Self-management interventions to improve mobility after stroke: an integrative review

Abstract

Purpose

To review the evidence around self-management interventions used to improve mobility post-stroke.

Materials and methods

An integrative review was carried out. Eight databases were searched from 1992 to July 2021 using keywords based on the PICOS strategy. Two reviewers independently screened and extracted the relevant data. Quality of studies was assessed and a quantitatively led narrative synthesis of data, supported by qualitative evidence, was then conducted.

Results

Twenty-four studies with 823 participants were reviewed. Self-management strategies such as patient education, providing information, goal setting, problem-solving, action planning, self-monitoring, and social support were integrated with rehabilitation therapy to improve mobility post-stroke. The reviewed studies showed improvements in functional mobility and walking ability, self-efficacy, participation in physical activity, and quality of life to various extents. Participants in qualitative studies considered the self-management interventions as a valuable addition to their therapy and perceived the improvement in their mobility following them.

Conclusion

There is some evidence that self-management interventions help to improve mobility outcomes post-stroke. Heterogeneity of data in the studies made meta-analysis impossible. Most of the identified studies examined the feasibility and fidelity of the interventions and further research is warranted to examine the efficacy of these interventions to improve functional mobility post-stroke.

Implications for rehabilitation

• Self-management interventions can improve mobility-related outcomes, which are considered a priority goal for many stroke survivors.

• Survivors valued their participation in self-management integrated care programmes and linked that to the perceived improvement in their rehabilitation outcomes.

• Self-management interventions such as patient education, goal-setting, self-monitoring, and professionals/carers support have been found to improve mobility outcomes for stroke survivors.

• The outcomes that benefited to a slight extent using self-management were; functional mobility, walking ability (speed, distance, and endurance), and a number of steps per day.

Keywords:

• Stroke
• rehabilitation
• self-management
• functional mobility
• walking
• community care
• exercise

Introduction

A stroke can impact individuals’ lives to varying extents [1]. The motor impairments might restrict the functional mobility of survivors and affect their physical activity and psychosocial status post-stroke [2,3]. Walking, an important form of functional mobility can be a challenge for about 75% of individuals recovering from a stroke [2,4]. Walking is considered a crucial goal for stroke survivors’ mobility after stroke [5]. Therefore, therapy plans should include the most effective strategies that help survivors to achieve their functional mobility goals.

Rehabilitation post-stroke aims to improve survivors’ functional recovery and to increase their capabilities for community reintegration [6]. During stroke rehabilitation, it has been evident that high intensity of physical activity training can lead to better functional outcomes such as walking [7]. A meta-analysis of the studies of the effectiveness of walking training post-stroke found that intensive walking training improved walking capacity and ability to self-care at different stages of stroke rehabilitation [8]. However, research has identified that rehabilitation services in the UK have not been able to offer intensive walking training for stroke survivors [5,9]. Patients are discharged home with early discharge strategies with a shortened hospital stay. However, there is a lack of adequate support, feeling of being lost or abandoned, or receiving- poor care at home following discharge [5,10]. Considering the need to improve mobility and independence after discharge and the limited availability of therapy resources following discharge there is a need to consider strategies that can empower patients. Promoting patients’ role of self-management can help them to cope better with the consequences and achieve their priority goal of walking.

The main principle of Self-Management (SM) is to empower individuals with knowledge, skills, and confidence that can increase survivors’ capabilities for functional rehabilitation and reintegration into the community [11]. Supporting strategies are taught to enable patients to overcome challenges after discharge from hospital care [12,13]. SM was defined by the National Clinical Guideline for Stroke in the UK as the “actions and confidence of individuals to manage the medical and emotional aspects of their condition in order to maintain or create new life roles” (5,p.28). SM can support rehabilitation efforts during early supported discharge from acute care and has been positively associated with functional and psychosocial outcomes of rehabilitation during both subacute and chronic stages of stroke recovery [11,14]. Therefore, it has been recommended by clinical guidelines as a considerable part of therapy planning [5]. Based on the above evidence, increasing patients’ SM skills might positively influence functional mobility outcomes post-stroke, hence, there is a need to explore the characteristics of the most effective SM strategies relevant to functional mobility.

In the rehabilitation literature, strategies used to enhance self-management behaviours range from providing patients with information or education about their conditions to more complex strategies such as involving a patient in goal setting, decision making, problem-solving, or self-monitoring during their rehabilitation process [15,16]. Theoretically, incorporating multiple strategies in an intervention can be most effective especially when behaviour change needs to be targeted [11,16]. Previously, reviews have evaluated the effectiveness of SM as a holistic approach that explores aspects of rehabilitation such as functional ability, participation, self-efficacy, and quality of life [11,16]; yet there has been no review of SM interventions for improving mobility post-stroke. This review aims to provide an evidence synthesis of the SM-based interventions that can be used by rehabilitation professionals to improve patients’ mobility post-stroke.

Objectives

To collect, review, synthesise evidence around self-management interventions to improve mobility and the relevant outcomes that were studied within stroke rehabilitation.

Review question

What are the effects and outcomes of self-management interventions related to the rehabilitation of mobility post-stroke?

Methods

An integrative review of relevant studies was undertaken [17]. A subtle realist perspective was assumed for this review that focuses on the common reality experienced across individuals. This perspective supports the combination of different methodologies to explore participants’ perceptions of reality [18,19]. The study protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO Registration # CRD42019151051).

Search strategy

A systematic search was undertaken on the following electronic databases from January 1992 until July 2021: PubMed, MEDLINE, EMBASE, PsycINFO, AMED, CINAHL, Cochrane Database of Systematic Reviews, and Web of Science. The search terms included: ‘stroke’, ‘rehabilitation’, ‘self-management’, ‘walking’, and ‘functional mobility. These terms were exploded using synonyms from relevant literature and identification of MeSH terms from PubMed (see Supplementary Appendix A for a detailed list). Further, searching was undertaken using electronic search engines including Google Scholar and ScienceDirect. We used limiters to search for the articles that were published in English, related to humans, and when keywords were found within abstracts. References from shortlisted articles and a search of grey literature including publications from conference proceedings were also conducted as a secondary search. If the initial search had returned any PhD theses we searched for articles published from this work. We wrote to authors of theses and conference abstracts to check for any relevant publications.

