COMPARISON BETWEEN CLINICAL AND ULTRASONOGRAPHIC PARAMETERS IN PREDICTING DIFFICULT AIRWAY: AN OBSERVATIONAL PROSPECTIVE STUDY

Various clinical tests are applied at the bedside to predict diﬃcult airway during the pre-anaesthec examina�on


INTRODUCTION
Difficult airway encompasses various scenarios involving difficult face-mask ven la on, inability for placing supraglo c airway devices, limited glo s view on laryngoscopy, mul ple a empts at tracheal intuba on, or 1 poorly iden fiable surgical airway landmarks.Difficult 2 laryngoscopy is said to be the main cause of difficult airway.This can be encountered by anesthesiologists, emergency physicians, intensivists and paramedical health care workers.Unan cipated difficult airway may be challenging to manage and may significantly increase morbidity and Circumference either alone or in combina ons.Despite the assessment tools, the incidence of unan cipated difficult 5 airway can be 2-8 %.These clinical parameters may be difficult to assess in trauma, uncoopera ve or unconscious pa ents.
The use of ultrasonography, as an addi onal tool, in periopera ve and cri cal care se ng can help health care providers to predict difficult airway using various parameters.Being noninvasive, reproducible, portable and at the point of care helps to guide further management of the airway.Ultrasonography of the airways is feasible, informa ve and can be rapidly assessed.Predic on of difficult airway results in an cipa on and prepara on to prevent morbidity and mortality in the opera ng rooms, intensive care units and emergency departments.Predic ng difficult laryngoscopy is important because 30-40% of these cases have difficult tracheal 7 intuba on .The ra onale of this study was to provide data regarding use of ultrasonography for airway evalua on.
The objec ve of this study was to compare the predic ve ability of various clinical parameters with that of ultrasonographic parameters.This study will help in decision making among anesthe sts during pre-anaesthe c assessment of the airway.

METHODOLOGY
A er obtaining permission from Ins tu onal Review Commi ee (IRC) and informed wri en consent, this prospec ve, cross-sec onal and observa ve study was carried out in 200 pa ents (ASA I and II) who required general anesthesia with endotracheal intuba on.The sample size was calculated according to the study by Rana et al who found the incidence of difficult intuba on to be 12.5%.Using Fisher's formula [n = t × P (1 -P)/m where n = required sample size; t = confidence level at 95% (standard value of 1.96); P = 0.125; m = margin of error at 5% (standard value 0.05)].The sample size was calculated to be 168.We enrolled 200 pa ents, to allow for probable dropout.Pa ents unwilling to par cipate, pregnant women, pa ents with oral and cervical pathology, mouth opening less than 3 cm, limited cervical spine mobility and pa ents unable to follow commands were excluded.
In the preopera ve holding area both clinical and ultrasonographic parameters were assessed and recorded.The rou ne airway assessment included mouth opening, Modified Mallampa (MMP) grade, Upper lip bite test (ULBT), thyromental distance and cervical mobility.The pa ents mee ng the inclusion criteria were further evaluated using ultrasonography.
The clinical parameters recorded were MMP grade, ULBT and thyromental distance in si ng and relaxed posi on.The ultrasonographic parameters were recorded using high frequency linear probe (6-13 MHz) and low frequency curvilinear probe (2-5 MHz) (Sonosite® M-Turbo, Fujifilm® Sonosite, Bothell, WA, USA).The pa ents were asked to lie supine with head and neck in neutral posi on.The linear probe was placed at the floor of mouth, midway between mentum of the mandible and hyoid bone to visualize the tongue and maximum transverse tongue width (Tw) was measured.The linear probe was then used to measure so ssue distance from skin to hyoid bone (DSHB).The hyoid bone was iden fied in transverse plane as hyper echoic inverted U-shaped structure with posterior acous c shadowing.In the neutral posi on curvilinear probe was placed sagi al at the floor of mouth to measure length of the tongue (TL)from p of tongue to base of tongue .The curvilinear probe was then rotated by 90 degrees and maximum tongue height(Th) was measured transverse plane.In the same neutral posi on hyomental distance (HMDn) was measured with curvilinear probe.Now the pa ent was asked to maximally extend the head without raising the shoulders.The curvilinear probe was again placed sagi al at the floor of the mouth and hyomental distance was measured (HMDe).The tongue volume was calculated as result of mul plica on of tongue length (TL), height (Th) and width (Tw).The hyomental distance ra o (HMDR) was computed as ra o between HMDe and HMDn.The hyomental distance ra o was chosen because in previous studies it has been shown that pa ent's age, height and BMI may have an impact on absolute hyomental 9 distance values.
The pa ents were then induced as per the ins tu on protocol and laryngoscopy was performed using appropriate size Macintosh blade and Cormack-Lehane(CL) grade was assessed without external laryngeal manipula on by an anesthe st not aware of the clinical and ultrasonography findings .The laryngoscopy was classified as easy (CL Grade 1 and 2) or difficult (CL Grade 3 and 4).The trachea was intubated using appropriate size endotracheal tube.The The MS Excel 2010 and SPSS 22 (SPSS Inc., Chicago, IL, USA) were used for data entry and analysis.The results were averaged (mean ± standard devia on [SD]) for each parameter for con nuous data.The Chi-square test was used to determine the sta s cal difference between the easy and difficult airway.The sensi vity, specificity, posi ve predic ve value (PPV), and nega ve predic ve value (NPV) were calculated to assess the predic ve value of the tests.
The receiver opera ng characteris c (ROC) graphs were plo ed to assess the op mal cut-off scores, and the area under the curve (AUC) was calculated to assess the prognos c accuracy.

