Contributions of Counseling and Sound Generator Use in Tinnitus Retraining Therapy: Treatment Response Dynamics Assessed in a Secondary Analysis of a Randomized Trial

Abstract

Purpose: Tinnitus retraining therapy (TRT) has been widely used for 30 years, but its efficacy and the component contributions from counseling and sound therapy remain controversial. The purpose of this secondary analysis from the Tinnitus Retraining Therapy Trial (TRTT) was to compare treatment response dynamics for TRT (counseling and conventional sound generators) with partial TRT (pTRT; counseling and placebo sound generators) and standard of care (SOC; a patient-centered counseling control).

Method: The TRTT randomized 151 participants with primary tinnitus (no significant hearing or sound tolerance problems) to TRT, pTRT, or SOC, each of which encouraged use of enriched environmental sound. The primary outcome, mean change in Tinnitus Questionnaire score assessed at baseline and follow-up across 18 months, was normalized for a common baseline and fitted with an exponential model. Time constants were estimated to quantify and compare the treatment response dynamics, which were evaluated for statistical significance using bootstrap analyses.

Results: The change in response to TRT took less time to achieve than that for either pTRT or SOC, as demonstrated by time for normalized Tinnitus Questionnaire scores to decline to 63% and 99% of baseline TRT values: 1.2 months (95% CI [0.2, 1.9]) and 5.7 months (95% CI [0.9, 9.0]), respectively. Corresponding SOC values were 2.7 months (95% CI [1.5, 4.1]) and 12.4 months (95% CI [6.9, 19.0]), while those for pTRT were 2.2 months (95% CI [1.2, 3.4]) and 10.1 months (95% CI [5.7, 15.9]). The differences were significant for TRT versus SOC (p = .020), borderline significant for TRT versus pTRT (p = .057), but nonsignificant for pTRT versus SOC (p = .285). The magnitude of the asymptotic treatment response did not differ significantly among groups.

Conclusion: Sound generator use in TRT increases treatment efficiency (beyond any advantage from enriched environmental sound) without affecting treatment efficacy (determined by counseling).

Trial registration: ClinicalTrials.gov NCT01177137.

Figures

Figure 1.

Fitted exponential models to TQ change scores. TQ change scores as a function of treatment visit for paired comparisons of the treatment groups indicated in the inset legend of each panel. Best-fitting exponential models and associated confidence limits are superimposed on each pair of change scores. Larger negative TQ change scores represent greater improvement over the course of an intervention. Time constants and corresponding TQ change scores also are denoted for each treatment group. SOC = standard of care (patient-centered counseling control); τ63 and τ99 = time constants derived from the exponential model quantify the time points (in months from baseline) at which the TQ change scores improved by 63% and 99%, respectively; TRT = tinnitus retraining therapy (TRT counseling and sound generator use); CI = confidence interval; pTRT = partial tinnitus retraining therapy (TRT counseling and placebo sound generator use); TQ = Tinnitus Questionnaire.

Figure 2.

Differences in treatment response dynamics for TRT and pTRT highlight the contributions of counseling and sound generator use in the TRT intervention. Shown here are the exponential models from Figure 1 fitted to the TQ change scores as a function of time in intervention for TRT (blue curve) and pTRT (red curve); corresponding τ99 time constants and TQ99 change scores are marked on the respective exponential models, and these values are listed in the inset legend. Shading represents overall differences in the treatment responses to TRT and pTRT. The contribution to TRT from counseling, effectively presented alone when paired with placebo sound generators in the pTRT intervention (shown by red curve), is quantified by the magnitude of the TQ99 change score, which is denoted as the “counseling effect.” The contribution to TRT from sound generator use is represented by the time difference (in months) between the τ99 value for TRT and that for pTRT; this time difference is denoted as the “sound generator effect.” Note that the magnitude of change in the TQ change scores at asymptote (i.e., the time at which the intervention is effectively complete as defined at τ99) is nearly identical for the TRT and pTRT curves (reflecting a common counseling contribution), whereas the time at which the TRT curve becomes asymptotic precedes that for the pTRT curve by 4.4 months (representing the contribution to the TRT intervention from sound generator use, which was not effective in the pTRT intervention). TRT = tinnitus retraining therapy (TRT counseling and sound generator use); τ99 = time constants derived from the exponential model quantify the time points (in months from baseline) at which the TQ change scores improved by 99%; TQ = Tinnitus Questionnaire; pTRT = partial tinnitus retraining therapy (TRT counseling and placebo sound generator use).

Figure A1.