Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Protocol - Intervention

Implementation interventions to promote the uptake of evidence‐based practices in stroke rehabilitation

Authors' declarations of interest

Version published: 06 March 2017 Version history

https://doi.org/10.1002/14651858.CD012575

This is not the most recent version

view the current version 15 October 2020
Collapse all Expand all

Abstract

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

The primary objective of this Cochrane Review is to assess the effects of implementation interventions to promote the uptake of evidence‐based practices (including clinical assessments and treatments recommended in evidence‐based guidelines) in stroke rehabilitation.

A secondary objective is to assess the effects of implementation interventions that are tailored towards identified barriers to change compared to non‐tailored interventions in stroke rehabilitation.

Background

Healthcare systems globally, strive for improvements in service delivery to enhance patient outcomes. Implementation interventions are a key component in achieving this aim. Recent decades have seen important advances in the field of stroke with the emergence of strong evidence for approaches to stroke recovery (Langhorne 2011; Lindsay 2014). Despite this, there have been significant delays in implementing evidence into clinical practice (Bayley 2012; Walker 2013a), and stroke survivors often do not receive care based on the best available evidence (Hall 2013; Intercollegiate Stroke Working Party 2015; National Stroke Foundation 2016).

Implementation interventions in stroke rehabilitation have received heightened attention due to persistent evidence‐practice gaps, and the need for stroke services to operate in an efficacious and fiscally responsible manner. Initiatives to promote the uptake of evidence, such as the use of clinical practice guidelines, have been investigated in systematic reviews in allied health (Hakkennes 2008; Thomas 1999), and in stroke care specifically (Donnellan 2013). Although these empirical studies supported the use of guidelines in changing the process and outcome of care, no substantial evidence is available to inform the selection of guideline implementation strategies.

Menon 2009 conducted a systematic review exploring the use of implementation strategies by rehabilitation professionals and found disparities in the literature, with some suggestion of effectiveness of multi‐faceted interventions versus passive dissemination from physiotherapy studies. A more recent review considered implementation strategies in allied health disciplines and found equivocal results for interventions used; the low methodological quality of included studies preventing the authors from recommending one implementation strategy over another (Jones 2015). Previous reviews have focused on a restricted group of health professional disciplines, and no review to date has addressed implementation specifically in stroke rehabilitation. A review of the effectiveness of implementation interventions in stroke rehabilitation will provide valuable information to stakeholders to address uncertainty regarding where to direct resources to improve the care and health outcomes for stroke survivors.

Description of the condition

Stroke is a leading cause of death and adult disability internationally (Mendis 2013). The Global Burden of Disease study reveals an increasing prevalence of stroke, caused by an epidemiological transition of increased risk factor prevalence and population ageing (Leyden 2013). Of the 15 million people worldwide who experience a stroke annually, one‐third die, and another third live with ongoing disability (WHO 2011). Those experiencing resultant disability may have impairments in physical, sensory, cognitive, or communication capacities. Poststroke disabilities negatively impact on quality of life and have major economic and societal costs (Cadilhac 2009). The overall global burden of stroke, measured as disability‐adjusted life years (DALYs) lost, is great and increasing (Feigin 2014). In Australia alone, the annual burden of disease cost is AUD 49.3 billion (Deloitte Access Economics 2016a).

Rehabilitation aims to enable individuals with stroke to reach and maintain their optimal functional levels by providing skills and tools needed to attain independence and self‐determination (WHO 2015). The contemporary approach to stroke rehabilitation is being transformed by a greater understanding of the brain's ability to reorganise following injury (neuroplasticity), and previous timeframes for therapy and expected recovery are no longer restricted to immediately after stroke (Carey 2012; Korner‐Bitensky 2013). The setting for stroke rehabilitation can range from acute inpatient care to outpatient and community settings, and services are often provided to patients in the context of a multidisciplinary team, with goal‐setting a key feature (Langhorne 2011). Clinical practice guidelines, a synthesis of research evidence, are used by health professionals to inform practice in stroke rehabilitation (Donnellan 2013). Rehabilitation to treat poststroke disability has significant individual benefits, along with economic benefits, in reducing DALYs lost (Turner‐Stokes 2016). However, little is known about the cost‐effectiveness of available treatment options or resource utilisation costs.