Eligibility criteria

The following eligibility criteria for studies were identified using the PICOS (Participants, Intervention, Comparison, Outcome, Study design) acronym. The eligibility criteria have been set out below based on the PICOS.

Participants

Studies that included results detailing separate (if mixed samples) and identifiable analysis of stroke survivors with any type or severity of a stroke, who have received any form of self-management support interventions to improve their mobility at any time after stroke. Studies were excluded if they recruited stroke survivors under 18 years of age or used self-management specifically for stroke prevention.

Interventions

The review included studies that used self-management strategies that targeted improved mobility either individually or in clusters. Studies were excluded if they involved psychosocial and/or behavioural interventions in isolation without targeting self-management behaviour. For example, (tele) rehabilitation or cognitive therapy that did not encourage self-management behaviours were excluded.

Comparison groups

Articles were included whether they had a comparator, usual care or no comparator intervention. We did not specify any inclusion criteria for the comparator treatments (all types) provided to the control groups of participants.

Outcomes

Studies were included if they provided qualitative or quantitative data related to functional mobility (individual's physiological ability to move independently and safely to perform ADLs in a different environment [20]) outcomes, in addition to other self-management-related outcomes such as psychological, social, or general well-being measures.

Study designs

There was no restriction on the types of study design for inclusion in this review.

Study selection

The identified results from the databases search were transferred and saved into the RefWorks reference management software (ProQuest LLC, 2020). The duplicates were removed and an initial screening of titles was done by a reviewer (AS). This was followed by a screening of abstracts which was done independently by reviewers (AS, SR and DG). Irrelevant studies were excluded at this stage based on the inclusion criteria. The full-text screening was done by two independent reviewers (AS and DG). Conflicts between the two reviewers were resolved by a third reviewer (SR) at each stage of screening. A PRISMA Flow Diagram was used to summarise the results of the screening.

Quality appraisal

To assess the risk of bias in the selected studies, the Cochrane's Risk of Bias Tool (the updated version) was used to critically appraise the randomised controlled trials [21]. The Risk Of Bias In Non-Randomised Studies of Interventions (ROBINS-I) was used to evaluate the non-controlled trials [22]. The Mixed Methods Appraisal Tool (MMAT) was used for the quality assessment of the mixed-methods studies [23]. The Critical Appraisal Skills Programme (CASP) was used for the critical appraisal of studies with other designs [24]. Studies with poor quality were not excluded from the review as we planned to describe all possible effects of self-management interventions on the stroke population. The Grading of Recommendations, Assessment, Development and Evaluations (GRADE) framework was used to rate the overall strength of evidence around outcomes, based on the quality of the reviewed quantitative studies [25]. Two reviewers (AS and DG) assessed the quality of the included studies independently and a third reviewer solved the discrepancy between their results (SR).

Data extraction

A data extraction template was adopted from the Cochrane Data Extraction and Assessment Template for each study and a summary table of the extracted information from all studies was created. This was done independently by two reviewers (AS and DG) and checked by a third reviewer (SR). The extracted data included information about participants' characteristics, study methods, interventions, outcomes and findings. Specific details about the strategies and protocols of self-management interventions used in the studies were extracted. Ten study authors were contacted for any missing data relevant to the review question.

Data synthesis

The data analysis process is suggested to follow the steps of data reduction, data display, data comparison, conclusion drawing, and verification [17,26]. Step 1 data reduction: Data from articles were reduced to a manageable format by initially classifying them into two main groups, quantitative and qualitative. Step 2 data display: The key elements of the studies were extracted using tabulation techniques (see Table 1 and Supplementary Appendix B). The articles were presented within the data extraction table in alphabetical order. Step 3 data comparison: Identified themes from step 2 were compared using a constant comparison method to identify the similarities and differences in the contents of the included studies. The data was organised under key sections that served as overarching themes. Qualitative findings were independently analysed by AS and SR using thematic analysis [51] (see Supplementary Appendix B). Findings were compared and conflicts were discussed between the reviewers until they reached a consensus. Each theme included subthemes that covered different aspects of participant's perspectives regarding that theme (see Supplementary Appendix B). Conclusions were presented by integrating the thematic analysis with the quantitative data sections supported by the qualitative findings under the themes.

Table 1. Summary table of the reviewed studies.

