Neuromodulation (desynchronisation) for tinnitus in adults

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

Main objective

To evaluate the clinical effectiveness of neuromodulation therapies that explicitly target pathological synchronous activity to reduce tinnitus perception in order to reduce tinnitus distress.

Secondary objectives

To evaluate the effectiveness of neuromodulation therapies in terms of reduced co-morbid symptoms and improved quality of life.

To review the physiological changes and adverse events associated with different neuromodulation (desynchronisation) therapies.

Background

Tinnitus is defined as the perception of sound in the absence of an external source (Jastreboff 2004). It is typically described by those who experience it as a ringing, hissing, buzzing or whooshing sound and is thought to result from abnormal neural activity at some point or points in the auditory pathway, which is erroneously interpreted by the brain as sound. Tinnitus can be either objective or subjective. Objective tinnitus refers to the perception of sound that can be also heard by the examiner and is usually due to blood flow or muscle movement (Eggermont 2010). Most commonly, however, tinnitus is subjective; the sound is only heard by the person experiencing it and no source of the sound is identified (Jastreboff 1988).

Subjective tinnitus affects 10% of the general population, increasing to as many as 30% of adults over the age of 50 years (Davis 2000; Møller 2000). It can be experienced acutely, recovering spontaneously within minutes to weeks, but is considered chronic and unlikely to resolve spontaneously when experienced for six months or more (Hahn 2008; Hall 2011; Rief 2005).

In England alone there are an estimated ¾ million GP consultations every year where the primary complaint is tinnitus (El-Shunnar 2011), equating to a major burden on healthcare services. For many people tinnitus is persistent and troublesome, and has disabling effects such as insomnia, difficulty concentrating, difficulties in communication and social interaction, and negative emotional responses such as anxiety and depression (Andersson 2009; Crönlein 2007; Marciano 2003). In approximately 90% of cases, chronic tinnitus is co-morbid with some degree of hearing loss, which may confound these disabling effects (Fowler 1944; Sanchez 2002). An important implication in clinical research, therefore, is that outcome measures need to distinguish benefits specific to improved hearing from those specific to tinnitus.

This is one of a number of tinnitus review protocols produced by the Cochrane Ear, Nose and Throat Disorders Group, which use a standard Background. The following paragraph (‘Description of the condition’) is based on an earlier review by the same lead author and reproduced with permission: Hoare 2014.

Description of the condition

Diagnosis and clinical management of tinnitus

There is no standard procedure for the diagnosis or management of tinnitus. Practice guidelines and the approaches described in studies of usual clinical practice typically reflect differences between the clinical specialisms of the authors or differences in the clinical specialisms charged with meeting tinnitus patients’ needs (medical, audiology/hearing therapy, clinical psychology, psychiatry), or the available resources of a particular country or region (access to clinicians or devices, for example) (Biesinger 2010; Cima 2012; Department of Health 2009; Hall 2011; Henry 2008; Hoare 2011). Common across all these documents, however, is the use or recommendation of written questionnaires to assess tinnitus and its impact on patients by measuring severity, quality of life, depression or anxiety. Psychoacoustic measures of tinnitus (pitch, loudness, minimum masking level) are also recommended. Although these measures do not correlate well with tinnitus severity (Hiller 2006), they can prove useful in patient counselling (Henry 2004), or by demonstrating stability of the tinnitus percept over time (Department of Health 2009).

Clinical management strategies include education and advice, relaxation therapy, tinnitus retraining therapy (TRT), cognitive behavioural therapy (CBT), sound enrichment using ear-level sound generators or hearing aids, and drug therapies to manage co-morbid symptoms such as insomnia, anxiety or depression. The effects of these management options are variable and they have few known risks or adverse effects (Dobie 1999; Hoare 2011a; Hobson 2010; Martinez-Devesa 2010; Phillips 2010).

