Abstract
Objective
Unanticipated difficult tracheal intubation is a significant source of morbidity and mortality in anesthetized patients. A number of modules have been developed to predict difficult airways, but they are often complex in nature. We combined the modified Mallampati score (M), thyromental distance (T), anatomical abnormality (A), and cervical mobility (C) into a single scoring system with the acronym M-TAC, and evaluated it against Mallampati scoring.
Methods
We prospectively analyzed 500 adult patients of the American Society of Anesthesiologists (ASA) class I or II, scheduled for elective surgery under general anesthesia. Preoperative airway assessments using M-TAC were performed, all of which were given a score. Anesthesiologists, blinded to the pre-anesthetic airway assessment, performed laryngoscopy and graded the laryngoscopic view as per Cormack and Lehane's classification. For the study purpose, difficult laryngoscopy was defined as Cormack and Lehane Grade 3 or 4 of laryngoscopic view.
Results
An M-TAC score ≥ 4 had a significantly higher sensitivity (96% vs. 72%) and specificity (86% vs. 78%) with a high positive predictive value (44% vs. 28%) and a very low false negative value (2% vs. 15%) in comparison with Mallampati scoring (p < 0.05). Analysis of the receiver operating characteristic (ROC) curve for predicting difficult laryngoscopy revealed an area under the curve of 0.83 (95% CI = 0.78–0.88) for Mallampati scoring and 0.94 (95% CI = 0.92–0.96) for M-TAC scoring system.
Conclusion
The M-TAC scoring system has provided a higher sensitivity and specificity in predicting difficult laryngoscopy in comparison with Mallampati classification.
Keywords
Laryngoscopy: difficult; Mallampati classification; Intubation, intratracheal;
1. Introduction
Unanticipated difficult intubation is not only a threat to patient's life, but often evaluates the skill of an experienced anesthesiologist. Even though the reported incidence of unanticipated difficult intubation in anesthesia is rare, it often leads to disastrous respiratory complications.1 Thus, to predict a possible difficult intubation in time is of the essence. Many scoring systems have hitherto been developed to identify possible difficult intubation, none of which stands out to be solely the best.2, 3, 4, 5, 6, 7, 8, 9 Thus, suggestions for a combination of these tests have evolved over time for a better prediction.10, 11, 12, 13, 14, 15
We ventured to combine Mallampati score with some other anatomical factors to develop a new and simpler clinical prediction model for a better predictive ability.
2. Methods
We designed a scoring system, modified Mallampati score (M), thyromental distance (T), anatomical abnormality (A), and cervical mobility (C) (M-TAC), by assembling four of the most widely practiced tests for evaluation of difficult airways. ‘M’ denotes modified Mallampati class of oropharyngeal view.3, 16While being seated, each patient was asked to open his or her mouth wide and protrude the tongue maximally without phonation; the view was classified as shown in Table 1.4
TAC consists of three other factors: T = thyromental distance; A = anatomical abnormality of face/neck/oral cavity, and C = cervical mobility range (flexion and extension). Thyromental distance (T) was measured along a straight line from the thyroid notch to the lower border of the mandibular mentum, with the head fully extended and the mouth closed.17 The range of cervical mobility was evaluated according to the method suggested by Wilson et al. 2Patients were asked to extend the neck fully and a pencil was placed vertically on the forehead. While the pencil was held firmly in position, the neck was flexed and the angle was measured. Thereafter, T, anatomical abnormalities (A) of face, neck or oral cavity, and cervical mobility were each classified into three grades as shown in Table 1. The allotted grades were the respective scores of M-TAC (i.e., Grade II means score 2) (Table 1). Therefore, on evaluation of the airway, if we found Mallampati Grade 2, T Grade 1, anatomical abnormality Grade 0, and cervical mobility Grade 2, then this airway was documented as M2T1A0C2 or M-2 TAC-3 with a total M-TAC score of 5.
The study protocol was approved by the “Ethics Committee for Human Studies” of Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow (India) and informed written consent was obtained from all the participants. Five hundred consecutive adult patients aged 18–72 years, irrespective of sex, of the American Society of Anesthesiologists (ASA) physical status I or II, scheduled for elective surgeries under general anesthesia from December 2008 to February 2010, were enrolled for the study. Patients with an obvious difficult airway (fractured mandible or cervical spine disorder, obstructive airway tumor, edentulous patients, mouth opening <3 cm etc.), or those who refused to participate, were excluded.