Study/Country/Design	Participants	Interventions	Theoretical frameworks/SM strategies	Outcomes	Results	Methodological quality
Baskett et al. [27] New ZealandRCT	N = 100 (mean age 71.7 (IG) and 67.8 (CG) in yrs). Stroke survivors after discharge. Subacute stage.	IG: OT and PT visit once a week and exercises prescribed. Encouraged to exercise several times a day. Therapists guided the self-directed therapy. CG: Outpatient or day hospital therapyDuration: 18 months. Format: One-to-one. Delivered by: PT/OT at home for the intervention group.	TF: Unclear. SM strategies: Goal setting/ action planning/self-monitoring.	Gait speed, Limb function, Activities of daily living, Time with therapists, Mood of subjects and caregivers, Anticipated outcomes at entry, Perceived outcome at exit.	No statistical differences between the control and experimental groups in any outcome measures.	High risk of bias (CRoB). No blinding. Used surrogate quality of life measures in screening for mood disorders.
Broetz & Birbaumer [28] GermanyCase report	54-year-old male who suffered cerebral bleeding with left hemiparesis 25 years ago (chronic stage).	Over the 18-month period the programme included: Daily intensive exercises × 1 week. One-hour sessions/ week. Goal-setting. Self-control cues for exercises Supervision and feedback after successful trials. Intensive training week repeated at 6 months and 12 months. Duration: 18 months. Format: One-to-one. Delivered by: Physiotherapist.	TF: Behavioural physiotherapy incorporating motor learning. SM strategies: Goal setting, Motivation, Supervision, Feedback and Reward.	Walking capacity, Goal attainment, andChange of symptoms.	Rhythmic walking improved from 74 steps/ minute to 112 steps/minute, Walking speed increased from 8.65 s to 6.48 s, and Running for 80 M.	High risk of bias (CASP). Case report of only one participant, high possibility of selection, performance, and detection bias.
Chumbler et al. [29] USAA prospective, single-blinded RCT	N = 48(mean age 67.1–67.7 yrs). Participants with ischemic or haemorrhagic stroke. Subacute stage. Discharged to the community. Not cognitively impaired. Able to follow a 3-step command. Discharge motor Functional Independence Measure (FIM) score (18–88).	IG: STeleR intervention: 3 home visits, 5 telephone calls, In-home messaging device for instructions on functionally based exercises and adaptive strategies. CG: usual care Duration: 3 monthsFormat: One-to-one. Delivered by: Therapist via home visit or telephone calls	TF: The Verbrugge and Jette’s model for the disablement process. SM strategies: Goal settingAction planningSelf-monitoring.	Telephone Version of Functional Independence Measure Function scales of the Late-Life Function and Disability Instrument (LLFDI).	The outcomes improved at 6 months for the STeleR group and declined for the usual care group4 of the 5 (LLFDI) component subscales (P0.05),	Low risk of bias (CRoB). Sample had majority males and wasn't inclusive for stroke population.
Dean et al. [30] UKMixed methods study	N = 45 (mean age 70 (IG) and 71 (CG) in yrs). Stroke survivors with mobility issues. Discharged from rehabilitation services at least 1 month before. Subacute and chronic stages. Possessing sufficient physical, cognitive, and communication ability.	IG: ReTrain which included: A one to one to 1 session followed by ten, twice-weekly group classes, Cardiovascular and strength work, Self-management strategies, Functional activities, Goal setting, and Bespoke home-based training programme. CG: Exercise after stroke advice booklet plus usual care. Duration: 3 monthsFormat: Group and one-to-one. Delivered by: Personal trainers on the UK Register of Exercise Professionals.	TF: Action for Rehabilitation following Neurological Injury (ARNI). SM strategies: EducationGoal settingAction planningProblem-solvingSelf-monitoringGroup and social support.	Primary outcomes: Rivermead Mobility Index, Timed Up and Go, Modified Patient-Specific Functional Scale, 7-day objective physical activity levels using wrist-worn accelerometer, and Physical activity diary. Secondary outcomes: Stroke Self-Efficacy Questionnaire, Fatigue Assessment Scale, EuroQol- 5 Dimension (EQ-5D-5L).	The trial was not powered to detect differences in any outcomes between trial arms or over time.	Low risk of bias (MMAT). It failed to identify a clearprimary outcome for a definitive RCT. Small sample that did not have a wide age range or ethnic diversity. This was a pilot study, not designed to demonstrate generalisability.
DePaul et al. [31] CanadaMixed methods study	N = 10 (mean age 62.3; SD 11.7 years). Inpatient stroke rehabilitation Acute and subacute stages. Functional Ambulation Classification scores (2-4). Functional Independent Measure ≥ 40.	IMPACT: Starts in the hospital and continues after discharge. Usual therapyPhysiotherapy −8 sessions × 2 weeks Duration: 14 days. Format: One-to-one. Delivered by a trained PT.	TF: Social cognition theory and Self-determination theory. SM strategies: Education and professional supportGoal setting Decision makingAaction planningSelf-monitoring and Feedback.	6 min Walking Test, 5-meter Walk Test, Functional Independence Measure (total and motor), Berg Balance Test, Functional Ambulation Category, Chedoke-McMaster Stroke Assessment (CMSA leg and foot). Therapists’ and patients’ perceptions were qualitatively collected.	Gait speed improved by an average of 0.32 m/s (SD 0.27 m/s) on 5-m walk test. 6MWT improved on an average of 85.0 m (SD 64.0) over the 2-week intervention period. All patients indicated that the programme helped them increase their activity and indicated they would continue to practice walking-related activities beyond the coaching period.	High risk of bias (MMAT). Small and inconclusive sample. Single therapist and single setting. Single-group study design. Not powered to detect outcomes changes. Amount of training wasn't accurately reported across all participants.
Ezeugwu & Manns [32] CanadaSingle group, longitudinal intervention study	N = 34 (mean age = 64.6 yrs). Within 2 weeks of discharge from inpatient facilities. Subacute stage. Able to stand up from a chair. Walk at least 5 meters with or without gait aids. Able to understand 2-step commands.	STand Up Frequently From Stroke (STUFFS) Sedentary time was interrupted with upright activities (standing and walking) at home and in the community. A motivational wrist-worn activity monitor was used throughout the intervention. Duration: 8-weeks Format: One-to-one. Delivered by PT at home.	TF: Social cognitive theory. SM strategies: Action planningSelf-monitoring.	Chedoke-McMaster Stroke Assessment (CMSA) for leg and foot, 5-meter walk test, Timed Up and Go, Sedentary behaviour and physical activity (time sedentary, standing, and stepping, the number of steps, sit-to-stand transitions) were measured using the activPAL, The Montreal Cognitive Assessment (MoCA), Stroke Impact Scale, and Self-Efficacy for Exercise Scale Outcomes were assessed with 1-week baseline, 8-week intervention, and 8-week follow-up.	Sedentary time decreased by 54.213.7 min/day (p < .01) at post intervention and 26.814.0 min/day (p < 0.07) at follow-up, Significant improvements in walking speed, (p < .05). The number of steps and time spent stepping were not statistically different across both time periods.	Some concerns about risk of bias (CASP). The sample was limited to specific stroke population. No control group. The duration of the intervention was too short to achieve change in behaviour.
