Deep brain stimulation for Tourette's syndrome

Objectives

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

To assess the efficacy and harm of deep brain stimulation for motor symptoms, with psychiatric and behavioural comorbidities, either individually or in combination, in adults and adolescents with Tourette's syndrome compared to placebo, sham intervention, or the best available behavioural and pharmacological treatment.

Background

Description of the condition

Tourette's syndrome (TS), also known as Gilles de la Tourette’s syndrome, is a childhood‐onset condition with a prevalence of 0.3% to 0.9% in children (Knight 2012; Scharf 2015), and 0.002% to 0.08% in adults (Levine 2019), which presents with unwanted, involuntary movements (motor tics) and sounds (phonic tics (Johnson 2023)). Although the exact pathophysiology of TS is still not fully understood, growing evidence from genetics (Hirschtritt 2015; Yang 2021; Yu 2019), and neuroimaging (Nielsen 2020; Worbe 2015), suggests that it is a neurodevelopmental disorder with multifactorial causes (Hartmann 2018). TS is diagnosed from clinical signs and symptoms, according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM‐5), as having two or more motor tics and at least one phonic tic for at least a year, and beginning before the age of 18 years, which should not be due to medicine or other medical conditions (APA 2013). Although a family history is common (nearly 84%) and can help in diagnosis, it is not a necessary criterion for TS (Kurlan 2014; Ludolph 2012).

The motor and phonic tics of people with TS can vary in type, number, frequency, intensity, complexity, and interference (Szejko 2022a). The widely used Yale Global Tic Severity Scale (YGTSS) provides a helpful evaluation of symptoms for clinicians (McGuire 2018). Both motor and phonic tics have been further categorised into simple and complex tics, and people with TS may present with combinations of different phenotypes (Eapen 2015; Szejko 2022a). Simple motor tics are defined as rapid, brief, repetitive movements that involve a single muscle group (such as eye blinking and snapping), while complex motor tics exhibit slower, purposeless but co‐ordinated patterns of behaviours that involve multiple muscle groups, such as facial grimace, head gestures, bending, and rotating (Johnson 2023; McGuire 2018). Simple phonic tics are non‐word vocalisations (such as throat clearing, sniffing, humming, and animal noises); complex phonic tics include words, phrases, statements, and combinations of sounds (Johnson 2023; McGuire 2018). Socially inappropriate copropraxia (e.g. motor symptoms, such as obscene gestures), and coprolalia (e.g. outburst of obscene words) are observed in approximately 28.1% of the TS population; disabling tics (injurious behaviours) are observed in approximately 17%. These cause problems and even permanent harm or disfigurement (Baizabal‐Carvallo 2022; Cheung 2007; Sambrani 2016; Stafford 2020). TS and comorbid psychiatric or behavioural symptoms can result in potentially life‐threatening outcomes, and are defined as malignant TS (Cheung 2007). Premonitory urges, or unpleasant sensations often precede tics (Hallett 2015; Schubert 2021). The severity of tics tends to worsen under emotional and environmental challenges, and improve during focused attention and activities (Barnea 2016). People with TS may also express involuntary or semi‐voluntary (voluntary responses to urges or sensation) tics (Jankovic 1997); some are able to suppress their tics to varying degrees (Conelea 2018).

Psychiatric and behavioural disorders are commonly combined with TS. Previous studies suggested that 88% of the TS population experience at least one type, and 58% experience two or more psychiatric comorbidities during their lifetime (Hirschtritt 2015; Solís‐García 2021). Attention deficit hyperactivity disorder (ADHD) and obsessive‐compulsive disorder/behaviour (OCD/OCB) are the two most common comorbidities of TS; depression, anxiety, sleep disorders, self‐injurious behaviour, autism spectrum disorder, and other comorbidities are also observed (Andrén 2022; Hirschtritt 2015; Sambrani 2016). People with TS and psychiatric comorbidities experience a worse quality of life, and an even greater risk of mortality than those without psychiatric comorbidities (Huisman‐van Dijk 2019; Meier 2017).

In general, due to the unusual nature of a number of TS symptoms, people with this disorder often feel shunned in social settings. The symptoms place severe restrictions on quality of life and social engagement, further reinforcing the need to understand the most effective treatments.

