Types of studies

We will include randomized controlled trials or quasi‐randomized controlled trials (e.g. studies where allocation is made by medical record number). We will search for studies that compare virtual reality interventions with either an alternative intervention or no intervention. Multi‐arm studies (studies that compare different types of virtual reality with a control group) will also be included. Cross‐over trials will be excluded.

Types of participants

We will include studies of adult participants (aged 18 years or older) with a confirmed diagnosis of any type of MS (primary progressive, secondary progressive, relapsing‐remitting, and progressive‐relapsing MS), with any level of severity and at any stage of the disease. Current diagnostic criteria for MS are the revised McDonald Criteria (Thompson 2018). Depending on the type of MS, different weights are attributed to different types of information (e.g. clinical history, neurological examination, lumbar puncture, magnetic resonance imaging, evoked potentials) in order to make the diagnosis. The relative weight of these information types has changed over time with different versions of the diagnostic criteria. What has not changed is that MS is in principle a clinical diagnosis, made by a clinical neurologist. Studies that were carried out in a mixed sample of participants will be eligible if they report data on people with MS as a separate subgroup.

Types of interventions

We will include studies using VR interventions that meet the following definition: "an artificial, computer‐generated simulation or creation of a real‐life environment or situation allowing the user to navigate through and interact with" (Baus 2014). Studies using any form of non‐immersive, semi‐immersive or full‐immersive VR, and studies using commercially available gaming consoles, will be included. We will include all VR interventions that use more than one treatment session and evaluate any intensity and duration of VR. We will classify the VR interventions based on the level of immersion (full‐, semi‐ and non‐immersive) and the dosage of intervention. We will examine the included studies and choose a threshold regarding the total hours of intervention to enable comparisons between interventions of higher and lower doses interventions.

The comparison group may receive either an alternative intervention or no intervention. Since there is a broad range of alternative interventions, we will consider these to be any intervention that is designed to be therapeutic at the impairment, activity, or participation level without the use of virtual reality.

Types of outcome measures

The expected time points of evaluation of the outcomes will be as follows.

1. Within one week prior to the start of the intervention (pre‐outcome measures)

2. Within one week after the end of the intervention (post‐outcome measures)

3. Follow‐up measurements (more than one week after the intervention)

Electronic searches

We will identify trials through systematic searches of the following electronic bibliographic databases.

1. Cochrane Central Register of Controlled Trials (CENTRAL), in the Cochrane Library

2. MEDLINE (PubMed, from 1996 onwards)

3. Embase (from 1980 onwards)

4. Web of Science (from 1900 onwards)

5. PEDro (from 1999 onwards)

6. CINAHL (from 1984 onwards)

7. IEEE Xplore

The preliminary search strategy for MEDLINE (PubMed) (Appendix 1) will be adapted for use in the other databases. We will also conduct a search of ClinicalTrials.gov (www.ClinicalTrials.gov), the WHO International Clinical Trials Registry Platform (ICTRP) Search Portal (apps.who.int/trialsearch/) and the ISRCTN registry (www.isrctn.com/) for ongoing or unpublished trials. We will search all databases from their inception to the present, and we will impose no restriction on publication status. We will only include trials published in English, Dutch, French and German.

Searching other resources

In an effort to identify trials potentially missed through the database searches, trials that are ongoing or planned and trials from ‘grey literature’, we will perform an expanded search. This will include the following:

1. checking the reference lists of all included studies, texts and any relevant reviews on this topic;

2. contacting authors and researchers active in this field;

3. use of Cited Reference Search (Web of Science and PubMed) for forward‐tracking of important articles;

4. use of Google Scholar alerts for new results matching our search;

5. searching for unpublished trials or grey literature using the SIGLE database (searching for dissertations, conference proceedings, reports and trial registries).