Haworth et al. [33] UKAn exploratory, single blind RCT	N = 44 ( mean age 40.9 (IG) and 42.3 (CG) in yrs). Patients with neurological conditions including stroke. Within 4 weeks of discharge from in-patient care. Subacute stage. Able to carry out exercise independently. Sit and ransfer independently.	IG: Exercise and education programme twice weekly over four weeks. Education involved lectures about the benefits of exercises and self-management mobility, strength and balance exercises CG: Standard follow-up care including medical review clinics and outpatient therapy. Duration: 4 weeks. Format: Group.Delivered by: by research therapist at specialist neurosciences centre.	TF: Self-efficacy. SM strategies: EducationGroup intervention.	Primary outcomes: Exercise Efficacy Scale and Human Activity Profile. Secondary outcomes: 10-metre timed walkShort Form 36 Health Survey, Hospital Anxiety and Depression Scale, UWIST Mood Adjective Checklist, Motricity Index, Frenchay Arm Test, Rivermead Mobility Index, and a visual analogue scale. Data were collected at baseline and at 6, 12 and 24 weeks.	At 6 weeks, significant improvements in exercise efficacy in the treatment group (p = 0.001). Moderate improvements inactivity levels were noted for both study groups with no between-group differences. No significant improvements in SF 36 ‘physical health’ Follow-up at 12 and 24 weeks suggested that the improvements noted in exercise self-efficacy, activity levels and quality of life were maintained and were independent of the exercise programme.	Some concerns about risk of bias (CRoB).Small sample size. The intervention needed to be extended to accurately measure HAP. Weaknesses within the chosen measures.
Huijbregts et al. [34] CanadaMixed methods study	N = 18 (mean age 61.8 (IG) and 65.6 (CG) in yrs). Medical diagnosis of stroke. Completed active rehabilitation. Chronic stage. Living in the community.	IG: Telehealth MOST intervention was implemented using videoconferencing. MOST self-management programme included: Discussion and exercise. Goal-setting andproblem-solving, Exercise, and community- reintegration skills. CG: Waiting list control group. Duration: 18 sessions over 9 weeks. Format: Group therapy. Delivered by: Trained health professional facilitators.	TF: Unclear. SM strategies: EducationProblem solvingGoal settingSocial support.	Berg Balance Scale (BBS), The Reintegration to Normal Living Index, The Stroke-Adapted Sickness Impact Profile, Chedoke-McMaster Stroke Assessment Activity Inventory, Goal Attainment Scaling, and The Geriatric Depression Scale.	There was a significant difference in BBS scores between the T-MOST group and the WLC group (mean difference −4.27, 95%CI: −6.66 to −1.87).	High risk of bias (MMAT). Small sample size No randomisation High risk of selection bias due to recruitment strategy. Baseline differences between groups were not clear.
Kanai et al. [35] JapanPre–post interventional study	N = 22 (mean age 62.5 yrs). Acute ischemic stroke. Able to walk without assistance or gait aid Within seven days of admission.	IG: 40–120 min of a supervised rehabilitation program 5–6 times a week until discharge. Instruction on the self-monitoring approach and discussed physical activity (PA) targets, Exercises included body stretches, weight resistance exercises, aerobic exercise. Accelerometer worn 24 h/day until discharge Duration: 11 days. Format: One-to-one. Delivered by: PT at hospital	TF: Self-efficacy. SM strategies: EducationGoal settingAction planningSelf-monitoring.	The daily number of steps was taken as an index of Physical Activity andSelf-efficacy for physical activity (SEPA) scale.	PA during the intervention was higher than that at the baseline measurement (5709.4 ± 2236.1 vs. 2813.9 ± 1511.9 steps/day, p < 0.001). Self-efficacy for PA during the intervention was also higher than that at the baseline measurement (76.4 ± 18.8 vs. 58.9 ± 29.0 points, p < 0.001).	High risk of bias (CASP). Very small sample size. No control group. Was not statistically powered to detect changes in the outcomes. Used only the number of steps/day as the measure of PA. Did not assess psychological factors or pre-morbidity exercise habits. Did not conduct functional tests at discharge.
Kendall et al. [36] AustraliaRCT	N = 100 (mean age 65.96 yrs). First stroke. Subacute stage. No dementia or psychiatric illness. Sufficient English language. Expectation of discharge to their homes.	IG: SM education programme targetting mobility and exercise. Duration: 7 weeks. Format: Group Delivered by professionals in community settings.	TF: The Chronic Disease Self-Management Course was used to operationalise the concept of psychosocial skill expansion. SM strategies: EducationGoal settingAction planningGroup interaction and Support.	Stroke Specific Quality of Life scale (SSQOL) and the Self-efficacy Scale.	The study identified a significant impact of the intervention on the quality of family roles, self-care, work productivity, and functioning in daily activities within SSQOL (p = 0.05). The control and intervention groups differed significantly in self-efficacy levels across all times, F(1, 93) = 9.45, p = 0.003, including pre-intervention.	Some concerns about risk of bias (CRoB). Minimal data was available about the clinical status of the participants. People with language difficulties were excluded.
Malagoni et al. [37] Italy Feasibility and pilot RCT	N = 12 (mean age 66.5 ± 11.9 yrs). People with chronic stroke. Ability to ambulate independently at least for 10 meters on level ground. FAC > 3. Ability to perform TUG.	IG: Structured home-based exercise programme with 2 10-minute sessions/day (6 days/week) of intermittent walking. CG: Standard supervised group therapyprogramme. Duration: 10 weeks. Format: Group and individualised sessions at home. Delivered by: Physiotherapists.	TF: Unclear. SM strategies: Self-monitoring and feedback.	Functional outcome measures included 6-Minutes Walking Test, Timed Up and Go, Stair Climb tests, and Physical activity domain of the SF-36 questionnaire.	6-minute walk test and physical activity domain of the SF-36 significantly increased in both groups (p = 0.03). Timed Up and Go improved in both groups, significantly for the home-based exercise group (p = 0.03) while stair climb remained stable for both groups.	High risk of bias (CRoB). Small sample size No blinding
Mansfield et al. [38] CanadaQuasi-experimental feasibility-controlled trail	N = 26 (mean age 62.1 (IG) and 67.1 (CG) in yrs). Any type of stroke. Subacute stage.	IG: PROPEL, included group exercise and self-management. Supervised and individualised aerobic training 3 times a week, An inpatient module and an outpatient module. CG: received in- and outpatient rehabilitation Duration: 6 weeks. Format: Group. Delivered: by a PT and a healthcare student in an outpatient rehabilitation setting.	TF: Transtheoretical model + social cognitive theory and self-efficacy. SM strategies: Goal settingAction planningProblem solvingSelf-monitoring	Physical activity level was assessed using an accelerometer, heart rate monitor, and the Physical Activity Scale for Individuals with Physical Disabilities (PASIPD), the Short Self-Efficacy for Exercise (SSEE) scale, the Short Outcome Expectation for Exercise (SOEE) scale, and the Barriers to Being Active Quiz (BBAQ).	PROPEL participants had a higher participation in physical activity than COMP participants (Hedge’s g ≥ 0.5),reported higher SSOE scores (Hedge’s g: 0.50) and fewer significant barriers to physical activity in the BBAQ than COMP participants (Hedge’s g: −1.18).	Serious risk of (ROBINS-I). Extremely low recruitment rate. Groups were not comparable at baseline. Patients with communication difficulties were excluded. Only 34% of daily heart rate data were available.
