A comparison of threshold-based fitting strategies for nonlinear hearing aids

In recent years, wide dynamic range compression (WDRC) has been used with increasing success. To optimize the fit with this type of hearing aid circuitry, subjective measures of loudness growth often are used. Unfortunately, these type of measures cannot be performed with infants, young children, and some elderly individuals. The primary purpose of this study was to compare the fitting recommendations of two recently described threshold-based procedures for fitting nonlinear hearing aids (DSL 4.0 and FIG6) to the use gain settings of satisfied adult hearing aid users for whom the fitting was based on subjective measures of loudness growth. Because it cannot be assumed that the use settings for adults will be appropriate for young children, a secondary goal was to quantify the audibility of speech at the use settings derived from loudness growth measures.

Forty-nine adult hearing aid users with mild to severe sensorineural hearing loss participated in this study. For all subjects, loudness growth measures were used to optimize the fit of a 2-channel WDRC hearing aid. The use gain at 50 and 80 dB SPL was compared with the gain recommended by DSL, FIG6, and the manufacturer's threshold-based fitting algorithm.

In general, both DSL and FIG6 prescribed more gain than actually was used by these hearing aid wearers. These discrepancies increased as a function of frequency, and differences in excess of 20 dB were observed in some cases. The manufacturer's algorithm provided a closer approximation to the use gain than either DSL or FIG6. Utilizing these use gain values, an Aided Audibility Index (AAI) was calculated for soft, average, and loud speech across four degrees of hearing loss, ranging from mild to severe (12 conditions). Transfer functions for continuous discourse and nonsense syllables were applied to yield estimated intelligibility scores. For the higher context speech materials, estimated intelligibility was > or = 85% for nine of the 12 conditions. For low-context speech materials, estimated intelligibility was > or = 85% for only three of the 12 conditions.

Results suggest that the gain recommendations provided by both DSL and FIG6 exceeded the gain actually used by the adult hearing-impaired subjects in this study. Gain recommendations from the manufacturer's algorithm provided a closer approximation to the use gain values of these subjects. These findings suggest that, for adult hearing aid users who cannot perform loudness judgments reliably, the manufacturer's algorithm would be expected to provide a closer approximation to loudness-based use gain values than either DSL or FIG6. However, AAI calculations revealed that the gain recommendations from this algorithm produce adequate audibility of speech only if one assumes linguistic competence. When AAI values are transformed to predict the intelligibility of low-context speech materials, it appears that the degree of audibility may not be appropriate for prelingually hearing-impaired children with more than a moderate hearing loss.