A randomized single-blind controlled trial of a prototype digital polytherapeutic for tinnitus

Abstract

Objective: This randomized single-blind controlled trial tested the hypothesis that a prototype digital therapeutic developed to provide goal-based counseling with personalized passive and active game-based sound therapy would provide superior tinnitus outcomes, and similar usability, to a popular passive sound therapy app over a 12 week trial period.

Methods: The digital therapeutic consisted of an app for iPhone or Android smartphone, Bluetooth bone conduction headphones, neck pillow speaker, and a cloud-based clinician dashboard to enable messaging and app personalization. The control app was a popular self-help passive sound therapy app called White Noise Lite (WN). The primary outcome measure was clinically meaningful change in Tinnitus Functional Index (TFI) between baseline and 12 weeks of therapy. Secondary tinnitus measures were the TFI total score and subscales across sessions, rating scales and the Client Oriented Scale of Improvement in Tinnitus (COSIT). Usability of the US and WN interventions were assessed using the System Usability Scale (SUS) and the mHealth App Usability Questionnaire (MAUQ). Ninety-eight participants who were smartphone app users and had chronic moderate-severe tinnitus (>6 months, TFI score > 40) were enrolled and were randomly allocated to one of the intervention groups. Thirty-one participants in the USL group and 30 in the WN group completed 12 weeks of trial.

Results: Mean changes in TFI for the USL group at 6 (16.36, SD 17.96) and 12 weeks (17.83 points, SD 19.87) were clinically meaningful (>13 points reduction), the mean change in WN scores were not clinically meaningful (6 weeks 10.77, SD 18.53; 12 weeks 10.12 points, SD 21.36). A statistically higher proportion of USL participants achieved meaningful TFI change at 6 weeks (55%) and 12 weeks (65%) than the WN group at 6 weeks (33%) and 12 weeks (43%). Mean TFI, rating and COSIT scores favored the US group but were not statistically different from WN. Usability measures were similar for both groups.

Conclusions: The USL group demonstrated a higher proportion of responders than the WN group. The usability of the USL therapeutic was similar to the established WN app. The digital polytherapeutic demonstrated significant benefit for tinnitus reduction supporting further development.

Keywords: clinical trial; digital therapeutic; serious game; sound therapy; tinnitus.

Figure 1

Consort flow chart for participant recruitment and retention.

Figure 2

Example screenshots for (A) the USL intervention (i) Menu, (ii) Passive therapy sounds, (iii) AOIL task, (iv) Tracking task. (B) The WN intervention. (i) Menu, (ii) Passive therapy sounds, (iii) Sound control, (iv) Sound mixing.

Figure 3

Audiogram for per protocol participants. USL group (color, n = 31) WN group (black, n = 30). Mean thresholds and standard error bars are shown.

Figure 4

(A) Change in TFI total score between baseline and 12 weeks for each group. The horizontal line indicates average TFI change for each group. (B) Responder analysis. The proportion of the two groups with reduced TFI of (>5, 13, 20, and 30 points) at 12 weeks of trial (*P < 0.05, **P < 0.01). A change of >13 points is considered clinically meaningful.

Figure 5

TFI score across sessions. (A) Per-protocol. (B) Intent-to-treat. USL group (dashed line) WN group (solid line) (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.001). Mean scores and standard error bars are shown.

Figure 6

TFI subscale scores across sessions. (A) Intrusive, (B) control, (C) cognitive, (D) sleep, (E) auditory, (F) relaxation, (G) quality of life, and (H) emotional distress. USL group (dashed line) WN group (solid line) (*P < 0.05, **P < 0.01, ***P < 0.001). Mean scores and standard error bars are shown.

Figure 7

Rating scales scores across sessions. (A) Problem, (B) strong, (C) uncomfortable, (D) annoyance, (E) ignore, and (F) unpleasant. USL group (dashed line) WN group (solid line) (*P < 0.05, **P < 0.01). For consistency with other figures mean and standard error scores are shown, the non-parametric statistics shown are based on rankings.

Figure 8

(A) COSIT scores (USL group, Blue; WN group, Orange). (B) System Usability Scale (SUS) scores. (C) MHealth app usability questionnaire (MAUQ) scores. Mean scores and standard error bars are shown.

Figure 9

Effect size (Cohen's d) for recent clinical trials at 12 weeks (3 months) of intervention. T30 stimulator trial (14), Multimodal (MM) sound and tongue stimulation trial (17), Tinnitus Retraining therapy (TRT) Partial TRT and Standard of Care trial (16), and current trial (WN, USL, intent-to-treat data).