Diagnostic value of different screening tests in isolation or combination for predicting difficult intubation: A prospective study

INTRODUCTION

Unanticipated difficult tracheal intubation is one of the common challenges faced by the anaesthesiologists. The reported incidence of a difficult laryngoscopy or endotracheal intubation varies from 1.5% to 13% in patients undergoing surgery.[1] The failure of anaesthesiologists to establish airway following the induction of general anaesthesia is one of the most frequent causes of anaesthesia related adverse events. Numerous investigators have attempted to predict difficult intubation using simple bedside physical examination. Nevertheless, the diagnostic accuracy of these screening tests have varied from trial to trial, probably because of the difference in patient characteristics, difference in the incidence of difficult intubation, inadequate statistical power and difference in test thresholds.[2] Question remains as to whether a combination of these bedside tests can improve predictive accuracy of difficult intubation.

The aim of our study was to evaluate the diagnostic value of sternomental distance (SMD), thyromental distance (TMD), inter incisor gap (IIG), upper lip bite test (ULBT) and Mallampati grading (MPG) in isolation and in combination for predicting difficult intubation.

METHODS

A prospective double-blind observational study was conducted from January 2011 to June 2011 on all American Society of Anaesthesiologists I and II physical status patients posted for elective surgery under general anaesthesia after approval by Institutional Ethical Committee and written informed consent by the patients. Total number of patients was 484. They were pre-operatively assessed by an anaesthesiologist not involved in intubation for MPG, SMD, TMD, IIG and ULBT [Table 1] assessment and measurements were obtained. Patients with obvious head and neck pathology, edentulous patients, mass in the mouth, body mass index (BMI) >40, protruding upper incisors (total of 34) were excluded from the study. Finally, 450 patients were included to participate in the study.

Table 1.

Diagnostic predictors

On the day of the surgery after premedication with intravenous glycopyrrolate 0.004 mg/kg and fentanyl 2 μg/kg, induction was carried out with thiopentone sodium 5 mg/kg. Succinylcholine 2 mg/kg was used to facilitate endotracheal intubation. After adequate relaxation, laryngoscopy was performed by an experienced anaesthesiologist who was not aware of the preoperative airway assessment. Laryngoscopy was performed with the head in the sniffing position using a Macintosh blade, and the laryngoscopic view was determined using the Cormack–Lehane (CL) grading system [Table 2]. Laryngoscopic grades III and IV were considered to have difficult intubation.

Table 2.

Laryngoscopic view

Statistical analysis was performed using SPSS software version 12 (SPSS Inc. Chicago, IL, USA). To test the predictive power of each parameter for identifying difficult laryngoscopy, sensitivity (S), specificity (Sp), positive predictive value (PPV) and negative predictive value (NPV) were calculated. The association between different predictors and difficult laryngoscopy was evaluated using the Chi-square test. To determine the discriminative power of individual tests and the combination, receiver operating characteristic (ROC) analysis was done and the area under the curve (AUC) with 95% confidence interval was calculated.

RESULTS

A total of 450 patients were evaluated out of which 320 were men (71%) and 130 were women (29%) with ages ranging from 18 to 65 years. There was no statistical difference between age, sex, BMI and easy or difficult laryngoscopy grades [Table 3]. Incidence of difficult intubation in our study population was 7% (32 patients). CL grade I was seen in 56% of patients, CL II in 37%, CL III in 7%, and no patients had CL grade IV. All the patients could be intubated either using external manipulation of the larynx or by bougie guidance. ULBT had the highest sensitivity, specificity, PPV, NPV, accuracy and area under ROC amongst the individual tests [Table 4]. The combination of ULBT and MPG had the highest sensitivity (78%) and ULBT and TMD had the highest PPV (64%) and specificity 98% with highest AUC (0.79). Combination of ULBT, MPG and TMD gave the highest sensitivity, specificity and PPV [Table 5].

Table 3.

