A report on the development and validation of first-ever perceptual speech evaluation tool for head and neck cancer (HNC) patients.

Abstract

9116 Background: Although there are many perceptual voice evaluation scales available but to date there is no scale available for speech evaluation in HNC patients. The aim of this study was to develop and validate first ever speech-specific scale for perceptual evaluation of speech in HNC patients. Methods: Based on an extensive literature review and discussion in a multidisciplinary setting 5 speech parameters (intelligibility, articulation, speech rate, nasality and asthenia) and overall grade of speech impairment were selected and evaluated for the development and validation of this scale. Speech samples of 117 subjects (65 HNC patients and 52 age and sex matched healthy volunteers) were recorded using a standard protocol consisting reciting specific words and a standard text passage. All samples were independently judged and rated by 3 experienced raters (speech and language therapists) and were re-rated 12 weeks apart for establishing test-retest reliability. Internal consistency, test-retest reliability and Inter-rater reliability of different parameters of this scale and overall grade were determined using Cronbach's alpha and Spearman's rank correlation coefficients. Construct validity was tested using Spearman's rank correlation coefficient. Group validity was tested by Mann-Whitney U-test. Results: The Cronbach's alpha coefficients for internal consistency ranged from 0.87-0.90 for connected speech and 0.79-0.84 for words. The Spearman's correlation coefficients for test-retest reliability of these parameters for connected speech varied between 0.38-0.87 while for words the values ranged between 0.50-0.71. These values for inter-rater reliability of connected speech were between 0.55-0.99, while for words they ranged between 0.40-0.57. For construct validity, the Spearman's correlation coefficient between overall grade of connected speech and overall speech question of SHI, Speech domain of SHI and speech question of UWQOL ranged between 0.35- 0.55. Conclusions: GAAIN-R scale is a valid and reliable tool for speech evaluation in HNC patients. This scale may be used in other groups of patients' as well especially those with neurological diseases. No significant financial relationships to disclose.