Can tongue thickness measured by ultrasonography predict difficult tracheal intubation?

Transcription

1 British Journal of Anaesthesia, 118 (4): (2017) doi: /bja/aex051 Respiration and the Airway Can tongue thickness measured by ultrasonography predict difficult tracheal intubation? W. Yao* and Bin Wang Department of Anesthesiology, The First Affiliated Hospital of Wannan Medical College, Wuhu, Anhui, China *Corresponding author. yaowdmd@163.com Abstract Background. Increased tongue thickness is likely to be associated with difficult airways. However, no methods to evaluate tongue thickness were available. Currently, tongue thickness can be measured by ultrasonography. The present study investigated the predictive value of tongue thickness to predict difficult tracheal intubation. Methods. Adult patients undergoing tracheal intubation and general anaesthesia were enrolled in the study. Tongue thickness was assessed using submental ultrasonography in the median sagittal plane before anaesthesia. Airway assessments were conducted. Ratios of tongue thickness to thyromental distance were calculated to investigate the potential predictive value of their combination. The primary outcome was difficult tracheal intubation. A multivariable logistic regression and receiver operating characteristic curve analysis were used. Results. In total, 2254 patients were analysed. One hundred and forty-two (6.3%) patients experienced difficult laryngoscopy, and 51 (2.3%) patients experienced difficult tracheal intubation. Increased tongue thickness (>6.1 cm) was an independent predictor for difficult tracheal intubation [sensitivity 0.75, 95% confidence interval (CI) ; specificity 0.72, 95% CI ]. An area under the curve of 0.78 (95% CI ) for predicting difficult tracheal intubation was calculated. Increased ratios of tongue thickness to thyromental distance (>0.87) presented a considerable area under the curve (0.86, 95% CI ), sensitivity (0.84, 95% CI ), and specificity (0.79, 95% CI ). Conclusions. Tongue thickness measured by ultrasonography and its ratio to thyromental distance present significant capacities to predict difficult tracheal intubation. Clinical trial registration. ChiCTR-RCS Key words: airway management; intubation, intratracheal; laryngoscopy; tongue; ultrasonography Difficult airway remains a potential risk for patients undergoing general anaesthesia and a challenge for anaesthetists. 12 There have been no effective methods to predict difficult airway accurately. 3 7 Seeking a more accurate method is still necessary for anaesthetists. Anaesthetists know that increased tongue thickness affects the performance of laryngoscopy and tracheal intubation and increases the risk of difficult airway. 8 9 However, there are no convenient and accurate recommended measurements to evaluate tongue thickness. Imaging techniques, such as threedimensional computed tomography, X-ray, and magnetic resonance imaging, display the anatomical features of the upper airways well and have been recommended for evaluation of difficult airway However, it is difficult to use these techniques in airway research and clinical applications because of their high cost and potential harm to the body. The low incidence of difficult airways and the large sample demands for research further exacerbate the difficulty in applicatio of these traditional imaging methods in clinical practice and research. The modified Mallampati test reflects tongue volume to some Editorial decision: February 5, 2017; Accepted: Month 0, 0000 VC The Author Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please journals.permissions@oup.com 601

2 602 Yoa and Wang Editor s key points Ultrasonography may be useful in predicting difficult tracheal intubation, by measuring the thickness of the tongue. Increased tongue thickness (>6.1 cm) indicates an increased risk of difficult tracheal intubation. extent, but its limited predictive power 4 and requirement that patients perform mandatory actions decrease the application value for predicting difficult airway, especially in unconscious patients. 