Description of the intervention

Deep brain stimulation (DBS) is a minimally invasive therapy that uses implanted electrodes and an implantable pulse generator to deliver an electrical current directly to a specific part of the brain, especially the deep nucleus (Krauss 2021; Meng 2023). After surgery, it provides lasting stimulation for therapeutic effects. Its parameters are adjusted individually to achieve greater control of symptoms and reduce adverse events (Johnson 2019). It has shown promising effectiveness and an acceptably low risk of harm for several neurological conditions (Fenoy 2014; Wang 2020; Wang 2021; Wang 2023). Behavioural therapies, such as Comprehensive Behavioural Intervention for Tics (CBIT) and pharmacological treatments, such as atypical antipsychotics (i.e. aripiprazole, tiapride, and risperidone) have been effective for managing tic symptoms in people with TS (Johnson 2023; Pringsheim 2019a; Pringsheim 2019b; Roessner 2022).

Although the majority of people with TS have mild symptoms that improve with age, resulting in a favourable prognosis after behavioural and pharmacological treatments, either alone, or in combination (Pringsheim 2012; Pringsheim 2019a; Rizzo 2018), some people continue to experience ongoing and severe treatment‐refractory symptoms (Kious 2016; Macerollo 2016), or unacceptable side effects from medications (Jakobsen 2016). For these people, DBS has emerged as an increasingly popular intervention, with promising efficacy (Martinez‐Ramirez 2018; Szejko 2022b; Welter 2017). Data from the International Tourette's Syndrome Deep Brain Stimulation Public Database and Registry suggest that more than 350 people with TS have received DBS. Previous studies showed a mean improvement of 45.1% in YGTSS scores, but also important adverse events, based on 185 participants after one‐year follow‐up (Martinez‐Ramirez 2018). After a mean follow‐up of 33.7 months, 110 participants continued to show significant improvement in tics and OCB (Johnson 2019). Several randomised controlled trials (RCTs) also showed significant improvement in tics and psychiatric symptoms of efficacy, harm, and tolerability, but with mixed results (Baldermann 2021; Kefalopoulou 2015; Müller‐Vahl 2021; Welter 2017).

Currently, there are no definite conclusions about the optimal DBS targets for TS (Johnson 2023; Szejko 2022b). The most often used targets are in the centromedian thalamic region, i.e. the centromedian nucleus‐parafascicular complex (CM‐Pf), or the centromedian nucleus‐nucleus ventrooralis internus complex (CM‐Voi), and the anteromedial or posteroventrolateral part of the globus pallidus (GPi); several other targets have been suggested, such as the anterior limb of the internal capsule (ALIC), nucleus accumbens (NA), or a combination of multiple targets (Martinez‐Ramirez 2018; Szejko 2022b). Some studies suggest that different targets show comparable efficacy (Baldermann 2016; Wehmeyer 2021), while others reported differences amongst them, especially regarding their efficacy in treating psychiatric and behavioural symptoms (Servello 2020; Welter 2008). Adaptive closed‐loop DBS, based on neurophysiological biomarkers, could be a promising future direction for TS treatment, by providing on‐demand optimised stimulation (Cagle 2022).

How the intervention might work

Despite the widespread use of DBS, the underlying mechanisms of action are still debated, and remain elusive (Ashkan 2017; Okun 2012). It has been suggested that the most plausible mechanism of DBS is multifactorial and non‐exclusive, not through a single unifying mechanism (Herrington 2016). Some possible hypotheses include immediate neuromodulation effects (such as local and network‐wide electrical and neurochemical effects, and modulation of oscillatory activity through balancing excitatory and suppressed effects), synaptic plasticity, and long‐term neuronal reorganisation as neuroprotection and neurogenesis (Anderson 2004; Birdno 2008; Kim 2015; Kokkonen 2022; Okun 2012; Stefani 2017). Only a few studies have examined the potential therapeutic mechanisms of DBS in TS (Hartmann 2018; Johnson 2023; Szejko 2022b). Neuroimaging, neurobiology, and neurophysiology studies have revealed differences in functional brain connectivity (Nielsen 2020), potential abnormal cortico‐basal ganglia circuits or networks (Kataoka 2010), and neural oscillations from thalamic or pallidum regions (Neumann 2018; Zauber 2014), amongst which DBS may exert modulatory effects (Cagle 2020; Cagle 2022; Zhu 2019). Based on cohorts of people with TS who underwent DBS, Johnson and colleagues conducted several important studies. They found that the location of stimulation relative to structural anatomy alone may not predict response (Johnson 2019), the structural connectivity of stimulation targets could mediate symptom improvement, and the networks involved in tic improvement (Johnson 2020), and fibre pathway activation contributed to variable outcomes for people with TS (Johnson 2021). This network modulation effect has also been suggested in several other studies of DBS for people with TS (Baldermann 2022; Ganos 2022; Jo 2018; Morishita 2021). DBS‐induced alterations in neurotransmitters (e.g. striatal dopaminergic transmission) have also been discovered in human (Kuhn 2012; Vernaleken 2009), and animal studies (Rusheen 2023). The mechanisms of DBS in treating other neurological conditions may also be potential references for TS treatment (Casagrande 2019). Improvements in motor and psychiatric symptoms of TS may involve several different complex therapeutic mechanisms for DBS, requiring further exploration (Johnson 2023).

