Prediction of hearing levels from acoustic reflex thresholds in persons with high-frequency hearing losses.

This study examined the precision of the bivariate method in subjects with high-frequency sensorineural hearing loss. The current bivariate data effectively separated normal hearing subjects from those with pure tone averages of greater than 32 dB HL, in a manner consistent with the results of Popelka and Trumpf (1976) and Margolis and Fox (1977b). However, for persons with high-frequency losses the prediction of hearing levels from acoustic reflex thresholds (ARTs) appears to be complicated. Moderate hearing losses involving 500, 1000 and 2000 Hz ("speech frequencies") as well as higher frequencies were identified on the basis of elevated average ARTs for 500, 1000 and 2000 Hz. Normal ears (pure tone averages of less than 30 dB HL) were isolated from others on the basis of position on the bivariate graph. Those with (1) normal hearing in the "speech frequencies" and a high-frequency loss and (2) a mild loss in the "speech frequencies" and a high-frequency loss, could be separated from those with normal hearing by location on the bivariate graph, and from those with moderate (or worse) losses on the basis of average ART for tones. Consideration of these findings is useful in the evaluation of patients at risk for high-frequency loss, such as patients with noise exposure, and is particularly useful in cases of suspected functional impairment within this population. A modification of the bivariate method is suggested which extends its application to patient populations with a large incidence of high frequency sensorineural hearing loss.