13 Until now, the tongue thickness has not been well explored in predicting difficult airway. Ultrasonography may bring a new change in difficult airway research because it is non-invasive, convenient, and inexpensive. Previous clinical practice and observations have found that ultrasonography can image a patient s tongue and accurately measure tongue thickness in the sniffing position. This variable may reflect the internal characteristics of a patient s upper airway anatomy while requiring less patient collaboration than traditional techniques. However, whether tongue thickness assessed by ultrasonography is a useful predictor of difficult airway is not known. Therefore, this study was designed to evaluate the predictive value of accurately measured tongue thickness using ultrasonography for predicting difficult tracheal intubation and difficult laryngoscopy. Methods The Ethics Committee of Yijishan Hospital of Wannan Medical College approved the research protocol, and written informed consent forms were obtained from the patients before enrolment in the study. This prospective observational study was conducted from May 1, 2014 to December 31, 2014 and from January 1, 2016 to May 31, 2016 at our institution. Patients who were undergoing elective surgery with general anaesthesia were included. The following inclusion criteria were used: (i) 18- to 90-yr-old patients; (ii) ASA score of I, II, or III; (iii) undergoing general anaesthesia requiring tracheal intubation; (iv) no upper airway anatomical deformity, trauma, or tumour; (v) no identified difficult airway or difficult airway history that required the patient to be awake during tracheal intubation; and (vi) no subglottic airway stenosis. The following exclusion criteria were used: (i) patients who dropped out of the study; (ii) modified anaesthesia protocol or cancellation of tracheal intubation for a nondifficult-airway reason; and (iii) missing patient characteristics or study variables (e.g. age, sex, and modified Mallampati test score). Only the patients who were assigned to the operation rooms in a specified area were enrolled because of the large number of patients in our institution but the limit of two sonographers and two ultrasound machines. Airway assessments All enrolled patients underwent an ultrasonographic measurement of tongue thickness in a transitional waiting hall before transport to the operating room. A low-frequency convex array probe (S8; SonoScape Corp LP, Shenzhen, China) was used for ultrasonography. Two experienced sonographers performed the ultrasonographic measurements. The following ultrasonographic methods were used to measure tongue thickness. The patient assumed a supine position with the neck stretched. To obtain a uniform status of the tongue, all of the patients were asked to keep their mouth closed and to place the tongue tip slightly touching the incisors, with the tongue relaxed and with no phonation. The probe was placed under the chin in the median sagittal plane (Fig. 1A) and adjusted to obtain the entire tongue outline clearly on the screen. This image was frozen. The maximal vertical dimension from the tongue surface to the submental skin was measured and defined as the tongue thickness (Fig. 1B). A special team performed preoperative classical difficult airway assessments, including the modified Mallampati test, the thyromental distance, and the inter-incisor distance for all patients, early on the preoperative day. The modified Mallampati test was measured as described previously The view was graded as follows: Grade 1, visible soft palate, fauces, uvula, and pillars; Grade 2, visible soft palate, fauces, and uvula; Grade 3, visible soft palate and base of the uvula; and Grade 4, soft palate not visible. Grades 3 and 4 were deemed predictors of difficult airway. Thyromental distance was measured as the straight distance between the upper border of the thyroid cartilage incisure and the bony point of the mentum with the head and neck stretched. Inter-incisor distance was obtained when the patient maximally opened his or her mouth. A higher modified Mallampati test score, thicker tongue, and lower thyromental distance suggested a greater risk of difficult airway; therefore, the ratios of the modified Mallampati test to thyromental distance (in centimetres) and tongue thickness (in centimetres) to thyromental distance (in centimetres) were calculated. Moreover, in order to investigate the relationship between tongue thickness and other predictors, correlation coefficients were calculated. Induction of general anaesthesia Airway evaluations were completed, and the patients were transported to the operating room. The patients were placed in a supine position, and standard monitors were applied. Oxygen for inhalation was administered through a facemask. General anaesthesia was induced using midazolam (0.05 mg kg 1 ), fentanyl (0.004 mg kg 1 ), propofol (1 2 mg kg 1 ), and vecuronium (0.1 mg kg 1 ). Mask ventilation was administered when the patients stopped communication. Patients were placed in the sniffing position after 3 min of ventilation, and one of the attending anaesthetists, who had 5 yr of experience, performed the laryngoscopies. A Macintosh number 3 or number 4 laryngoscope blade was used. External laryngeal pressure or manipulation was permitted to improve the view of the glottis. The tracheal tube size was selected based on the anaesthetist s clinical experience. Study end points The primary end point was difficult tracheal intubation, which was evaluated based on the intubation result after every intubation. Difficult tracheal intubation was defined as an insertion of the tracheal tube using conventional laryngoscopy that required more than two attempts, lasted >10 min, or required an alternative technique. 