Self-Management Support Interventions for Stroke Survivors: A Systematic Meta-Review

Background There is considerable policy interest in promoting self-management in patients with long-term conditions, but it remains uncertain whether these interventions are effective in stroke patients. Design Systematic meta-review of the evidence for self-management support interventions with stroke survivors to inform provision of healthcare services. Methods We searched MEDLINE, EMBASE, CINAHL, PsychINFO, AMED, BNI, Database of Abstracts of Reviews for Effectiveness, and Cochrane Database of Systematic Reviews for systematic reviews of self-management support interventions for stroke survivors. Quality was assessed using the R-AMSTAR tool, and data extracted using a customised data extraction form. We undertook a narrative synthesis of the reviews' findings. Results From 12,400 titles we selected 13 systematic reviews (published 2003-2012) representing 101 individual trials. Although the term ‘self-management’ was rarely used, key elements of self-management support such as goal setting, action planning, and problem solving were core components of therapy rehabilitation interventions. We found high quality evidence that supported self-management in the context of therapy rehabilitation delivered soon after the stroke event resulted in short-term (< 1 year) improvements in basic and extended activities of daily living, and a reduction in poor outcomes (dependence/death). There is some evidence that rehabilitation and problem solving interventions facilitated reintegration into the community. Conclusions Self-management terminology is rarely used in the context of stroke. However, therapy rehabilitation currently successfully delivers elements of self-management support to stroke survivors and their caregivers with improved outcomes. Future research should focus on managing the emotional, medical and social tasks of long-term survivorship.


Introduction
The incidence of stroke continues to rise in low-and middle-income countries, [1] and although it is now declining in high-income countries, demographic changes and improved survival means the overall numbers of people living with stroke is high and likely to increase. [2] One in 20 adults in high income countries are now affected by stroke, [1] and one in three stroke survivors are left permanently disabled, placing a large burden on health and social care. [3][4][5] Promotion of self-management is a core response of healthcare systems globally to the challenge of long-term condition (LTC) survivorship. [5][6][7] Currently, available support for selfmanagement ranges from the provision of disease-specific information via a website or leaflet, [8] to extensive generic programmes such as the UK Expert Patient Programme, which aims to promote behavioural change by building the confidence of individuals to manage their condition and the biopsychosocial impact of living with a LTC. [9] We adopted the holistic definition of self-management proposed by the US Institute of Medicine. [10] "Self-management is defined as the tasks that individuals must undertake to live with one or more chronic conditions. These tasks include having the confidence to deal with medical management, role management and emotional management of their conditions." Medical, role and emotional tasks have been described by Corbin and Strauss as the core components of the management of LTCs. [11] Self-management support in the context of stroke survivorship should therefore aim to empower individuals with the skills to: (1) manage medical tasks (e.g. secondary stroke prevention); (2) maintain or change behaviours or life roles (e.g. dress oneself, return to work); and (3) deal with emotional consequences of stroke survival (e.g. post-stroke depression). To facilitate these, Lorig and Holman identified five core self-management skills: problem solving; decision making; appropriate resource utilisation; forming a partnership with a healthcare provider; and taking necessary actions. [12] Self-efficacy, an individuals' confidence in their ability to carry out a certain task or behaviour, is commonly viewed as the mediator between the acquisition of self-management skills, and the enactment of self-management behaviours (see Fig 1). [13] To inform healthcare systems seeking to promote self-management, we performed a metareview of existing systematic reviews investigating stroke self-management support. The broad perspective that can be achieved by a meta-review makes the outputs particularly relevant for informing policy or clinical practice. [14] This meta-review is part of a systematic overview of the evidence for self-management support of LTCs commissioned by the National Institute for Health Research Health Services and Delivery Research Programme. [15] Methods