Why it is important to do this review

TS is a complex neuropsychiatric movement disorder that is commonly associated with psychiatric behavioural comorbidities, either alone, or in combination (Johnson 2023; Pandey 2018). With a variety of clinical manifestations and treatments, it is recognised as being at the crossroads between neurology, psychiatry, neurosurgery, and paediatrics, and requires multidisciplinary management (Müller‐Vahl 2022; Pringsheim 2019a). DBS is increasingly performed for severe, treatment‐refractory TS symptoms (Kious 2016; Macerollo 2016), or for those who experience unacceptable side effects (Jakobsen 2016). Several RCTs have reported outcomes for people with TS who received DBS (Baldermann 2021; Kefalopoulou 2015; Müller‐Vahl 2021; Welter 2017). However, critical questions remain unanswered about the efficacy of DBS in improving motor symptoms and psychiatric and behavioural comorbidities, and its harm and tolerability for people with a variety of TS symptoms (Johnson 2023; Schrock 2015; Szejko 2022b). Several systematic reviews and meta‐analyses have analysed DBS for TS; they all included both observational studies and RCTs (Baldermann 2016; Coulombe 2018; Wehmeyer 2021). Detailed complete effects, i.e. efficacy in tics and comorbidities, harms, quality of life, cognition, and tolerability, of DBS for TS were not systematically summarised in these studies.

Considering the potential risk of bias and uncertainty of the evidence, clinicians still need the synthesis of RCTs, with comprehensive evidence‐based rigorous findings to inform clinical decisions. The effects of different brain targets of DBS for TS are still a matter of debate, and newly completed trials and trials on emerging types of DBS, such as closed‐loop DBS, have not been reviewed in previous systematic reviews (Baldermann 2016; Johnson 2019; Martinez‐Ramirez 2018). Cochrane reviews use rigorous methodologies and are crucial sources to help answer clinical and research questions. With increasing evidence on DBS as a treatment for TS, a Cochrane review is important to help health professionals, people with TS and their families, and policymakers make evidence‐based clinical decisions, and solve issues of uncertainty (Cumpston 2023).

Objectives

To assess the efficacy and harm of deep brain stimulation for motor symptoms, with psychiatric and behavioural comorbidities, either individually or in combination, in adults and adolescents with Tourette's syndrome compared to placebo, sham intervention, or the best available behavioural and pharmacological treatment.

Methods

Criteria for considering studies for this review

Types of studies

We will include parallel‐group, cross‐over, or cluster‐randomised controlled trials (RCTs), of any duration, evaluating the efficacy, harm, or tolerability of deep brain stimulation (DBS) versus placebo, sham intervention, or the best available behavioural and pharmacological treatment in people with Tourette's syndrome (TS).

We will consider both open‐label and blinded (regardless of the type of blinding for participants, assessors, or both) trials. We will also consider summaries of unpublished RCTs, and RCTs published as abstracts only, after contacting the trial authors for non‐published data.

For cross‐over trials, a major concern is whether there is a sufficient wash‐out period to reduce the potential influence of carry‐over and microlesional effects. We will evaluate this and give it adequate analysis (see Unit of analysis issues).