Four fellow anesthesiologists participated in this study; two performed pre-anesthetic assessment of the airway, and two performed laryngoscopy under general anesthesia. All information in relation to the pre-anesthetic check up was shared between the anesthesiologists, except for the data relating to airway evaluation specific to this study and M-TAC scoring. For safety reasons, the anesthesiologists providing anesthesia in the operation theatre were always equipped with a difficult airway cart and Difficult Airway Society guidelines.18
All patients were premedicated with oral lorazepam (0.04 mg/kg) and ranitidine (150 mg) the night before and 2 hours before anesthesia, respectively. Standard fasting guidelines were observed in all patients. Monitors for electrocardiogram (ECG) lead II and V, noninvasive blood pressure, heart rate and peripheral oxygen saturation were applied before induction. Following pre-oxygenation for 3 minutes, anesthesia was induced with intravenous midazolam (0.04 mg/kg) and fentanyl (2 μg/kg) and propofol (2–2.5 mg/kg). Adequacy of mask ventilation was assessed under 1–2% sevoflurane in 100% oxygen. When the anesthesia level was satisfactory, vecuronium bromide (0.1 mg/kg) was administered intravenously and IPPV continued for 3 minutes. Laryngoscopy was performed in sniffing position using a Macintosh laryngoscope and the best possible laryngoscopic view was obtained after confirmation by a second anaesthesiologist. Difficult laryngoscopy was defined as the view observed corresponding to Grade 3 or 4 of the Cormack and Lehane (CL) laryngoscopic view.19
Three attempts at endotracheal intubation were allowed before the act was considered as a failure. In this situation, the participants followed the next step of the algorithm using the secondary intubation plan. If the airway could not be secured even with these, we continued with laryngeal mask airway (LMA) and other efforts to maintain ventilation and oxygenation. If intubation and ventilation could not be achieved, cricothyrotomy was reserved as the last resort.
2.1. Statistical analysis
The sensitivity, specificity, positive predictive value, and negative predictive value in each group were calculated. The data were analyzed using the Student t test for continuous variables and Fisher's Exact test or Yates Chi-square test, as appropriate, for noncontinuous variables. The predictive accuracy of the studied parameters was compared by measuring the area under the receiver operating characteristic (ROC) curve (AUROC). The package SPSS 11.5 (SPSS Inc., 233 South Wacker Drive, 11th Floor, Chicago, IL 60606-6412) was used for statistical analysis and statistical significance was defined as p < 0.05.
3. Results
A total of 500 patients (292 males and 208 females) were analyzed. The median age = 46 years (range = 18–72 years), weight = 54 kg (range = 42–110 kg), and height = 159 cm (range = 146–186 cm).
Following induction of anesthesia, ventilation with a bag and mask was easy in 496 patients and assistance from a second person was required in four (0.8%) patients. Fifty-three (10.6%) patients had a difficult laryngoscopic view; CL-3 grade in 45 (9.0 %) and CL-4 grade in eight (1.6%) patients were noted (Table 2).
One hundred and thirty five patients had a modified Mallampati class of 3 and 4, of whom only 38 had difficult laryngoscopy (28%). In contrast, 15 out of 365 patients (4.1%) had difficult laryngoscopy with a modified Mallampati class of 0, 1 or 2. In comparison, none of the 279 patients with an M-TAC score of ≤2 had CL-3 or 4. Again, none of the patients had difficult laryngoscopy with a modified Mallampati class 3 and a TAC score of 0 (M3TAC0), whereas all had difficult laryngoscopy when a modified Mallampati class 3 was associated with a TAC score of ≥3 (M3TAC3).
Again, 40% of the patients had a difficult laryngoscopic view with a Mallampati class 4 and a TAC score of 0, whereas all patients had a difficult view when a Mallampati class 4 was associated with a TAC score of >1.
Twenty-eight patients (32%) had difficult laryngoscopy with an M-TAC score of 4 and 82% of patients scored ≥5 in M-TAC (Table 2).
Four patients who required two persons to manage the mask ventilation, had an M-TAC score of 2; the laryngoscopic view, as observed, was CL-1 (n = 1) or CL-2 (n = 3) and no difficulty was encountered during tracheal intubation.