Search Strategy and Selection Criteria
Informed by preliminary scoping of the literature, our basic search strategy was; 'self-management support terms' AND 'stroke terms' AND 'systematic review terms'. Self-management support search terms included "confidence", "self-efficacy", "responsib Ã ", "autonom Ã ", "educat Ã ", "knowledge", "(peer or patient) ADJ1 (support or group)" and "(lifestyle or occupational) ADJ1 (intervention Ã or modification Ã or therapy)" as well as relevant MeSH terms (see Supporting information: S1 Table for full search strategy).
We searched MEDLINE, EMBASE, CINAHL, PsychINFO, AMED, BNI, Cochrane Database of Systematic Reviews, and Database of Abstracts of Reviews for Effectiveness from January 1993 to June 2012. We also hand-searched the journals BioMed Central Systematic Review, Health Education and Behaviour, Health Education Research, Journal of Behavioural Medicine and Patient Education and Counseling. A forward citation search was performed on all included reviews using ISI Proceedings (Web of Science), and all included publication reference lists were screened.
Eligibility criteria were: systematic reviews which searched for randomised controlled trials (RCTs); included individuals with a clinical diagnosis of stroke; reviewed interventions which focused on, or incorporated, strategies to support self-management (as defined above) delivered to stroke survivors, their caregivers, or both; and included outcomes on healthcare service use, health outcomes, health behaviour, quality of life, or self-efficacy of stroke survivors. We excluded: non-English publications; reviews which included a range of study designs or conditions unless they provided separate data for RCTs with stroke survivors; mono-component interventions (e.g if focused on acquiring a specific skill as opposed to broader self-management skills); or if only carer-related outcomes were reported. Quantitative Systematic Meta-Review of Stroke Self-Management Following training to establish consistent practice, the initial screening of titles and abstracts was performed by one reviewer (HLP, EE, or GP) with a 10% check by a second reviewer (HP or ST), with good inter-rater agreement (96%). Full text screening was undertaken by two reviewers (HLP, EE or GP) working independently with 81% agreement; any disagreements were re-screened by the third reviewer, and 10% were checked by a fourth reviewer (HP or ST).

Quality Appraisal, Data Extraction, Outcomes and Relevance
The quality of all included reviews was appraised using the R-AMSTAR tool, [16] by one reviewer (HLP) with a 10% check by a second reviewer (GP). A review was defined as high quality (score>40), reasonable quality (score 31-39), or low quality (score<30). (See Supporting information: S2 Table for the R-AMSTAR quality criteria).
Data were extracted by one reviewer (HLP) using a piloted data extraction table and the completed tables were checked by a second reviewer (HP) for accuracy with disagreement resolved by discussion.
We extracted the findings and conclusions as synthesised by the authors of the reviews, and specifically avoided going back to the individual RCTs. However, the aims of both the included reviews and the RCTs they included did not always completely match the aims of our metareview. We therefore assessed the potential relevance of the individual RCTs to our aim and used this, in combination with the quality assessment results, to guide the weight we attached to the conclusions of each review.
Primary outcomes of interest were those we anticipated might benefit most from a selfmanagement intervention: (1) activities of daily living (ADL); (2) extended activities of daily living (extended ADL); (3) self-efficacy; (4) community reintegration, ability to participate in work, leisure or social activities; and (5) quality of life (QOL). Secondary outcomes were; cognitive function, mood, compliance, use of care services, and poor outcome(s) or death. See Table 1 for outcome measure definitions.

Data Synthesis
Based on our preliminary scoping work, we expected substantial heterogeneity amongst included reviews, several of which would themselves include a heterogeneous group of RCTs. We therefore planned to undertake a narrative synthesis. Interpretation of results was facilitated by discussion amongst the multidisciplinary study team and an end-of-project national workshop.
the RCT interventions included in each review and an explanation of why we considered that these interventions represented components of self-management support. See Supporting information: S4 Table for further detail. Seven reviews explored interventions based on therapy rehabilitation, [17][18][19][20][21][22][23] though the focus of the interventions varied. Hoffman 2010, and Poulin 2012, looked at interventions designed specifically for people with cognitive impairment. [18,21] The remaining reviews explored therapy rehabilitation generally, [17,20] or occupational therapy (OT) specifically. [19,22,23] Self-management components in the therapy-based interventions included: problem solving; remediation training; goal setting; information provision; support with adaptive equipment; liaison with other services; and training in ADL. The majority of interventions were home-based and delivered to individuals on a face-to-face basis, though other models included delivery in an outpatient rehabilitation centre, or group setting. Delivery of the therapy rehabilitation was initiated soon after the acute stroke event in five reviews, [18][19][20]22,23] and later in stroke recovery (six months to more than one year) in two reviews. [17,21] Outcomes were measured between one week and 12 months after the end of the intervention period.
The remaining six reviews looked at various self-management support interventions including referral to stroke liaison workers, [24] information provision, [29] self-efficacy enhancement, [26] patient held records, [25] and caregiver problem solving. [27] Rae-Grant 2011 was the only review that explicitly examined self-management programmes. [28]  Frenchay Activities Index, Nottingham Extended ADL, Lawton Independent ADL scale, other unspecified EXTENDED ADL scales

Self-efficacy
The confidence that an individual has in their own ability to perform a specific task or behaviour Recovery efficacy (REFFI), Self-efficacy to perform, Selfefficacy scale

