The Voice Analysis as a Preoperative Prediction Method of a Difficult Airway

Description

Proper airway management is essential for safe anesthesia practice. In particular, ventilation with an orofacial mask and endotracheal intubation are a crucial step in general anesthesia. The prevalence of a difficult airway is approximately 2.2% although it can vary greatly according to different groups of patients. It is defined, according to the American Society of Anesthesiology, as the clinical situation in which an experienced anesthesiologist has difficulty with ventilation using a face mask, during endotracheal intubation or both. Although other definitions also exist and include difficulty in ventilation or placement of supraglottic devices, difficult laryngoscopy defined by a 3 or 4 grade on the Cormack-Lehane scale or multiple attempts at intubation. A case of an unexpected difficult airway can lead to minor complications, such as loss or damage to teeth, pharynx or entail severe injuries, such as aspiration of gastric content, laryngeal lesions, respiratory failure, cardiorespiratory arrest or even death. Furthermore, complications increase considerably when there are no alternative plans and multiple repeated attempts at intubation are made.

Thus, the factors that mostly contribute to complications when an unanticipated difficult airway is found are due to deficiencies in identification, communication, preoperative planning and lack of training. An unpredicted difficult airway is a situation in which decision making is difficult, management is complex, and it is considered a crisis situation.

The importance of evaluation and management of the airway is such that different study groups within the main scientific societies of anesthesiologists are dedicated to studying and implementing protocols with the aim of avoiding these risky situations. A difficult airway should be detected during the pre-anesthetic visit, thus allowing the anesthesiologist to evaluate the patient, identify risk factors, advance to possible complications, and prepare an appropriate anesthetic plan. This means organizing both material and personal help, informing the patient; changing the type of anesthesia or even postponing elective surgeries to schedule them when the patient is correctly optimized making sure that the means in which the surgical procedure is safe.

Regardless of the surgical procedure and the initial anesthetic plan, the evaluation of the airway should always be performed with all patients since any sedation or regional anesthesia can be converted into a general anesthesia and require control of the airway.

In recent years, an increase in the efficiency of health procedures without the loss of safety or quality is expected. An attempt to protocolize a remote preoperative evaluation (telephone call) has been made which patients undergoing low complexity procedures. However, the most important limiting factor in remote pre-anesthetic evaluation is the inability to perform the airway assessment properly. Although there are risk factors that can be detected with a medical history such as a previous record of a difficult airway, obstructive sleep apnea syndrome, hoarseness, dysphonia or obesity, the traditional evaluative predictive tests require the physical presence of the patient and are based on the physical examination of the patient such as thyroid-chin distance, cervical mobility, mouth opening and subluxation capacity, or the assessment of pharyngeal structures by the Mallampati test consisting of the direct visualization of the upper airway. However, no test can predict a difficult airway based on a single characteristic in the patient's absence.

The combined evaluation of several risk factors obtains greater sensitivity than when analyzed in isolation. This is the case of the Arn Test, which consists of a multifactorial index that offers a sensitivity and specificity greater than 90% to predict a difficult airway when a score greater than or equal to 11 is obtained. In this context, new evaluation methods have appeared, such as ultrasound or other imaging tests, which intend to correlate the anatomical images of the airway with the presence of a difficult airway. But, the detection of a difficult airway depends not only on the morphology, but also on its functional characteristics. With the introduction of new technologies many attempts to develop other methods to predict a difficult airway such as facial recognition based on image processing have been made, however they have not succeeded yet. Other investigators propose the analysis of voice parameters as a reflection of anatomical and functional characteristics of the superior airway. Specifically, studies carried out in the field of maxillofacial surgery describe how the expansion of the maxilla affects the widening of the upper airway and, consequently, in the formation of the vowels and how this translates into a variation in the properties of the voice, such as frequency or amplitude. Accordingly, the investigators propose that voice characteristic analysis could reflect the airway's anatomy and be able to predict a difficult airway.

Building on this, the investigators aim to develop a voice airway assessment method that replaces anthropometric parameters evaluated in traditional tests to predict a difficult airway, facilitating a remote airway evaluation.

Voice recording will be made through a smart phone application to patients who are going to undergo general anesthesia and require orotracheal or nasotracheal intubation, by direct laryngoscopy, at the pre-anesthetic visit. The day of the surgical procedure, the result of the intubation - whether or not a difficult airway exists- will be registered.

The records of the mobile application database will be downloaded, the voice signal will be processed, and parameters related to frequency, morphology and perturbation will be extracted employing Matlab as these are considered continuous variables, to determine the statistical significance of the differences within parameters. The non-parametric test of Kolmogorov-Smirnov will be used for an easy and a difficult independent airway groups. Thus, these variables will be introduced into several classification algorithms obtained by combined methods using machine learning in order to predict the classification of patients according to the Cormack scale grade and sensitivity and specificity will be determined to assess their ability to predict a difficult airway. The area under the receiver operating characteristic curve will be used to assess the ability of the method to predict difficulty.

The recruitment period will take place over a period of 12 months or until the estimated sample size is reached. The total duration of the study will be approximately 6 months years, ending with a total of 800 patients.

Details