Pathophysiology

Most people with chronic tinnitus have some degree of hearing loss (Ratnayake 2009), and the prevalence of tinnitus increases with greater hearing loss (Han 2009; Martines 2010). The varying theories of tinnitus generation involve changes in either function or activity of the peripheral (cochlea and auditory nerve) or central auditory nervous systems (Henry 2005). Theories involving the peripheral systems include the discordant damage theory, which predicts that the loss of outer hair cell function, where inner hair cell function is left intact, leads to a release from inhibition of inner hair cells and aberrant activity (typically hyperactivity) in the auditory nerve (Jastreboff 1990). Such aberrant auditory nerve activity can also have a biochemical basis, resulting from excitotoxicity or stress-induced enhancement of inner hair cell glutamate release with upregulation of N-methyl-D-aspartate (NMDA) receptors (Guitton 2003; Sahley 2001).

In the central auditory system, structures implicated as possible sites of tinnitus generation include the dorsal cochlear nucleus (Middleton 2011; Pilati 2012), the inferior colliculus (Dong 2010; Mulders 2010), and the auditory and non-auditory cortex (discussed further below). There is a strong rationale that tinnitus is a direct consequence of maladaptive neuroplastic responses to hearing loss (Møller 2000; Mühlnickel 1998). This process is triggered by sensory deafferentation and a release from lateral inhibition in the central auditory system allowing irregular spontaneous hyperactivity within the central neuronal networks involved in sound processing (Eggermont 2004; Rauschecker 1999; Seki 2003). As a consequence of this hyperactivity, a further physiological change noted in tinnitus patients is increased spontaneous synchronous activity occurring at the cortical level, measurable using electroencephalography (EEG) or magnetoencephalography (MEG) (Dietrich 2001; Tass 2012; Weisz 2005). Another physiological change thought to be involved in tinnitus generation is a process of functional reorganisation, which amounts to a change in the response properties of neurons within the primary auditory cortex to external sounds. This effect is well demonstrated physiologically in animal models of hearing loss (Engineer 2011; Noreña 2005). Evidence in humans, however, is limited to behavioural evidence of cortical reorganisation after hearing loss, demonstrating improved frequency discrimination ability at the audiometric edge (Kluk 2006; McDermott 1998; Moore 2009; Thai-Van 2002; Thai-Van 2003), although Buss 1998 did not find this effect. For comprehensive reviews of these physiological models, see Adjamian 2009 and Noreña 2011.

It is also proposed that spontaneous hyperactivity could cause an increase in sensitivity or ‘gain’ at the level of the cortex, whereby neural sensitivity adapts to the reduced sensory inputs, in effect stabilising mean firing and neural coding efficiency (Noreña 2011; Schaette 2006; Schaette 2011). Such adaptive changes would be achieved at the cost of amplifying ‘neural noise’ due to the overall increase in sensitivity, ultimately resulting in the generation of tinnitus.

Increasingly, non-auditory areas of the brain, particularly areas associated with emotional processing, are also implicated in bothersome tinnitus (Rauschecker 2010; Vanneste 2012). Vanneste 2012 describes tinnitus as “an emergent property of multiple parallel dynamically changing and partially overlapping sub-networks”, implicating the involvement of many structures of the brain more associated with memory and emotional processing in tinnitus generation. However, identification of the structural components of individual neural networks responsible for either tinnitus generation or tinnitus intrusiveness, which are independent of those for hearing loss, remains open to future research (Melcher 2012). One further complication in understanding the pathophysiology of tinnitus is that not all people with hearing loss have tinnitus and not all people with tinnitus have a clinically significant hearing loss. Other variables, such as the profile of a person’s hearing loss, may account for differences in their tinnitus report. For example, König 2006 found that the maximum slope within audiograms was higher in people with tinnitus than in people with hearing loss who do not have tinnitus, despite the ‘non-tinnitus’ group having the greater mean hearing loss. This suggests that a contrast in sensory inputs between regions of normal and elevated threshold may be more likely to result in tinnitus.