Community reintegration
The ability of individuals to reintegrate into their society, including participation in leisure or social activities or work, where relevant  Quantitative Systematic Meta-Review of Stroke Self-Management Quantitative Systematic Meta-Review of Stroke Self-Management (Continued) Quantitative Systematic Meta-Review of Stroke Self-Management Quantitative Systematic Meta-Review of Stroke Self-Management What self-efficacy enhancing interventions influence mobility, ADL, depression and HRQL?
Self-efficacy enhancing interventions for stroke patients.
Interventions must aim to increase confidence in one's ability to perform a task or specific behaviour. Interventions must also be feasible and suitable to be delivered in nursing practice.
Self-efficacy is the confidence in one's ability to perform a task or specific behaviour. A high sense of self-efficacy leads to desired outcomes.   Table) gives the results of the R-AMSTAR quality assessment and the judgements made on the relevance of the individual RCTs included within the reviews. R-AMSTAR scores ranged from 24 to 42 out of a possible total of 44. In seven reviews, [17,19,20,23,24,27,28] the majority of RCTs were deemed to be self-management interventions and so the review findings were judged to be highly relevant to our review aim. Table 4 documents the findings of each review and our interpretation of these results. The only review that searched for interventions described as self-management, [28] did not identify any RCTs delivered to stroke survivors, suggesting that there is a paucity of evidence exploring the concept of 'self-management' within stroke care.

Intervention Results
The interventions described in the different reviews were diverse but six could be grouped as therapy-based interventions. We present a synthesis of our findings below, considering first the primary and then secondary outcomes and (where relevant) sub-group analyses. Rae-Grant, 2011 [28] None identified 0 0 27

Smith, 2008 [29]
RCTs 17 9 40 Relevance of the interventions reported in the RCTs included in the systematic reviews was assessed on the basis of the detail provided in the review report. The quality of reporting details about the interventions varied between the reviews so that some judgement was required. doi:10.1371/journal.pone.0131448.t003 Quantitative Systematic Meta-Review of Stroke Self-Management Quantitative Systematic Meta-Review of Stroke Self-Management (Continued) Quantitative Systematic Meta-Review of Stroke Self-Management Found no RCTsno outcomes to report.
No RCTs were identified which studied the use of patient held medical records in stroke survivors. This highlights an area of potential stroke SM where more primary research is required. (Continued)

Chronic Disease SM Course
Only the chronic disease selfmanagement course definitely met our definition of SM support and that showed positive results on a range on health-related quality of life outcomes. However, the results from this review must be taken with caution as each subgroup represents a single study.
Found no RCTsno outcomes to report.
There is an absence of RCTs explicitly investigating stroke selfmanagement.
We take active, but not passive, information provision to be SM support.
Community reintegration 0 QoL 0 2°Mood ++ Clinically small benefit of information provision on depression (WMD -0·52; 95% CI, 0·93 to -0·10; P = 0·01) Active information provision significantly more effective than passive information for depression (P < 0·02 for all the trials), and anxiety (P < 0·05 for trials reporting dichotomous data, P < 0·01 for trials reporting continuous data) This review provides evidence that active information has a positive impact on anxiety and depression in stroke survivors Therapy Rehabilitation: Primary Outcomes The primary outcome of ADL was assessed in six reviews of therapy rehabilitation, with four of these reviews overlapping substantially in the RCTs included. [19,20,22,23] Two high quality, highly relevant reviews, [19,20] and two reviews of reasonable quality, [22,23] reported some evidence, [20,22,23] or strong evidence, [19] of beneficial effect on ADL five weeks to 12 months after intervention delivery. One review of reasonable quality which had an overlap of just one RCT found no effect on ADL. [18] The only review to search for therapy rehabilitation delivered one year post-stroke found no beneficial effect on ADL. [17] Outcomes for extended ADL were also reported in six reviews, [17,[19][20][21][22][23] with three identifying some, [19,22,23] two finding strong, [20,21] and one review finding no evidence of benefit. [17] Of these, two looked exclusively at interventions delivered in the late phase of stroke recovery; one finding no benefit, [17] and the other (based on a single study) finding strong evidence of benefit. [21] Two reviews reported measures of community reintegration, [22,23] both of reasonable quality, and both identified a significant trend favouring therapy intervention.
The three highest quality reviews, all of high relevance, reported QOL outcomes, however none demonstrated any significant benefit. [17,19,20] Mood was assessed in four reviews, including the three highest quality reviews, with no significant benefits reported. [17,19,20,23] One high quality review assessed service use and found no intervention effects. [20] Compliance was reported in one review of reasonable quality which found a significant, positive effect in one RCT. [21] Cognitive function was reported in two lower quality reviews, both of low relevance, with one finding positive effects in one RCT, [21] and the other finding no effect. [22] Therapy Rehabilitation: Secondary Outcomes The composite measure of poor outcome (deterioration in ADL, dependence/institutional care or death) was reported in the three highest quality reviews, all finding significant beneficial effects. [17,19,20] Other Models of Self-Management Support A high quality review of interactive information provision (see Table 2 for specific examples) found strong evidence of a beneficial impact on mood, though the effect was small and of doubtful clinical significance. [29] A lower quality review of interventions to enhance self-efficacy found that a chronic disease self-management course had a significant positive effect on QOL. [26] Based on only one RCT, a lower quality review exploring problem solving delivered to caregivers identified positive influences on community reintegration. [27] The remaining two reviews identified no RCTs of stroke survivors. [25,28] The review of stroke liaison workers, whilst finding no overall benefit in subjective health status, identified a significant effect on QOL for the sub-group of interventions with an emphasis on education and information provision. [24]

