the glottis.
Our study suggests how the anaesthesiologist might take best advantage from a pre-operative TFEL when it is performed as routine pre-operative evaluation on other grounds. Routine pre-operative TFEL is carried out in a number of situations, the most notable examples being ENT surgery and thyroid surgery, but also procedures that jeopardise the recurrent laryngeal nerve, such as thoracic malignancy or aortic arch aneurysm surgery.27 In these cases, data about the ease of glottis visualisation are almost invariably overlooked. Recording the ES-TP could provide the anaesthesiologist with cheap and effective information, and in situations with potential dificulties, TFEL could even be requested by the anaesthesiologist in the pre-operative evaluation workup as an additional tool in difficult airway management.
Acknowledgements relating to this article
Assistance with the study: thanks to Francesco Pilolli and Roberta Torricelli for their collaboration with the article.
Financial support and sponsorship: this study had institutional funding, Department of Anaesthesiology, Vita-Salute University, San Raffaele Hospital, Milan; Department of Otorhinolaryngology – Head and Neck Surgery, Vita-Salute University, San Raffaele Hospital, Milan.
Conflicts of interest: none.
1. Timmermann A, Bergner UA, Russo SG. Laryngeal mask airway indications: new frontiers for second-generation supraglottic airways. Curr Opin Anaesthesiol
2. Martin LD, Mhyre JM, Shanks AM, et al. 3,423 Emergency tracheal intubation at a university hospital: airway outcomes and complications. Anesthesiology
3. Cook TM, Scott S, Mihai R. Litigation related to airway and respiratory complications of anaesthesia: an analysis of claims against the NHS in England 1995–2007. Anaesthesia
4. Mallampati SR, Gatt SP, Guigino LD, et al. A clinical sign to predict difficult tracheal intubation: a prospective study. Can Anaesth Soc J
5. Shiga T, Wajima Z, Inoue T, et al. Predicting difficult intubation in apparently normal patients: a meta-analysis of bedside screening test performance. Anesthesiology
6. Ilper H, Grossbach A, Franz-Jager C, et al. Thyromental distance (‘Patil’) revisited: knowledge and performance of a basic airway screening tool among European anesthetists. Anaesthesist
7. Khan ZH, Kashfi A, Ebrahimkhani E. A comparison of the upper lip bite test (a simple new technique) with modified Mallampati classification in predicting difficulty in endotracheal intubation: a prospective blinded study. Anesth Analg
8. el-Ganzouri AR, McCarthy RJ, Tuman KJ, et al. Preoperative airway assessment: predictive value of a multivariate risk index. Anesth Analg
9. Vannucci A, Cavallone LF. Bedside predictors of difficult intubation: a systematic review. Minerva Anestesiol
10. Apfelbaum JL, Hagberg CA, Caplan RA, et al. Practice guidelines for management or the difficult airway: an updated report by the American Society of Anesthesiologists Task Force on Management of the Difficult Airway. Anesthesiology
11. Cohen SM, Pitman MJ, Noordzij JP, et al. Evaluation of dysphonic patients by general otolaryngologists. J Voice
12. Nacci A, Matteucci J, Romeo SO, et al. Complications with fiberoptic endoscopic evaluation of swallowing in 2,820 examinations. Folia Phoniatr Logop
13. O’Dell K. Predictors of difficult intubation and the otolaryngology perioperative consult. Anesthesiol Clin
14. Torre C, Zaghi S, Camacho M, et al. Hypopharyngeal evaluation in obstructive sleep apnea with awake flexible laryngoscopy: validation and updates to Cormack–Lehane and modified Cormack–Lehane scoring systems. Clin Otolaryngol
15. Terris DJ, Hanasono MM, Liu YC. Reliability of the Muller maneuver and its association with sleep-disordered breathing. Laryngoscope
16. Yentis SM, Lee DJH. Evaluation of an improved scoring system for the grading of direct laryngoscopy. Anaesthesia
17. Adnet F, Borron SW, Racine SX, et al. The intubation difficulty scale (IDS): proposal and evaluation of a new score characterizing the complexity of endotracheal intubation. Anesthesiology
18. Osman A, Sum KM. Role of upper airway ultrasound in airway management. J Intensive Care
19. Yao W, Wang B. Can tongue thickness measured by ultrasonography predict difficult tracheal intubation? Br J Anaesth
20. Fulkerson JS, Moore HM, Anderson TS, et al. Ultrasonography in the preoperative difficult airway assessment. J Clin Monit Comput
21. Yentis SM. Predicting difficult intubation – worthwhile exercise or pointless ritual? Anaesthesia
22. Krage R, van Rijn C, van Groeningen D, et al. Cormack–Lehane classification revisited. Br J Anaesth
23. Yamamoto K, Tsubokawa T, Shibata K, et al. Predicting difficult intubation with indirect laryngoscopy. Anesthesiology
24. Rosenblatt W, Ianus AI, Sukhupragarn W, et al. Preoperative endoscopic airway examination (PEAE) provides superior airway information and may reduce the use of unnecessary awake intubation. Anesth Analg
25. Guo Y, Feng Y, Liang H, et al. Role of flexible fiberoptic laryngoscopy in predicting difficult intubation. Minerva Anestesiol
26. Lu Y, Jiang H, Zhu YS. Airtraq laryngoscope versus conventional Macintosh laryngoscope: a systematic review and meta-analysis. Anaesthesia
27. Ishii K, Adachi H, Tsubaki K, et al. Evaluation of recurrent nerve paralysis due to thoracic aortic aneurysm and aneurysm repair. Laryngoscope