Description of the intervention

Neuromodulation is taken to imply a type of prescribed intervention where the desired outcome is to “interfere on some level with the nervous system of a patient so as to modify it to affect benefit for the patient” (Arle 2011). Neuromodulation therapies, typically involving electrical stimulation of the cortex, deep brain or vagus nerve, have been used in the treatment of neuropsychiatric conditions including Parkinson’s disease, epilepsy, depression and intractable pain (Adamchic 2014; Bergey 2013; Levy 2010; Riva-Posse 2013). For tinnitus, where pathological neuronal activity might be associated with auditory areas of the brain, interventions using either electrical or acoustic stimuli separately, or paired electrical and acoustic stimuli, have been proposed as treatments. Interventions using electromagnetic stimuli, such as repetitive transcranial magnetic stimulation (rTMS), are considered in another Cochrane review, Meng 2011, and will not be reviewed here.

Potential therapies where electric stimuli are delivered to achieve neuromodulation include those using surface electrodes (e.g. transcranial direct or alternating current stimulation; tDCS, tACS; Antal 2004; Zaehle 2010), using electrodes implanted epidurally (De Ridder 2004), or vagal nerve stimulation (VNS) using surgically implanted (Engineer 2013) or transcutaneous (tVNS; Kreuzer 2014; Lehtimäki 2013) electrodes, or cochlear implants (Song 2013). Where there is sufficient residual hearing, neuromodulation therapies using only acoustic stimuli delivered via ear-level devices may be used (e.g. Co-ordinated Reset (CR®) Neuromodulation; Tass 2012). A further class of therapy uses both acoustic and electrical stimuli to drive neuromodulation (e.g. an ear-level device delivering acoustic stimuli paired with electrical vagus nerve stimulation (VNS); Engineer 2013).

This review will focus on such neuromodulation therapies that demonstrably, or are hypothesised to, interrupt pathological synchronous cortical activity associated with tinnitus. Whilst both increased spontaneous firing and synchrony of firing within and across assemblies of cortical neurons are considered neural correlates of tinnitus (Roberts 2010), increased synchrony is considered a catalyst for tinnitus that propagates tinnitus-related neural activity within and beyond auditory areas (Noreña 2013). This pathological synchrony associated with tinnitus can be represented in the oscillatory brain activity recorded with EEG or MEG so a physiological effect of any intervention targeting this mechanism can be measured objectively.

How the intervention might work

Neuromodulation therapies should deliver a permanent reduction in tinnitus percept by driving the neuroplastic changes necessary to interrupt abnormal levels of oscillatory cortical activity and restore typical levels of activity. This change in activity should alter or interrupt the tinnitus percept (reduction or extinction) and this should be concomitant with a change in the level of self reported tinnitus handicap.

At a cortical level the effects of neuromodulation therapies to be reviewed should result in changes in oscillatory activity. Exactly what neurophysiological change is necessary for a change in tinnitus perception is not entirely clear. Using MEG, Weisz 2005 and Weisz 2007 found increased oscillatory power in delta (1 Hz to 3 Hz) and gamma frequency bands (40 Hz to 90 Hz), and decreased power in the alpha band (8 Hz to 12 Hz), in people with tinnitus compared to a control group without tinnitus. Differences were most prominent in the temporal regions. In their next study, they showed there to be a reduction in delta but not gamma band activity during residual inhibition (Kahlbrock 2008). Adjamian 2012 similarly reported that (MEG) delta band power was higher in participants who had tinnitus and normal hearing compared to participants who reported no tinnitus or hearing loss, and that masking the tinnitus percept reduced delta activity. In this study, however, gamma band activity did not correlate with tinnitus or hearing loss. Recently, Schlee 2014 explored auditory alpha activity using MEG, confirming it to be reduced in tinnitus and further demonstrating that variability in alpha related to duration of tinnitus. EEG studies corroborate some of these effects but not others. For example, Tass 2012 associated residual inhibition (reduced tinnitus percept) with reduced delta band power, but Ashton 2007 reported increased gamma power but no abnormality in delta and alpha band power in people with tinnitus. Taken together, changes in delta band oscillatory activity seem to provide the best predictor of changes in tinnitus percept, but changes in gamma band oscillatory activity should be interpreted with caution.

