Voice pathology: Assessment of Voice and Analysis of the Disease Burden, Chaitanya Gadepalli, Doctor of Medicine (MD), The University of Manchester.2016. Aims: The global aim of this project is to investigate what we can learn from existing and new voice quality assessment techniques. The objectives are as follows: 1) To investigate problems arising from voice quality assessment by professionals, such as inter and intra rater consistency. 2) To investigate whether measurements of voice quality according to the GRBAS scale can be made by computer. 3) To investigate the impact of self-reported reflux symptoms, on voice quality. 4) To compare the occurrence of voice problems in teachers and non-teachers using self-assessment questionnaires. Methods: A database of voice samples was collected from 107 subjects in a quiet room using standardised digital recording equipment. The subjects were also asked to provide self-assessments of their Voice Handicap Index (VHI-10) and Reflux Symptom Index (RSI) scores. A group of five experienced Speech & Language Therapists (SLTs) were recruited to independently give GRBAS scores to these voice samples. Our own specially designed software presented the voice samples in randomised order to each rater and allowed the rater to enter the scores. About 20% of the samples were repeated randomly to allow both inter and intra-rater consistency to be assessed. To compare the occurrence of voice problems in teachers and non-teachers, questionnaires were distributed in person and via survey monkey. 454 teachers from 44 schools and 304 non-teachers participated. The response rate was 67.9 % from primary schools, 41.2 % from secondary schools and 40% from the non-teachers. Questions consisted of general demographics, VHI-10 and questions relating to voice problems. The survey allowed us to compare the results of self-assessment with voice quality assessment by professionals and computer. Results: The consistency of GRBAS scoring by our trained professionals was found to be substantial according to Cohen Kappa and excellent according to the Intra Class Correlation (ICC) coefficient for all GRBAS components. The computer program developed in collaboration with the School of Computer Science produced encouraging results, showing that GRBAS scores produced by a computer were close to the gold standard scores in statistical cross-validation testing. It was found that VHI-10 and RSI scores correlated strongly with the Grade component of our gold standard GRBAS scores. All components of GRBAS, except roughness, showed strong correlation with total VHI-10 score. Roughness (R) showed only moderate correlation. In the survey of voice problems, 30% of teachers and 9% of non-teachers had reported some problems with their voice. VHI-10 scores showed that 27% of teachers and 44.7 % of non-teachers did not perceive themselves as having voice problems. Conclusion: The consistency of GRBAS scoring by our trained professionals was found to be reasonable and the VHI-10 and RSI self-assessments were found to be reasonably consistent with the GRBAS scoring. This gives us confidence in both the GRBAS scoring and the self-assessments. Our results have also established the feasibility of GRBAS voice quality analysis by computer. Our VHI-10 self-assessments indicated that voice disorders are an occupational health issue for teachers, causing a disease burden to a significant number of teachers within the group we studied. We have identified risk factors in teachers.
|Date of Award||31 Dec 2017|
- The University of Manchester
|Supervisor||Barry Cheetham (Supervisor) & Jarrod Homer (Supervisor)|
- Disease burden