Endotracheal intubation was accomplished either with a stylet (n = 26) or a Frova's bougie (n = 19) in patients with a CL-3 laryngoscopic view, whereas accomplishment could be achieved with fiber-optic laryngoscope (n = 4), intubating laryngeal mask airway (n = 2) and retrograde intubation (n = 2) in patients with a CL-4 laryngoscopic view. Conventional laryngoscopy was successful in the rest.
The sensitivity (96%) and specificity (86%), positive predictive value (44%) and negative predictive value (99%) of M-TAC ≥ 4 were highly significant in comparison with M3 and M4 in predicting difficult laryngoscopy (p < 0.05) (Table 3).
Analysis of the ROC curve for predicting difficult laryngoscopy revealed an area under the curve of 0.83 (95% CI = 0.78– 0.88) for Mallampati classification and 0.94 (95% CI = 0.92–0.96) for the M-TAC scoring system (Fig. 1).
Download full-size image
4. Discussion
We observed that our study model (M-TAC) possessed better sensitivity and specificity and produced significantly higher positive and negative predictive values for the prediction of difficult laryngoscopy in comparison with Mallampati scoring. In the published literature,2, 11, 12, 13, 14, 15, 16 we found that no model was as sensitive and predictive as the present one. The above fact was also supported by ROC curve analysis, which showed that the area under curve (AUC) for the M-TAC score was comparable to that of the Mallampati scoring system.
Mallampati scoring is one of the most commonly used clinical tests to predict difficult laryngoscopy.3 However, simple bedside tests, such as the Mallampati test,3, 20 the modified Mallampati test,9 T,21 or C2 have been found to be of limited use in predicting difficult laryngoscopy, when each individual was investigated in isolation, as it has long been realized that difficult laryngoscopy is a multifactorial problem. Thus, effective prediction requires a combination of tests.22, 23
A recent meta-analysis found the combination of Mallampati scoring and T to be the most accurate predictor of difficult intubation, but with a considerably low sensitivity of 36% (95% CI = 14–59%) only.22
As the anatomy of the head and neck region plays a leading role in deciding the fateful profile of the airway, especially associated with influential abnormality, we tried to accommodate A into this new scoring system, along with three other techniques, namely, the Mallampati test, T and C, which forms the acronym M-TAC.
Mallampati grading is the most commonly performed test for airway assessment and therefore the “M” of the acronym M-TAC comes from Mallampati classification. Although Mallampati class 3 has been classified as a risk factor for a difficult airway, we found a score of M3 alone is not able to stratify the risk of difficult laryngoscopy adequately, as all of our patients (n = 48) who had M3T0A0C0 had an easy laryngoscopic view (CL-1 = 12, CL-2 = 36). A score of TAC2 or more was invariably associated with higher Mallampati grades (Grade 3 or 4), thereby increasing the M-TAC score to ≥4, a risk factor for difficult laryngoscopy. Conversely, a higher Mallampati score was not necessarily associated with higher TAC grades. A total M-TAC score of ≥4 was predictive of difficult intubation.
Although obesity is not an independent risk factor for difficult airway, it is one of several factors that need to be considered during pre-anesthetic evaluation.24 Our four patients with a mean body mass index of 32 (1.8) and M-TAC score of ≤2 (M1TAC1, M2TAC0) required two persons to manage mask ventilation; the laryngoscopic view observed was CL-1 (n = 1) and CL-2 (n = 3); no difficulty was encountered during endotracheal intubation. It shows that if the M-TAC score is <3, obesity per se is not a risk factor of difficult laryngoscopy; however it may be a risk factor for difficult mask ventilation.
We found, five patients who had a Mallampati class 4 with a TAC score of 0 (M4T0A0C0), in whom the laryngoscopic view observed was CL-2 in three and CL-3 in two patients and none of the patients had a CL-4 grade. All of these patients were successfully intubated using either a stylet or a Frova's bougie. We found that the Mallampati test was not a very sensitive indicator to predict difficult laryngoscopy, as a score of M3 or M4 with a TAC score of 0 (only 2/53 had difficult laryngoscopy) was associated with a significantly higher percentage of false positive results.
