Sources of variability contributing to test-retest differences in threshold-based articulation indices

Area of Science:

• Audiology
• Speech-Language Pathology
• Hearing Science

Background:

• Accurate audiological assessments are crucial for effective hearing rehabilitation.
• The Articulation Index (AI) is a tool used to predict speech intelligibility based on hearing thresholds.
• Understanding the reliability of AI calculations, particularly test-retest differences, is essential for clinical practice.

Purpose of the Study:

• To investigate test-retest variability in Articulation Index (AI) calculations.
• To determine how clinical threshold-based AI calculation methods influence test-retest differences.
• To analyze the impact of hearing thresholds and speech spectrum on AI test-retest reliability.

Main Methods:

• Calculated AI using a nine-frequency-band method on 209 ears across two separate days.
• Determined standard deviations of test-retest differences for AI values.
• Examined the influence of hearing thresholds and speech spectrum on AI variability.

Main Results:

• Test-retest standard deviations for AI were greatest when AI values were near 50%.
• Variability was minimal when AI values were near 0% or 100%.
• Increased hearing threshold changes significantly elevated test-retest differences in AI.

Conclusions:

• Clinical threshold-based AI calculations exhibit test-retest variability, particularly at mid-range AI values.
• The relationship between hearing thresholds and speech spectrum significantly impacts AI reliability.
• Increasing the number of measured thresholds may offer marginal improvements in AI test-retest consistency.