Abstract

This is the prospective randomized study of 70 patients done in Seth GS Medical College and KEM Hospital Mumbai in out patients department of anaesthesia for prediction of difficult intubation by 10 tests. All measurement were done by finger method and scale method by expert and observer. Data were analyzed statistically and results were noted to find out interobserver reliability. Conclusion –Results were graded from best to least reliable.

Introduction

The most important and prime aim of the anaesthesiologists is to maintain the airway during induction of general anaesthesia. For predicting the difficult intubation 10 tests are recommended. We have found that results of airway assessment varied from person to person, i.e. INTER-OBSERVER RELIABILITY. The higher the reliability the lower the error variance. To address this concern we examined the inter-observer reliability of ten tests used to predict difficult tracheal intubation and ranked these tests in terms of their repeatability from best to least reliable.

Aims and Objectives

1. To determine the reliability of these tests by finger method and scale method.
2. To get the most reliable method.
3. To assess reliability with Wilson score and modified Mallampatti classification.
4. Pre-operative airway evaluation, have least variance.

Material and Methods

This is a prospective study done in the year 2005 January to June. Following institutional approval, 70 patients were randomly selected for assessing difficult intubation by 10 standard tests.

Inclusion Criteria

Age > 18 years < 75 years; ASA I / II; Patients coming for repeat surgeries with documented history of difficult intubations in previous surgery.

Exclusion Criteria

Age < 18 years or > 75 years; Edentulous patients; Patients with facial trauma.; congenital anomalies of the face; and gross swelling of face or neck. Unstable cervical spine. Recently operated patients.

All patients underwent for the 10 tests.

1. Interincisor gap is less than 4 cm, expected difficult intubation.
2. Slux i.e. Protrusion of mandible: Protrusion was graded as follows:
   Grade I : If lower incisors could be protruded anterior to upper incisors.
   Grade II : If upper and lower incisors touch each other.
   Grade III : If lower incisors could not be brought forward to touch the upper incisors.
   
3. Thyromental distance: A distance of more than 6 cm. normal.
4. Mandibular ramus length: Normal > 6 cm.
5. Sternomental Distance: It should be > 12.5 cm.
6. Atlanto Occipital Extension: More than 350 is normal.
7. Chin protrusion test: It should be more than 2.5 cm.
8. Profile classification: It is an imaginary line through the most prominent part of brow and maxilla while viewing the patient in profile i.e. from the side

The grades are -
I. Protruding  : prominence of chin in front of the line
II. Neutral  : prominence of chin at the line
III. Retrognathic : prominence of chin behind the line

9. Modified Mallampati test:

Grade I : Soft palate, uvula, fauces and pillars visible
Grade II : Soft palate, uvula, fauces visible, but pillars obscured.
Grade III : Soft palate only visible.
Grade IV : Soft palate not visible.

10. Best view by any possible way.

In our study we have used two Raters, observer O, and expert E. Both of them have performed the above mentioned ten tests and their readings were noted. Each performed the tests by finger method and scale method. The four tests- MPC, Best view, Slux, PC were performed individually and grades were noted.
Both Raters ranked the results into three grades: 1/2/3.
Table 1 showing the grades of Raters.

Table 1

Total sum of the grades of the ten tests was calculated, which was 10 to 30.

Wilson’s score was noted.

Statistical Analysis

The data were analyzed using three statistical methods alpha, kappa and interclass correlation coefficient. The values were varying 0 to 1.

Scores were taken as > 0.75 excellent, 0.4 – 0.75 moderate and £ 0.4 poor interobserver reliability. In order to compare the clinical performance sensitivity and specificity were calculated. Correlation was determined between 10 tests and laryngoscopic view.

Observations and Results

Age of the patients varied from 18 yrs to 72 yrs, 56 patients were males and 14 were females, Mean height of the patients was 162.87 cm with SD of 5.93 cm, Mean weight of the patients was 59 kgs. with SD of 12.48 kg (Figs. 1 and 2). (Tables 2, 3, 4 and 5).