Sub-Group Results
Sub-groups of therapy-based interventions that appeared to have most impact on primary outcomes included comprehensive occupational therapy (as opposed to specific skills training) on ADL, [22] and face-to-face training groups (as compared to video conferenced or computerbased interventions) on problem solving self-efficacy. [21] Targeted interventions were associated with significant increases in the outcome of primary focus, but tended not to be associated with benefits in other domains. [23] These sub-group results are from reviews of reasonable quality.
Walker's 2004 high quality review of therapy rehabilitation found that effects varied by age; older patients appeared to gain more benefit in extended ADL skills than those who were younger. [23] Those with the most severe disability were found to gain least from the support interventions: stroke liaison workers reduced dependence in individuals with mild to moderate, but not severe disability. [24] Therapy rehabilitation achieved a non-significant improvement in community reintegration for patients with lower levels of dependency. [23] Discussion

Summary of Principal Findings
We found little evidence specifically using the terminology 'self-management' in the stroke literature. However, core elements of self-management support including problem solving, decision making, and goal setting are delivered to stroke survivors and their caregivers within the context of therapy rehabilitation. High quality evidence demonstrates that therapy rehabilitation incorporating these elements delivered soon after a stroke improves ADL and extended ADL and reduces the risk of 'poor outcome'. There is some evidence that early rehabilitation facilitates reintegration into the community. The limited evidence related to therapy rehabilitation delivered a year or more after the index stroke suggests some benefits on extended ADL and risk of 'poor outcome'.
The reviews exploring other forms of self-management support found evidence to suggest that active information provision has a small, beneficial effect on mood, educational support from stroke liaison services can improve QOL, and caregiver problem solving facilitated community reintegration.
The strength of evidence for these findings is summarised in Fig 3.

Strengths and Weaknesses of the Study
In addition to adhering to recommended systematic review search strategies, a strength of our methodology was the regular meetings between team members, whose multidisciplinary backgrounds encompassed public health, primary care and health psychology, enabling a balanced interpretation. By undertaking a meta-review we were able to synthesise the evidence relating to a broad range of different approaches to addressing our topic of interest, thus providing a convenient overview for policy makers, commissioners of healthcare services and clinicians to inform decisions on the provision of supported self-management for people living with the effects of a stroke. [30] However, meta-reviews of systematic reviews have some intrinsic limitations. We were reliant upon the review authors providing accurate and detailed descriptions of RCTs, and re-synthesis of materials already synthesised risks further loss of detail. To address these issues we appraised the quality of all reviews using R-AMSTAR, [16] and used these scores alongside relevance scores, to inform the weighting of evidence. Additionally, where reviews did not provide adequate narrative descriptions of interventions, we referred to tabulated details in appendices where present.
This meta-review was a commissioned, policy-focused 'rapid' review, meaning that screening and data extraction were conducted by one reviewer, and not two reviewers working independently. Whilst we acknowledge this as a potential weakness, we ensured all reviewers were trained before commencing screening, conducted a 10% check of all screening, and report agreement levels. Data extraction forms were also checked by a second reviewer to ensure data integrity. Quantitative Systematic Meta-Review of Stroke Self-Management A widely encountered problem for many review authors was the heterogeneity of RCTs which limited, or prevented, meta-analysis. This also presented challenges for our meta-review. Whilst we planned to conduct a narrative synthesis from the outset, the heterogeneity of the reviews within the 'other self-management' category limited the conclusions we could draw. However, we identified seven reviews exploring therapy rehabilitation, providing a more convincing depth of evidence.
Self-management is only one component of therapy, and the benefits we observed may relate to other aspects of the rehabilitation programme. However, in the context of a complex intervention, such as supported self-management, it is rarely possible to isolate the impact of one component from the clinical context. Our inclusion criteria ensured that all the reviews we included explicitly included trials that evaluated aspects of self-management support, and we excluded reviews reporting mono-component interventions focussing on developing a specific task. In addition, many of our outcomes reflected self-management skills such as coping with daily living and reintegration into the community.
We excluded reviews where we were unable to extract RCTs separately from other study designs. This restricted the number of reviews we were able to include, and may have resulted in the omission of important evidence. On the other hand, our strict inclusion criteria ensured that included reviews provided a high level of relevant evidence.

