Predicted and measured word-recognition scores unmask distortion in the impaired auditory system

The Speech Intelligibility Index (SII) predicts word-recognition performance under a variety of listening conditions. Although it reasonably predicts average performance it does not accurately predict the performance of individual hearing-impaired listeners. This is because the SII accounts for the attenuation component of hearing loss but not deficits in suprathreshold processing (distortion). This study sought to develop a method for measuring the distortion component of hearing loss and provide guidelines for interpreting word-recognition scores. Two large databases were mined for pure-tone thresholds and word-recognition scores (total n = 9928 ears). SII was calculated and scores were predicted based on transfer functions relating SII to word-recognition score. Cases were selected with complete air- and bone-conduction thresholds and only cases with normal hearing and sensorineural hearing loss were included. Word-recognition scores were parsed into hearing-loss groups based on pure-tone threshold averages. Distributions of differences between measured and predicted scores become broader and more asymmetric with increasing hearing loss progressing toward larger negative differences (predicted score higher than measured score) with increasing hearing loss. A classification scheme was developed to categorize differences into Normal, Low, Moderate, and High distortion. Distortion categories may be useful in counseling patients and may inform the hearing-instrument selection process.