Introduction
Difficult mask ventilation (DMV) is a major challenge in routine anesthetic practice and its prediction during preoperative visits facilitates planning of appropriate airway management [1,2]. The risk of DMV is increased in cases of obesity [3]. Patients with obesity have reduced oxygen reserve [4], which makes them more vulnerable to serious hypoxia in the case of DMV. Predicting DMV, particularly in patients with obesity, by using bedside tests remains a key focus in the field of anesthesia. Guidelines for airway evaluation are available and are updated regularly; however, unanticipated DMV is still encountered in some cases, and no reliable measure for ruling out DMV is available at present [5].
Ultrasound-derived airway measurement has gained popularity in recent years. These metrics reflect anterior neck soft tissue thickness and have shown good ability for predicting difficult intubation [6,7]. However, data on their ability to predict DMV are lacking. Among these measurements is the skin-to-hyoid distance (SHD), which have been investigated as a predictor of DMV but yielded conflicting results in populations without obesity [8,9]. Nevertheless, data regarding its predictive value in patients with obesity remain scarce. We hypothesized that the SHD could accurately predict DMV in this population.
Thus, in this study, we assessed the ability of ultrasound-measured SHD to predict DMV in patients with obesity.
Materials and Methods
This prospective observational study was conducted at a university hospital from October 2023 to December 2024, following approval from Cairo University Research Ethics Committee (MD-229-2023). Written informed consent was obtained from all participants before their enrollment in the study. This study was performed in line with the principles of the Declaration of Helsinki (2013).
Participants were adult patients with a body mass index (BMI) > 35 kg/m2 and American Society of Anesthesiologists physical status (ASA-PS) of II–III, scheduled for elective surgery under general anesthesia.
We excluded cases in whom intravenous induction of anesthesia and mask ventilation were not planned in accordance with the ASA guidelines. This applied to patients who required awake tracheal intubation from the beginning, such as patients with a documented history of failed intubation and those with apparent airway abnormalities (e.g., airway mass). Patients with suspected difficult laryngoscopy and who met 1 of the following criteria were excluded: suspected DMV, risk of rapid desaturation, or risk of aspiration of gastric contents. Furthermore, edentulous patients and pregnant women were excluded from the study.
In the pre-anesthetic room, an experienced anesthetist assessed the patients’ airways. The following assessments were performed: modified Mallampati test, thyromental distance measurement, sternomental distance measurement [10], upper lip bite test [11], mouth opening [12], and neck mobility. The STOP-Bang questionnaire [13], which consists of 8 questions related to the presence of snoring, tiredness, observed apnea, high blood pressure, BMI > 35 kg/m2, age > 50 years, neck circumference > 40 cm, and male sex, was administered. Each positive response scored 1 point.
Ultrasound examination was performed by an experienced anesthetist (who had performed at least 50 similar examinations previously) using a Butterfly iQ handheld ultrasound (Butterfly Network, Inc.) connected to an iPhone. This anesthetist was blinded to the findings of the airway tests. Ultrasound examination was performed with the patient in the supine position and the head in the neutral position (Fig. 1). The operator performed a craniocaudal scan of the neck with the probe placed on the transverse axis. The hyoid bone was identified, and the minimum distance between the skin and hyoid bone was measured (Fig. 1). This measurement was repeated 3 times, and the values were averaged for each patient.
In the operating room, standard monitors were applied. Pre-oxygenation was provided for 3 min before induction of anesthesia, with the head in the ramped position. Anesthesia was induced with 2 mg/kg propofol and 2 μg/kg fentanyl. Rocuronium (0.6 mg/kg) was administered after loss of consciousness. The adequacy of mask ventilation was assessed after neuromuscular blockade administration. All drug dosages were based on the patient’s lean body weight.
Mask ventilation was performed for 2 min using an appropriately sized face mask. The adequacy of face mask ventilation was confirmed by appropriate chest expansion and the presence of a capnographic wave. Tracheal intubation was performed with the patient in the sniffing position, using an appropriately sized Macintosh blade. The correct placement of the tracheal tube was confirmed by observing an appropriate capnographic wave. Mask ventilation and tracheal intubation were performed by an experienced anesthetist (> 3 years’ experience) who was blinded to the ultrasound measurement results. External laryngeal manipulation was performed as required.
The difficulty of mask ventilation was evaluated using a 4-level Han scale [14]: grade 1, easily ventilated with a mask; grade 2, ventilated by mask with the aid of an oral airway/adjuvant; grade 3, ventilation with the mask was inadequate, unstable, or required 2 providers; and grade 4, inability to provide ventilation by mask. Grades 3 and 4 were considered DMV.
The Cormack–Lehane classification of the laryngoscopic view [15] and items for assessing the Intubation Difficulty Scale (IDS) [16] were recorded. A Cormack–Lehane classification > 2 was considered as difficult laryngoscopy, and an IDS score > 5 was considered as difficult intubation.