RESULTS
200 pa ents were examined using both physical and ultrasonographic parameters, which included 133 females and 67 males.The demographic pa ern and distribu on of pa ents in easy and difficult laryngoscopy group is shown in Table 1.The distribu on of CL grade is shown in figure1.Since CL grade III and IV were allo ed as difficult airway group, the incidence of difficult airway was 15.5%.The 3 pa ents with CL grade IV were intubated on repeated a empts with help of gum elas c bougie and change of laryngoscope blade.We did not encounter "could not ven late, could not intubate (CVCI)" scenarios.Similar to the clinical parameters, ROC was plo ed for the Ultrasonographic parametersU lising receiver opera ng curves, the cutoff value for predic ng difficult laryngoscopy 3 for tongue volume was found to be 100.08cmwith sensi vity of 93.5% and specificity of 72.85%.The AUC for TV was found to be 0.90(95% CI: 0.83-0.966,p=0.00) which shows it to be a fairly good test to predict difficult laryngoscopy.The cut off value for HMDR was found to be 1.09 with sensi vity of 64.5% and specificity of 88.2%.The AUC for HMDR was 0.885(95% CI: 0.817-0.953,p=0.00) which depicts a fairly good predic ve u lity.The AUC for DSHB was found to be 0.732(95% CI: 0.636-0.872,p=0.0014) and cut off value of 0.75cm with sensi vity of 67.75% and specificity of 66%.

DISCUSSION
The use of various predic ve parameters can be of immense value in reducing morbidity and mortality in cases of both an cipated and unan cipated difficult airway.These parameters enable anesthe sts to be aware and hence prepared to iden fy and manage difficult airway scenarios.
Various clinical parameters rely on pa ents being awake and able to obey commands assess mouth opening, Modified Mallampa grade, thyromental distance, upper lip bite test, and neck circumference and neck movement.These tests 10 have been used with variable success rates.
The availability of USG in the OR and ICUs has provided its 6 use in management of the airway.The USG can be used to measure anterior neck so ssue thickness at levels of hyoid bone, vocal cords, epiglo s, trachea, cricothyroid membrane, predict endotracheal tube sizes , predict post extuba on stridor , visualiza on of cricothyroid membrane during [11][12][13][14][15] cricothyrotomy and diagnosing laryngeal abnormali es.
In another meta-analysis, the various bedside clinical parameters for difficult airway had limited or inconsistent predic ve values.The authors suggested use of mul ple or 17 combined parameters to predict difficult airway.We found similar trend in our study with low specificity as MMP had specificity of 54.8%, specificity of ULBT and TMD was 19.4% and 22.6% respec vely.

Ultrasonographic parameters and CL grade Predic on
In a study conducted by Ohri R. et al. in 50 adult pa ents, the correla on between tongue volume and difficulty in laryngoscopy was evaluated using real me 2D ultrasonography and found that larger tongue volume as measured by USG are associated with more difficult 18 laryngoscopy.This was based on the clinical assump on that size and volume of tongue might directly affect laryngoscopy and view of glo s because of insufficient room in submandibular space.This finding is similar to that of our study.
Another study conducted by Wojtczak et al. in five obese and seven morbidly obese adult pa ents, performed submandibular USG found that the tongue volumes did not 19 differ sta s cally in easy and difficult laryngoscopy groups.This finding is different from our study and the difference could be because of normal BMI in our pa ent popula on.
In another similar study conducted by Andruszkiewicz et al. in 199 adult pa ents using submandibular USG, tongue volume was higher in difficult laryngoscopy group which 9 was sta s cally significant.These findings were similar our study but with the notable difference in method used for calcula ng the tongue volume.We used both sagi al and ventral measurements of the tongue length, width and height and their mul plica on product to calculate tongue volume which was different and less me consuming as to the method used, where in the authors used product of tongue cross sec onal area(by tracking the borders of the tongue ) and width.

