Reliability of perceptions of voice quality: evidence from a problem asthma clinic population.

INTRODUCTION Methods of perceptual voice evaluation have yet to achieve satisfactory consistency; complete acceptance of a recognised clinical protocol is still some way off. MATERIALS AND METHODS Three speech and language therapists rated the voices of 43 patients attending the problem asthma clinic of a teaching hospital, according to the grade-roughness-breathiness-asthenicity-strain (GRBAS) scale and other perceptual categories. RESULTS AND ANALYSIS Use of the GRBAS scale achieved only a 64.7 per cent inter-rater reliability and a 69.6 per cent intra-rater reliability for the grade component. One rater achieved a higher degree of consistency. Improved concordance on the GRBAS scale was observed for subjects with laryngeal abnormalities. Raters failed to reach any useful level of agreement in the other categories employed, except for perceived gender. DISCUSSION These results should sound a note of caution regarding routine adoption of the GRBAS scale for characterising voice quality for clinical purposes. The importance of training and the use of perceptual anchors for reliable perceptual rating need to be further investigated.