19 No more than four intubation attempts via the application of a Macintosh laryngoscope blade were permitted in our routine clinical anaesthesia procedure to ensure patient safety, and the operating time for each attempt was no longer than 1 min. Mask ventilation was used for at least

3 Tongue thickness predicts difficult intubation 603 Fig 1 (A) The curved probe was placed under the chin in the median sagittal plane. (B) The probe was adjusted to obtain the entire tongue outline clearly in the screen. Tongue thickness was measured as the maximal vertical dimension from the tongue surface to the submental skin. The tongue thickness of one patient was 5.6 cm. (C) When performing laryngoscopy, the linear array probe was placed in the median sagittal plane above the hyoid. (D) The compressed tongue thickness was measured as the vertical distance from the blade to the skin above the hyoid. The tongue thickness was compressed to 2.7 cm. 1 min before the next intubation attempt to ensure that the oxygen saturation was 98%. The secondary end point was difficult laryngoscopy. The glottic view was scored using Cormack Lehane grades during each laryngoscopy. 20 Grade 3 or 4 was deemed a difficult laryngoscopy. Difficult mask ventilation events (which were defined as the inability to establish facemask ventilation despite multiple airway adjuvants and two-handed mask ventilation) 21 and failure to intubate were documented. The anaesthetist immediately sought help from the difficult airway management team if a difficult or emergency airway was encountered. The following strategies for difficult airways were provided (there was no limit on the priority order): (i) fibrebronchoscopy-guided intubation; (ii) videolaryngoscopy (without guide slot, VL300L; UE Medical Corp., China)-assisted intubation; (iii) intubation or ventilation with laryngeal mask airways; (iv) lighted stylets or light wands; and (v) percutaneous thyrocricocentesis to establish a surgical airway rapidly. 7 The person who performed the ultrasonography was blinded to the results of the preoperative classical difficult airway assessments and tracheal intubation. The person who performed the classical difficult airway assessments was also blinded to the results of the ultrasonography and tracheal intubation. The person who performed the tracheal intubation was blinded to the results of the classical difficult airway assessments and ultrasonography. Reliability test To investigate the variations between the two sonographers, reliability tests between the two sonographers were performed. Twenty volunteers were examined by the two sonographers independently. The volunteers were required not to relay any further information to the other sonographer. In addition, to investigate the tongue thickness relationship between the natural situation in the sniffing position and the compressed situation under laryngoscopy, the tongue thicknesses of 20 patients were also examined by one sonographer when the patients were in the sniffing position and when they were undergoing laryngoscopy. A linear array probe was placed in the median sagittal plane above the hyoid to capture the image. The vertical distance from the blade to the skin overlying the hyoid bone was measured and deemed the compressed tongue thickness (Fig. 1C and D; the selection of this measuring line was in consideration that the line is easily located, and we speculated that it might predominantly influence the laryngoscopy).

4 604 Yoa and Wang Statistical analysis Statistical analysis was performed using SPSS software version 16.0 (SPSS, Chicago, IL, USA) and MedCalc version 12.7 (MedCalc Software, Maria-kerke, Belgium). Continuous variables and data with a normal distribution are expressed as means (SD), and categorical variable data are presented as frequency distributions. Ratios are expressed as the values and their 95% confidence intervals (CIs). Univariate comparisons between patients with and without difficult laryngoscopy or patients with and without difficult tracheal intubation were performed using Mann Whitney U, Student s t, v 2 or Fisher s exact tests. Variables that were significant in univariate logistic regression were included in a multivariable logistic regression. The remaining variables (screened according to