Abstract:
OBJECTIVE : A new approach to classify oral–nasal balance disorders based on instrumental measurements was developed based on linear discriminant analysis (LDA) of nasalance scores of simulated oral–nasal balance disorders by de Boer and Bressmann. The current study aimed to apply the newly developed functions to clinical data to investigate the applicability of this new approach.
DESIGN : Retrospective diagnostic accuracy study.
SETTING : Tertiary university hospital.
PARTICIPANTS : Fifty-five Dutch-speaking Flemish children (age 4-12 years) with normal (n = 20), hypernasal (n = 18), hyponasal (n = 12), or mixed nasality (n = 5).
INTERVENTIONS : Nasalance scores of an oral and a nasal text were used to calculate 3 sets of LDA function scores. Predicted classification was consecutively based on the function values of the group centroids originally determined by de Boer and Bressmann and adapted LDA functions and group centroids based on clinical data.
MAIN OUTCOME MEASURES : Discriminatory power of the linear discriminant formulas.
RESULTS : Based on the original LDA functions, 56% of the speech samples matched the perceptual classification. Applying a correction factor for age and language differences resulted in a 67% correct classification, although 83% of the hyponasal samples were ranked as “normal resonance.” Rederivation of the LDA functions based on current clinical data resulted in an 80% correct classification.
CONCLUSIONS : The new approach of classifying oral–nasal balance disorders based on a combination of nasalance scores was promising. However, further clinical research is needed to refine the LDA functions and group centroids before clinical application is possible.
