Evidence review

Advice

Clinical and technical evidence
Costs and resource consequences
Strengths and limitations of the evidence

Evidence review

Clinical and technical evidence

RESET trial

RESET was an exploratory, patient‑blinded, randomised placebo‑controlled trial (Tass et al. 2012; NCT00927121). In the study, 63 patients were treated at 2 sites in Germany, and randomised to 4 treatment groups and an active placebo group. The 4 intervention groups received different doses of Acoustic CR Neuromodulation delivered by a portable acoustic device with comfortable earphones for 4–6 hours per day (groups 1–3, using different sequences of sounds) or 1 hour per day (group 4). The treatment groups and outcomes are reported in tables 1 and 2. The main outcome measure was change in Tinnitus Questionnaire (TQ) and visual analogue scale (VAS) for annoyance and loudness after 12 weeks of treatment with Acoustic CR Neuromodulation. Whole‑head electroencephalograms (EEGs) were recorded to look for changes in tinnitus‑related activity.

Most of the outcomes of the RESET study were compared with baseline measurements rather than the placebo group. Results from the RESET study are reported in table 2.

Loudness and annoyance as measured by VAS after 12 weeks of treatment with Acoustic CR Neuromodulation was statistically significantly reduced in the active group 1 compared with placebo (p<0.05) while the patient was receiving stimulation. Improvement in annoyance as measured by VAS for the active group 3 was statistically significant (p<0.05) compared with placebo after the stimulation was turned off for 2.5 hours.

The mean difference in TQ scores after 12 weeks of treatment was −12.4 (standard deviation [SD] ±8.9) for group 1 and −8.4 (SD±7.1) for the placebo group. The difference in TQ between the groups of 4.0 points was not compared statistically.

The authors conducted a post‑hoc analysis (Tass et al. 2013) by pooling results into 'effective' and 'ineffective' stimulation groups. TQ scores for patients allocated to receive effective stimulation (groups 1 and 3 combined) were statistically significantly reduced compared with ineffective stimulation (groups 2, 4 and 5 combined) after 12 weeks of treatment (p=0.0076). Loudness and annoyance as measured by VAS were statistically significantly reduced in the effective stimulation group compared with the ineffective group.

Fifteen mild to moderate adverse events and 2 serious adverse events were reported. Thirteen adverse events occurred during the blinded phase of the study and 2 occurred during the unblinded long‑term extension phase. Eight of the adverse events were considered to be treatment related, of which 3 were associated with a transient increase of tinnitus loudness (although the authors did not report to which group the patients belonged). The 2 serious adverse events were not associated with the study treatment.

A further 6 articles were identified that related to the RESET study. Two of these were letters relating to the key paper by Tass and colleagues on the RESET study outlined below (Rucker and Antes 2013; Tass et al. 2013). In their letter, Tass and colleagues report additional analyses including analysis of covariance (ANCOVA) using baseline values for annoyance and loudness as measured by VAS, and TQ, as covariates between the groups. The authors reported statistically significant group differences for loudness and annoyance as measured by VAS (p<0.05) and TQ (p<0.001) in 'off‑stimulation', although they did not explore which individual comparisons were statistically significant (Tass et al. 2013).

The remaining 4 articles used data from the RESET study to address other research questions (Adamchic et al. 2012a; Adamchic et al. 2012b; Adamchic et al. 2013; Silchenko et al. 2013). Adamchic et al. (2012a) looked at whether a change in tinnitus pitch correlated with a change in loudness and annoyance scores as measured by VAS; the authors also investigated whether changes in brain synchrony correlated with pitch change. Adamchic et al. (2012b) used loudness and annoyance scores as measured by VAS from the RESET trial to study whether they were valid measurements of tinnitus severity. Adamchic et al. (2013) compared EEG recordings from the RESET study from patients whose tinnitus responded well to neuromodulation against healthy controls. The authors studied whether the treatment shifted EEG patterns closer to normal physiological EEG patterns, or whether it was associated with substantially different, non‑physiological patterns. Silchenko et al. (2013) considered the impact of Acoustic CR Neuromodulation on neural brain networks underlying the perception of tinnitus using EEG measurements before and after neuromodulation. This study used data from patients whose condition responded well to the treatment in the RESET study, and found that after treatment, the pathological interactions in the brain were diminished in 'good responders'.