Because of the risk of bias, we will exclude quasi‐experimental studies (with a control group that was 'as good as random' but still assigned to the intervention in a non‐randomised process, such as alternating participants sequentially, or assigning by date of birth (Bärnighausen 2017)), non‐randomised clinical trials, observational studies (cohorts, case‐controls, case series), case reports, letters, reviews, and non‐clinical studies.

Types of participants

We will include adults, aged 18 years and over, and adolescents, aged 13 to 17 years, in any setting, with a clinical diagnosis of TS and motor and phonic tics. Because there is a potential for discrepancies in characteristics and disease severity, and a potential difference in outcomes between adults and adolescents, we will analyse their data separately.

Because the DSM‐5 (APA 2013), and other recent diagnostic criteria for TS (i.e. DSM‐5‐TR (APA 2022)) would not have been used in earlier studies, we will adopt a pragmatic approach to the definition of TS. We will accept that participants who were included in RCTs with a diagnosis of TS, and who were evaluated on recognised TS criteria or a validated and fit‐for‐purpose TS‐specific severity scale, did, in fact, have TS. It is not necessary that they also have psychiatric or behavioural comorbidities.

If studies only include and report separately on a subset of relevant participants, they will still be eligible. If separate data are not available, we will contact the study authors to request relevant data.

We will impose no restrictions on the number of participants or the number of recruitment centres.

Types of interventions

Interventions

Participants will receive any type of DBS (open‐loop DBS or adaptive closed‐loop DBS designs), independent of the target nucleus, the device used, or the stimulation parameters. After being turned on, DBS will provide lasting stimulation during the intervention periods; its parameters will be routinely adjusted individually, based on symptoms and adverse events.

Comparators

Depending on the data available, we will compare DBS with either (1) placebo, (2) sham intervention, or (3) the best available behavioural, e.g. Comprehensive Behavioural Intervention for Tics (CBIT) or pharmacological treatment, such as aripiprazole, tiapride, or risperidone; their dose, frequency, or duration will depend on each participant's history and severity of symptoms.

As required by MECIR standards, we will list all treatment arms of each study in the characteristics of included studies table (Higgins 2023a).

Types of outcome measures

Reporting one or more of the outcomes listed here in the trial is not an inclusion criterion for the review (Higgins 2023a). There will be no restrictions on the follow‐up periods.

After considering clinically meaningful time points, we will group time points into prespecified intervals to represent short‐term (≤ 6 months), medium‐term (> 6 months and ≤ 12 months), and long‐term (> 12 months) follow‐up (Kefalopoulou 2015; Szejko 2022b).

We will consider participant attrition, data available for synthesis, and most concerns of the participants, clinicians, and decision makers to be crucial in the short‐term (≤ 6 months), when providing information for promoting therapeutic effects.

The outcomes must be assessed using validated assessment tools for TS by participants, clinicians, or both; subjective evaluation of clinical status will be assessed with medical records.

We anticipate that different studies may report outcomes measured with different validated tools. If studies report data for an outcome measured on more than one of these scales, we will rank the outcome measures as listed, based on the literature on core outcome sets, and clinical use. However, we will include all validated assessments, either self‐reported or reported with the assistance of the interviewer.

Primary outcomes

1. Symptoms of motor and phonic tics

   1. Yale Global Tic Severity Scale (YGTSS) – motor and phonic scores, i.e. tic scores (Leckman 1989)

   2. Tourette's Syndrome Global Scale (TSGS (Harcherik 1984))

   3. Clinical findings, such as videotape tic counts

   4. Tic Symptom Self‐Report (TSSR (Leckman 1988))

   5. Modified Rush Video‐Based Tic Rating Scale (MRVS (Goetz 1999))

2. Outcomes of psychiatric or behavioural comorbidities (or both) and emotional states

   1. Yale‐Brown Obsessive‐Compulsive Scale (Y‐BOCS (Goodman 1989))

   2. Barratt Impulsiveness Scale (BIS (Patton 1995))

   3. Hamilton Depression Scale (HAM‐D (Hamilton 1960))

   4. Beck Depression Inventory (BDI (Beck 1961))

   5. Self‐rating Depression Scale (SDS (Zung 1965))

   6. Montgomery–Åsberg Depression Rating Scale (MADRS (Montgomery 1979))

   7. Hamilton Anxiety Scale (HAM‐A (Hamilton 1959))

   8. State‐Trait Anxiety Inventory (STAI (Spielberger 1970))

   9. Beck Anxiety Inventory (BAI (Beck 1988))

   10. Self‐rating Anxiety Scale (SAS (Zung 1971))

   11. Conners’ Adult Attention Deficit Hyperactivity Disorder Rating Scale (CAARS (Conners 1999))

   12. Brief Psychiatric Rating Scale (BPRS (Lachar 1999))

3. Harms – measured by harms per participant

   1. surgery‐related adverse events, e.g. infection, haemorrhage, stroke, and death