Evidence for Interventions
Therapy rehabilitation supports self-management. In the relatively new and emerging field of stroke self-management, the term 'self-management' was poorly recognised and infrequently utilised. As reviewers this challenged us to think reflexively and adaptively about what it really meant to self-manage. Lorig and Holman describe five skills central to self-management. These include supporting the acquisition of problem solving skills, decision making, and taking action (goal setting, or action plans), all prominent features of many stroke rehabilitation programmes. [12] In contrast to action plans in other LTCs which focus on planning for clinical emergencies, for example managing acute asthma, [31] 'taking action' in the context of stroke focuses on setting goals towards task accomplishment. The on-going symptoms of stroke survival means self-management must support individuals to cope with and adapt to disability; core aims of therapy provision.
A described element of self-management support is the forming of a patient/healthcare provider partnership. [12] Whilst this was not explicitly described in the reviews of therapy-based rehabilitation, it is a key feature in the work of OTs and other allied therapists, and may therefore be implicit in the therapy-based interventions. [32] The remaining skill described by Lorig and Holman is the ability to find and utilise resources. The provision of such information is a prominent feature of stroke liaison interventions, [24] and has been identified by stroke survivors and their caregivers as a useful service. [33] The commonalities between stroke rehabilitation programmes and self-management support have also been recognised by Jones, who noted that the aims of rehabilitation often involved increasing problem-solving self-efficacy, constructing action plans, and making decisions, all prominent elements of self-management support. [34] A stated goal of OT is to promote a sense of self-efficacy. [32] Self-efficacy beliefs are an acknowledged mediator of selfmanagement, [19] further supporting a significant role for OT in supporting self-management. Whilst our meta-review demonstrates the specific value of therapists in the context of stroke, effective implementation of self-management requires a whole systems approach in which an integrated healthcare organisation actively promotes collaborative/communicative relationships between enabled patients and motivated healthcare professionals. [35] Medical, role and emotional management. Our definition of self-management support encompasses medical, role and emotional management. [10] The main beneficial effects identified in this review (ADL, extended ADL and 'poor outcome') reflect the needs of stroke survivors in the early phase of adjustment. Our parallel synthesis of the qualitative evidence on the experiences of stroke survivors highlights the long-term and frustrating process of adjustment after stroke, and the consequent feelings of increasing social isolation. [15] The early focus mediated by rehabilitation therapists on basic function-related goals needs to merge into later interventions which support reintegration into society through supporting the adoption of more meaningful societal roles. Our data provide less clear evidence as to what format this late phase support should take, though some positive effects on community reintegration and mood were identified. [22,23,27] Emotional tasks involve being able to deal with psychological responses such as poststroke depression; only one review found a (clinically small) significant benefit on mood. [29] Current (often therapy-based) interventions are not providing adequate support to enable individuals to self-manage emotional tasks, and future interventions should address this gap.
Medical tasks were rarely explored in the included reviews, but such tasks provide the foundation of secondary stroke prevention and modification of risk factors is an important element of self-management. Lawrence and colleagues found lifestyle interventions such as diet modification and smoking cessation could affect positive behavioural change in stroke survivors; [36] more explicit support to enable individuals to adopt such behaviours should therefore be considered in future self-management support interventions.

Conclusions and Implications
In contrast to conditions such as asthma and diabetes in which the concept of self-management has been widely explored, evaluated and recommended by guidelines, [31,37] self-management terminology is rarely used in the context of stroke. However, therapy rehabilitation currently successfully delivers elements of self-management support to stroke survivors and their caregivers.
UK national clinical stroke guidelines now recommend offering all patients training in self-management skills, acknowledging the benefits to be gained by providing such support. [38] Those developing stroke self-management support interventions should recognise and respond to the changing needs of stroke survivors as they progress from the acute stroke event through early rehabilitation to long term survivorship. This should include supporting selfmanagement of more complex social roles as well as empowering stroke survivors to manage emotional and medical tasks. Research is needed to explore a new model of stroke self-management which is integrated across secondary, primary, and community care and adopts a whole systems perspective.
Supporting Information S1