Olney et al. [39] CanadaRCT	N = 72 (mean age 63.5 (IG) and 65.8 (CG) in yrs). Chronic stroke survivors. Able to walk a total of 15 min with rests (with or without assistive devices except a 4-point walker). Able to tolerate activity for 45 min with rests. No severe coronary artery disease.	IG: One week supervised instruction program followed by a 9-week unsupervised home programThe unsupervised group had exercises prescribed to them and written instructions to monitor and progress.. CG: 10-week supervised strengthening and conditioning program (supervised) Duration: 10 weeksFormat: group format for the supervised sessions and 1 to 1 for the unsupervised sessions. Delivered by: therapist at rehabilitation centre for the supervised group and at home for the unsupervised group.	TF: Unclear. SM strategies: Education and Self-monitoring.	Primary Outcome: Walking speed (6MWT). Secondary Outcomes: Human Activity Profile (HAP) adjusted activity score, Medical Outcome Study 36-Item Short-Form survey (SF-36), Physiological Cost Index, and lower extremity muscle strength.	The 6MWT speed increased significantly in both groups and remained significantly improved at one year. The HAP showed an increasing trend only in the supervised group and was significant at one year. The SF-36 Physical component summary score increased significantly in the supervised group and remained improved at one year; the unsupervised group showed significant improvement at one year.	Low risk of bias (CRoB). It was not possible to blind participants and testers. The generalisability of the findings might be limited due to restricted eligibility criteria of participation.
Paul et al. [40] UKQuasi-experimental pilot study	N = 23 (mean age 55.3 (IG) and 56.3 (CG) in yrs). Single unilateral stroke. Discharged from rehabilitation. Chronic stage. Able to walk independently with or without using an aid or orthosis. Able to comprehend instruction.	IG: STARFISH program included a mobile phone app-based intervention, with evidence-based behaviour change techniquesThe intervention started with 1 week of baseline step count using phone accelerometer, followed by use of the STARFISH application over a 6-week period. CG: Usual care. Duration: 6 weeks. Format: Group. Delivered: through a mobile app.	TF: Behaviour Change Techniques (BCTs). SM strategies: Feedback, Self-monitoring and Social support	Physical activity (steps/day), Ten-Meter Walk Test, Sedentary time, Fatigue Severity Scale, Instrumental Activity of Daily Living Scale, Stroke Specific Quality of Life Scale, and Psychological General Well-Being Index.	The average daily step count increased by 39.3% in the intervention group and reduced by 20.2% in the control group (p = 0.005 for group–time interaction). Similar patterns of data and group–time interaction were seen for walking time (p = 0.002) and fatigue (p = 0.003).	Moderate risk of bias (ROBINS-I). Sample was not inclusive for stroke population. Duration of the intervention was short. Assessor was not blinded. No randomisation. The sample was heterogeneous in PA. Mobile phone accelerometer needs validation.
Portz et al. [41] USAMixed-methods study	N = 13 (mean age 73.23 yrs). Individuals with chronic stroke (> 6 months). Had a fall or current fof. Completion of post-stroke rehabilitation. Able to stand, at risk for falls (≤ 46 on the Berg Balance Scale). Able to understand English. Score of ≥ 4 out of 6 on the short Mini Mental Status Exam.	IG: MYOT a yoga-infused self-management intervention. One hour of group self-management education followed by one hour of yoga (2 times per week). Duration: 8 weeks. Format: Group. Delivered by an OT and a research student	TF: Self-efficacy and Problem solving. SM strategies: EducationGoal settingAction planning/Problem solvingSelf-monitoringGroup discussion and activities.	Quantitative Outcomes: Endurance (6-MWT), Strength (30-Second Chair Stand test), and Gait Speed (10-Meter Walk test). Qualitative outcomes: Participant perceptions of changes in physical fitness and PA were collected	Physical fitness outcome measures including endurance and lower and upper body strength significantly (p < .02) improved. Based on qualitative results, participants expressed positive changes in endurance, strength, gait speed, flexibility, and balance. They also reported improvements in walking ability and duration and expressed a desire to continue yoga and increase levels of exercise.	Some concerns about risk of bias (MMAT). The long-term effect of the programme was not measured. It wasn’t clear why some patients did not notice any change.
Preston et al. [42] AustraliaSingle-group, pre-post intervention study	N = 20 (mean age 68 yrs). Discharged home from an acute stroke unit. Subacute stage. Able to walk 10 m across flat ground without any aids at a speed of ≥ 0.8 m/s. Scored ≥ 24 on the Mini Mental State Examination. Resided within 30 km of the hospital. No moderate to severe receptive aphasia.	Five 60-minute home-based sessionsCarer or family member involved. The sessions included: education, goal setting, barrier identification, self-monitoring, and feedback. Duration: 3 months. Format: One-to-one. Delivered by: Physiotherapist.	TF: Behaviour Change Techniques (BCTs). SM strategies: EducationGoal settingAction planningProblem solving Self-monitoring and feedback.	Primary Outcomes: Duration of moderate physical activity in min/day and counts of physical activity in steps/day. Secondary Outcomes: Self-efficacy, walking ability, participation, and quality of life.	At 3 months, participants completed 27 min/day (95% CI 4–49) more moderate physical activity than at baseline and 16 min/day (95% CI − 10–42) at 6 months. Participants perceived the self-management program to be useful and there were few adverse events (muscle soreness, falls, and hospitalisation).	Some concerns about risk of bias (CASP). No randomisation. No control group. There were challenges around using Sensewear armband to collect PA data.
Richardson et al. [43] CanadaCase report	N = 15 (mean age 65.5 yrs). Chronic stroke survivors. First ischemic stroke. Able to walk at least 2.44 mts. Able to transfer to a chair. Some active movement in the UE during recovery. Cognitive function score of 26 points on the Montreal Cognitive Assessment (MoCA). No receptive aphasia. Can communicate in English.	START: Self-Management and Task-Oriented Approach to Rehabilitation Training. Duration: 10 sessions Format: One-to-one. Delivered by physiotherapist at home.	TF: Self-efficacy. SM strategies: Goal settingAction planningProblem solvingSelf-monitoring.	TUG, Barthel Index, Stroke Impact Scale, Patient Activation Measure, modified version of Patient Satisfaction Questionnaire-18.	Significant changes were observed in the TUG (p < 0.01), the Barthel Index (p < 0.01, and the Stroke Impact Scale (p < 0.001).	High risk of bias (CASP). Very small sample size and not inclusive. High possibility of bias as there was no randomisation or control group.
Scrivener et al. [44] AustraliaPre-post single group study	N = 14 (mean age 57 ± 14.2 yrs). Adults with stroke. Discharged from rehabilitation. Chronic stage. Could walk at > 0.4 m/s with/without an aid.	TASK programme. A video-guided exercise programme. The programme comprised of two phases: in phase one, four weeks of the program was supported by weekly supervised sessions at the rehabilitation centre. In phase two, four weeks of the program was completed at home without direct supervision. Format: One-to-one. Delivered by physiotherapist at home.	TF: Self-efficacy. SM strategies: Education/self-monitoring.	Physical performance, physical activity, and exercise self-efficacy were measured at baseline and 4 and 8 weeks.	Walking speed improved over the duration of the program (mean difference −0.12 m/s, 95% CI −0.22 to −0.02, p = 0:02). Self-efficacy and physical activity did not change over the duration of the program.	High risk of bias (CASP). The sample size for this study was small and diverse in characteristics. High possibility of bias as there was no randomisation or control group.