   2. hardware‐related adverse events, e.g. hardware failure

   3. stimulation‐related adverse events, e.g. dysarthria, dyskinesia, and weight gain

We will assess the type of adverse event, its individual proportions, treatments, and prognoses (see Unit of analysis issues).

Secondary outcomes

1. Other clinical status and findings

   1. Yale Global Tic Severity Scale – total and impairment scores

   2. General clinical findings

   3. Drug usage and tolerance

   4. Health status index

2. Quality of life

   1. Gilles de la Tourette's Syndrome–Quality of Life (GTS‐QoL (Cavanna 2008))

   2. Short Form 36 (SF‐36) Quality‐of‐Life Questionnaire (Ware 1992)

3. Cognitive states – based on applicable age

   1. Wechsler Intelligence Scale for Children (WISC (Raiford 2018))

   2. Wechsler Adult Intelligence Scale (WAIS (Wechsler 2008))

   3. Montreal Cognitive Assessment (MoCA (Nasreddine 2005))

   4. Mini‐Mental State Examination (MMSE (Tombaugh 1992))

   5. Binet‐Simon Test (BST (Roid 2003))

4. Tolerability – proportion of participants who withdrew from the study or discontinued DBS due to adverse events

5. Cost‐effectiveness – evaluated when available; cost‐effectiveness of DBS for TS

Search methods for identification of studies

Electronic searches

We will conduct systematic searches of the following electronic bibliographic databases. We will start our searches in 1999, as the first reported DBS procedure for TS was performed in 1999 (Vandewalle 1999). We will not impose any language or publication restrictions. We will translate reports as needed.

• Cochrane Central Register of Controlled Trials (CENTRAL; latest issue) in the Cochrane Library

• MEDLINE Ovid (1999 to present)

• Embase Ovid (1999 to present)

• Web of Science (1999 to present)

The search strategy for MEDLINE Ovid is shown in Appendix 1; we will adapt it for use in other databases, combined with the Cochrane Highly Sensitive Search Strategies for identifying RCTs, when available (Lefebvre 2024).

Searching other resources

We will search the following trial registries for ongoing or unpublished trials.

• US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov)

• World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch)

We will cross‐check the reference lists of all eligible studies and relevant systematic reviews for potential studies. We will also contact study authors, principal investigators, and DBS manufacturers for ongoing trials.

Data collection and analysis

Selection of studies

Two review authors (SW and YZ) will independently select the studies. They will independently screen all titles and abstracts that were identified in the search, selecting those that potentially satisfy the inclusion criteria of the review. Then, they will retrieve the full text of the studies identified from the screening procedure, and further decide on inclusion or exclusion. Finally, the two review authors will cross‐check the included studies; discrepancies will be resolved by discussion, or if necessary, by consulting a third senior review author (YL) to make the final decision.

The detailed screening and selection process will be presented in a PRISMA flow diagram (Liberati 2009). The reasons for exclusion will be reported in the characteristics of excluded studies tables. Multiple reports of the same study will be considered and collated as an entire study under a single reference ID. We will use EndNote™ software to manage records and assist in deduplication (EndNote 2018).

Data extraction and management

Two review authors (SW and YZ) will independently extract the following data from each included study, in duplicate, to pre‐standardised data extraction electronic forms using Excel. The data extraction electronic forms will be piloted with a sample of three representative studies. After that, they will cross‐check their forms; disagreements will be resolved by consensus, and by involving a third review author (YL) to make the final decision. If needed, we will contact the authors of the included studies to request missing data.

