Acoustic Measures of Voice Perturbation Offer Limited Value as Standalone Indicators of Laryngeal Penetration or Aspiration.

Abstract

OBJECTIVE: To identify which, if any, acoustic measures are associated with swallowing safety. DESIGN: Prospective, original research. SETTING: Primarily outpatient. PARTICIPANTS: A total of 34 adults (N=34) referred for a videofluoroscopic swallowing study (VFSS) were audio-recorded completing a sustained vowel and reading a sentence before the VFSS and again after swallowing each bolus. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Swallowing safety was initially quantified using the Penetration-Aspiration Scale with ratings then defined dichotomously (eg, no penetration/aspiration vs penetration and/or aspiration). Acoustic measurements were calculated in Praat and included jitter (relative average perturbation [RAP]), shimmer (local), noise-to-harmonic ratio, and cepstral peak prominence (CPP). RESULTS: Before bolus swallowing, RAP and shimmer (local) predicted penetration or aspiration (P=.036, P=.046, respectively) when included in a model with diagnosis and sex, whereas changes in CPP significantly predicted penetration or aspiration after bolus swallowing (P=.02) when included in a model with diagnosis and bolus consistency. However, receiver operating characteristic analyses were poor (P>.05), indicating these acoustic measures did not perform well as independent classifiers of airway invasion. CONCLUSIONS: Although RAP, shimmer (local), and CPP may predict subsequent airway invasion in a larger model, they do not appear to be effective predictors when used alone. Noise-to-harmonic ratio was not predictive of penetration or aspiration in any of the models. Results suggest that acoustic voice features are insufficient as standalone indicators of airway invasion risk or events.