Sullivan et al. [45] USpre-post test follow up design	N = 11 (mean age 60.4 yrs). Community-dwelling stroke survivors. 21 years old or older. Ambulators. Stroke onset 6 months or more before participation (chronic stage).	A pedometer monitored, community-based intervention was used for 6 weeks. A briefverbal educational module was provided to eachparticipant addressing the benefits of exerciseafter stroke, pedometer use, and completion ofdaily exercise diaries. Duration: 6 weeks. Format: One-to-one. Delivered by: therapist at home through telephone coaching sessions.	TF: Self-efficacy. SM strategies: EducationGoal setting and decision makingSelf-monitoring.	Primary Outcomes: Walking speed (10-mWT) and endurance (6MWT). Secondary Outcomes: the Activities-Specific Balance Confidence Scale, Stroke Impact Scale–16 (SIS-16), and a pedometer satisfaction survey.	There were low effect sizes for changes in SIS-16 (0.312) and 6MWT (0.293). Increasing steps correlated with increased perception of physical function.	Some concerns about risk of bias (CASP). Small sample size. The heterogeneity of participants may have affected results. Possibility of a ceiling effect for the steps/day. Pedometer was inaccurate with participants who walk slower than 0.5 m/s. Short duration of intervention.
Taylor et al. [46] CanadaPre-Post intervention feasibility study	N = 12 (mean age72.7 yrs). Stroke survivors between 3 and 18 months post-stroke Subacute and chronic stages.	Moving On after STroke (MOST): MOST involves sessions with 1 h of discussion followed by 1 h of exercise (cardiovascular, range of motion and strength components). Participants learn about stroke-related issues, but more importantly, learn problem-solving and goal-setting skillsTwo sessions/week. Duration: 9 weeks. Format: Group with patients and carers. Delivered by: Clinicians via videoconference.	TF: Unclear. SM strategies: EducationGoal settingProblem solvingGroups with support.	Walking endurance (6MWT) Short-Term Goals Goal Attainment Scaling,Berg, Berg Balance Scale, Activity-Specific Balance Confidence Scale), Geriatric Depression Scale and Reintegration to normal living), .	Pre–post improvements were seen in goal setting, mood, balance, balance confidence, and walking endurance. Participants reported that the program provided people with stroke as well as caregivers with greater awareness of stroke, increased social support, and improved ability to cope.	Some concerns about risk of bias (CASP). No control group. No randomisation. Insufficient numbers to compare the health outcomes of the telehealth participants to those in the group receiving face-to-face care.
Treger et al. [47] IsraelRCT (single blinded)	N = 56 (mean age 56.6 (IG) and 62.2 (CG) in yrs). Subacute stage . Ability to walk independently (with or without walking aids). FIM score ≥ 60 at mid-hospitalisation evaluation.	IG: Received detailed instructions for 14 exercises and was instructed to walk for at least 30 min daily. Follow up phone calls. CG: Usual care. Duration: 3 months. Format: One-to-one. Delivered by: a physiotherapist.	TF: Unclear. SM strategies: EducationSelf-monitoring/Review follow up.	The 10-metre walk test (10MWT), 6-minute walk test (6MWT), and Timed Up & Go (TUG) were performed twice; once at the end of the inpatient rehabilitation period and three months later.	A significant improvement was shown in the intervention group in 10MWT (p = 0.008), 6MWT (p < 0.001) and TUG (p = 0.009).	Moderate risk of bias (CRoB). Participants were not blinded. Participants received instuctions from therapist without involvement in goal-setting or planning.
Van den berg et al. [48] AustraliaRCT (proof of concept)	N = 63 (mean age 64.5 (IG) and 70.1 (CG) in yrs). Stroke survivors within subacute stage. With mobility problems (FAC score < 5). Caregiver to participate in the programme. Had sufficient cognition to take part and no depression (HADS score < 11).	IG: CARE4STROKE: Caregiver-mediated exercises with e-health support Telerehabilitation services. Participants had 30 min, 5x a week Exercises executed by a caregiver plus usual care. CG: Usual care. Duration: 8 weeks. Format: One-to-one. Delivered by: Physiotherapist and caregiver at rehabilitation centre or nursing home and continued at home if patient discharged before finish.	TF: UnclearSM strategies: Self-monitoringCaregivers’ support.	Primary outcome: The SIS-mobility domain. Secondary outcomes: Length of stay, Motor impairment, Strength, Walking ability, Balance, Extended activities of daily living, Psychosocial functioning, Self-efficacy, QoL, and fatigue. Additionally, caregiver’s self-reported fatigue, symptoms of anxiety, self-efficacy, and strain were assessed.	Intention-to-treat analysis showed no between-group difference in SIS- mobility; Per-protocol analysis, examining those who were discharged home with telerehabilitation showed a trend toward improved mobility (p = 0.06), significantly improved extended ADLs scores at week 8 (−3.6, p = 0.01) and week 12 (3.0, p = 0.03).	Low risk of bias (CRoB). Small sample size. The 2 groups were not comparable at baseline. Study did not provide definitive measures of effectiveness. Participants were not blinded. 11 patients dropped out (high attrition). Study was not designed or powered to detect adverse events.
Vloothuis et al. [49] NetherlandsRCT (proof of concept)	N = 66 (mean age 59.26 (IG) and 60.53 (CG) in yrs). 34 inpatients with stroke. Subacute stage. Lived independently before the stroke. Discharged home. Able to follow instructions (MMSE > 18 points). FAC Score < 5. Had a caregiver to assist.	IG: CARE4STROKE: Caregiver-mediated exercises with e-health support Telerehabilitation services. Participants had 30 min, 5x a week Exercises executed by a caregiver plus usual care. CG: Usual care. Duration: 8 weeks. Format: One-to-one. Delivered by: therapist and caregiver at rehabilitation centre or nursing home and continued at home..	TF: Assumptions that more training is better (Motor control theory). SM strategies: Goal settingCaregivers’ support.	Primary outcomes: SIS-mobility domain . Secondary outcomes: Motor impairment, Strength, Walking ability, Balance, Mobility and Extended ADL of patients, Caregiver strain of caregivers, and Mood, Self-efficacy, Fatigue and QoL of both patients and caregivers. Outcomes were assessed at baseline, 8 and 12 weeks after randomisation.	No significant between-group differences were found regarding SIS-mobility after 8 and 12 weeks.	Low risk of bias (CRoB). Variability in the decision for discharge. Patients were not included at fixed times after stroke. Study may be under powered.
White et al. [50] AustraliaMixed methods study	N = 22 (mean age 65.76 yrs). Community dwelling stroke survivors. Chronic stage. Not accessing other rehabilitation services at the time of enrolment.	MASTERSTROKE A secondary prevention stroke group program with two 2-h sessions weekly (1 h for education and 1 h for exercise). The exercise component incorporated fitness, strength, mobility and balance and education focused on secondary stroke prevention whilst also providing chronic condition self-management support. Duration: 9 weeks. Format: Group.Delivered by: PT and multi-professional community-based team.	TF: Unclear. SM strategies: EducationGroup support.	Timed Up and Go (TUG), Six Minute Walk Test, Participants’ perceptions were qualitatively collected.	TUG (mean change = 3.89, p = 0.00), and quality of life scores (mean = 0.25, p = 0.01). Qualitative responses results confirmed increases in knowledge about stroke and exercise tolerance, successfulness of a group program and lifestyle modification post-stroke.	Some concerns about risk of bias (MMAT). Small sample size. Was not statistically powered, nor controlled. The assessors were also the therapists who implemented the programme. Qualitative interviews were limited to one per person that could limit the ability to capture changes over time and sustainability of modifications.