Results

As per our statistical analysis a - coefficient for finger method was minimum in Mandibular ramus length i.e. 0.7035 and maximum in chin protrusion test, 0.9558, and with scale method was minimum in thyromental distance 0.8419 and maximum

Table 2 : Ratio of mean of differences

For all tests the mean of difference is significantly different (P < 0.01) some systemic rater bias exist , this amount of systemic bias may not be clinically relevant.

Fig. 1 : Reliability between observer and expert for finger method and scale method a coefficient – for finger Method :Minimum in M R L – 0.7035. Maximum in CPT – 0.9558.a coefficient – for scale Method :Minimum in TMD 0.8419. Maximum in IG – 0.9563.	Fig. 2 :Inter class correlation coefficients mixedeffects model (consistency definition) *ICC is - Minimum 0.8419 in TMD, Maximum 0.9563 in IG.

Table 3 :Comparison of SLUX, PC, MPC, BV between observer and expert with agreement tested by Kappa statistic	Distribution of disagreement ratio for various measurements
Kappa Coefficient : Minimum with MPC 0.41, Maximum with BV 0.869	Disagreement Ratio: Minimum with BV – 12.86%. Maximum with MPC – 41.43%

in interincisor gap, 0.9563. a - coefficient increased by scale method in all distances. The increase is maximum in interincisorgap interclass correlation coefficient (ICC) was minimum in thyromental distance, 0.8419 and maximum in interincisor gap, 0.9563.% Ratio of mean of differences and mean of all measurements was minimum in Atlanto occipital extension –0.73% and maximum in interincisorgap. 2.41%, but they were less than 15%, raters systemic bias was clinically insignificant.

Table 4 : Comparison of wilson’s score and ten test score between observer and expert by kappa statistic;	Table 5 : Icc and kappa coefficient of 6 true positive cases between observer and expert
6 cases were for difficult intubations.,all patients had normal intubation.

Kappa coefficient for Modified Mallampati Classification was minimum i.e. 0.416 and is maximum for Profile Classification, 0.781. Disagreement Ratio was minimum in Best view, 12.86% and maximum is Modified Mallampati Classification, 41.43%. Kappa value for Wilson score was 0.864 and for 10 test was 0.732 significant.

Wilson score has sensitivity 100%. In observer and expert, specificity 40%.

Positive predictive value 66.67% and negative predictive value 100%.

10 test scores have sensitivity 100%. In observer and expert, specificity 40%.

Positive predictive value 66.67% and negative predictive value 100%.

Discussion

The ten tests were explained to Raters and to patients.

