Limited Interchangeability of Cochlear Tuning Estimates Across Fast Psychophysical Tuning Curve, Distortion-Product...

Auditory frequency selectivity is a critical determinant of speech intelligibility in noise. Although fast psychophysical tuning curves (fPTCs) and otoacoustic emission (OAE)-based metrics are often interpreted as indirect estimates of cochlear tuning, they differ in measurement pathway and physiological source. It remains unclear whether these methods provide interchangeable estimates at the individual-listener level. This study evaluated the correlation, absolute agreement, and Bland-Altman limits of agreement among behavioral and OAE-based tuning estimates to determine whether one method can serve as a reliable proxy for another in clinical or translational assessments.

A secondary analysis was performed on a dataset of 67 adults with hearing profiles ranging from normal to moderate high-frequency loss. Frequency selectivity was quantified at 2, 4, and 8 kHz using three methods: fPTCs, distortion product OAE-level ratio function estimates, and stimulus-frequency otoacoustic emission (SFOAE) phase-gradient delays. Tuning sharpness, quantified as the equivalent-rectangular-bandwidth quality factor (QERB), was compared across methods using Pearson correlations, intraclass correlation coefficients, and Bland-Altman limits of agreement. The impact of high-frequency thresholds and signal to noise ratio on method discrepancy was also modeled.

Intermethod associations were weak and inconsistent across frequencies. Absolute agreement was poor, with intraclass correlation coefficient [(2,1)] values near or below zero across method pairs. Bland-Altman analyses showed wide limits of agreement, indicating substantial individual-level disagreement. At 4 kHz, the fPTC-SFOAE comparison showed limits of agreement from approximately -2.11 to 1.53 log units, corresponding to multiplicative limits of approximately 0.12× to 4.63× on the original QERB scale. Higher high-frequency thresholds and lower SFOAE signal to noise ratio were associated with greater intermethod disagreement.

Behavioral and OAE-based tuning metrics were not interchangeable at the individual-listener level. The weak correlations, poor absolute agreement, and wide limits of agreement suggest that fPTC, distortion product OAE-level ratio function, and SFOAE phase estimates should be treated as complementary rather than redundant measures. Future work should determine whether intermethod mismatch provides diagnostic information beyond any single tuning metric.