Nasalance-Based Preclassification of Oral–Nasal Balance Disorders Results in Higher Agreement of Expert Listeners’ Auditory-Perceptual Assessments: Results of a Retrospective Listening Study

Abstract

Objective: Reliable perceptual and instrumental assessment of oral–nasal balance disorders is a persistent problem in speech-language pathology. The goal of the study was to evaluate whether nasalance-based preclassification of oral–nasal balance disorders improves listener agreement. Design: Retrospective listening study. Setting: Tertiary university hospital. Participants: Fifty-four randomly selected recordings of patients with repaired unilateral cleft lip and palate (UCLP). Three experienced speech-language pathologists participated as expert listeners. Interventions: Two listening experiments were based on nasalance scores and audio recordings of speakers with repaired UCLP. The speakers were preclassified as normal, hypernasal, hyponasal, or mixed based on their nasalance scores. Initially, the listeners determined the diagnostic category of the oral–nasal balance for 62 audio recordings (8 repeats). Six months later, they listened to 38 of the recordings (6 repeats) along with a spreadsheet indicating the nasalance-based categories for the oral–nasal balance. The listeners confirmed, or rejected and corrected, the nasalance-based preclassification. Main Outcome Measures: Intralistener, interlistener agreement, and agreement between listener categories and nasalance-based oral–nasal balance categories. Results: In the first study, the agreement between the listeners’ diagnostic category and the nasalance-based category was 45.1% and the interlistener agreement was 36.7%. In the second study, the agreement between the listeners’ category and the nasalance-based category was 67.1% (75% agreement for the correct nasalance-based categories and 41.7% for the misclassifications), and the interlistener agreement was 85.4%. Conclusions: Preclassification of oral–nasal balance disorders based on nasalance scores may help listeners achieve better diagnostic accuracy and higher agreement.