N: Sample size; SM: Self-management; TF: Theoretical Framework; CRoB: Cochrane's Risk of Bias Tool; ROBINS-I: The Risk Of Bias In Non-Randomised Studies of Interventions; MMAT: The Mixed Methods Appraisal Tool; CASP: The Critical Appraisal Skills Programme; RCT: Randomised Controlled Trial; OT: Occupational Therapist; PT: Physiotherapist; FAC: Functional Ambulation Category; PA: Physical Activity.

Results

Study characteristics

A total of 1903 studies were found from the databases searched. Secondary searches resulted in 45 articles. After screening, twenty-four studies conducted between January 1999 and July 2021 were included in this review (Figure 1 PRISMA flow diagram). Nine studies were randomised controlled trials [27,29,33,36,37,39,47–49], two quasi-experimental studies [38,40], one single group longitudinal study [32], five pre-post designs [35,42,44–46], and two case reports [28,43]. Five studies used mixed-methods designs to examine the feasibility of different SM interventions [30,31,34,41,50].

Figure 1. PRISMA flow diagram of study selection [53].

Process of removing duplicates and results of screening titles, abstracts, and full texts of eligible studies.

Studies were conducted in different countries: Canada (7 studies), Australia (5 studies), UK (3 studies), USA (3 studies), Netherlands (1 study), Japan (1 study), New Zealand (1 study), Israel (1 study), Italy (1 study) and Germany (1 study). Participants were recruited from various settings involving various stages of recovery defined as acute (less than 2 weeks after stroke), subacute (within 6 months from the onset of stroke) and chronic (more than 6 months after onset). Eleven studies were recruited from acute and subacute stages [27,29,31–33,36,38,42,47–49], ten studies from the chronic stage [28,34,37,39–41,43–45,50], one study recruited only from the acute stage (3 ± 1 days from stroke onset) [35], and two studies recruited participants including both subacute and chronic stage of recovery [30,46].

Description of participants

A total of 823 participants (523 men and 300 women) with a mean age of 63.7 years (age range: 40.9-73.2 years) were included in the selected studies. Ninteen studies indicated that participants had to have mild to moderate impairments ((FIM ≥ 40) or were able to walk independently or with minimal assistance (FAC < 5)) to be eligible for participation in the selected studies [27–33,35,37,39–45,47–49]. Three studies had recruited survivors with ischemic stroke [35,43,47], six studies had a mix of ischemic and haemorrhagic [29,30,32,37,48,49] and one study had only one participant with haemorrhagic stroke [28]. Other studies did not specify the stroke type of their participants.

The availability of a caregiver was an eligibility criteria for inclusion in four studies where the role of caregivers was a part of the intervention [36,46,48,49]. In five studies, the presence of caregivers was optional [27,31,32,38,42] and the rest of the studies did not indicate any role of caregivers or family members. Table 1 shows participant's characteristics in each study.

Quality of evidence

The highlights of the methodological quality critique of studies can be seen in Table 1. Six studies [30,31,34,35,46,50] were not powered enough to detect changes in the outcomes due to the small sample size. In addition to sampling, five studies [28,34,38,43,44] had a high risk of bias across different categories including selection, performance, and detection bias. Three of the RCTs [27,33,37] had a high risk of performance bias because the blinding of assessors and participants was impossible due to the nature of the applied intervention. In two studies [32,35], there was a high risk of detection bias because the assessors and/or participants were not blinded. Also, there was a high risk of selection bias in three studies [28,38,43] because of their recruitment methodology which had targeted specific participants. There were only six studies (RCTs) that had an overall low to moderate risk of bias [29,30,39,47–49] and one non-randomised controlled trial that had a moderate risk of bias [40]. The risk of detection and performance bias was unclear in certain studies [36,43] as they did not provide any information about the blinding. In some studies, it was not possible to make a judgment on the risk of bias because of a lack of adequate information [38,40].