The neural mechanisms of neuromodulation differ according to the intervention. For example, tDCS affects change via a sustained polarising or depolarising effect on neuronal membrane potentials, and is observed to modulate gamma band activity (Antal 2004). On the other hand, tACS is hypothesised to affect neuromodulation via the up- and down-regulation of synapses, and there are pilot data to suggest this affects changes in alpha band activity (Ironside 2013), but a reproducible effect on oscillatory activity has yet to be demonstrated (Vanneste 2012a).

Neuromodulation therapy using only acoustic stimuli emerged from work to develop ‘co-ordinated reset’ (CR) technology using deep brain stimulation for the treatment of Parkinson’s disease, which is now at proof of concept (Adamchic 2014). CR® neuromodulation relies on a phase reset of the oscillatory dynamics through repeated patterns of stimulation of neuronal subpopulations which, over time, change mean synaptic weight (spike timing dependent plasticity) to promote stable desynchronised activity (Tass 2012a). Tass 2009 confirmed the possibility of sustained desynchronisation between neuronal populations after multi-site electrical stimulation in a rat model of epilepsy, and using a computational modelling approach predicted how a desynchronising sound sequence could be used to interrupt or ‘unlearn’ pathological synchronous network activity associated with tinnitus (Tass 2012a).

The potential for neuromodulation therapies that use a combination of acoustic and electrical stimuli, where temporal coincidence of the two stimuli is essential, developed from promising observations in the animal literature. In a study involving rats with noise-induced hearing loss and a presumed noise-induced tinnitus in the region of 8 kHz to 10 kHz, Engineer 2011 showed that a regime of acoustic stimulation using multiple tones outside the tinnitus frequency region, paired with VNS, either prevented or reversed both behavioural and physiological markers of tinnitus. This included reversal of increased neural synchrony in the primary auditory cortex. Acoustic stimuli were necessary to target neural populations within the auditory cortex, but VNS was concluded essential to promoting plastic change through a “synergistic action of multiple neuromodulators” (Engineer 2011), with a likely significant role for acetylcholine (Engineer 2013). Acoustic stimulation or VNS alone were insufficient for these changes to occur. This approach has shown safety and efficacy in humans. De Ridder 2012 implanted electrodes on the left vagus nerve of 10 patients with severe chronic tinnitus, and delivered VNS paired with tones for 2.5 hours per day over 20 days. Some patients reported an improvement in tinnitus severity immediately after therapy, and across a subgroup tested with EEG, delta and theta band power was decreased during therapy trials compared to sham trials. tVNS paired with notched music has also shown proof of concept (Lehtimäki 2013), however, consistent with the animal literature, tVNS without concurrent acoustic stimuli appears to have no effect on tinnitus (Kreuzer 2014).

Why it is important to do this review

Currently there is no single treatment that has a replicable long-term effect on tinnitus (specifically, a reduction in the tinnitus perception) that is superior to placebo. Clinical management of tinnitus is still largely symptomatic, to reduce intrusiveness or reduce the negative emotional reactions that lead to distress. In contrast, neuromodulation therapies aim to interrupt neuronal activity associated with tinnitus to effect a permanent change in this activity, with a consequent reduction in tinnitus distress. Claims about permanent reductions in tinnitus, if substantiated, would set these therapies apart from other options, such as hearing aids or sound generators, which might only provide benefit during wearing. However, approaches to neuromodulation are varied and evolving. It is likely that studies will use different protocols, so there is a need to document systematically how developments translate to changes in estimates of effect.

Evidence for the relative effectiveness of neuromodulation therapies needs to be carefully accounted, particularly given the large placebo effect of many proposed tinnitus interventions. At the same time, interventions that claim to modulate neural activity cannot be regarded as benign and adverse events or reactions need careful reporting. There has never been a systematic review on the effects of neuromodulation by direct electrical and/or acoustic stimulation for the treatment of tinnitus. This review is important, therefore, as a synthesis of the data that currently exist and as a document that will monitor progress and research quality in the field, to inform future activity and decision-making in research and in health care.