Selection of studies

Two review authors (EDK and ES) will independently screen the titles and abstracts of the search results to identify studies for possible inclusion. If there are any disagreements, a third author will be asked to arbitrate (DB). The same two authors will then independently assess the eligibility of studies based on full text. We will document the reasons for excluding reports. We will resolve any disagreement through discussion or, if required, we will consult a third person (DB). We will record the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' table (Liberati 2009).

Data extraction and management

We will design and use a structured and standardized data collection form for study characteristics and outcome data which has been piloted on at least one study in the review. Two review authors (EDK and ES) will independently extract study characteristics from included studies. We will contact the primary authors of potentially eligible studies to provide data and clarification if the required data are absent, ambiguous, or reported insufficiently. In case data remain missing after these efforts, we will assess the impact in terms of introducing bias to our analyses, and either reject the study or solely include sufficiently reported elements.

We will extract the following study characteristics.

1. Methods: study design, study setting, inclusion criteria and exclusion criteria, method of allocation, risk of bias

2. Participant details: descriptive characteristics including age, sex, type of MS, disease duration, severity of condition (according to Kurtzke’s Expanded Disability Status Scale (EDSS) score (Kurtzke 1982)), sample size and number of dropouts

3. Interventions: description of the intervention and comparison according to the 'TIDieR' checklist (Hoffmann 2014)

4. Outcomes: outcomes specified and collected, and time points reported, adverse events

Two review authors (EDK and ES) will independently extract outcome data from included studies. We will resolve disagreements by consensus or by involving a third person (DB). One review author (EDK) will transfer data into the Review Manager 5.4 file (RevMan 2020). We will double‐check if data are entered correctly by comparing the data presented in the systematic review with the data extraction form. The results will be presented in the ‘Summary of findings’ table.

Assessment of risk of bias in included studies

Two review authors (EDK and ES) will independently assess risk of bias for each study using version two of the Cochrane 'Risk of bias' tool (RoB2), as outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019b). Disagreements between authors will be resolved by discussion or by involving another author (DB). We will assess the risk of bias of a specific result of a trial according to the following domains.

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

6. Overall bias

We will use the signalling questions and algorithms in the RoB2 tool to rate each domain as 'low risk of bias', 'some concerns' or 'high risk of bias'. We will provide information from the study, together with a justification for our 'Risk of bias' judgment, in the ‘Risk of Bias’ table. The response options for the overall 'Risk of bias' judgment are the same as for individual domains (‘low risk of bias’, ‘some concerns’ or ‘high risk of bias’), and will be considered as follows:

1. 'low' if the study is judged to be at low risk of bias for all domains;

2. 'high' if the study is judged to be at high risk of bias in at least one domain, or to have some concerns for multiple domains;

3. 'some concerns' if the study is judged to raise some concerns in at least one domain, but not to be at high risk of bias for any domain.

The risk of bias will be assessed for the five outcomes included in our 'Summary of findings' tables (see Types of outcome measures). The effect of interest will be ‘the effect of assignment’. We will manage the 'Risk of bias' assessment by using the RoB2 Microsoft Excel tool (available at www.riskofbias.info) and store the consensus decisions for the signalling questions as supplemental data.

Measures of treatment effect

Two review authors will independently classify the outcome measures in terms of the domain assessed (gait function, balance function, upper limb function, cognitive function, fatigue, global motor function, activity limitation, participation restriction and quality of life, and adverse events). Outcomes measured during the VR intervention will not be included. If a study presents more than one outcome measure for the same domain, we will include all outcome measures. In a sensitivity analysis, we will investigate whether the choice to use the outcome measure most frequently used across studies has an impact on our conclusion. We will analyse dichotomous data as risk ratios with 95% confidence intervals, and continuous data as mean difference (MD) or standardized mean difference (SMD) with 95% confidence intervals. We will enter data presented as a scale with a consistent direction of effect.