Demographic data

Table 4.

Individual parameters

Table 5.

Combination of parameters

DISCUSSION

An ideal bedside predictor should have the ease of applicability, reliability, accuracy of prediction and minimal interobserver variation. The five tests we used were SMD, TMD, IIG, ULBT and MPG which gave us a good assessment of neck extension, mandibular space, mouth opening, jaw protrusion, TM joint movement and pharyngeal space. However, each of these predictors has their own limitations, and no single screening test can be 100% sensitive and specific. MPG has often been criticised for its low specificity and sensitivity and also its lack of reproducibility due to interobserver variability.[3,4,5] SMD and TMD are both quantitative in nature and are influenced by heterogenicity, variety of test thresholds, obesity and neck swellings. IIG is also quantitative and gives high false positive values. ULBT cannot be performed in edentulous patients and with protruding upper incisors.[2,6] Sensitivity and PPVs are important for an ideal screening test. Sensitivity of the test in the present study is the power of the test to find out the maximum number of difficult laryngoscopies and PPV is the predictive power of the test so that patients supposed to have normal airway are not subjected to a difficult airway protocol.

The incidence of difficult intubation in our study was 7% (32 patients). Out of the predictors we used, ULBT could predict 16 (50%) patients of difficult intubation and turned out to be the single best predictor in terms of sensitivity (50.00%), specificity (98.56%), PPV (72.72%), NPV (96.26%), accuracy (95.11%) and AUC (0.837). These results are similar to the study conducted by Allahyary et al.[7] and Khan et al.[8] MPG could predict difficult intubation only in 10 patients, having a low sensitivity of 31.25% and PPV of 30.30%. This was similar to the study conducted by Allahyary et al.[7] and Hester et al.[9] Both TMD and SMD had low sensitivity and higher specificity but PPV of TMD was higher than SMD and MPG. IIG had a very low sensitivity in predicting difficult intubation.

Since ULBT had the highest predictive power its combination with different parameters were evaluated. ULBT with MPG had the highest sensitivity of 78% and ULBT with TMD had the highest PPV of 64% and specificity of 98%, which was similar to study by Allahyary et al.[7] Our study showed an increase in the sensitivity for the combination of ULBT with TMD or SMD, which was unlike the study by Richa et al.[10] which had a low sensitivity for these combinations (18.7% and 15.6%, respectively). The study by Khan et al.[8] had lower PPV for these combinations compared to our study. Combination of ULBT with IIG did not further increase the predictive power because of high false positivity with IIG.

When three parameters were combined, the combination of either TMD or SMD with ULBT and MPG increased the sensitivity to 81.25%; however, ULBT + MPG + TMD had a higher PPV. Thus, a combination of ULBT, MPG and TMD had the highest sensitivity of 81%, specificity of 95% and PPV of 53% making it a good bedside screening test for difficult intubation. Combining other predictors did not further increase the predictive power. NPV of all the tests was more than 95% which proves that almost every test can predict easy intubation accurately than predicting difficult intubation.

We also calculated the AUC of ROCs with 95% confidence interval to find the discriminating power of the predictive test because it was independent from the incidence of difficult intubation. ULBT had AUC of 0.837 which was similar to the study by Khan et al.[8] however, it was only 0.604 in the study by Eberhart et al.[11] which could be due to the variability of clinical experience in the anaesthesiologist performing the endotracheal intubation. We found the highest AUC of 0.79 for combination of ULBT and TMD.

Limitations of our study were that we did not evaluate all the existing parameters of difficult intubation, and our study population was limited.

CONCLUSION

Upper lip bite test is useful as a single bedside predictor of difficult intubation because it is easy to perform, lacks interobserver variability and does not need any special equipment to perform it. Practically, ULBT with MPG can be the best combination to predict difficult intubation in bedside practice. Combination of ULBT, MPG and TMD can be a reliable clinical prediction model for difficult endotracheal intubation.