Objectives

Main objective

To evaluate the clinical effectiveness of neuromodulation therapies that explicitly target pathological synchronous activity to reduce tinnitus perception in order to reduce tinnitus distress.

Secondary objectives

To evaluate the effectiveness of neuromodulation therapies in terms of reduced co-morbid symptoms and improved quality of life.

To review the physiological changes and adverse events associated with different neuromodulation (desynchronisation) therapies.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs).

Types of participants

Adults (≥ 18 years) with subjective tinnitus.

Types of interventions

Therapies that are explicitly used for the modulation of neural activity associated with tinnitus to deliver a sustained desynchronisation of that activity. Based on the current literature this includes six forms of neuromodulation intervention: transcranial direct current stimulation (tDCS), transcranial alternating current stimulation (tACS), vagal nerve stimulation (VNS), transcutaneous vagal nerve stimulation (tVNS), cochlear implantation, acoustic neuromodulation therapy, and paired electrical and acoustic stimulation therapy. We will update this list as and if new forms of the intervention are proposed. We will consider each form of intervention separately in the review.

The primary comparison of interest is neuromodulation therapy versus a placebo or no intervention control.

Of secondary interest is the relative effectiveness of individual neuromodulation therapies compared to each other or to current standard therapy options, which include hearing aids, sound generators, psychological or education-based therapies, or combinations of therapies.

Types of outcome measures

We will analyse the following outcomes in the review, but they will not be used as a basis for including or excluding studies.

Primary outcomes

• innitus symptom severity, measured as the global score on a multi-item tinnitus questionnaire listed in Table 1. We will update this list on an ongoing basis whenever other questionnaires are introduced.

Table 1. Tinnitus Questionnaires.

Questionnaire (author, year)	Number of items and sub-scales	Psychometric properties
Tinnitus Handicap Inventory (Newman 1996)	25 items, 3 subscales	a = 0.93 for total scale
Tinnitus Functional Index (Meikle 2012)	25 items, 8 subscales	a = 0.97 for total scale
Tinnitus Handicap Questionnaire (Kuk 1990)	27 items, 3 subscales	a = 0.93 for total scale
Tinnitus Questionnaire (Hallam 1996)	52 items, 5 subscales	a = 0.91 for total scale; for subscales a = 0.76 to a = 0.94
Tinnitus Reaction Questionnaire (Wilson 1991)	26 items, 4 subscales	a = 0.96 and a test-retest correlation of r = 0.88
Tinnitus Severity Scale (Sweetow 1990)	15 items	Not reported

Secondary outcomes

• Generalised anxiety

• Generalised depression

• Generalised quality of life

All measured using multi-item questionnaires.

• Neurophysiological change (change in function as measured by MEG or EEG)

• Adverse effects

Search methods for identification of studies

We will conduct systematic searches for randomised controlled trials. There will be no language, publication year or publication status restrictions. We may contact original authors for clarification and further data if trial reports are unclear or unpublished, and we will arrange translations of papers where necessary.

Electronic searches

We will identify published, unpublished and ongoing studies by searching the following databases from their inception: the Cochrane Register of Studies online (the Cochrane Ear, Nose and Throat Disorders Group Register and CENTRAL, current issue); PubMed; EMBASE; CINAHL; LILACS; KoreaMed; IndMed; PakMediNet; CNKI; AMED; PsycINFO; Web of Science; ClinicalTrials.gov; ICTRP; Google Scholar and Google.

We will model subject strategies for databases on the search strategy designed for CENTRAL (Appendix 1). Where appropriate, we will combine subject strategies with adaptations of the highly sensitive search strategy designed by The Cochrane Collaboration for identifying randomised controlled trials and controlled clinical trials (as described in theCochrane Handbook for Systematic Reviews of Interventions Version 5.1.0, Box 6.4.b. (Handbook 2011)).

Searching other resources

We will scan the reference lists of identified publications for additional trials and contact trial authors if necessary. In addition, we will search PubMed, TRIPdatabase, The Cochrane Library and Google to retrieve existing systematic reviews relevant to this systematic review, so that we can scan their reference lists for additional trials. We will search for conference abstracts using the Cochrane Ear, Nose and Throat Disorders Group Trials Register and EMBASE.