The incidences of difficult laryngoscopy and intubation were determined. Demographic data, including patient age, sex, weight, BMI, ASA-PS, and the presence of beards, were also recorded.
The primary outcome was the ability of the SHD to predict DMV. Secondary outcomes were identifying predictors of DMV and comparing the ability of SHD to predict DMV with other predictors.
Statistical Analysis
The sample size was calculated using MedCalc software version 14 (MedCalc software bvba). The calculation was based on detecting an area under the receiver operating characteristic curve (AUC) of 0.75, with a null hypothesis AUC of 0.5, an expected incidence of DMV of 5% [17], a power of 90%, and a significance level (alpha) of 0.05. Therefore, a minimum of 300 patients, including at least 15 cases of DMV, was required.
Statistical Package for the Social Sciences (SPSS) (version 23) for Microsoft Windows (IBM Corp.), MedCalc (version 14), and R software (version 4.4.1, https://www.r-project.org/) were used for data analysis. Patients were divided according to the difficulty of mask ventilation (easy mask ventilation and DMV). Categorical data are expressed as frequency (%) and were analyzed using the chi-square test or Fisher’s exact test, as deemed appropriate. Continuous data were tested for normality using the Shapiro–Wilk test and are presented as either mean ± standard deviation or median (quartiles) according to the data distribution. Univariate analysis was performed to identify predictors of DMV. Odds ratios and 95% CIs for these predictors were determined. When complete separation was detected (i.e., presence of a beard), Firth’s penalized likelihood approach was applied using the logistf package in R. Multivariate logistic regression analysis was conducted using the STOP-Bang score, modified Mallampati test, upper lip bite test, and SHD to identify the independent predictors of DMV. The ability to predict DMV was determined using AUC analysis. Values with the highest sensitivity and specificity as determined using the Youden index were identified. Positive- and negative predictive values, and positive- and negative likelihood ratios were also calculated. The AUC of different predictors was compared using the DeLong test. Internal cross-validation of the prediction of DMV by the SHD was performed using R software. To ensure a robust evaluation of the model, we performed a stratified 5-fold cross-validation, which was repeated 10 times. This approach was chosen to stabilize the performance estimates, particularly given the small number of DMV cases (n = 22). Stratified sampling was applied to maintain class distribution across folds. For each fold, a generalized linear model with binomial logistic regression was trained using SHD as the predictor variable. The model was assessed based on its ability to distinguish between DMV and non-DMV cases, with the AUC as the primary performance metric. The intraclass correlation coefficient (ICC) and 95% CI were calculated for the 3 SHD measurements obtained for each patient (absolute agreement, 2-way model). Statistical significance was set at P < 0.05.
Results
Of the 349 patients screened for eligibility, 23 were excluded; thus, 326 patients were included in the study and their data were available for the final analysis (Fig. 2). The number of patients with DMV, difficult laryngoscopy, and difficult intubation was 22/326 (6.7%), 31/326 (9.5%), and 8/326 (2.5%), respectively. One participant was classified as grade 4 on the Han scale for DMV. The number of patients with severe obesity (BMI 35.0–39.9 kg/m2), morbid obesity (BMI 40.0–49.9 kg/m2), and super-obesity (BMI ≥ 50.0 kg/m2) was 115/326 (35.3%), 147/326 (45.1%), and 64/326 (19.6%), respectively.
Patients with DMV were predominantly male and had higher weight, ASA classification, STOP-Bang score, modified Mallampati grade, upper lip bite class, and SHD than those with easy mask ventilation (Table 1). Multivariate analysis showed that the SHD, in addition to the STOP-Bang score and modified Mallampati grade, is an independent predictor of DMV (Table 2).
The AUC (95% CI) for predicting DMV by the SHD was 0.88 (0.84–0.92). An SHD > 1.9 cm had a positive-predictive value of 27%, while its negative predictive value was 99% (for SHD ≤ 1.9 cm) (Table 3 and Fig. 3). Using stratified cross-validation, the cross-validated AUC (95% CI) for SHD for predicting DMV was 0.88 (0.86–0.90). The AUC of the SHD was higher than that of the upper lip bite test (P = 0.003) and was comparable to those of the STOP-Bang score and the modified Mallampati test.
The ICC (95% CI) for the SHD was 0.99 (0.99–0.99).
Discussion
In this study, we revealed that ultrasound-measured SHD can predict DMV in patients with obesity. The SHD was greater in patients with DMV than in those with easy mask ventilation. The SHD showed good ability to predict DMV. An SHD ≤ 1.9 cm can exclude DMV with 99% accuracy. The high accuracy of SHD in predicting DMV could be related to the fact that the hyoid bone is connected to the tongue and larynx; therefore, increased soft tissue deposits in this area would decrease airway compliance and increase the risk of airway obstruction after anesthesia induction.
