Digital CBT & app-based therapeutics for Tinnitus: What Does the Evidence Show?
Digital Therapeutics · Ranked #4 on our Top 10 (stable)
Current evidence
CBT's evidence, delivered by app — the fastest-maturing corner of the field, with the first sham-controlled wins.
Evidence strength: Moderate (3/5) · Tier 2 — promising · Evidence independence: Mixed.
Smartphone-delivered structured tinnitus programs — CBT-based courses, sound tools and guided coaching. Examples with real trial evidence: Kalmeda (Germany, prescribable DiGA), Oto (UK), SilentCloud (CE-marked, FDA 510(k)-cleared), MindEar, and audiologist-guided iCBT platforms.
Did tinnitus loudness improve?
None shown. All app RCT primary outcomes are distress instruments (TFI/THI/TQ); none demonstrates loudness reduction.
Did tinnitus distress improve?
Strong. Kalmeda RCT N=187: d=1.1 (vs delayed treatment — inflates vs sham designs). Japan 2026 double-blind sham-app RCT: THI −20.4 between groups at week 16. Oto RCT: between-group 9 TFI points at 6 months, 95% CI [2, 16] (unblinded, waitlist). iCBT: d=0.46.
Loudness means the tinnitus percept itself became quieter (psychoacoustic matching or loudness ratings). Distress means questionnaire scores such as THI/TFI, sleep, anxiety or quality of life improved — the sound may be unchanged. A THI/TFI improvement is never evidence the tinnitus got quieter.
How strong is the evidence?
Multiple independent RCT wins and a first sham-controlled success earn 3/5; the category is heterogeneous — the score applies to the tested products, not to app stores at large.
Has the result been independently replicated?
Limited independent. Positive results across several independent products/groups (Kalmeda, Kyorin, Oto, iCBT), each product mostly single-trial.
Independence: Some trials manufacturer-involved (Kalmeda, Kyorin, Oto co-funding); iCBT research largely independent.
What are the limitations?
- Most apps on the market have zero evidence — the tested few are the exception
- DEFINE trial (app vs therapist) still unpublished
- Several key trials are small or manufacturer-sponsored
Conflicts of interest: Kalmeda and Kyorin trials are manufacturer-involved; Oto trials are academic-company collaborations.
What should patients know about safety?
Safety evidence: Strong. No meaningful harms reported; main risk is delaying proper audiological evaluation.
Is it available?
Regulatory status: Varies by product. Kalmeda: permanently listed in Germany's DiGA registry (#350, permanent since Dec 2021 — prescribable, reimbursed). SilentCloud: FDA 510(k) K221125 (Jan 2023) — cleared as a tinnitus MASKER ('relief... while using the device', fitted by a hearing professional), plus EU MDR CE mark per the manufacturer. Most other apps are unregulated wellness products — check each one.
Availability: App stores; SilentCloud cleared; most others wellness-tier · Europe: Kalmeda prescribable in Germany; others via app stores · Cost (approx.): Free–$30/month typical; Kalmeda reimbursed in Germany
What the studies found
Kalmeda app (mobile CBT-based therapy): RCT behind Germany's DiGA listing
Walter U, Pennig S, Kottmann T, et al. · PLOS Digital Health · 2023 · N=187 · RCT
Significant TQ reduction vs control, Cohen d = 1.1 (p<0.001). Kalmeda is permanently listed in the German DiGA registry (#350, permanent since Dec 2021; prescribable, reimbursed) on this basis.
Control was 3-month DELAYED treatment (waitlist-type), which inflates effect sizes vs sham-controlled designs — the d=1.1 is not comparable head-to-head with sham-controlled trials. A 9-month extension RCT was published in JMIR 2025.
Smartphone tinnitus app: double-blind sham-controlled RCT (Japan)
Wasano K, Kawasaki T, Goto F, et al. · JAMA Otolaryngology–Head & Neck Surgery · 2026 · N=60 · Double-blind sham-controlled RCT, 3 centers
At week 16: THI −16.8 (active) vs +3.6 (sham — the sham arm slightly worsened, which flatters the difference); between-group −20.4 points (95% CI −28.2 to −12.6); meaningful improvement 70% vs 27%.