P<0.05) were deemed independent predictors. Receiver operating characteristic curve analysis and area under the curve (AUC) were used to examine the discrimination capacity of variables to predict difficult tracheal intubation or difficult laryngoscopy. The AUC of the multivariable logistic regression model was calculated. The Youden index 22 (the maximal difference between sensitivity and 1 specificity, associated with the optimal cut-point) was used to determine the criteria for independent predictors to predict difficult tracheal intubation or difficult laryngoscopy. Odds ratios, sensitivities, specificities, positive predictive values, negative predictive values, and their 95% CIs for independent predictors were calculated. Krippendorff s a value was used to evaluate the reliability between the two sonographers measurements, and Student s paired t-test was used to analyse their difference. Correlations between continuous variable data or ordinal category data were analysed with a Pearson correlation analysis or a Spearman correlation analysis as appropriate. All comparisons were twosided tests, and P<0.05 was considered statistically significant. In order to reduce the impact of logistic regression model overfitting based on a 3% incidence of primary end-point events of our previous observations in the same institution, 23 we enlarged the sample to >2000 subjects, and made sure that the ratio of the number of patients who suffered end points to the number of potential predictors was not < Results In total, 2480 patients were enrolled in the study during the study period. Two hundred and twenty-six patients were excluded, and 2254 patients were included for analyses. Table 1 shows the basic characteristics. Figure 2 shows the outcomes of the patients. There were 142 (6.3%) patients who experienced difficult laryngoscopy, and 51 (2.3% of all patients, compared with that of the previous sample-size-estimation study, no significant difference, P¼0.11) of these patients also experienced difficult tracheal intubation. Forty-nine patients were successfully intubated (10 patients via four attempts, 32 patients via a videolaryngoscope, five patients via fibrebronchoscopy, and two patients via a lighted stylet), and two patients experienced intubation failure. Two patients with difficult mask ventilation experienced difficult laryngoscopy and difficult tracheal intubation but did not experience ventilation or intubation failure. The Spearman correlation coefficient of tongue thickness to other predictors was 0.02 (95% CI ) for the inter-incisor distance, 0.07 (95% CI ) for the thyromental distance, 0.14 (95% CI ) for the modified Mallampati test, 0.31 (95% CI ) for the sex, 0.16 (95% CI ) for the age, and 0.38 (95% CI ) for the BMI, respectively. Prediction of difficult intubation and difficult laryngoscopy The AUCs of the predictors (including the ratio of tongue thickness to thyromental distance and the ratio of modified Mallampati test to thyromental distance, excluding sex for its dichotomy) for difficult intubation and for difficult laryngoscopy are presented in Table 1. After univariate analysis and multivariate logistic regression, we identified four independent predictors of difficult intubation (advanced age, reduced thyromental distance, thicker tongue, and reduced inter-incisor Table 1 Univariate comparisons of preoperative characteristics between patients with and without difficult laryngoscopy or patients with and without difficult tracheal intubation. *All patient characteristics were compared using Mann Whitney U-tests for continuous variables and v 2 or Fisher s exact tests for categorical variables. Data are shown as the means (SD) or counts. AUC, area under the receiver operating characteristic curve; CI, confidence interval Variable Difficult laryngoscopy Difficult tracheal intubation Yes (n¼142) No (n¼2112) P-value* AUC (95% CI) Yes (n¼51) No (n¼2203) P-value* AUC (95% CI) Sex (male/female, n) 102/40 957/1155 < / /1178 <0.001 Age [yr; median (range)] 61 (28 82) 49 (18 83) < ( ) 61 (30 80) 50 (18 83) < ( ) BMI (kg m 2 ) 23.3 (3.3) 22.8 (3.5) ( ) 23.7 (3.7) 22.8 (3.5) ( ) Modified Mallampati test 98/44 911/1201 < ( ) 37/14 972/1231 < ( ) >2 (yes/no; n) Inter-incisor distance (cm) 3.5 (0.6) 4.1 (0.6) < ( ) 3.2 (0.5) 4.1 (0.6) < ( ) Thyromental distance (cm) 7.0 (0.8) 7.5 (0.8) < ( ) 6.8 (0.6) 7.5 (0.8) < ( ) Tongue thickness (cm) 6.2 (0.5) 5.8 (0.5) < ( ) 6.4 (0.4) 5.9 (0.5) < ( ) Ratio of modified 0.45 (0.15) 0.32 (0.14) < ( ) 0.47 (0.15) 0.33 (0.15) < ( ) Mallampati test to thyromental distance (cm) Ratio of tongue thickness (cm) to thyromental distance (cm) 0.90 (0.13) 0.79 (0.11) < ( ) 0.94 (0.10) 0.80 (0.11) < ( )

