Clinical Assessment of Functional Hearing Deficits: Speech-in-Noise Performance

The clinical evaluation of hearing loss, using a pure-tone audiogram, is not adequate to assess the functional hearing capabilities (or handicap) of a patient, especially the speech-in-noise communication difficulties. The primary objective of this study was to measure the effect of elevated hearing thresholds on the recognition performance in various functional speech-in-noise tests that cover acoustic scenes of different complexities and to identify the subset of tests that (a) were sensitive to individual differences in hearing thresholds and (b) provide complementary information to the audiogram. A secondary goal was to compare the performance on this test battery with the self-assessed performance level of functional hearing abilities.

In this study, speech-in-noise performance of normal-hearing listeners and listeners with hearing loss (audiometric configuration ranging from near-normal hearing to moderate-severe hearing loss) was measured on a battery of 12 different tests designed to evaluate speech recognition in a variety of speech and masker conditions, and listening tasks. The listening conditions were designed to measure the ability to localize and monitor multiple speakers or to take advantage of masker modulation, spatial separation between the target and the masker, and a restricted vocabulary.

Listeners with hearing loss had significantly worse performance than the normal-hearing control group when speech was presented in the presence of a multitalker babble or in the presence of a single competing talker. In particular, the ability to take advantage of modulation benefit and spatial release from masking was significantly affected even with a mild audiometric loss. Elevated thresholds did not have a significant effect on the performance in the spatial awareness task. A composite score of all 12 tests was considered as a global metric of the overall speech-in-noise performance. Perceived hearing difficulties of subjects were better correlated with the composite score than with the performance on a standardized clinical speech-in-noise test. Regression analysis showed that scores from a subset of these tests, which could potentially take less than 10 min to administer, when combined with the better-ear pure-tone average and the subject's age, accounted for as much as 93.2% of the variance in the composite score.

A test that measures speech recognition in the presence of a spatially separated competing talker would be useful in measuring suprathreshold speech-in-noise deficits that cannot be readily predicted from standard audiometric evaluation. Including such a test can likely reduce the gap between patient complaints and their clinical evaluation.