Table 1 Summary of the RESET trial: Tass et al. (2012)

Study component	Description
Objectives/hypotheses	To measure an improvement in tinnitus symptoms.
Study design	Randomised, single‑blind, placebo‑controlled trial (proof‑of‑concept).
Setting	Two treatment centres in Germany (start date reported as November 2011 in trial database; NCT00927121). Patients were treated for 12 weeks with different durations of treatment of Acoustic CR Neuromodulation and compared with an active placebo group.
Inclusion/exclusion criteria	The study included adults with chronic (≥6 months) tonal, subjective tinnitus, and able to hear stimulation tones (hearing impairment up to 50 dB within frequency of 0.125–12.0 kHz). Patients were excluded if they suffered from Meniere's disease, auditory hallucinations, symptomatic hearing disorders, tinnitus due to temporomandibular joint disorders, brainstem diseases, psychiatric disorders or objective tinnitus.
Primary outcomes	Tinnitus distress level measured by a German version of the TQ. Loudness and annoyance of tinnitus as measured using a VAS.
Statistical methods	No formal sample size calculation reported. Patients were allocated to unequal groups including a small placebo group. ITT analysis was used. LOCF was used where data were missing, although no details of missing data were reported. Exploratory statistics reported and outcomes were compared mostly to baseline rather than placebo. Confidence intervals were not reported. Unusual methods were used to 'normalise' baseline differences and the unusual subgroup matching procedure. No adjustment was made to the p value cut‑off for significance to correct for the increased risk of detecting a significant change when carrying out multiple testing.
Participants	Ninety nine patients from 8 centres were screened and 63 were randomised to 5 groups (2 treatment centres). Groups 1 (n=22), 2 (n=12), 3 (n=12) and 4 (n=12) received difference durations of treatment of Acoustic CR Neuromodulation delivered using a portable acoustic device and comfortable earphones. Group 5 (n=5) received placebo stimulation. Groups 1, 2 and 3 received stimulation for 4–6 hours daily (either continuously or split into sessions no shorter than 1 hour). Groups 4 and 5 received stimulation for a maximum of 1 hour each day. The pattern of sounds delivered to patients in groups 1–4 was based on a specific algorithm to match the patients' tinnitus tones. The authors reported that groups 1 and 3 received 'effective stimulation' and that groups 2, 4 and 5 received 'ineffective stimulation'. It is not clear whether this pooling was post‑hoc. Patients were assessed at weeks 1, 4, 8, 12 and 16 (patients were blind to treatment allocated during this phase). After 12 weeks there was a 4‑week pause in treatment. Patients were offered an optional unblinded long‑term extension of the treatment during which they received the same stimulation pattern as that applied to group 1 in the blinded phase. During the long‑term extension patients were assessed every 4 weeks for 24 weeks.
Results	Change in VAS loudness and annoyance after 12 weeks of treatment for group 1 was statistically significant compared with placebo (p<0.05) for 'on‑stimulation' (measured 15 minutes after beginning stimulation). Change in VAS annoyance for group 3 was statistically significant (p<0.05) compared with placebo for 'off‑stimulation' (measured 2.5 hours after stopping stimulation). VAS loudness and annoyance reduced statistically significantly compared with baseline in groups 1 and 3 (p≤0.01) in the on‑ and off‑stimulation conditions after 12 and 16 weeks respectively. VAS loudness/annoyance scores after 12 weeks of treatment did not reduce significantly compared with baseline in the placebo group. Mean TQ scores were significantly reduced compared with baseline in groups 1 (p<0.0001), 2 (p<0.05), 3 (p<0.01) and 4 (p<0.01) for weeks 12 and 16. There was no reduction in mean TQ scores compared with baseline for the placebo group. The authors did not compare reduction in TQ scored for the intervention groups with the placebo group, as would be expected. The difference between TQ mean scores for group 1 and placebo was 4.0 points (not compared statistically). The authors pooled the results from group 1 and 3 into an 'effective stimulation group' (n=34) and pooled results from groups 2, 4 and 5 into an 'ineffective stimulation group' (n=29). TQ scores were reduced in both groups after 12 weeks of treatment compared with baseline (p<0.0001). VAS loudness reduced in both groups (on‑stimulation) (p<0.001) but only the effective group showed a significant reduction off‑stimulation. VAS annoyance results showed a similar trend. The authors did not compare the differences between the groups. Following an unblinded 24‑week long‑term extension period during which Acoustic CR Neuromodulation was applied the authors report that 40% of patients showed an improvement in TQ of ≥15 points, 35% showed a TQ improvement of between 6 and 14 points, 23% were unchanged (±5 points), and 2% worsened by ≥6 points. VAS loudness and annoyance after 40 weeks were not reported.
Conclusions	Using Acoustic CR Neuromodulation for 12 weeks resulted in a statistically significant improvement in VAS annoyance and loudness and TQ scores compared with baseline measurements for patients allocated to receive effective stimulation. No statistically significant change from baseline was observed in the placebo group. Change in VAS loudness and annoyance for group 1 was statistically significant compared with placebo.
Abbreviations: CR, co‑ordinated reset; ITT, intention to treat; LOCF, last observation carried forward; LTE, long‑term extension; n, number of patients; TQ, Tinnitus Questionnaire; VAS, visual analogue scale.