Description of the intervention

An implementation intervention is defined as any strategy aimed at increasing the use of research‐based knowledge in healthcare practice. Various implementation interventions can be used in stroke rehabilitation. The Cochrane Effective Practice and Organisation of Care (EPOC) Group has categorised such interventions in a taxonomy of delivery arrangements, financial arrangements, governance arrangements, and implementation strategies (EPOC 2015a). Examples of interventions include creation of new multidisciplinary teams or changing facilities or equipment (delivery arrangements), using pay for performance or external funding (financial arrangements), instituting policies for regulating training by health professionals (governance arrangements), and use of audit and feedback or local opinion leaders (implementation strategies). Interventions tailored to prospectively identify barriers to change are more likely to improve professional practice compared with no intervention or dissemination of guidelines (Baker 2015).

How the intervention might work

Implementation interventions aim to produce change in people's behaviour or the environments in which they operate, or both. Implementation interventions may target change at one or more levels (e.g. individual health professionals, teams, organisations, system) and may be tailored to overcome identified barriers to implementation. For example, audit and feedback, which involves providing a summary of clinical performance to health professionals over a specified period of time, is hypothesised to work by changing health professionals' awareness and beliefs about their current practice and subsequent consequences, changing perceived subjective norms, self‐efficacy, or by directing attention to a set of specific tasks (Ivers 2012). The use of opinion leaders is another recognised implementation intervention, where an individual in a socially influential position within a system is able to promote and affect behavioural change through informal leadership. This implementation strategy is proposed to work via persuasive communication and interpersonal skills, where opinion leaders assist others to identify best‐practice evidence and then catalyse change (Flodgren 2011). While evidence about the effects of implementation interventions across various settings, health professional groups, and clinical areas is growing (Squires 2014), there is a paucity of data on the mechanisms of action, effectiveness, and cost‐effectiveness of many implementation interventions.

Why it is important to do this review

Currently, stakeholders in stroke rehabilitation do not have a synthesised evidence base to guide implementation. The growing international interest in the use of implementation strategies for stroke care has been highlighted by a recent study of implementation methods used by 10 European countries, conducted by the European Implementation Score Collaboration (Di Carlo 2015). Results from a systematic review of implementation in rehabilitation conducted by Jones 2015, demonstrate an evolving awareness of implementation within rehabilitation professions, evidenced by the increase in implementation studies published in recent years.

Although broad reviews of implementation interventions across all health settings and clinical areas have been published (Flodgren 2011; Forsetlund 2009; Grimshaw 2005; Ivers 2012; Mickan 2011), there is variation across settings, populations, and behaviours, which results in difficulty interpreting which strategies are likely to be most effective in stroke rehabilitation. Rather, these reviews suggest that implementation interventions are likely to be efficient under different circumstances and for different clinicians. Additionally, though the efficacy of multi‐component strategies has been questioned more recently (Squires 2014), this conflicts with evidence from certain rehabilitation professions where multi‐component interventions demonstrated effectiveness in changing health professionals' behaviour (Menon 2009). A contemporary review of research evidence in this area is required.

Other rehabilitation fields have begun to investigate implementation strategies for specific practice areas, for example, spinal cord injury (Noonan 2014), however, to our knowledge there are no systematic reviews of implementation interventions in stroke rehabilitation. This is despite the fact that a large proportion of stroke survivors have unmet rehabilitation needs (National Stroke Foundation 2012), and that particular implementation interventions may be more effective in the implementation of complex, skill‐based stroke rehabilitation therapies. Despite advances in stroke rehabilitation research and evidence‐based care, translation into the clinic and community remains a challenge (Walker 2013a; Walker 2013b).

This Cochrane Review will incorporate a previous Cochrane Review on the topic of in‐hospital care pathways for stroke when considering the evidence for implementation strategies to embed care pathways within a health system (Kwan 2004). It will also compliment another Cochrane EPOC Review on implementation interventions in acute stroke units (Luker 2017).

Objectives

The primary objective of this Cochrane Review is to assess the effects of implementation interventions to promote the uptake of evidence‐based practices (including clinical assessments and treatments recommended in evidence‐based guidelines) in stroke rehabilitation.

A secondary objective is to assess the effects of implementation interventions that are tailored towards identified barriers to change compared to non‐tailored interventions in stroke rehabilitation.

