Evaluation of accuracy of cervical auscultation for clinical assessment of dysphagia

Cervical auscultation (CA) is frequently used to clinically assess dysphagia. CA can distinguish respiratory and swallowing sounds in dysphagic patients from those in normal subjects. This study was done to evaluate the accuracy of CA.Thirty-seven patients with head and neck cancer were studied. Respiratory and swallowing sounds were detected by two accelerometers and recorded on VHS tape with videofluorographic (VF) images.Forty-four sound samples were edited and presented to six examiners through headphones. Each sound sample included expiratory sounds before swallowing, sounds produced by swallowing 5 ml barium, and expiratory sounds immediately after swallowing.Without referring to VF images, each examiner evaluated and classified the sound samples into one of five categories: normal (N), acceptable (Ac), residue (R), penetration (P), and aspiration (A).To evaluate the accuracy of CA, we assessed the rate of agreement between auditory evaluation and VF images.The following results were obtained:1. The average accuracy was 55.3%, but the agreement rate of auditory evaluation with VF images within an error of one category averaged 79.5%.2. When N and Ac were regarded to have no dysphagia, and R, P, and A were regarded to have dysphagia, the agreement rate between auditory evaluation and VF images averaged 83.5% in the two groups.3. The agreement rate of A with coughing averaged 89.6%, and the agreement rate of A without coughing averaged 34.8%. However, the agreement rate of auditory evaluation of A without coughing with VF images averaged 76.5% for dysphagia. These results show the effectiveness of CA as a clinical tool for diagnosing dysphagia.