1. Interincisor gap coefficient for Interincisor gap is 0.9563 and mean of differences is 2.41%, means that inter observer reliability for Interincisor gap is excellent. The study done by Keyvan Karkout et al6 found ICC for Interincisor gap as 0.93 and mean of difference as 2.8%, results are comparable. Ailello G Metcalf1 study showed that in patients with temporomandibular joint diseases where mouth opening is restricted, direct laryngoscopy and hence intubation was difficult. They found that the visualization of larynx was impossible if interincisor gap was less than 2.5 cms. Wilson et al found that interincisor gap < 5 cms was associated with difficult intubation. In one patient whose interincisor gap was 0.8 cm by scale method by expert Rater, we had to plan for awake fibreoptic intubation. Post induction, the laryngoscopy revealed a Cormack Lehane grade of IV. The other patients whose interincisor gap was less than 4 cm and greater than 2.5 cm, direct laryngoscopy was possible but Cormack Lehane grading was III. It indicates that interincisor gap is a good predictor of difficult intubation with excellent interobserver reliability.
2. Subluxation Slux gives idea about temporomandibular joint mobility, receding mandible as well as buck teeth. Wilson et al showed that slux 0 or > 0 i.e. grade II and III is associated with difficult intubation i.e. mandible at upper incisor level or receding have increased risk. We have found k value for slux 0.417 and disagreement ratio 15.71. So it has poor interobserver reliability as compared with Keyvan Karkouti6 et al study where they found k value of 0.66 and disagreement ratio 12% which means moderate reliability.
3. Thyromental distance gives idea about anterior mandibular space. ICC for thyromental distance was 0.8419 mean of differences 1.74% which mean thyromental distance has moderate interobserver reliability. Study done by Keyvan Karkouti et al6 found that thyromental distance has moderate interobserver reliability with ICC 0.89 and mean of differences 3.4%. The results of both studies are equivalent, standard deviation of the measurement of thyromental distance was 1.0 cm. in males and 2.2 cm. in females by scale method. This may be due to prominent thyroid cartilage in males, enabling the Raters to take more accurate readings. Hence, variability is less in males because of less error. In thyromental distance, thyroid cartilage maybe difficult to identify, especially in females. So the variance increases.
4. Mandibular ramus length is an indicator to assess mandibular space. Study done by H.C.Chou5 et al showed mandibular ramus length is one of the important factors in assessment of difficult intubation. In our study mandibular ramus length ICC 0.8492, mean of differences —1.52% which mean moderate IOR. Results are comparable with Keyvan Karkouti6 et al.Exact identification of angle of mandible especially in obese patients may be difficult and the measurement of mandibular ramus length may give different readings to the two raters.
5. Sternomental distance is a test used to assess neck movements. As per our results sternomental distance ICC 0.9017 and mean of differences 1.22% means moderate IOR interobserver reliability. Keyvan Karkouti6 et al found ICC 0.67 and mean of differences 7.0%. moderate interobserver reliability. The head extension may vary. Hence variance increases.

Study done by Ramdhani9 et al showed that sternomental distance less than 13.5 cm is a predictor of difficult intubation. D’Savva4 has shown sternomental distance as best predictor of difficult intubation if less than 12.5 cm.

In our study sternomental distance was more than 12.5 cm in all cases.

6. Atlanto occipital extension. Study done by Youdas12 and other study by Chow FL, Duncan3 showed Atlanto occipital extension more than 60° was associated with easy laryngoscopy, 30-60° have increased risk and less than 30° maybe very difficult intubation, Kappa value for Atlanto occipital extension, we have found is 0.9501 and Keyvan Karkouti6 et al was 0.66, moderate interobserver reliability.
7. In chin protrusion test

   Study done by Bellhouse CP, Pore2 showed that chin protrusion test measures the mandibular space. Anterior distance of mandible less than 2.5 cm increases risk. In our study ICC value for chin protrusion test is 0.9504 and mean of difference is 2.24%. excellent inter observer reliability as compared to ICC 0.89 and mean of difference 3.4% in study done by Keyvan Karkouti6 excellent reliability in both studies.

8. Profile classification which gives idea about receding mandible, has moderate interobserver reliability with moderate disagreement ratio. Wilson et al11 have given importance to receding mandible which can be measured by Profile’s classification. Receding jaw has definite correlation with increased difficulty in intubation. Our study showed k value of 0.781 and disagreement ratio 12.86 which means moderate reliability as compared to Keyvan Karkouti et al.6
9. Mallampatti’s classification: We have used modified Mallampatti’s classification. It has shown poor interobserver reliability with high disagreement ratio. Effects of posture, phonation and observer was shown by Tham et al;10 Effect of pregnancy, by Pilkington et al;8 Effect of Phonation, by Oater et al;7 Till date Mallampatti’s classification is the most widely used primary test to assess difficult intubation in routine practice. Our study showed k value of 0.416 and disagreement ratio 41.43%, poor interobserver reliability. This may be due to effect of posture, phonation or swallowing movements or phasic change of tone of pharyngeal muscles.
10. Best view gives an idea whether external manipulations done under anaesthesia would help to visualise the glottic opening especially when intubation is difficult due to large tongue or decreased oro pharyngeal space. The effect of tongue in or out, phonation, neck movement which gives MPC I/II view is seen as best view. As per MPC identification may vary giving moderate interobserver reliability and low disagreement ratio.