Methods

Criteria for considering studies for this review

Types of studies

We will include randomised trials, including cluster and stepped wedge randomisation for cluster trials, non‐randomised trials, controlled before‐after studies, and interrupted time series studies. Non‐randomised trials will be required to have at least two intervention and two control sites to be considered eligible for inclusion (EPOC 2016b). Interrupted time series studies will be required to have a clearly defined intervention point, and at least three data points before and three after the intervention. Controlled before‐after studies will be required to have contemporaneous data collection, and at least two intervention and two appropriate control sites to meet inclusion criteria. We decided to include non‐randomised study designs due to the acknowledged complexity of stroke rehabilitation interventions and the difficulty with feasible randomisation in certain situations (Craig 2012). Where available, we will also include economic evaluations, such as cost‐effectiveness analyses, cost‐utility analyses, and cost‐benefit analyses, conducted alongside the designs specified above.

We will include full‐text studies, conference abstracts, and unpublished data, and will include studies irrespective of their publication status and language of publication.

Types of participants

Participants will be any qualified healthcare professional providing rehabilitation for stroke survivors, where qualified health professionals form the majority (> 50%) of the study population. Examples of health professionals involved in stroke rehabilitation include doctors, nurses, occupational therapists, physiotherapists, speech therapists, dieticians, social workers, psychologists, and pharmacists; although we will consider studies with any health professional working in the area of stroke for inclusion in the review. We will also include health professionals in postgraduate training, such as resident doctors. We will exclude studies focused on undergraduate students.

Defining evidence‐based practices in stroke rehabilitation for study inclusion

In the absence of a classification system for stroke rehabilitation interventions, for the purposes of this review, we will define evidence‐based practices in stroke rehabilitation by the therapy being provided, in conjunction with the setting it is provided in.

Defining therapy provided in stroke rehabilitation

We will use the World Health Organization (WHO) definition of rehabilitation to guide inclusion, where rehabilitation is: "a process aimed at enabling (individuals) to reach and maintain their optimal physical, sensory, intellectual, psychological and social functional levels. Rehabilitation provides disabled people with the tools [strategies] they need to attain independence and self‐determination" (WHO 2015). It is recommended that stroke rehabilitation involves principles of individualised goal‐setting, the involvement of a multidisciplinary team, provision of education, and encouragement of long‐term self‐management (Langhorne 2011). A range of therapies may be delivered in the context of stroke rehabilitation. Studies considered for inclusion will involve provision of evidence‐based practices in stroke rehabilitation, as defined above, and will be aimed at either the impairment level (for example, muscle power and tone, cognitive processing, or speech and language deficits), the activity level (for example, mobility, dressing, or meal preparation) or the participation level (leisure activities or employment) (WHO 2011). Addressing psychosocial issues is also of importance in stroke rehabilitation (Dewey 2007), as such, we will also include studies involving the provision of social or psychological‐based therapy (for example, counselling).

Defining therapy settings in stroke rehabilitation

Stroke rehabilitation is care provided to a patient once they are medically stable; it may be commenced soon after stroke (24 to 48 hours) and occur in acute, subacute, or community settings. We will define rehabilitation settings as the following (Turner Stokes 2015).

  • Inpatient settings: rehabilitation is provided in the context of 24‐hour care, in either a hospital ward or a specialist rehabilitation unit.

  • Outpatient or day treatment settings: rehabilitation is provided in a hospital context, a specialist subacute rehabilitation facility, or a community venue.

  • Domiciliary or home‐based settings: rehabilitation is provided in a patient's home or local community.

We will consider for inclusion qualified health professionals delivering stroke rehabilitation in any of the settings outlined above. Given the related Cochrane Review on implementation interventions in acute stroke units (Luker 2017), we will exclude studies focused solely on acute stroke units from this review. Should studies report on implementation interventions across settings (for example, for staff working in both acute stroke units and hospital rehabilitation wards), we will attempt to extract data from rehabilitation settings separate to acute stroke settings.

Types of interventions

We will include interventions designed to increase the uptake of evidence‐based practices (including clinical assessments and treatments recommended in evidence‐based guidelines) in stroke rehabilitation, and to bring about changes in the behaviour of healthcare professionals, or stroke services, or both. We will use the Effective Practice and Organisation of Care (EPOC) taxonomy to group interventions according to the following (EPOC 2015a).

  • Delivery arrangements: interventions aimed at producing change through considering how, when, and where healthcare is organised and delivered, and who delivers healthcare; for example, quality and safety systems, triage, transportation services, care pathways, or telemedicine.

