Reliability and Validity of Speech Evaluation in Adductor Spasmodic Dysphonia

Abstract

The aim of this study was to evaluate speech in patients with adductor spasmodic dysphonia (ADSD) by perceptual evaluations and acoustic measures, and to examine the reliability and validity of these measures. Twenty-four patients with ADSD and 24 healthy volunteers were included in the study. Speech materials consisted of three sentences constructed from serial voiced syllables to elicit abductor voice breaks. Three otolaryngologists rated the degree of voice symptoms using a visual analog scale (VAS). VAS sheets with five 100-mm horizontal lines were given to each rater. The ends of the lines were labeled normal vs severe, and the five lines were labeled as overall severity of each of the four speech symptoms (strangulation, interruption, tremor and strained speech). Nine words were selected for acoustic analysis, and abnormal acoustic events were classified into one of the three categories. To evaluate the intra- and inter-rater and intermeasurer reliabilities of the VAS scores or acoustic measures, Pearson r correlations were calculated. To examine the validity of perceptual evaluations and acoustic measures, the sensitivity and the specificity were calculated. Pearson r correlation coefficients for overall severity showed the highest intra- and inter-rater reliabilities. For acoustic events, the intrameasurer reliabilities were r = .645 (frequency shifts), r = .969 (aperiodic segments), and r = 1.0 (phonation breaks), and the intermeasurer reliability ranged from r = .102 to r = 1.0. Perceptual evaluation showed high sensitivity (91.7%) and specificity (100%), whereas acoustic analysis showed low sensitivity (70.8%) and high specificity (100%). Both perceptual evaluation and acoustic measures alone were found likely to overlook patients with true ADSD.