Data collection and analysis

Selection of studies

DJH and JAH will independently review all studies retrieved to determine their eligibility for inclusion in the review. Any disagreements between authors will be discussed between all four authors until a consensus is reached.

Data extraction and management

GWS, JAH and DW will independently extract data using a purposefully designed data extraction form. We will pilot the data extraction form on a subset of articles and revise it as indicated before formal data extraction begins.

Information to be extracted will include: trial design, setting, methods or randomisation and blinding, power, inclusion and exclusion criteria, type of intervention and control, treatment duration, treatment fidelity, type and duration of follow-up, and outcome measures and statistical tests.

Data to be extracted will include: baseline characteristics of participants (age, sex, duration of tinnitus, tinnitus severity, tinnitus loudness and pitch estimates, details of co-morbid hearing loss, anxiety or depression), and details of any attrition or exclusion.

Outcome measure data to be extracted will include: group mean and standard deviation at pre- and post-intervention and follow-up, and results of any statistical tests of between group comparisons.

We will also contact authors where further information is required that is not contained within the trial publication or in an accessible database. If not reported or provided by the authors we will estimate standard deviations in RevMan 2014 using the available data, such as standard errors, confidence intervals, P values and t values. Where data are only available in graph form, authors will make and agree numeric estimates.

After independent data extraction by GWS, JAH and DW, all authors will review the extracted data for disagreements, and revisit and discuss the relevant studies as required to reach a final consensus.

Assessment of risk of bias in included studies

DW and GWS will undertake assessment of the risk of bias of the included trials independently, with the following taken into consideration, as guided by theCochrane Handbook for Systematic Reviews of Interventions (Handbook 2011):

• sequence generation;

• allocation concealment;

• blinding;

• incomplete outcome data;

• selective outcome reporting; and

• other sources of bias.

We will use the Cochrane ‘Risk of bias’ tool in RevMan 5.3 (RevMan 2014), which involves describing each of these domains as reported in the trial and then assigning a judgement about the adequacy of each entry: ‘low’, ‘high’ or ‘unclear’ risk of bias.

Measures of treatment effect

We will analyse dichotomous data as risk ratios (RR) with 95% confidence intervals (95% CIs). We will summarise continuous outcomes as mean difference (MD) with 95% CI. We will use standardised mean difference (SMD) (Cohen’s d effect size (ES)) when different questionnaires are used to measure the same outcome. A positive effect size indicates that the treatment group achieved better outcomes than the control group.

Unit of analysis issues

For parallel-group RCTs the unit of analysis will be the group mean. However, some studies included in the review may involve clustering (for example, a group counselling intervention) or compare more than two intervention groups. To avoid unit of analysis errors we will consider alternative analyses for cluster-randomised trials and for studies with more than two intervention groups. For cluster-randomised trials we will adopt approximate analyses - effective sample sizes (Donner 2002). For studies with more than two intervention groups we will either combine groups to create a single pair-wise comparison or, if this is not appropriate, select the most relevant pair of interventions for comparison.

Dealing with missing data

Where possible we will contact the original investigators to request missing data. If this option is not available then we will clearly state the methods used to deal with the missing data. We will also perform sensitivity analyses to assess how sensitive results are to reasonable changes in the assumptions that are made. We will address the potential impact of missing data on the interpretations made in the review’s discussion.

Assessment of heterogeneity

We will calculate heterogeneity of aggregated effect sizes using Cochran’s Q statistic (Chi² test with K-1 degrees of freedom, where K is the number of studies) and the I² statistic (percentages of around 25%, 50% and 75% of I² would mean low, medium and high heterogeneity, respectively (Huedo-Medina 2006).

Assessment of reporting biases

For each neuromodulation intervention, we will investigate potential publication bias and the influence of individual studies on the overall outcome identified in this review. We will search for and request study protocols for the included studies and, where available, we will evaluate whether there is evidence of selective reporting.

