PP303 - Comparisons of Speech Intelligibility Index Measures and Conversion Factors

Abstract:
Speech Intelligibility Index (SII) has recently gained attention as studies suggest it is useful as a criterion for hearing aid fittings for children with mild hearing loss and for referral for a cochlear implant evaluation. There are numerous platforms that measure SII from the Count-the-Dots audiogram to automated probe-microphone systems like the Audioscan Verifit 2. Unfortunately, not all methods are accessible for all hearing care professionals. The current study compares the SII output across different methods for various audiograms and provides correction factors among them. We hope this study helps professionals to utilize SII regardless of their limited resources.

Summary:

Introduction
The Speech Intelligibility Index (SII) estimates the amount of speech information audible to a listener in unaided or aided conditions. SII has recently gained attention as recent studies suggest that professionals should conduct hearing aid fittings for children with mild hearing loss and refer for cochlear implant evaluations based on the unaided and aided SII scores, respectively (McCreery et al., 2020; Ching, T. Y. et al., 2015). There are numerous platforms that measure SII, and each method varies in accessibility. Recently, it has become commonly available in computerized audiometry and probe-microphone systems. The current study aims to compare the SII (in %) obtained using three common methods for audiograms varying in severity and configuration and to provide conversion factors if there is a significant difference.

Methods
Three common SII calculation platforms were used to compare the SII scores: Count-the-Dots audiogram (Killion & Mueller, 2010), Otosuite audiometry software (Natus, 2011), and Verifit 2 probe microphone system (Audioscan, 2022). A total of 120 audiograms were created using five different severities and four different configurations of hearing loss. Each audiogram was entered into each platform for comparisons of unaided SIIs. Descriptive statistics, analysis of variance (ANOVA) with multiple groups, and Pearson correlations were used to describe results and the relationships among them. If applicable, a conversion factor was created to relate the three methodologies.

Results
Preliminary results using 20 audiograms of 4 configurations and 5 severities showed strong significant correlations in SII output: Count the Dots vs. the Otosuite (r = 0.999, p < 0.001), Count the Dots vs. Verifit 2 with 65dB SPL input levels (r = 0.970, p < 0.001), and Otosuite vs. Verfit 2 with 55dB SPL input levels (r = 0.971, p < 0.001). Correlation decreased when Otosuite was compared to Verfit 2 at 75 dB SPL (r = 0.779, p < 0.001). For severities greater than moderately severe, all three methods provided 0% SIIs. For other severities, Verifit 65 dB SPL input always provided the highest SIIs as compared to the Count the Dots (t = 3.52, p = 0.002) and Otosuite methods (t = 4.31, p < 0.001). SIIs obtained for ‘sloping’ configurations had the greatest difference (on average 10% more), while ‘steeply sloping’ configurations had the smallest difference (on average 3% more). Based on these data, the following correction factor was calculated: SII using Count the Dots + 6.5 % = estimated SII obtained using the Verifit 65 dB SPL input.

Discussion
The current study compared SIIs using methods from least to most equipment required for hearing care professionals. The SIIs using the most accessible method, the Count-the-Dots audiogram, were strongly correlated and about 6% lower (i.e., more conservative) than SII using the Verifit 65dB SPL. We hope this study provides professionals in various working environments with an easy comparison table in various SII methods, so that they can conduct hearing aid referral or fitting and cochlear implant evaluation referral efficiently and with easily accessible tools.