  • Financial arrangements: interventions aimed at producing change through insurance schemes, considering how funds are collected, how services are purchased, and the use of targeted financial incentives or disincentives; for example, user fees or out‐of‐pocket payments, voucher schemes, or pay for performance.

  • Governance arrangements: interventions aimed at producing change through considering rules or processes that affect the way in which powers are exercised, particularly with regard to authority, accountability, openness, participation, and coherence. Examples are stakeholder involvement in policy decisions, accreditation of healthcare professionals, and policies that regulate liability for health professionals.

  • Implementation strategies: interventions aimed at producing change in healthcare organisations, the behaviour of healthcare professionals, or the use of health services by healthcare recipients; for example, audit and feedback, educational materials, educational meetings, educational outreach visits, local opinion leaders, and tailoring.

We will compare interventions to usual practice, that is, no implementation intervention or to a different implementation intervention compared to the experimental group.

Types of outcome measures

We will include studies reporting on the following outcomes.

Primary outcomes

  • Quality of care

    • Objective measures of healthcare provider adherence to evidence‐based practice, operationalised as the extent to which the health professional gave recommended advice or delivered clinical practice guideline recommended interventions (e.g. as measured by a case note audit tool).

Secondary outcomes

  • Patient outcomes

    • Objective measures of patient health behaviour, that is, adherence to recommended treatment.

    • Objective measures of patient health status and well‐being, including physical health and treatment outcomes, psychological health and psychosocial outcomes (e.g. activity of daily living (ADL) measures, stroke impact scales, quality of life measures).

  • Healthcare provider outcomes

    • Objective measures of healthcare providers reported intention to change behaviour and other hypothesised mediators, including knowledge, attitudes, and beliefs and skills.

    • Objective measures of healthcare provider satisfaction, other stakeholder or patient satisfaction, including experience of care.

  • Resource use outcomes

    • Economic outcomes, including, cost‐effectiveness, where effectiveness is measured according to the primary outcome (e.g. cost per quality‐adjusted life year (QALY) or disability‐adjusted life year (DALY)).

    • Resources needed to provide the intervention, for example, human resources, training, equipment, and supplies.

    • Where available, we will also evaluate economic measures related to the longer‐term consequences of successful or unsuccessful implementation.

We will consider impacts on equity (that is, differential impacts on disadvantaged populations) for all outcomes. We will also consider short‐term and long‐term outcomes, as available, in order to assess sustainability of any effects.

Adverse effects

We will report any undesirable effects reported in included studies, for example, adverse clinical effects, in the review.

Search methods for identification of studies

Electronic searches

A search strategy has been developed, guided by implementation literature regarding terminology and database searching (Lokker 2010; Mckibbon 2010), and in consultation with review authors and the EPOC Information Specialist. We will search the Cochrane Database of Systematic Reviews (CDSR) and the Database of Abstracts of Reviews of Effects (DARE) for related systematic reviews. We will search the following databases for primary studies.

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

  • MEDLINE, Ovid SP (1946 to present).

  • Embase, Ovid SP (1947 to present).

  • PsycINFO Ovid SP (1967 to present).

  • CINAHL (Cumulative Index to Nursing and Allied Health Literature), EBSCOhost (1980 to present).

  • PDQ‐Evidence (www.pdq‐evidence.org).

We will adapt the MEDLINE search strategy for other databases using appropriate and individualised syntax and terminology (Appendix 1). Due to the multiple, competing, and variable terms used to describe implementation, we will use proximity operators in addition to truncations.

We will use two methodological filters to limit results; the Cochrane Highly Sensitive Search Strategy (sensitivity‐ and precision‐maximising version) to identify randomised trials in MEDLINE (Higgins 2011), and an EPOC methodology filter to identify non‐randomised trial designs.

Searching other resources

Grey literature

We will conduct a grey literature search to identify studies not indexed in the databases listed above.

We will document additional sources, if any, in the review.

Trial registries

We will search the following registries for unpublished and in‐progress studies.

  • The World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp/en/).

We will conduct citation tracking for included studies and will perform reference checking on all included studies.

We will provide appendices for all strategies used, including a list of sources screened and relevant reviews/primary studies reviewed.