Poudel D et al
In a study conducted by Rana et al. in 100 pa ents, using point of care ultrasound found that HMDR had strong nega ve correla on with CL grading with an area under the curve (AUC) of 0.871 and regression coefficient of − 0.466 8 (95% CI : -0.956 to -0.786) .These findings were similar to our study which showed similar nega ve correla on with regression coefficient of -0.410.In a similar study sta s cally significant difference between HMDR in easy and difficult laryngoscopy groups was noted 9 which is same as that in our study.Various studies have been conducted to assess so ssue thickness at the anterior neck and its significance in predic ng difficult airway.In a study conducted by Yadav NK et al. the sensi vity and specificity of so ssue thickness at the level of hyoid bone was 68% and 73% respec vely, in 20 neutral posi on.The AUC for the same was 0.72(95% CI: 0.61-0.82).In our study, the AUC for DSHB was found to be 0.732(95% CI: 0.636-0.872,p=0.0014), sensi vity of 67.75% and specificity of 66% which are comparable.In a pilot study conducted by Adhikari et al. thickness of so ssue at the hyoid bone and thyrohyoid membrane level 21 was greater in pa ents with difficult laryngoscopy.In a study conducted by Wu et al, the predic ve value of anterior neck so ssue mass at the level of hyoid, thyrohyoid membrane and anterior commisure and found these to be 22 independent predictors of difficult laryngoscopy.

CONCLUSIONS
This study demonstrate that ultrasound derived parameters may be useful to predict difficult laryngoscopy and difficult airway.Tongue volume and hyomental distance ra o can have moderate predic ve value and can be evaluated using point of care ultrasonography Further studies are needed to develop a model incorpora ng both clinical and ultrasonographic measurements.

LIMITATIONS OF THE STUDY
There are several limita ons in our study.Due to the low incidence of difficult airway, the number pa ents in difficult and easy laryngoscopy group were not equal.This inequality might have some effect on the predic ve validity of the tests used.Furthermore ultrasonography is a user dependent tool and some bias might have crept in.The conduct of laryngoscopy and visualiza on of glo s is a complex procedure and many factors like skill and experience of anesthe st, presence of secre ons and anatomic variability might play a role.
physical predictors of difficult airway have been in use with varying degree of significance.The most commonly used airway assessment parameters include Modified Mallampa Score(MMP), Upper Lip Bite Test, Inter-Incisor Gap, Thyromental Distance, Hyomental distance, Sternomental distance, Neck Mobility and 4 Among the clinical parameters, MMP grade had sensi vity of 89.9%, specificity of 54.8%, Posi ve predic ve value (PPV) of 92%, Nega ve predic ve value (NPV) of 70.8%.The receiver opera ng curve (ROC) was plo ed which showed area under curve (AUC) of 0.753, which was highest among the clinical parameters.The Upper Lip Bite Test (ULBT) had sensi vity of 88.2%, specificity of 19.4%, Posi ve predic ve value (PPV) of 86.9%, Nega ve predic ve value (NPV) of 66.7%.ROC was plo ed which showed AUC of 0.588.The Thyromental Distance (TMD) had sensi vity of 80.9%, specificity of 22.6%, Posi ve predic ve value (PPV) of 86.4%, Nega ve predic ve value (NPV) of 29.2% and ROC showed AUC of 0.563.

Table 1 :
Demographic distribu on of pa ents Original Research Ar cle Poudel D et al anesthesia was maintained and surgery proceeded according to the standard protocol.Difficulty in intuba on, change of blade or use of external manipula on or gum elas c bougie or CVCI scenarios (Could not Ven late, Could not Intubate) was noted.The CVCI scenarios are among the most dreaded condi ons where in the bag and mask or endotracheal intuba on are impossible and may result in significant hypoxia with organ damage or mortality.

Table 2 :
Various parameters used for airway evalua on