Comparison of electromyography, sound, bioimpedance, and high-resolution manometry for differentiating swallowing and vocalization events.

Abstract

Non-invasive surface recording devices used for detecting swallowing events include electromyography (EMG), sound, and bioimpedance. However, to our knowledge there are no comparative studies in which these waveforms were recorded simultaneously. We assessed the accuracy and efficiency of high-resolution manometry (HRM) topography, EMG, sound, and bioimpedance waveforms, for identifying swallowing events. Six participants randomly performed saliva swallow or vocalization of "ah" 62 times. Pharyngeal pressure data were obtained using an HRM catheter. EMG, sound, and bioimpedance data were recorded using surface devices on the neck. Six examiners independently judged whether the four measurement tools indicated a saliva swallow or vocalization. Statistical analyses included the Cochrane's Q test with Bonferroni correction and the Fleiss' kappa coefficient. Classification accuracy was significantly different between the four measurement methods (P<0.001). The highest classification accuracy was for HRM topography (>99%), followed by sound and bioimpedance waveforms (98%), then EMG waveform (97%). The Fleiss' kappa value was highest for HRM topography, followed by bioimpedance, sound, and then EMG waveforms. Classification accuracy of the EMG waveform showed the greatest difference between certified otorhinolaryngologists (experienced examiners) and non-physicians (naive examiners). HRM, EMG, sound, and bioimpedance have fairly reliable discrimination capabilities for swallowing and non-swallowing events. User experience with EMG may increase identification and interrater reliability. Non-invasive sound, bioimpedance, and EMG are potential methods for counting swallowing events in screening for dysphagia, although further study is needed.