Brains tern Electric Response Audiometry: Estimation of the Amount of Conductive Hearing Loss with and without Use of the Response Threshold

Abstract

Three aspects of brainstem response audiometry were investigated in the present study. (1) The brainstem response threshold was compared with the pure-tone audio-gram in 40 patients with conductive hearing loss. The brainstem response threshold has a one-to-one relationship with the mean of the pure-tone thresholds at 2 and 4 kHz. The correlation coefficient in this comparison is 0.84 and the standard error of the estimate is 8.3 dB. Taking into account corresponding results in cochlear hearing loss [Drift et al.: Audiology 26: 1-10, 1987] it is concluded that the brainstem response threshold provides a good estimate of the amount of peripheral hearing loss, independent of the type of hearing loss. (2) It was shown [Drift et al.: Audiology 27: 260-270, 1988] that different types of peripheral hearing loss can be distinguished reliably with brainstem response audiometry. Parameters relevant for this distinction were the horizontal shift of the latency-level curve (1(L) curve), that of its derivative and the response threshold. In the clinical situation measurement of the response threshold is not always possible due to restlessness of the patient. To simulate this situation we randomly truncated the lower parts of the 1(L) curves of quiet patients. The test group consisted of 22 adult normally hearing subjects, 79 patients with cochlear hearing loss, 40 with conductive hearing loss and 22 with mixed hearing loss. Linear discriminant analysis was applied to the horizontal shift of the 1(L) curve and of its derivative. The brainstem diagnosis 'normal hearing' correctly excludes a conductive hearing loss in 98% of the cases and the brainstem diagnosis 'cochlear hearing loss' does so in 79%. The brainstem diagnosis 'conductive hearing loss' correctly predicts a conductive component of hearing loss in 94% of the cases and the brainstem diagnosis 'mixed hearing loss' does so in 90%. The distinction between cochlear hearing loss and normal hearing is not reliable, neither is the distinction between conductive and mixed hearing loss. (3) The amount of the conductive component of hearing loss can be estimated by the horizontal shift of the 1(L) curve. Statistical comparison with the mean of the air-bone gaps at 2 and 4 kHz gave a correlation coefficient of 0.77, a standard error of the estimate of 9.7 dB, and a slope of the regression line of 0.93. An overestimation of about 7 dB has to be taken into account in case of mixed hearing loss.

Key Words:

• BERA
• Conductive hearing loss
• Response threshold
• Latency-level curve
• Derived latency-level curve
• Horizontal shift
• Pure-tone audiogram