1. Methods: authors, contacts, locations, and centres to conduct the study, date of the study, study design (methods in randomised; parallel‐group, cross‐over, or cluster; and open‐label or blinded), study setting (arms), wash‐out period (for cross‐over trials), withdrawals, and follow‐up periods

2. Participants: recruitment methods; inclusion, exclusion, and diagnostic criteria, number randomised, number and reasons for withdrawals, number and reasons for loss to follow‐up; number analysed, and participant characteristics (age, sex, and others)

3. Interventions: number of participants, dropouts and reasons, type of intervention (conventional open‐loop DBS, or adaptive closed‐loop DBS designs), target nucleus, device used, stimulation parameters, and co‐interventions

4. Comparisons: number of participants in each arm, dropouts and reasons, and types of comparisons

5. Outcomes: definition of outcomes, time point measurements (detailed information as previously stated in Types of outcome measures), missing outcomes, and validation of measurement tools

6. Notes: funding information, conflicts of interest, and ethical approval

The final data will be transferred and entered into Review Manager (RevMan) by SW and cross‐checked by YZ and MW (RevMan 2024).

Assessment of risk of bias in included studies

Two review authors (JZ and MW) will independently assess the risk of bias in the included studies, using the Cochrane RoB 2 tool for RCTs (Higgins 2023b; Sterne 2019). They will use the RoB 2 online Excel (www.riskofbias.info), and illustrate the summary with figures. Any disagreements will be resolved by discussion, or consultation with a third review author (FM).

We will assess the main outcomes at short‐term (≤ 6 months) follow‐up:

1. symptoms of motor and phonic tics

2. outcomes of psychiatric or behavioural comorbidities (or both) and emotional states

3. harms

In cases of missing information, we will contact the trial authors for clarification. We will use the signalling questions in the RoB 2 Excel tool and classify the risk of bias for each domain as low risk of bias, some concerns, or high risk of bias. The overall assessment will be the least favourable assessment across the domains of bias (Higgins 2023b; Sterne 2019).

1. bias arising from the randomisation process

2. bias due to deviations from intended interventions

3. bias due to missing outcome data

4. bias in measurement of the outcome

5. bias in selection of the reported result

If we include cluster‐RCTs and cross‐over RCTs, we will also assess the risk of bias domain specific to cluster‐RCTs, Bias arising from the timing of identification and recruitment of participants.

If we include cross‐over RCTs, we will also assess the risk of carry‐over effects, Bias due to deviations from intended interventions; period effects, Bias arising from the randomisation process; and the possibility of selective reporting of first period results, Bias in selection of the reported result (Higgins 2023b).

Measures of treatment effect

We will report the following measures of treatment effect from the included studies. We prefer to extract continuous outcomes when possible. We will convert all the data by measuring their effect with a consistent direction of effect (Higgins 2023c). The effect of the assignment to the interventions at baseline will be of interest in quantifying; we will undertake an intention‐to‐treat (ITT) analysis rather than a per‐protocol (PP) analysis, whenever possible (Hernán 2017).

Dichotomous data

We will record the number of events and corresponding number of participants in each study arm for each outcome to calculate the risk ratio (RR) with a 95% confidence interval (CI).

Continuous data

We will record the mean, standard deviation (SD), and total number of participants in each study arm to calculate the mean difference (MD) with 95% CI when all studies use the same measure for an outcome. If more than one study uses different scales or tools to measure the same outcome, we will use the standardised mean difference (SMD) with 95% CI.

If studies report time‐to‐event outcomes, we will use the hazard ratio (HR) with 95% CI. If data are not reported in a format that can be entered directly into a meta‐analysis, we will convert them to the required format, using the information in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2023c). If none of these are possible, we will use median and interquartile ranges (IQRs) to present the data.

Unit of analysis issues

Individual participants should be the analytical unit. When participants are allocated into multiple control groups, e.g. Control 1 and Control 2, we will attempt to obtain participant‐level data, and summarise them as one control group to avoid double counting. Similarly, adverse events should be counted at the individual level, as the number and proportion of participants with adverse events. If we include studies with multiple arms versus the control group, we will combine all arms into a single pairwise comparison.

We will analyse data from parallel‐group RCTs, preferably with an ITT approach. For cross‐over trials, we will include the first phase data, treating them as parallel‐group data. Although there is no consensus for the precise duration of the carry‐over effect of DBS, it has been proposed that a one‐week therapeutic wash‐out should be considered in DBS trials for movement disorders (Charles 2014; Hacker 2023). Therefore, we will only use the second phase data when there has been a wash‐out period of at least one week following stimulation, or studies did not provide stimulation in the first phase and their data have been adjusted for correlation factors using paired analysis.

