Predicting Hyperacusis in Tinnitus Patients

Abstract

The goal of this study was to develop a model that predicts whether tinnitus patients would report hyperacusis based upon their audiometric thresholds and loudness discomfort levels (LDLs), measured at 1000, 2000, 4000, and 8000 Hz. Follow-up LDLs subsequent to tinnitus retraining therapy (TRT) were used to predict hyperacusis subsequent to treatment. Pretreatment and TRT follow-up audiometric thresholds and LDLs were analyzed for a group of 68 tinnitus patients who self-reported sound tolerance complaints and a group of 58 tinnitus patients who denied sound tolerance problems. A logistic regression model was implemented to determine which of the eight audiological measures (i.e., four threshold and four LDL measures) were essential in predicting the hyperacusis or tinnitus-only categorization. Predictions from the logistic regression model are compared with three versions of a published categorization scheme. The success of each model varied depending on the hearing-loss configuration, with poorest performance for sloping hearing losses. The logistic regression model was the best model overall. The logistic regression model predicted improvement for 54 of 68 (79%) self-reported hyperacusis patients following TRT treatment. The logistic regression model improves the prediction of hyperacusis in tinnitus patients and successfully predicts improvement in the majority of tinnitus patients following TRT treatment.