5 Tongue thickness predicts difficult intubation 605 Patients meeting the inclusion criteria (n = 2654) Patients declining to participate (n = 174) Patients enrolled (n = 2480) Patients successfully included for analysis (n = 2254) Patients excluded Missing values (n = 195) Laryngeal mask use (n = 16) Operation cancellation (n = 15) Difficult intubation (n = 51) Difficult laryngoscopy (n = 142) Cormack Lehane level level 3: n = 135; level 4: n =7 Outcome of intubation Success with 4 attempts (n = 10) Success with video laryngoscope (n = 32) Success with fibrobronchoscopy (n = 5) Success with lighted stylet (n = 2) Intubation failure in 2 cases: (LMA use (n = 1); awake intubation (n = 1)) Fig 2 Study flow chart and patient outcomes. distance), and six independent predictors of difficult laryngoscopy (male sex, advanced age, reduced thyromental distance, thicker tongue, increased modified Mallampati test, and reduced inter-incisor distance). No correlation coefficient >0.4 was observed between any two of these predictors. Receiver operating characteristic curve analysis of this logistic regression model demonstrated an AUC of 0.95 (95% CI ) for difficult intubation and an AUC of 0.87 (95% CI ) for difficult laryngoscopy. Figure 3 shows the receiver operating characteristic curves of tongue thickness, modified Mallampati test, thyromental distance, ratio of tongue thickness to thyromental distance, and ratio of modified Mallampati test to thyromental distance. Tables 2 and 3 show the optimal criteria as determined by the Youden index of predictors to predict difficult intubation and difficult laryngoscopy, and show the odds ratios, sensitivities, specificities, positive predictive values, negative predictive values, and their 95% CIs. Results of the reliability test The tongue thickness of 20 volunteers (11 males, nine females) was 60 (SD 5) mm for one of the two sonographers measurements and 61 (5) mm for the other sonographer s measurement. Between the two sonographers measurements, Krippendorff s a value was 0.95 (95% CI ). There was no significant difference between the two sonographers with regard to their measurements (P¼0.149). Twenty patients (nine males, 11 females) were examined by one sonographer when the patients were in the sniffing position and when they were undergoing laryngoscopy. The tongue thickness in the natural situation was 60 (6) mm, whereas in the compressed situation it was 27 (4) mm. A Pearson correlation coefficient of 0.72 (95% CI ) was observed. Discussion This study demonstrates that tongue thickness, as measured accurately using ultrasonography, is a useful predictor for predicting difficult airway. Multivariable logistic regression demonstrates that tongue thickness may be an independent predictor of both difficult laryngoscopy and difficult tracheal intubation, and its predictive power is similar to the modified Mallampati test and thyromental distance. It seems generally accepted that increased tongue thickness will influence the modified Mallampati test, but the minor correlation (r¼0.14) between tongue thickness and the modified Mallampati test does not fully support this opinion. The thickness of the tongue is not stable, and it is easy to generate variation between measurements. However, the reliability test shows that a Krippendorff s a of 0.95 demonstrates good stability between examiners. Obviously, the tongue can be compressed by a laryngoscope. But the present data show that the retained tongue thickness under laryngoscopy has a considerable correlation (r¼0.72) with the tongue thickness in the natural situation. Increased tongue size or supra-epiglottic mass affects the performance of tracheal intubation. 89 Our study confirmed the association between difficult airway and tongue thickness using ultrasonographic measurements. Ezri and colleagues 25 demonstrated that the soft tissue thickness of the anterior neck as quantified by ultrasound was an indicator of difficult laryngoscopy in a select group of obese

6 606 Yoa and Wang A 100 B Sensitivity Sensitivity specificity specificity TT,AUC: 0.69 ( ) TT,AUC: 0.78 ( ) TMD,AUC: 0.68 ( ) MMT,AUC: 0.68 ( ) Ratio of MMT to TMD, AUC: 0.72 ( ) Ratio of TT to TMD, AUC: 0.75 ( ) TMD, AUC: 0.75 ( ) MMT, AUC: 0.70 ( ) Ratio of MMT to TMD, AUC: 0.76 ( ) Ratio of TT to TMD, AUC: 0.86 ( ) Fig 3 Receiver operating characteristic curve analysis of airway assessment tests and their areas under curve (AUCs; value and its 95% confidence interval) for predicting difficult laryngoscopy (A) and difficult tracheal intubation (B). MMT, modified Mallampati test; TMD, thyromental distance; TT, tongue thickness. Table 2 Variable values to predict difficult tracheal intubation (n¼2254). *Only independent variables, ratio of tongue