Reliability in Hearing Threshold Prediction in Normal-Hearing and Hearing-Impaired Participants Using Mixed Multiple ASSR.

Abstract

The rapidly evolving field of hearing aid fitting in infants requires rapid, objective, and highly reliable methods for diagnosing hearing impairment. The aim was to determine test-retest reliability in hearing thresholds predicted by multiple auditory steady-state response (ASSRthr) among normal-hearing (NH) and hearing-impaired (HI) adults, and to study differences between ASSRthr and pure-tone threshold (PTT) as a function of frequency in each participant. ASSR amplitude versus stimulus level was analyzed to study ASSR growth rate in NH and HI participants, especially at ASSRthr. Mixed multiple ASSR (100% AM, 20% FM), using long-time averaging at a wide range of stimulus levels, and PTT were recorded in 10 NH and 14 HI adults. ASSRthr was obtained in 10 dB steps simultaneously in both ears using a test-retest protocol (center frequencies = 500, 1000, 2000, and 4000 Hz; modulation frequencies = 80-96 Hz). The growth rate at ASSRthr was calculated as the slope (nV/dB) of the ASSR amplitudes obtained at, and 10 dB above, ASSRthr. PTT was obtained in both ears in 1 dB steps using a fixed-frequency Békésy technique. All of the NH participants showed PTTs better than 20 dB HL (125-8000 Hz), and mean pure-tone average (PTA; 500-4000 Hz) was 1.8 dB HL. The HI participants exhibited quite symmetrical sensorineural hearing losses, as revealed by a mean interaural PTA difference of 6.5 dB. Their mean PTA in the better ear was 38.7 dB HL. High ASSRthr reproducibility (independent of PTT) was found in both NH and HI participants (test-retest interquartile range = 10 dB). The prediction error was numerically higher in NH participants (f ≥1000 Hz), although only a significant difference existed at 1000 Hz. The median difference between ASSRthr (dB HL) and PTT (dB HL) was approximately 10 dB in the HI group at frequencies of 1000 Hz or greater, and 20 dB at 500 Hz. In general, the prediction error decreased (p < 0.001) with increasing hearing threshold, although large intersubject variability existed. Regression analysis (PTT versus ASSRthr) in HI participants revealed correlation coefficients between 0.72-0.88 (500-4000 Hz) and slopes at approximately 1.0. Large variability in ASSRthr-PTT versus frequency was demonstrated across HI participants (interquartile range approximately 20 dB). The maximum across-frequency difference (ASSRthr-PTT) in an individual participant was 50 dB. HI participants showed overall significantly higher amplitudes and slopes at ASSRthr than did NH participants (p < 0.02). The amplitude-intensity function revealed monotonically increasing ASSRs in NH participants (slope 2 nV/dB), whereas HI participants exhibited heterogeneous and mostly nonmonotonically increasing ASSRs. Long-time averaging of ASSR revealed high ASSRthr reproducibility and systematic decrease in prediction error with increasing hearing threshold, albeit large intersubject variability in prediction error existed. A plausible explanation for the systematic difference in ASSRthr between NH and HI adults might be significantly higher ASSR amplitudes and higher overall growth rates at ASSRthr among HI participants. Across-frequency comparison of PTT and ASSRthr in an individual HI participant demonstrated large variation; thus, ASSR may not be optimal for, e.g., reliable threshold prediction in infants and subsequent fine-tuning of hearing aids.