In cluster‐randomised trials, the participating centre will be the unit of analysis, and we will analyse clusters using an estimate of the intracluster correlation coefficient (ICC) derived from the trial.

Dealing with missing data

We will contact investigators and study sponsors for additional missing data and unpublished data. If the data remain unavailable after one month has elapsed, we will only analyse the available data, and undertake a complete‐case analysis for studies with incomplete follow‐up (Deeks 2023).

We will undertake a sensitivity analysis as described in the Sensitivity analysis section to assess bias from missing data. We will also assess selective nonreporting or underreporting of results in the included studies.

Assessment of heterogeneity

We expect that the variability in the participant characteristics, e.g. different psychiatric and behavioural comorbidities, interventions, e.g. target nucleus, and outcomes will add to the clinical heterogeneity of the results. In addition, variability in study design, outcome measurement tools, and risk of bias may introduce methodological heterogeneity between the included studies. We will assess statistical heterogeneity (variability in the intervention effects, which is a consequence of clinical or methodological diversity, or both, amongst the studies) by visually inspecting forest plots for overlapping direction and magnitude of effects, and the degree of overlap between CIs, and will use the I² and Chi² (P value < 0.10 for significance) statistics to quantify inconsistency amongst the trials in each analysis. The interpretation of the I² statistic is as follows (Deeks 2023).

• 0% to 40%: might not be important

• 30% to 60%: may represent moderate heterogeneity

• 50% to 90%: may represent substantial heterogeneity

• 75% to 100%: considerable heterogeneity

Since we expect clinical and methodological heterogeneity between the included trials, we will use a random‐effect model for more conservative pooled estimates. If substantial statistical heterogeneity is identified, we will explore possible sources of heterogeneity using subgroup analyses. We will take into account any statistical heterogeneity when interpreting results, particularly when there is variation in the direction of effect.

Assessment of reporting biases

If we include at least 10 studies, we will use funnel plots to assess reporting biases for each outcome, and further test asymmetry with Egger's methods for continuous data (Egger 1997), and Begg's methods for dichotomous and time‐to‐event data (Begg 1994). When asymmetry exists, we will further investigate its reasons, such as publication bias, selection reporting bias, poor methodological design, and heterogeneity.

Data synthesis

We will follow the guidance provided by the Cochrane Handbook to undertake data synthesis using meta‐analysis (Deeks 2023), and synthesis without meta‐analysis (SWiM), either alone or in combination (McKenzie 2023). We will include all eligible studies in the primary analysis, regardless of their risk of bias. We will provide a narrative discussion of the risk of bias and explore the effect of bias with a Sensitivity analysis. We will also incorporate summary assessments of risk of bias into the GRADE assessment of the certainty of evidence for each important outcome.

We will conduct a meta‐analysis only when the included studies have adequately similar PICO elements with sufficient data (Deeks 2023). We will use a random‐effects model for a more conservative interpretation than the fixed‐effect approach, because we anticipate substantial heterogeneity between the included trials, due to varied tics, psychiatric, and behavioural symptoms in the included TS population.

Subgroup analysis and investigation of heterogeneity

We hope to explain some of the heterogeneity encountered in the included studies by undertaking these subgroup analyses, using the formal test for subgroup differences in RevMan for the primary outcomes, when possible (RevMan 2024). We note the limitations of subgroup analyses, including their observational nature and limited power to detect differences with fewer than 10 studies per category, which we will evaluate when performing subgroup analyses, and consider when interpreting the results.

1. Different psychiatric and behavioural comorbidities of participants (participants with different comorbidities may introduce variability in the effect)

2. Target nuclei, such as centromedian nucleus‐parafascicular complex, centromedian nucleus‐nucleus ventrooralis internus complex, anteromedial or posteroventrolateral part of the globus pallidus, anterior limb of the internal capsule, nucleus accumbens, and combinations of multiple targets (different target nuclei may introduce variability in the effect)

3. DBS types, such as conventional open‐loop DBS and adaptive closed‐loop DBS designs (different types of DBS designs may introduce variability in the effect)

4. Control interventions include a placebo, a sham intervention, and the best available behaviour or pharmacological treatment (different comparators may introduce variability in the effect)

5. Study design, such as parallel‐group, cross‐over, or cluster‐RCTs (different study designs may introduce variability in the effect).