thickness (in centimetres) to thyromental distance (in centimetres), and ratio of modified Mallampati test to thyromental distance (in centimetres) were included; the optimal criteria of the variables were determined by the Youden index. CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value. Variables* Odds ratio (95% CI) Sensitivity (95% CI) Specificity (95% CI) PPV (95% CI) NPV (95% CI) Age >52 yr 4.1 ( ) 0.75 ( ) 0.59 ( ) 0.04 ( ) 0.97 ( ) Inter-incisor distance 3.8 cm 22 (8.5 54) 0.90 ( ) 0.70 ( ) 0.07 ( ) 1.00 ( ) Thyromental distance 7.0 cm 6.4 (3.3 13) 0.78 ( ) 0.64 ( ) 0.05 ( ) 0.99 ( ) Tongue thickness >6.1 cm 7.7 (3.9 16) 0.75 ( ) 0.72 ( ) 0.06 ( ) 0.99 ( ) Ratio of modified Mallampati 5.3 ( ) 0.65 ( ) 0.76 ( ) 0.06 ( ) 0.99 ( ) test to thyromental distance (cm) >0.42 Ratio of tongue thickness (cm) to thyromental distance (cm) > (9.6 44) 0.84 ( ) 0.79 ( ) 0.09 ( ) 1.00 ( ) patients. However, the results of another study contradict these results. 26 Srikar and colleagues 27 examined anterior neck soft tissue thickness at five different levels in a short axis and found that ultrasonographic measurements of the anterior neck soft tissue thickness at the level of the hyoid bone and thyrohyoid membrane could be used to distinguish difficult laryngoscopy. In the present study, patients were examined by mid-sagittal plane submandibular sonography, which allowed the detection of the entire tongue outline and identification of the thickest point. Wojtczak and colleagues 14 investigated tongue volume in obese patients found that it did not differ significantly between groups with and without difficult laryngoscopy. These results seem contradictory to our study. However, if we consider that a lower thyromental distance or hyomental distance is associated with difficult laryngoscopy, the contradiction may be explained as follows: an increased tongue thickness and a shorter tongue, but not necessarily exhibiting an increased tongue volume, indicates a greater possibility of difficult laryngoscopy or difficult tracheal intubation. The incidence of difficult airway increased greatly when thicker tongues and shorter thyromental distances were both encountered. Previous studies also confirmed that the

7 Tongue thickness predicts difficult intubation 607 Table 3 Variable values for the prediction of difficult laryngoscopy (n¼2254). *Only independent variables, ratio of tongue thickness (in centimetres) to thyromental distance (in centimetres), and ratio of modified Mallampati test to thyromental distance (in centimetres) were included; the optimal criteria of variables were determined by the Youden index. CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value Variables* Odds ratio (95% CI) Sensitivity (95% CI) Specificity (95% CI) PPV (95% CI) NPV (95% CI) Male sex 3.1 ( ) 0.72 ( ) 0.55 ( ) 0.10 ( ) 0.97 ( ) Age >52 yr 3.4 ( ) 0.70 ( ) 0.60 ( ) 0.10 ( ) 0.97 ( ) Modified Mallampati test >2 2.9 ( ) 0.69 ( ) 0.57 ( ) 0.10 ( ) 0.97 ( ) grade Inter-incisor distance 4.0 cm 7.3 (4.5 12) 0.86 ( ) 0.54 ( ) 0.11 ( ) 0.98 ( ) Thyromental distance 7.0 cm 3.4 ( ) 0.65 ( ) 0.65 ( ) 0.11 ( ) 0.97 ( ) Tongue thickness >6.0 cm 3.3 ( ) 0.63 ( ) 0.66 ( ) 0.11 ( ) 0.96 ( ) Ratio of modified Mallampati 4.2 ( ) 0.63 ( ) 0.73 ( ) 0.13 ( ) 0.97 ( ) test to thyromental distance (cm) >0.42 Ratio of tongue thickness (cm) to thyromental distance (cm) > (5.0 10) 0.63 ( ) 0.81 ( ) 0.18 ( ) 0.97 ( ) combination of the modified Mallampati test and thyromental distance for predicting difficult airway is more effective than that of single application Therefore, we calculated ratios of the modified Mallampati test to thyromental distance (in centimetres) and ratios of tongue thickness (in centimetres) to thyromental distance (in centimetres) and used them to predict difficult laryngoscopy or difficult tracheal intubation. Previous studies using a combination of the modified Mallampati test and thyromental distance often used them in the logical OR manner. This method may increase the sensitivity, but it decreases the specificity. In contrast, the logical AND conjunction of the two variables may decrease the sensitivity, but it increases the specificity. The present data demonstrated that the ratios of the two variables may be a balanced selection. The AUC of the ratio of tongue thickness to thyromental distance for predicting difficult tracheal intubation or predicting difficult laryngoscopy was remarkable when comparing with that of tongue thickness, thyromental distance, and the ratio of the modified Mallampati test to thyromental distance. Previous studies and our research demonstrate that many factors affect the formation of a difficult airway. It is difficult to obtain a high predictive