Predictors of Response to Cognitive Behavioral Therapy in Patients With Tinnitus

Abstract

Clinical guidelines recommend cognitive behavioral therapy (CBT) as a treatment for tinnitus. However, patient response to CBT is variable, and currently, there are no known predictors of response to CBT treatment for tinnitus. To identify the clinical predictors of patient response to CBT for treatment of tinnitus. This was a secondary cohort analysis of a single-arm clinical study including adults with chronic bothersome tinnitus recruited from Washington University School of Medicine in St Louis (Missouri) from September 2019 to February 2023. Participants completed an 8-week group CBT program with a licensed clinical psychologist. Each week consisted of 2.5 hours of CBT, amounting to 20 hours of total CBT participation, primarily delivered through a virtual platform. Conjunctive consolidation was used to create a predictive classification system for response to CBT based on tinnitus bother and anxiety levels. Response to CBT was predefined as a 13-point or greater decrease in the Tinnitus Functional Index (TFI) survey score. The study sample included 88 adult patients (median [IQR] age, 59 [49-66] years; 47 [53%] females and 41 [47%] males) with chronic bothersome tinnitus, of whom 53 (60%) had at least 13-point decrease in TFI and were considered to be responders. In univariable and multivariable logistic regression analyses, high to moderate anxiety level and severe tinnitus bother were associated with treatment response (adjusted odds ratio: anxiety, 3.33; 95% CI, 0.90-12.30; tinnitus bother, 12.08; 95% CI, 1.48-98.35). The clinical stratification system showed good predictive and discriminative ability (χ2 for linear trend = 20.0; C statistic = 0.75; 95% CI, 0.65-0.85). The findings of this study show that assessment of bother and anxiety levels in patients with tinnitus may be useful for identifying those who are more likely to respond to CBT. Before incorporation into clinical practice, future research should externally validate this finding in a separate population.