Table 2 Summary of the RESET trial: Tass et al. (2012)

	Group 1 ^a	Group 2 ^b	Group 3 ^c	Group 4 ^d	Group 5 ^e (placebo)	Analysis
Randomised	n=22	n=12	n=12	n=12	n=5
Efficacy	n=22	n=12	n=12	n=12	n=5
VAS loudness (change from baseline [SD]) 'off stimulation'^f	−21.8 (19.2) p<0.001*	−2.1 (21.7) p=0.844*	−25.8 (25.3) p=0.004*	−6.7 (15.3) p=0.297*	−9.0 (18.8) p=0.500*
VAS loudness (change from baseline [SD]) 'on stimulation'^g	−37.3 (24.7) p<0.001*	−21.3 (25.3) p=0.020*	−29.6 (30.0) p=0.008*	−18.8 (23.7) p=0.025*	−9.0 (18.8) p=0.500*	G1 significant reduction compared with placebo (p<0.05).
VAS annoyance (change from baseline [SD]) 'off stimulation'^f	−18.0 (17.2) p<0.001*	−4.2 (24.6) p=0.611*	−28.8 (27.0) p=0.004*	−7.5 (16.7) p=0.281*	−2.0 (16.4) p=1.000*	G3 significant reduction compared with placebo (p<0.05).
VAS annoyance (change from baseline [SD]) 'on stimulation'^g	−32.7 (23.2) p<0.001*	−22.1 (33.5) p=0.057*	−31.7 (33.3) p=0.010*	−18.8 (23.7) p=0.020*	−8.0 (13.0) p=0.500*	G1 significant reduction compared with placebo (p<0.05).
TQ score (change from baseline [SD])	−12.4 (8.9) p<0.001*	−5.2 (8.0) p=0.027*	−15.5 (15.1) p=0.007*	−8.6 (7.0) p=0.003*	−8.4 (7.1) p=0.125*
Safety	15 mild to moderate AEs occurred in total: 13 AEs occurred during the blinded phase of the study and 2 AEs occurred during the long‑term extension phase. Eight were judged to be related to the treatment, of which 3 were associated with a transient increase of tinnitus loudness. These 3 patients continued to the long‑term extension phase of the study.
Patients reporting serious adverse events	2 SAEs were reported (abdominal pregnancy and avascular necrosis of the femoral head; not associated with treatment).
Abbreviations: AE, adverse event; EEG, electroencephalogram; n, number of patients; SAE, serious adverse event; SD, standard deviation; TQ, tinnitus questionnaire; VAS, visual analogue scale. * p values relate to a comparison with baseline measurements. ^aG1: Patients received stimulation for 4–6 hours a day (4‑tone sequence). ^bG2: Patients received stimulation for 4–6 hours a day (4‑tone sequence selected at random from 12 tones). ^cG3: Patients received stimulation for 4–6 hours a day (4‑tone sequence with the repetition rate controlled by EEG measurement). ^dG4: Patients received stimulation for 1 hour a day (4 tones per sequence). ^eG5: Patients received stimulation with a placebo tone for 1 hour a day. ^f'off stimulation': measurements were taken 2.5 hours after stopping Acoustic CR Neuromodulation. ^g 'on stimulation': measurements were taken 15 minutes after starting Acoustic CR Neuromodulation.

Wurzer and Weimann (2011)

Wurzer and Weimann presented data from a single‑arm observational study conducted in Germany and published as an abstract at the 2011 Tinnitus Research Initiative Conference. Details of this study are reported in tables 3 and 4. No full text, peer‑reviewed paper was available to accompany the abstract.

In the study, 70 patients with chronic tonal subjective tinnitus were treated with Acoustic CR Neuromodulation for at least 6 months. The authors report that 40% of participants showed at least a 15‑point reduction in TQ score, and 30% of patients showed between a 6‑ and 14‑point reduction. No statistical tests were reported in the abstract.