value using any single parameter. A combination of variables may improve the capacity to predict difficult airway significantly. However, too many variables (i.e. logical AND) would miss many difficult airways and might significantly reduce the sensitivity, while increasing the specificity Logistic regression analysis balances this contradiction to some extent and reserves the most valuable predictors. Our results demonstrate that the components of the multivariable logistic regression model for predicting difficult tracheal intubation are reduced thyromental distance, thicker tongue, advanced age, and reduced inter-incisor distance, with an AUC of 0.95 (95% CI ). Interestingly, the modified Mallampati test is not included in this model. This might reveal that compared with the modified Mallampati test, a thicker tongue has a more direct relationship with difficult intubation. In the present data, the BMI was also a weak predictor, which is consistent with a previous study. 6 However, we observed a significant correlation (r¼0.38, 95% CI ) between a thicker tongue and high BMI. This may be interpreted as that high BMI can cause a thicker tongue and impair tracheal intubation. Patients who have a high BMI but not a thicker tongue are likely to be intubated easily. That is to say, a thicker tongue might have a more direct relationship with difficult intubation than does a high BMI. This study also has limitations, such as the limited number of institutions participating in the study, the relatively narrow sample involved, and the fact that only a single race (Han Chinese) was included, with no children or teenagers and few excessively obese patients. The study quality was controlled using blinded methods, but it was difficult to achieve randomly selected patients. Patients with a higher risk of difficult airway might have exhibited greater willingness to participate in our study because of preoperative communication and education. We enlarged the sample to >2000 subjects, but the patients with a primary outcome are still limited and are barely enough to perform a multivariable logistic regression. Therefore, the results may exhibit some bias. In addition, even though the present study demonstrates a good reliability of ultrasonography for tongue thickness, we failed in performing a validity study of this method, such as comparison with computed tomography, magnetic resonance imaging, and so on, because of the limitation of our study conditions. A validation study of this method in predicting difficult airway is also absent. In summary, a thicker tongue evaluated by ultrasonography was an independent predictor for difficult tracheal intubation and difficult laryngoscopy. The ratio of tongue thickness to thyromental distance was another way to predict difficult airway. This airway assessment does not require patients to perform a mandatory action and thus may be useful in unconscious patients. Authors contributions Study design: W.D.Y. Conduct of the study, data collection, data analysis, and manuscript preparation: W.D.Y., B.W.

8 608 Yoa and Wang Acknowledgements We are thankful to our colleagues (Department of Anesthesiology and Intensive Care Unit, the First Affiliated Hospital Wannan Medical College) for help with samples. We arealsoverygratefultot.yu,h.wu,x.j.jin(departmentof Anesthesiology, The First Affiliated Hospital of Wannan Medical college, Wuhu, China), Z. B. Shen (Department of Ultrasonic Medicine, The First Affiliated Hospital of Wannan Medical college, Wuhu, China), and Y. H. Li (Department of Anesthesiology, The First Affiliated Hospital of Anhui Medical University, Hefei, China) for their contributions to the present study, which are nearly equal to those of authors. Declaration of interest None declared. Funding Wannan Medical College, Wuhu, China (No. WK2014F03); Education Department of Anhui Province, Hefei, China (No. KJ2015A149). References 1. Cook TM, MacDougall-Davis SR. Complications and failure of airway management. Br J Anaesth 2012; 109: i Cook TM, Woodall N, Frerk C; Fourth National Audit Project. Major complications of airway management in the UK: results of the Fourth National Audit Project of the Royal College of Anaesthetists and the Difficult Airway Society. Part 1: anaesthesia. Br J Anaesth 2011; 106: Yentis SM. Predicting difficult intubation worthwhile exercise or pointless ritual? Anaesthesia 2002; 57: Lundstrøm LH, Vester-Andersen M, Møller AM, Charuluxananan S, L Hermite J, Wetterslev J. Poor prognostic value of the modified Mallampati score: a meta-analysis involving patients. Br J Anaesth 2011; 107: Langeron O, Cuvillon P, Ibanez-Esteve C, Lenfant F, Riou B, Le Manach Y. Prediction of difficult tracheal intubation: time for