Acoustic and Perceptual Parameters of Swallow Sounds Associated with and Without Aspiration in Preterm Neonates: A...

Cervical auscultation (CA), used an adjuvant to the clinical feeding examination by Speech-Language Pathologists (SLPs), has high sensitivity and reliability for detecting aspiration in neonates and children. CA can only be considered reliable if there are clear distinctions in the acoustic and perceptual swallow sound profiles between swallow sounds associated with and without aspiration. We aimed to determine whether the acoustic and perceptual swallow sound profile differs between swallows associated with and without aspiration in preterm neonates. We simultaneously recorded video and swallow sounds during videofluoroscopic swallow studies. We used the Penetration-Aspiration Scale to define aspiration. Acoustic parameters were obtained using a Python program. Two SLPs independently rated the presence/absence of perceptual parameters from 1 s audio wavefiles. Differences between swallow types were analyzed using mixed linear regression modelling to account for clustered data, and a test of two proportions for acoustic and perceptual parameters, respectively. Penalized logistic regression analysis was used for perceptual sound predictors of aspiration. A total of 293 swallows (7 aspiration, 286 no aspiration) from 13 preterm neonates [69%; median birth gestation 31 (range = 24-36 weeks)] were analyzed. There were significant differences between these two swallow types for both acoustic and perceptual swallow sound parameters. Combinations of perceptual sound parameters (wet/stridor pre-swallow sounds; quick, loud, initial discrete sound, bolus transit sound, final discrete sound, glottal release sound during swallow sounds; and wheeze post-swallow sounds were most predictive of aspiration (area under the curve, 0.82: 95%CI0.60-1.00). In conjunction with a clinical feeding examination, CA-obtained sound features can likely help guide SLP's evaluation of aspiration risk in preterm neonates.