Data collection and analysis

Selection of studies

The EPOC Information Specialist will complete the electronic searches and provide authors with a deduplicated file of results and details of searches. Two review authors (LSC, MT) will independently screen all titles and abstracts identified and apply inclusion and exclusion criteria. Two review authors (LSC, MT) will independently screen full‐text articles and identify studies for inclusion and identify and record reasons for exclusion of the ineligible studies. Any disagreements or uncertainties regarding studies for inclusion will be resolved through discussion with a third review author (LMC, NAL or DO).

We will record the process of study selection in a PRISMA flow chart (Liberati 2009).

Data extraction and management

Two review authors (LSC, MT) will independently extract data from included studies using a modified version of the Cochrane Effective Practice and Organisation of Care (EPOC) Group data collection form (EPOC 2013a). We will pilot the form on a sample of included studies prior to formal use. Categories of items to be extracted include the following: country, setting (including time poststroke), study design, characteristics of participants, characteristics of the therapy, targeted behaviour change, implementation strategy (including rationale and theoretical underpinning, intervention components, mode/s of delivery, frequency, duration, provider characteristics, and tailoring), the comparison intervention, outcomes, costs, and any adverse events. We will rectify any disagreements with regards data extraction by consensus or third review author (LMC, NAL or DO) decision.

Assessment of risk of bias in included studies

Two review authors (LSC, MT) will independently assess the risk of bias through domain‐based evaluation, with any disagreements resolved through discussion with a third review author (LMC, NAL or DO).

Randomised trials, non‐randomised trials and controlled before‐after studies

We will assess the risk of bias using the Cochrane 'Risk of bias' tool (Higgins 2011), addressing six criteria: adequate sequence generation, adequate concealment of allocation, blinded assessment of primary outcome(s), adequately addressed incomplete outcome data, absence of selective outcome reporting, and absence of other sources of risk of bias. We will also assess four additional criteria specified by EPOC (EPOC 2016a): similar baseline characteristics, similar baseline outcome measures, reliable primary outcome measures, and adequate protection against contamination. For economic outcomes, we will use the Consensus on Health Economic Criteria (CHEC) list for assessment of methodological quality of economic evaluations (Evers 2005).

Interrupted times series studies

We will assess risk of bias using seven standard criteria (EPOC 2016a): the independence of the intervention of other changes, the shape of the intervention effect being prespecified, the unlikelihood of the intervention affecting data collection, adequate prevention of knowledge of allocated interventions, adequate consideration of incomplete outcome data, absence of selective outcome reporting, and absence of other sources of risk of bias. If the study has ignored trend changes and conducted a simple t‐test of intervention periods without additional support for this decision, we will not include the study in the review unless reanalysis is possible (EPOC 2015b).

We will judge studies as being of low, high, or unclear risk of bias for each domain listed above, in accordance with guidelines from the Cochrane Handbook for Systematic Reviews of Interventions and the Cochrane EPOC Group (EPOC 2016a; Higgins 2011).

We will represent findings for each study and outcome in a 'Risk of bias' summary figure. When considering treatment effects, we will give consideration to the risk of bias for the studies that contributed to the outcome.

Measures of treatment effect

Outcomes

We will separately analyse and report outcome data from different types of study designs. We will express healthcare provider behaviour outcomes as adherence to evidence‐based practice (including guideline recommended care). For studies reporting individual and summary measures of provider behaviour, we will use summary measures. When several outcomes are reported in a study, we will only extract results for the outcome(s) explicitly described as the primary outcome(s). When the primary outcome is not specified, we will use the outcome(s) described in the sample size calculation as the primary outcome. When the primary outcome is unclear or where the study describes several primary outcomes, we will calculate the median value across multiple outcomes. In result reporting, we will specify whether we have used the primary outcome or the outcome with the median value.

Outcomes measures of treatment effect for randomised trials, non‐randomised trials, and controlled before‐after studies

For all dichotomous outcomes, we will calculate the adjusted risk differences and 95% confidence intervals (CIs). For continuous outcomes, we will calculate mean differences (where the same tool is used to measure outcomes across studies) or standardised mean differences (where different tools are used to measure the same outcome across studies) and 95% CIs.

Outcomes measures of treatment effect for interrupted time series

We will measure interrupted time series of trends before and after intervention using regression analysis, with adjustment for autocorrelation. We will present results as changes along two dimensions: change in level, and change in slope; where change in level is the immediate effect of the intervention and change in slope is the change in trend from pre‐ to postintervention (EPOC 2015b).