a paradigm change. Anesthesiology 2012; 117: Lundstrøm LH, Møller AM, Rosenstock C, Astrup G, Wetterslev J. High body mass index is a weak predictor for difficult and failed tracheal intubation: a cohort study of 91,332 consecutive patients scheduled for direct laryngoscopy registered in the Danish Anesthesia Database. Anesthesiology 2009; 110: Frerk C, Mitchell VS, McNarry AF, et al. Difficult Airway Society 2015 guidelines for management of unanticipated difficult intubation in adults. Br J Anaesth 2015; 115: Koay CK. Difficult tracheal intubation analysis and management in 37 cases. Singapore Med J 1998; 39: Ovassapian A, Glassenberg R, Randel GI, Klock A, Mesnick PS, Klafta JM. The unexpected difficult airway and lingual tonsil hyperplasia: a case series and a review of the literature. Anesthesiology 2002; 97: Kanaya N, Kawana S, Watanabe H, et al. The utility of threedimensional computed tomography in unanticipated difficult endotracheal intubation. Anesth Analg 2000; 91: Greenland KB, Edwards MJ, Hutton NJ, Challis VJ, Irwin MG, Sleigh JW. Changes in airway configuration with different head and neck positions using magnetic resonance imaging of normal airways: a new concept with possible clinical applications. Br J Anaesth 2010; 105: 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: Levitan RM, Everett WW, Ochroch EA. Limitations of difficult airway prediction in patients intubated in the emergency department. Ann Emerg Med 2004; 44: Wojtczak JA. Submandibular sonography: assessment of hyomental distances and ratio, tongue size, and floor of the mouth musculature using portable sonography. J Ultrasound Med 2012; 31: Prasad A, Yu E, Wong DT, Karkhanis R, Gullane P, Chan VW. Comparison of sonography and computed tomography as imaging tools for assessment of airway structures. J Ultrasound Med 2011; 30: Kristensen MS, Teoh WH, Graumann O, Laursen CB. Ultrasonography for clinical decision-making and intervention in airway management: from the mouth to the lungs and pleurae. Insights Imaging 2014; 5: Mallampati SR, Gatt SP, Gugino LD, et al. A clinical sign to predict difficult tracheal intubation: a prospective study. Can Anaesth Soc J 1985; 32: Samsoon GL, Young JR. Difficult tracheal intubation: a retrospective study. Anaesthesia 1987; 42: Apfelbaum JL, Hagberg CA, Caplan RA, et al. Practice guidelines for management of the difficult airway: an updated report by the American Society of Anesthesiologists Task Force on Management of the Difficult Airway. Anesthesiology 2013; 118: Cormack RS, Lehane J. Difficult tracheal intubation in obstetrics. Anaesthesia 1984; 39: Kheterpal S, Martin L, Shanks AM, Tremper KK. Prediction and outcomes of impossible mask ventilation: a review of 50,000 anesthetics. Anesthesiology 2009; 110: Youden WJ. Index for rating diagnostic tests. Cancer 1950; 3: Yu T, Wang B, Jin XJ, et al. Predicting difficult airways: rule or 3-3 rule? Ir J Med Sci 2015; 184: Harrell FE Jr, Lee KL, Matchar DB, Reichert TA. Regression models for prognostic prediction: advantages, problems, and suggested solutions. Cancer Treat Rep 1985; 69: Ezri T, Gewürtz G, Sessler DI, et al. Prediction of difficult laryngoscopy in obese patients by ultrasound quantification of anterior neck soft tissue. Anaesthesia 2003; 58: Komatsu R, Sengupta P, Wadhwa A, et al. Ultrasound quantification of anterior soft tissue thickness fails to predict difficult laryngoscopy in obese patients. Anaesth Intensive Care 2007; 35: Adhikari S, Zeger W, Schmier C, et al. Pilot study to determine the utility of point-of-care ultrasound in the assessment of difficult laryngoscopy. Acad Emerg Med 2011; 18: Ittichaikulthol W, Chanpradub S, Amnoundetchakorn S, Arayajarernwong N, Wongkum W. Modified Mallampati test and thyromental distance as a predictor of difficult laryngoscopy in Thai patients. J Med Assoc Thai 2010; 93: Merah NA, Wong DT, Ffoulkes-Crabbe DJ, Kushimo OT, Bode CO. Modified Mallampati test, thyromental distance and inter-incisor gap are the best predictors of difficult laryngoscopy in west Africans. Can J Anaesth 2005; 52: Salimi A, Farzanegan B, Rastegarpour A, Kolahi AA. Comparison of the upper lip bite test with measurement of thyromental distance for prediction of difficult intubations. Acta Anaesthesiol Taiwan 2008; 46: 61 5

9 Tongue thickness predicts difficult intubation 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: Yildiz TS, Korkmaz F, Solak M, et al. Prediction of difficult tracheal intubation in Turkish patients: a multi-center methodological study. Eur J Anaesthesiol 2007; 24: Huh J, Shin HY, Kim SH, Yoon TK, Kim DK. Diagnostic predictor of difficult laryngoscopy: the hyomental distance ratio. Anesth Analg 2009; 108: Handling editor: Takashi Asai