Data assessing the accuracy of the SHD in patients with obesity are scarce. The results of a recent observational study by Tasdemir et al. [8] (n = 157) on the ability of the SHD to predict DMV in patients with obesity differed from ours. Tasdemir et al. reported lower accuracy (Tasdemir et al. AUC 0.70 versus 0.88 in our study) and a lower cutoff value (Tasdemir et al. 1.7 cm versus 1.9 cm in our study) than those determined in our study. The difference between our results and those of Tasdemir et al. may be related to the larger sample size used in our study, which also allowed for multivariate analysis and cross-validation of AUC analysis. Additionally, in the present study, we included patients with BMI > 35 kg/m2 because of the strong association thereof with obstructive sleep apnea, which is a major risk factor for DMV. In contrast, Tasdemir et al. included patients with BMI > 30 kg/m2. Consequently, our study had a higher prevalence of morbid obesity and super-obesity (211/326 [64.7%] in our study vs. 14/157 [8.9%] in Tasdemir et al.’s study), which may have contributed to the differences in results.
In patients without obesity, the accuracy of the SHD for predicting DMV remains controversial. Alessandri et al. [18] reported that the SHD had excellent accuracy (AUC = 0.93) for predicting DMV, whereas Bianchini et al. [9] and Anchalee et al. [19] reported that the SHD was not an independent predictor of DMV. Furthermore, the average SHD in patients with DMV in the study by Alessandri et al. was 1.14 ± 0.19 cm, which was lower than that reported in our study (2.01 ± 0.25 cm) and that of Tasdemir et al. (1.77 ± 0.36 cm). This suggests that the cutoff value differs between patients with and without obesity. Therefore, further studies are needed to confirm the cutoff value of SHD for predicting DMV across different BMI ranges.
We also report that the SHD demonstrated accuracy comparable to that of the STOP-Bang score for predicting DMV, but with higher sensitivity, which is a crucial factor in airway tests to minimize false-negative cases. Additionally, SHD measurement via ultrasound has a short learning curve and the measurement can be completed in less than a minute, even by trainees [20,21]. Unlike the STOP-Bang questionnaire, the SHD measurement does not require patient cooperation, making it applicable in other settings where patient engagement may be limited. This feature enhances the versatility of SHD and allows for effective assessments in diverse clinical situations.
Future studies are needed to determine whether a combination of the SHD and STOP-Bang score or any of their components could provide better accuracy in predicting DMV. DMV is a critical airway problem, and its seriousness increases if it is unanticipated before anesthesia. Thus, preoperative evaluation of the risk of DMV is essential for establishing an appropriate plan for airway management [2]. No single test can currently identify patients at risk of DMV; therefore, several airway features and measurements should be assessed. Ultrasound-derived airway measurements are recently gaining interest for the simplicity of these assessments, which allows bedside evaluation [2]. Several ultrasound measurements for airway evaluation have been introduced; however, none have been incorporated into existing guidelines. Furthermore, most airway ultrasound studies have focused on difficult laryngoscopy/intubation, and data for predicting DMV are scarce [22]. Obesity is a known DMV risk factor [23], highlighting the importance of assessing the risk of DMV in this population. The current study suggested that the SHD can be used to identify patients at risk of DMV. Having high negative predictive value makes SHD measurement an excellent screening test, which is of great importance in airway assessment, as false-negative cases (unanticipated DMV) should be avoided as much as possible. Our data showed that an SHD ≤ 1.9 cm can exclude DMV with 99% accuracy. Furthermore, SHD had the lowest negative likelihood ratio among the assessed airway tests, emphasizing its value for ruling out likely DMV. A negative likelihood ratio of 0.16 indicates that, if the SHD is ≤ 1.9 cm, the probability of encountering DMV decreases from 6.7% to approximately 1.1%.
This study had several advantages. First, this was the largest study to date to assess SHD in predicting DMV. Second, we strictly included patients with obesity, and more than 60% of our patients were classified as morbidly obese or super-obese. Third, we introduced the use of a handheld ultrasound probe for airway evaluation and demonstrated excellent intra-rater variability. Handheld ultrasound probes have several advantages as they are smaller in size than conventional ultrasound machines, increasing the feasibility of the examination, in addition to its “plug-and-play” properties, allowing measurement within a few seconds after switching on. Fourth, stratified k-fold cross-validation was used to validate the ability of the SHD to predict DMV. This approach reduces the risk of overfitting and provides a more generalizable estimate of model performance than does a single-sample AUC analysis. The high cross-validated AUC (0.88, 95% CI [0.86–0.90]) further supported the reliability of SHD measurement as a diagnostic test. While external validation remains ideal, these findings indicate the strong predictive performance of the SHD.
This study had some limitations. It was conducted at a single center. Additionally, we only included patients who underwent elective surgery. Therefore, further studies are required to confirm our findings in other populations and settings.
In conclusion, in patients with obesity, SHD measured using a handheld ultrasound probe is an independent predictor of DMV and can accurately predict DMV. An SHD ≤ 1.9 cm can exclude DMV with 99% accuracy.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