Table 3 Summary of the observational study: Wurzer and Weimann (2011)

Study component	Description
Objectives/hypotheses	To measure an improvement in tinnitus symptoms (using the Tinnitus Questionnaire) in patients treated with Acoustic CR Neuromodulation.
Study design	Single‑arm, observational study.
Setting	Specialised outpatient setting in Germany (no dates reported).
Inclusion/exclusion criteria	Patients with chronic tonal subjective tinnitus (0.2–10 kHz) with hearing loss <80 dB.
Primary outcomes	Change in TQ score from baseline following 6 months of treatment categorised as: 1) reduction of ≥15 points, 2) reduction of 6–14 points, and 3) worsening of >6 points.
Statistical methods	Descriptive statistics reported only.
Participants	70 patients received treatment. No sample size calculation reported.
Results	40% of patients showed a reduction of at least 15 TQ points after 6 months of Acoustic CR Neuromodulation treatment. 30% of patients reported a reduction of between 6 and 14 TQ points, and 1 patient's condition worsened by more than 6 TQ points.
Conclusions	70% of patients had a reduction in TQ score of more than 6 points after 6 months of treatment with Acoustic CR Neuromodulation. This study was reported as a conference abstract only and therefore the results should be interpreted cautiously.
Abbreviations: CR, co‑ordinated reset; n, number of patients; TQ, Tinnitus Questionnaire.

Table 4 Summary of the observational study: Wurzer and Weimann (2011)

	Acoustic CR Neuromodulation
Included	Did not report drop‑outs
Efficacy	n=70
Primary outcome: change in TQ score from baseline	40% of patients showed a reduction of ≥15 TQ points. 30% of patients showed a reduction of 6–14 TQ points. 1 patient worsened by >6 TQ points
Safety	Not reported
Patients reporting serious adverse events	Not reported
Abbreviations: n, number of patients; TQ, Tinnitus Questionnaire.

Two ongoing or in‑development trials of Acoustic CR Neuromodulation for tinnitus were identified in the preparation of this briefing:

NCT01541969: Evaluation of the CR Neuromodulation Treatment for Tinnitus (RESET2) – This trial has been completed. Hoare and colleagues (2012) have published the protocol in a peer‑reviewed journal.
NCT01435317: Acoustic Coordinated Reset (CR) Neuromodulation for the Treatment of Chronic Tonal Tinnitus ("RESET Real Life") (RRL) – This trial has completed recruitment. Although the manufacturer provided an unpublished interim report of the trial during the production of this briefing, data have not been included as the information has not yet been reported publically.

Costs and resource consequences

No published evidence relating to the cost or resource consequences of using Acoustic CR Neuromodulation in the NHS was identified for this briefing.

For existing NHS audiology clinics to provide Acoustic CR Neuromodulation treatment, at least one programming station would need to be purchased as well as the neuromodulation device itself. Audiologists would need training, and appropriate calibration needs to be considered. The manufacturer offers training and calibration.

Acoustic CR Neuromodulation needs an initial assessment during which the device is programmed by a trained audiologist to match the patient's tinnitus tone. Multiple follow‑up appointments are needed to reprogramme the device. There is uncertainty as to how existing audiology services would manage the additional use of their resources.

Supporting therapies such as counselling, sound enrichment and relaxation exercises would be needed alongside Acoustic CR Neuromodulation. This technology would not replace hearing aids in people whose tinnitus is associated with hearing loss.

Strengths and limitations of the evidence

The evidence discussed comes from one published proof‑of‑concept randomised controlled trial and one single‑arm observational study reported as a conference proceeding.

The randomised controlled trial by Tass et al. (2012) has several strengths including blinding of patients, effective allocation concealment and low drop‑out rates. These strengths reduce the risk of certain types of bias. The authors were clear that this study was the first to be conducted using human subjects and as such is proof‑of‑concept. Limitations include an unclear primary outcome description and the absence of a sample size calculation. The clinicians providing care were not blinded to the treatment allocation. Intervention groups were small and unequal, and the placebo group contained only 5 patients. A rationale for the choice of treatment regimens was not provided and the authors conducted unusual statistical analysis (such as a complex technique for normalising baseline differences). A key problem of the analysis of the trial was that most of the results were baseline controlled rather than compared with the placebo arm. In addition, post‑hoc pooling of results into 'effective' and 'ineffective' stimulation groups may not be appropriate. Although the authors reported statistically significant findings, they did not explore whether these would be considered clinically significant. With these issues in mind, the results should be interpreted cautiously. The study was also funded by the device manufacturer and the authors reported conflicts of interest, which may be a source of bias.

In a journal letter Rucker et al. (2013) raised a number of concerns about the design, reporting and analysis. Tass et al. (2013) responded with a rebuttal letter to the editor.

The single‑arm observational study by Wurzer and Weimann (2011) was reported as a conference abstract; it was not peer reviewed and provided limited information on the study. The results therefore should be interpreted with caution. In addition, the authors grouped and categorised TQ scores rather than reporting means or medians, and the abstract did not include any statistical comparisons. There is significant risk of bias in the study due to the absence of a control group, blinding or allocation concealment.

Neither study looked at whether improvements in patient‑reported tinnitus symptoms were maintained for more than 4 weeks after stopping treatment with Acoustic CR Neuromodulation. Long‑term outcomes would be important for people with tinnitus.