Unit of analysis issues

Given the nature of implementation research, which often focuses on health groups rather than individuals, we anticipate the inclusion of cluster‐designed studies. These studies are not randomised at an individual level and may introduce differences due to chance. Moreover, standard statistical methods based on an assumption of independence are inappropriate, as multiple observations within a cluster are usually positively correlated. We will therefore, evaluate the analysis methods of clustered studies by determining: the level of analysis (i.e. individual or cluster level) and the use of statistical correction (i.e. generalised estimating equations, mixed models (random‐effects) and multilevel models). Should unit‐of‐analyses issues present, we will conduct analysis adjusting for clustering. We will seek estimates of intra‐cluster correlation (ICC), an estimate of the similarity within and between clusters (Donner 1981), from study authors, using a database of ICCs (i.e. abdn.ac.uk/hsru/research/research‐tools/study‐design) or referring to estimates from other empirical research (Ukoumunne 1999). In addition, we will consider in analysis, temporal trends in stepped wedge cluster‐randomised trials, which may introduce the confounding effect of time (Hemming 2015).

Dealing with missing data

We will attempt to obtain additional necessary and unpublished information from the authors of primary studies through personal communication.

Assessment of heterogeneity

We will quantitatively assess outcome measures via the I2 statistic to determine the extent of heterogeneity, and whether the pooling of results in a meta‐analysis is appropriate. We will consider an I2 value greater than 50% suggestive of substantial heterogeneity (Higgins 2011); in this case, we will not conduct a meta‐analysis.

Assessment of reporting biases

If a meta‐analysis is indicated and we identify a sufficient number of studies (usually 10 or greater), we will consider funnel plots in the meta‐analysis to determine the potential for effects, such as publication bias.

Data synthesis

We will separately analyse and report outcome data from different types of study designs. It is not expected studies will use common outcome measures to evaluate implementation; given this, we will use the standardised mean difference to measure the effect size. If we find adequate numbers of similar outcomes across studies, we will use a graphical display, such as a whisker plot to represent the data. We will investigate the overall treatment effect by aggregating the results using a random‐effects model.

'Summary of findings' table and GRADE

We will create a 'Summary of findings' table for the outcomes of quality of care, objective measures of patient health behaviour, objective measures of patient health status and well‐being, and objective measures of healthcare providers' reported intention to change behaviour, in order to draw conclusions about the certainty of the evidence within the text of the review. If during the review process, we become aware of an important outcome that we failed to list in our planned 'Summary of findings' table, we will include the relevant outcome and explain the reasons for this is the section 'Differences between protocol and review'.

Two review authors (LSC, MT) will independently assess the certainty of the evidence for estimates of effect using the GRADE tool and GRADEpro software (GRADEpro 2015; Guyatt 2011). We will assess the certainty of evidence for outcomes in the 'Summary of findings' table as either high‐certainty, moderate‐certainty, low‐certainty, or very low‐certainty.

We will use the five GRADE considerations (risk of bias, consistency of effect, imprecision, indirectness, and publication bias) (Guyatt 2008), and will follow methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of interventions (Higgins 2011), and the EPOC worksheets (EPOC 2013b). We will resolve disagreements regarding GRADE ratings through discussion and provide justification regarding decisions to down‐ or upgrade the ratings using footnotes in the table with comments added to aid readers' understanding of the review, where necessary. We will use plain language statements to report these findings in the review (EPOC 2013b). We will consider whether there is any additional outcome information that we were unable to incorporate into meta‐analyses, note this in the comments, and state if it supports or contradicts the information from the meta‐analyses. If it is not possible to meta‐analyse the data, we will summarise the results in the text.

Subgroup analysis and investigation of heterogeneity

We will group studies for analysis according to:

  • study design;

  • intervention type according to EPOC taxonomy (EPOC 2015a);

  • population group, i.e. adults versus children; acute/subacute stroke survivors (≤ 6 months poststroke) versus chronic stroke survivors (≥ 6 months poststroke);

  • setting for stroke rehabilitation, i.e. acute setting, inpatient rehabilitation, outpatient or home‐based setting; and

  • tailored versus non‐tailored interventions.

Sensitivity analysis

We will conduct a sensitivity analysis based on risk of bias of included studies. We will exclude studies judged to be of low quality, that is of high risk of bias in the areas of concealment of allocation, blinded assessment of primary outcome, incomplete outcome data, and selective outcome reporting.