• Our study has shown Interobserver Reliability is increased by Scale method, which was reflected by increases in a coefficient value and significant rise in accuracy is seen in Interincisor gap measurement.
• Two test Interincisor Gap (ICC- 0.9563) Chin protrusion test (ICC- 0.9504) excellent interobserver reliability.

Two test Mallampatti classification (K -0.416) Slux (K - 0.417) poor interobserver reliability

Rest of the tests thyromental distance, sternomental distance, mandibular ramus length and profile’s classification have moderate inter observer reliability.

Wilson score and 10 tests score have comparable interobserver reliability

Study done by Keyvan Karkouti, Lorraine E, Ferris and Tina Meisami Fard23 had shown as follows :

• Interincisor gap and Chin protrusion test has excellent interobserver reliability.
• Mallampatti’s classification has poor interobserver reliability and
• Rest of the tests have moderate interobserver reliability

In clinical practice several factors may attribute to the lower reliability estimates. They may be patient related or Rater related.

Conclusion

Inter observer reliability is increased by Malleable scales than by finger method. Inter observer reliability is excellent in Interincisor gap. Chin protrusion test is the second and Atlantoccipital extension is the third test which has moderate inter observer reliability. Mallampatti’s classification and Slux has poor inter observer reliability. Wilson test and 10 test have comparable sensitivity in prediction of difficult intubation and both tests have excellent inter observer reliability.

References

1. Samsoon GTL, Young JRB. Difficult tracheal intubation: a retrospective study. Anaesthesia 1987; 42 : 487-90.
2. Frerk CM. Predicting difficult intubation. Anaesthesia 1991; 96 : 1005-8.
3. Wilson ME, Spiegelhalter D, Robertson JA, Lesers P. Predicting difficult intubation. British Journal of Anaesthesia 1988; 61 : 211-6.
4. Savva D. Prediction of difficult tracheal intubation. British Journal of Anaesthesia 1994; 73 : 149-53.
5. Ramadhani SAL, Mohamed LA, Rocke DA, Gouws E. Sternomental distance as the sole predictor of difficult Laryngoscopy in obstetric anaesthesia. British Journal of Anesthesia 1996; 77 : 312-6.
6. Chou HC, Wu TL. Mandibulo-hyoid distance in difficult Laryngoscopy. British Journal of Anaesthesia 1993; 71 : 335-9.
7. Marc Lewis, Berry CC. What is the best way to determine oropharyngeal classification and mandibular space length to predict difficult Laryngoscopy? Anesthesiology 1994; 81 : 69-75.
8. Aiello G, Metcalf I. Anaesthetic implication of temporomandibular Joint disease. Can J Anaesthesia 1992; 39 : 610-6.
9. Chow FL, Duncan PG, Code WE, Yip RW. Can bedside neck extension predict difficult intubation? Can J Anaesthesia 1993; 40 : 44.
10. Youdas JW, Carey JR, Garrett TR. Reliability of measurements of cervicle spine range of Motion comparison of three methods. Phesio Ther 1991; 71 : 98-106.
11. Tham EJ, Glidersleve CD, Sander LD, Vaughan RS. Effect of posture phonation and observer on mallampatti classification. British Journal of Anaesthesia 1992; 68 : 32-8.
12. Pilkington S, Carli F, Dakin MJ, et al. Increased Mallampatti score during pregnancy. British Journal of Anesthesia 1995; 74 : 638-42.
13. Comparison of Two methods for predicting Difficult Intubation JDL Oates; AD Macleod AD, Oates PD; FJ Pearsall FJ, Howies and GD Murray. British Journal of Anaesthesia 1991; 66 : 305-9.
14. Keyvan Karkouti, Lorraine, Tina Fard. Inter Observer Reliability of Ten Test Used for Prediction of Difficult Intubation. Canadian Journal of Anaesthesia 1996; 43 (6) : 554-9.

*Assoc. Professor, **PG Student, ***Professor and Head, KEM Hospital, Parel, Mumbai.