Per the authors, the first DOUBLE-BLIND sham-app-controlled RCT of a tinnitus digital therapeutic (earlier app RCTs were single-blind). Registration jRCTs032230359.
Oto app RCT vs waitlist (Flinders University)
Mui B, Muzaffar J, Chen J, Bidargaddi N, Shekhawat GS · Speech, Language and Hearing · 2025 · N=96 randomized (arm split per paper; overall dropout 21%) · Unblinded RCT vs waitlist
Between-group TFI improvement at 6 months: 9 points (95% CI [2, 16], p=.006) — verified against the paper's abstract. Significant time×treatment interaction at 6 months (p<.001, d=0.62) and 9 months (p=.002, d=0.54). Clinically meaningful (≥13-point) TFI reduction: 32% vs 12% at 6 months (z=2.20, p=.030). Overall dropout 21%.
A previous version of this record misreported the 6-month result as "16 points (p=.002, 95% CI 6–25)" — 16 is the upper bound of the actual CI [2, 16] and p=.002 belongs to the 9-month interaction; corrected 2026-09-04 after direct verification. Arm-level dropout and analyzed-N breakdowns reported earlier could not be re-verified against accessible text — treat as unverified. The PhD was co-funded by Oto Health and a co-author is affiliated with Oto Health. The larger DEFINE non-inferiority trial (N=198, vs therapist care) concluded Dec 2024 but remains unpublished.
Internet-based audiologist-guided CBT for tinnitus: US randomized controlled trial
Beukes EW, et al. · Journal of Medical Internet Research · 2022 · N=158 (79 iCBT / 79 monitoring) · RCT
Post-treatment TFI 36.6 vs 46.3 (Cohen d = 0.46). Clinically significant improvement in 57% vs 15% — by the paper's own reliable-change criterion, not the standard 13-point TFI threshold.
Control was weekly monitoring (no active comparator). 1-year follow-up paper reported maintained efficacy (PMID 36353148).
MindEar (Tinnibot) chatbot CBT: pilot RCT
Bardy F, et al. · Frontiers in Audiology and Otology · 2024 · N=30 randomized, 28 analyzed (14/arm) · Pilot RCT: chatbot iCBT alone vs chatbot + 4 psychologist video sessions (no untreated control)
TFI decreased 11.2 (chatbot-only) vs 20.0 (hybrid) post-treatment; clinically significant improvement 42% vs 64%, converging to 64% in both arms at 16 weeks; between-group differences not significant.
Pilot only: no untreated arm, small N, baseline imbalance, and no trial registration stated. First author co-founded the developer (Odio Tech) — sponsor-linked. Not indexed in PubMed at extraction time.
NICE guideline NG155: Tinnitus — assessment and management
National Institute for Health and Care Excellence · NICE (UK) · 2020 · N=National guideline · Clinical practice guideline (UK)
Amplification: OFFER with hearing loss affecting communication; CONSIDER with hearing loss without communication difficulty; DO NOT OFFER without hearing loss. Psychological therapies (stepped): digital tinnitus-related CBT → group interventions (MBCT/ACT/CBT) → individual CBT. Betahistine: 'Do not offer betahistine to treat tinnitus.' Sound therapy and neuromodulation: 'We were unable to make recommendations for practice in this area.'
Also advises against psychoacoustic pitch/loudness matching for assessment. Read from the official NICE pathway document (NBK557037 mirror). NICE's sound-therapy position (no practice recommendation) conflicts with AAO-HNSF's 'option' — shown as a disagreement, not resolved by this platform.
Related clinical trials
- Oto smartphone app vs standard care — Listed 'Recruiting' — registry stale (not updated since Oct 2024; primary completion date has passed)
Tracked trials are research in progress, not recommended treatments.
Related treatments
- TrueSilence / digital polytherapeutic (Searchfield) — Thirty years of Auckland tinnitus research just became a startup — personalized, gamified sound therapy by…
Evidence last reviewed: 2026-09-03 · Evidence included through: 2026-09-03 · Confidence: high. Open the interactive evidence profile.
This page summarizes published research for education. It is not medical advice; it cannot say what will work for any individual. Discuss treatment decisions with a qualified clinician.