The interobserver and test-retest variability of the dysphonia severity index.

OBJECTIVE The purpose of this study was to investigate the interobserver variability and the test-retest variability of the Dysphonia Severity Index (DSI), a multiparametric instrument to assess voice quality. METHODS The DSI was measured in 30 nonsmoking volunteers without voice complaints or voice disorders by two speech pathologists. The subjects were measured on 3 different days, with an interval of 1 week. RESULTS The difference in DSI between two observers (interobserver difference) was not significant. The intraclass correlation coefficient for the DSI was 0.79. The standard deviation of the difference between two duplicate measurements by different observers was 1.27. CONCLUSION Differences in measurements between different observers were not significant. The intraclass correlation coefficient of the DSI was 0.79, which is to be considered good. Differences in DSI within one patient need to be larger than 2.49 to be significant.