Unit of analysis issues

We will consider one unit of analysis issue in this review: the inclusion of studies with more than two intervention groups. If trials include multiple intervention groups, we will combine all relevant experimental intervention groups of the study into a single group, and all relevant comparator intervention groups into a single comparator group. To avoid double counting, we will pool the groups by combining the sample size and number of events for dichotomous outcomes and the means and standard deviations (SDs) for continuous outcomes, as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020).

Dealing with missing data

We will contact investigators or study sponsors in order to verify key study characteristics and obtain missing numerical outcome data if necessary. Where possible, we will use the Review Manager calculator to calculate missing standard deviations using other data from the trial, such as confidence intervals, based on the methods outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019a). Where this is not possible, and the missing data are thought to introduce serious bias, we will explore the impact of including such studies in the overall assessment of results by a sensitivity analysis.

Assessment of heterogeneity

We will assess heterogeneity by visual inspection of forest plots. We will use both the Chi² test and the I² statistic to quantify inconsistency among the trials in each analysis. We will consider the presence of substantial heterogeneity to be indicated by an I² value greater than 50%. If we identify substantial heterogeneity, we will report it and explore possible causes using prespecified subgroup analysis.

Assessment of reporting biases

If sufficient data are available, we will create and examine a funnel plot to explore possible publication bias. Our comprehensive search strategy, that also includes searching clinical trial registers, unpublished studies and ‘grey literature’, should assist in minimizing publication bias. We will investigate reporting bias by comparing trials with the preregistered protocol or, when the protocol is not available, by comparing the outcomes described in the methods section of trials with the results reported.

Data synthesis

We will undertake meta‐analyses only where this is meaningful, i.e. if the treatments, participants and underlying clinical question are similar enough for pooling to make sense. Where there are acceptable levels of heterogeneity, we will conduct a meta‐analysis with appropriate data using a random‐effects model in Review Manager 5.4. If, due to unacceptable heterogeneity, meta‐analysis is not appropriate, we will present a narrative summary of study results. Results of dichotomous (or binary) data will be reported as risk ratios (RRs) with 95% confidence intervals. For continuous data, we will report the treatment effect using standardized mean difference (SMD) with 95% confidence intervals where different scales were used for assessing the same outcome, and mean difference (MD) with 95% confidence intervals when studies used the same scales.

Subgroup analysis and investigation of heterogeneity

We plan to carry out subgroup analyses to determine whether outcomes vary according to the severity of MS (i.e. EDSS score of less than 6; EDSS score greater than 6), the type of intervention (i.e. level of immersion: non‐immersive, semi‐immersive, full‐immersive), and dosage of the intervention. We will compare subgroups using the ‘Test for Subgroup Differences’ in Review Manager.

Sensitivity analysis

We will conduct sensitivity analyses by excluding results with an overall assessment of 'high risk of bias'. We will consider and discuss the results of these analyses in comparison to the overall findings.

Summary of findings and assessment of the certainty of the evidence

We will create 'Summary of findings' tables for the following comparisons: 1) VR compared to conventional therapy, and 2) VR compared to usual care (thus provided as an additional intervention). The tables will include the primary outcomes (gait and balance function) as well as upper limb function, global motor function, quality of life and adverse events.

The certainty of the evidence will be assessed by two review authors (EDK and ES), using GRADE. The GRADE approach uses five domains (risk of bias, consistency of effect, imprecision, indirectness and publication bias) to assess the certainty of the body of evidence for each outcome included in the 'Summary of findings' table. The overall 'Risk of bias' judgment will be used to feed into the GRADE assessment. GRADE specifies four levels of certainty (high, moderate, low, very low). The body of evidence for randomized trials begins with a high‐certainty rating. Based on the five domains, the certainty of the evidence for a specific outcome can be downgraded. The reasons for downgrading will always be documented and classified as ‘no limitation’ (not important enough for downgrading), ‘serious’ (downgrading by one level), ‘very serious’(downgrading by two levels). We will use the online GRADEpro GDT tool to manage the certainty of evidence assessment.