Another commonly performed test for difficult laryngoscopy is TMD. However, it varies according to the patient size.21 When T is ≥6.5 cm with a fully extended atlanto-occipital joint, the laryngoscopic view is predicted to be easy.11 However, several studies have raised the issue of the reliability of T as an isolated predictor for difficult laryngoscopy.15, 21, 25, 26
It has long been realized that if we use more than one test to predict difficult intubation, predictive accuracy increases,10, 11, 12, 13, 14, 15 but the best possible combination is still to be answered. The ideal model for the prediction of difficult laryngoscopy requires high sensitivity and specificity, with few false positives and negatives. The consequence of false negative results may be deleterious and even life-threatening. Therefore, a decreasing false negative prediction rate is more important than falsely predicting difficult laryngoscopy in normal patients.
A more obvious question seems to be whether sensitivity and specificity are equally important. We believe that the purpose of any model should be the detection of as many patients as possible with a difficult airway, to minimize the potentially serious consequences of unanticipated difficult tracheal intubation. Therefore, a model with a high sensitivity and positive predictive value, along with highest negative predictive value, is required.
Krobbuaban et al studied T, the extent of mouth opening (interincisor gap), modified Mallampati class, and the range of neck movement for predicting difficult laryngoscopy in 550 adult patients.10 They observed the optimal cutoff point for T and interincisor gap to predict difficult laryngoscopy as 6.5 cm (sensitivity = 52%; specificity = 71%) and 3.5 cm (sensitivity = 39%; specificity = 69%), respectively. An interincisor gap of ≤3.5 cm, a T of ≤6.5 cm, neck movement ≤80 degrees, and a Mallampati class 3 or 4 were selected as being predictive of difficult laryngoscopy. They reported difficult laryngoscopy (CL Grade 3 or 4) in 12.5% of patients. In our study, the incidence of difficult laryngoscopy (n = 53) was 10.6%, which was comparable to those of other studies previously reported (9–12%).10, 11, 12 Our results might therefore favor a low negative predictive value and a high positive predictive value.
The Arné model,11 the Wilson model,2 and the Naguib model27 have been recommended, which are based on multivariate analysis of more than a dozen risk factors with variable success. The highest sensitivity was achieved with the Naguib model. Specifically, the sensitivity of this model was 81.4% (95% CI = 74–89%) compared with 40.2% (95% CI = 30–50%) for the Wilson model and 54.6% (95% CI = 45–65%) for the Arné model. The Naguib model was significantly more sensitive than the other two models, based on a pair-wise comparison using the McNemar test (p < 0.0001). The Cochran Q statistic value indicated that the three models differed significantly with respect to their predictive accuracy (p < 0.02). Both the Naguib model and the Arné model classified more intubations correctly (p = 0.01) than the Wilson model. Even though we have not compared our results with these quite cumbersome models, our model has a significantly higher sensitivity and predictability of difficult airway, with very few false negatives, than the abovementioned models.
Our study reflects a direct relationship between difficult laryngoscopy and M-TAC scoring. The higher the M-TAC score is, the greater the likelihood of difficult laryngoscopy will be. With an M-TAC score of ≥5, most patients are likely to have a difficult laryngoscopic view, whereas M with TAC ≥ 3 will lead to a difficult laryngoscopic view in all patients.
The limitation of our study is the subjectiveness of the data, as most of these criteria are rater dependent. Additionally, the results of this study conducted in an Asian population, may not stand true for other populations because of well-known ethnic variations, where a reassessment may be required. Further, anatomical factors like neck circumference, have not been included in this study to provide simplicity to the scoring system without compromising its effectiveness. In addition, some factors can affect more than one component of this study model, e.g., micrognathia plays a role in the anatomical abnormality section and leads to a shorter T. Similarly, macroglossia comes under the anatomical abnormity section and can affect Mallampati class. However, it is not only the micrognathia which affects the T; many other anatomical factors, including the position of the larynx, contribute to it. Also, a short T can be a surrogate for limited head extension, rather than small submandibular space.28 Thus, it is a fact that all anatomical factors which decide the ultimate airway outcome are interrelated. We considered some of these factors and tried to create a simple and effective scoring system.
In conclusion, several studies have evaluated different clinical risk factors, alone or in combination, to predict a difficult laryngoscopic view; however, none has produced a simple formulation for documentation and interpretation that has a high sensitivity and a low false negativity. Airway evaluation with a new acronym, M-TAC, is simple to document, inferential, and has a high sensitivity and predictive value, with a low false negativity, to identify difficult airways, especially when the score is ≥4.