Data synthesis

We will analyse separately the different forms of neuromodulation therapy. If more than one study is identified for a given form of neuromodulation therapy, and if combining studies is appropriate, we will use RevMan 5.3 to perform meta-analyses (RevMan 2014). We will pool data from randomised controlled trials using a fixed-effect model, except when heterogeneity is found. We will pool dichotomous data using the RR measure, while we will pool continuous data using the SMD measure, if more than one questionnaire is used to measure the same outcome. We will consider the psychometric properties of questionnaires with regard to their suitability for pooling. For our primary outcome (tinnitus symptom severity) data will only be included from multi-item questionnaires that show similar responsiveness and can be assumed to measure the same underlying construct of tinnitus symptom severity (high convergent validity) as other multi-item tinnitus questionnaires. We will take the same approach for secondary outcomes, although we anticipate that in both cases some essential information (construct validity, responsiveness) may be missing from the literature. DJH and JAH will independently examine the information available for each instrument, and categorise the instruments for potential inclusion in meta-analyses. GWS will collate these independent categorisations and all three authors will reach a consensus before any analyses are performed.

Subgroup analysis and investigation of heterogeneity

We intend to explore the following potential sources of heterogeneity using subgroup analyses: age, sex, hearing loss, baseline tinnitus severity, baseline anxiety or depression.

Sensitivity analysis

We will perform sensitivity analysis to explore whether any significant heterogeneity was a result of high risk of bias. If that is the case then we will exclude the lowest quality trials.

GRADE and ‘Summary of findings’

DJH and DW will use the GRADE approach to independently rate the overall quality of evidence for each outcome. JAH will then review the ratings and discuss any disagreements with DJH and DW, involving GWS as required until a consensus is reached. The quality of evidence reflects how confident we are that an estimate of effect is correct. We will apply this to our interpretation of results. The four possible ratings are: high, moderate, low, or very low. Rating as high-quality evidence implies confidence in the estimate of effect and that further research is very unlikely to change our confidence in the estimate. If a study is rated as very low quality this would imply that the estimate of effect is very uncertain.

The GRADE approach rates evidence from RCTs that do not have serious limitations as high quality. However, several factors can lead to the downgrading of the evidence to moderate, low or very low. The degree of downgrading is determined by the seriousness of the these factors: study limitations (risk of bias); inconsistency; indirectness of evidence; imprecision; and publication bias.

We will include a ‘Summary of findings’ table (Handbook 2011), and we will use the GRADE considerations to separately assess the quality of the body of evidence for each intervention and primary outcome, and to draw conclusions about the quality of evidence in the review.

Appendix 1. CENTRAL search strategy

#1 MeSH descriptor: [Tinnitus] explode all trees

#2 tinnit*

#3 #1 or #2

#4 MeSH descriptor: [Transcutaneous Electric Nerve Stimulation] explode all trees

#5 MeSH descriptor: [Vagus Nerve Stimulation] explode all trees

#6 MeSH descriptor: [Acoustic Stimulation] explode all trees

#7 MeSH descriptor: [Electric Stimulation] this term only

#8 MeSH descriptor: [Electric Stimulation Therapy] this term only

#9 tDCS or tACS or VNS or tVNS or TENS or TNS or ETS or TES

#10 neuromodulat* or “coordinated reset*” or “co-ordinated reset*” or cerbomed or NEMOS or desychronisation or InterX or Neuromomics or desychronization or neuro-modulat* or “neuro modulat*” or tinnipad or “Soterix Medical” or MicroTransponder or “Serenity System” or Cyberonics or DS8000 or neuroConn or T30 or serenade or soundcure or ANM or neurotherap*

#11 (vagus or vagal or transcutaneous or transcranial or Transdermal or Percutaneous or cutaneous or DC or AC or CR) near (stimul*)

#12 (direct or alternating) near current near stimul*

#13 Electrostimul* or neurostimulation or electro-stimul* or neuro-stimul* or “electro stimul*” or “neuro stimul*”

#14 “electrical stimul*” or “electric stimul*” or “acoustic stimul*” or “auditory stimul*”

#15 #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14

#16 #3 and #15