Sensitivity analysis

We will conduct the following sensitivity analyses to evaluate the robustness of the pooled estimates.

1. We will exclude studies that are at high risk of bias or have some concerns in the overall assessment of risk of bias, to test the stability of the results when risk of bias is not taken into account.

2. We will exclude studies with missing data, imputed data, or with more than 20% loss to follow‐up, to test whether these factors affect the results.

3. We will repeat the meta‐analyses using a fixed‐effect model, to see if and how the findings are different from a random‐effects model.

Summary of findings and assessment of the certainty of the evidence

We will include the summary evidence for these important outcomes, at short‐term (≤ 6 months) follow‐up, in a summary of findings table. We chose them as the most important information needed for clinicians, people with TS and their families, and policymakers (Schünemann 2023).

1. symptoms of motor and phonic tics

2. outcomes of psychiatric or behavioural comorbidities (or both) and emotional states

3. harms

The comparison of DBS with the best available behavioural or pharmacological treatment, alone or in combination, will be the key summary of findings table, since it represents the most important comparison. We will add summary of findings tables for: DBS versus placebo, and CBS versus sham intervention.

We will develop the summary of findings tables using GRADEpro GDT software, and use the GRADE approach to grade the certainty of the evidence (Atkins 2004; GRADEpro GDT; Guyatt 2008; Guyatt 2011; Schünemann 2003; Schünemann 2006; Schünemann 2023). We will justify, document, and incorporate judgements into the reporting of results for each outcome, and will draw conclusions about the certainty of the evidence within the text of the review. Three review authors (SW, YZ, and MW) will independently make judgements about evidence certainty based on the five GRADE considerations (risk of bias, consistency of effect, imprecision, indirectness, and publication bias). They will resolve disagreements by discussion, or involving a third review author (JZ). Since only RCTs will be included, we will only downgrade the evidence. We will include detailed explanations in the footnotes to support the judgements, and add comments to aid the reader’s understanding where necessary.

We have confirmed that the review authors for this review will not be trial authors on any included studies. If we include studies in the future, or prior to the completion of this review, in which they are trial authors, they will not participate in the risk of bias or GRADE assessments of these studies.

Appendices

Appendix 1. Preliminary MEDLINE Ovid search strategy

1. exp Tourette Syndrome/

2. exp Tic disorders/

3. exp Tics/

4. (Tourette Syndrome* or Tic disorder* or tic*).mp.

5. or/1‐4

6. exp Deep Brain Stimulation/

7. exp Electric Stimulation/

8. (Deep Brain Stimulation or Deep‐Brain Stimulation or DBS or Electric Stimulation or Neuromodulation).mp.

9. or/6‐8

10. 5 and 9

Contributions of authors

Shu Wang: design, data collation, data analysis, finish the manuscript, revision

Yuan Zhang: data collation, finish the manuscript, revision

Minzhong Wang: design, revision

Fangang Meng: design, revision

Yali Liu: design, revision

Jianguo Zhang: design, team leader, revision

All the authors listed reviewed and approved the manuscript.

Sources of support

Internal sources

• Beijing Tiantan Hospital, Capital Medical University, China

  Support to review authors Shu Wang, Fangang Meng, and Jianguo Zhang

• Beijing Children's Hospital, Capital Medical University, National Center for Children's Health, China

  Support to review authors Yuan Zhang and Yali Liu

• Shandong Provincial Hospital Affiliated to Shandong First Medical University, China

  Support to review author Minzhong Wang

External sources

• National Natural Science Foundation of China (81830033, 81971070), China

  Support to review authors Fangang Meng and Jianguo Zhang

• National Key Research and Development Program of China (2022YFC2405100, 2016YFC0105900), China

  Support to review authors Fangang Meng and Jianguo Zhang

Declarations of interest

Shu Wang: none

Yuan Zhang: none

Minzhong Wang: none

Fangang Meng: none

Yali Liu: none

Jianguo Zhang: none

These authors should be considered joint first author

New