Test-retest reliability of distortion-product thresholds compared to behavioral auditory thresholds

Abstract

When referred to baseline measures, serial monitoring of pure-tone behavioral thresholds and distortion-product otoacoustic emissions (DPOAEs) can be used to detect the progression of cochlear damage. Semi-logarithmic DPOAE input-output (I/O) functions enable the computation of estimated distortion-product thresholds (EDPTs) by means of linear regression, a metric that provides a quantitative estimate of hearing loss due to cochlear-amplifier degradation. DPOAE wave interference and a suboptimal choice of stimulus levels limit the accuracy of EDPTs. This work identifies the test-retest reliability of EDPTs derived from short-pulse DPOAE level maps (EDPTLM), a method that circumvents limitations associated with both wave interference and suboptimal choice of stimulus levels. The test-retest reliability was compared to that of EDPTs derived from semi-logarithmic I/O functions (EDPTI/O) and that of behavioral thresholds acquired with pure-tone audiometry (PTA) and modified Békésy tracking audiometry (TA) to provide a foundation for identifying and interpreting significant threshold shifts. The DPOAE-based auditory thresholds (EDPTLM and EDPTI/O) and behavioral thresholds (PTA and TA) were recorded seven times within three months at 14 frequencies with f2\xa0=\xa01-14\xa0kHz in 20 ears from ten subjects with normal hearing (4PTA0.5-4kHz\xa0<\xa020\xa0dB\xa0HL). To obtain EDPTLM, short-pulse DPOAEs were recorded using 21 L1,L2 pairs. Reconstruction of DPOAE growth behavior as a function of L1 and L2 using nonlinear curve fitting enabled the derivation of EDPTLM for each frequency. Test-retest reliability was determined using three different approaches: 1) centered thresholds, 2) average threshold differences, and 3) average absolute threshold differences, between each possible test session (N\xa0=\xa021). Test-retest reliability based on centered thresholds and average threshold differences showed no statistically significant difference between EDPTLM, EDPTI/O, PTA, and TA for the pooled analysis incorporating all stimulus frequencies. Average absolute threshold differences presented small but significant differences in test-retest reliability with median values of 3.00\xa0dB for PTA, 3.20\xa0dB for TA, 3.34\xa0dB for EDPTLM, and 3.51\xa0dB for EDPTI/O. A considerable frequency dependence of test-retest reliability was found; namely, the highest test-retest reliability was for EDPTLM at f2\xa0=\xa011\xa0-\xa014\xa0kHz. Otherwise, at lower frequencies, the highest test-retest reliability was for TA at f2\xa0=1\xa0-\xa02\xa0kHz. Overall, the test-retest reliability of EDPTLM was better than that of EDPTI/O and was similar to that for behavioral thresholds. Hence, deriving EDPTLM from individual level maps is a promising and sensitive method for objectively monitoring the state of the cochlea. Furthermore, the detection of an equidirectional threshold change at a single frequency in both EDPTLM and TA might allow reducing the threshold shift as indication of a follow-up examination from the clinical standard of 10\xa0dB down to 5\xa0dB. This stricter indicator might be beneficial when monitoring cochlear damage, for example ototoxicity, in the presence of (remnant) cochlear amplification at baseline.