In the mixed methods studies [28,43] reviewed, there was a high risk of selection bias as the complete outcome data for some participants were missing or not collected for follow-up. Withdrawal rates were not stated and the confounders were not accounted for [34,41]. There was a potential for measurement bias as the therapists who executed the study were also the assessors [50]. Also, the divergences and inconsistencies between quantitative and qualitative results were not adequately addressed. Using the Grading of Recommendations, Assessment, Development and Evaluation (GRADEpro) software [52], the certainty of evidence within the included studies was rated narratively for the main outcomes. Considering the methodological weaknesses and limitations of the majority of the studies to provide a robust judgment on the effectiveness of the SM interventions, the body of evidence has been graded ‘moderate, low, or very low’ on the GRADE system for the main outcomes (Supplementary Appendix C).

The themes arising from the integrative narrative synthesis are presented below.

1. Conceptualisation of self-management in stroke rehabilitation

The concept of SM was variably interpreted by the stakeholders (patients, professionals, researchers, and policy makers) within the qualitative components of the studies and authors of some studies included in this review. Qualitative studies indicate that survivors perceived SM as an approach that can help to increase their capabilities to manage their therapy at home or community, motivates them to undertake more training for mobility as implied in “I was doing three times a week, and I need to make it four or five times”, and hence promotes their independence and socialisation [6,7,27,30,31,41].

Policy makers and healthcare provider considered SM as sharing of responsibilities for rehabilitation services with patients and their caregivers [27]. From therapists’ perspective, the implementation of this concept in real practice has altered their role in service delivery and allowed for the transition of some responsibilities to patients.

Authors of some studies have narrowed their conceptual understanding of SM interventions to specific elements of its conceptual components such as a behaviour change of patients [28,32,40], self-efficacy [35], patient education [33], or self-directed therapy [27]. In the studies that applied SM strategies at the time of discharge from the hospital [27,29,30,39], SM was perceived mainly as a strategy to support the delivery of care at home or in the community and to support early supported discharge. During the chronic stage, SM strategies were suggested to increase the ability to manage the ongoing sequelae of stroke and risk factors for recurrent stroke [43,46,50].

2. Theoretical frameworks for self-management interventions

Using theories is very crucial for the development of complex interventions such as SM [54]. They can help understand how an intervention can be designed, implemented, and evaluated. The development of the interventions in this review was based on a variety of theoretical frameworks and process guidelines. All theories discussed within SM literature are described below. The social cognitive theory and theory of self-efficacy (SE) were commonly used to develop SM interventions, to mediate individuals’ behaviour of self-management, and to measure the effect of SM interventions on stroke survivors. SE in particular was used as a framework for the interventions in three studies [33,35,41]. Sullivan et al.’s [45] intervention were based on the effect of self-monitoring (a part of SE) to increase participants’ level of physical activity. One study used the knowledge-to-action cycle to develop and evaluate an SM and task-oriented intervention called START; however, the SM part of the intervention was developed based on self-efficacy [43].

The social cognitive theory was the foundation for the STUFFS intervention [32] and the PROPEL along with the Trans-Theoretical model [38] or self-determination theory [31]. The SteleR intervention was guided by Verbrugge and Jette’s model for the disablement process [29]. Kendall et al. [36] used the chronic disease self-management course to provide support for stroke survivors based on the concept of psychosocial skill expansion. Two studies used behaviour change techniques (BCTs) to increase the level of physical activity of participants [40,42]. Self-determination theory components were also used by Broetz and Birbaumer [28] in addition to the principles of behavioural physiotherapy to increase patient autonomy in daily training for walking. Multiple studies were not clear about the theoretical frameworks on which their interventions were developed [27,34,37,39,46,47,50]. Overall in studies that showed significant outcomes related to mobility the predominant theories used were identified as self-efficacy in three studies [42,43,45], social cognitive theory in one study [40], behaviour change techniques in three studies [29,38], and self-determination theory in two studies [30,46].

3. Self-management intervention strategies for mobility training

Mobility rehabilitation was directly targeted by SM intervention in eleven studies [28–31,37,40,45–49], while other studies included mobility indirectly as a part of physical activity training [27,33–35,38,39,41–44], sedentary behaviour change [32], stroke prevention [50], or within SM education about mobility [36].

The number and content of SM strategies varied across studies. The most common strategies included providing information, patient education, goal setting and decision making, problem-solving and action planning, identifying barriers, self-monitoring and feedback, and social support (Table 1). SM was accompanied by various types of therapy modalities including, usual care of physical and/or occupational therapy in most of the studies, yoga [41], behavioural physiotherapy [28], and aerobic exercises [35,38]. Incorporation of SM within exercise programmes enabled insights into potential options for participants as indicated by “It opened my eyes to what can be done” in the qualitative interviews [30,50]. The duration of the intervention varied from a minimum of one week [35] to a maximum of 3 months [27,29,30,42]. The longest duration was a case study that lasted for 18 months [28].

The content, intensity, and duration of therapies within SM training followed various clinical guidelines and were tailored based on the individual capacity of participants. Therapies consisted of fitness, mobility, strength, endurance, balance, range of motion, and stretching exercises. Most of the interventions had one or two sessions per week. The average number of sessions per week was 2.5 sessions and the average duration of each session was 54 min. The details of exercises dosage were not clearly reported in some studies [27,33,43]. Some participants in the qualitative reports indicated that uncertainty of dosage was evident in their participation in SM programmes and that the trainers were not specific about the exercise dosage at home, implied in “I could never pin him down to how long that should be for though” [30].

The SM interventions were delivered to the participants in a group format in eight studies [33,34,36,38,40,41,46,50] and in a one-to-one format in thirteen studies [27–29,31,32,35,42–45,47–49]. In three studies [30,37,39] the SM interventions were delivered in groups and in one-to-one format at some points during the intervention. Participants perceived that group sessions were more beneficial for them to share their experiences as it “was good to talk to them[peers] and find out how they’ve managed” and to increase socialisation [50]. They believed that these sessions allowed them to compare themselves with other participants and appreciate their own circumstances and personal progress.

The majority of interventions were delivered to the participants in their homes [27–29,32,37,39,40,42,43,45,47] or in community settings [30,33,36,38,41,46,50]. One intervention was delivered to the hospitalised patients to promote their physical activity including mobility [35]. Three interventions started in the hospital and continued at home after patient discharge [31,48,49]. SM has been used with several forms of technology for easier delivery of instruction, feedback, communication, and follow-up with therapists. Examples of applied technology include telerehabilitation/telehealth, e-health, mobile apps, and phone calls. Telerehabilitation was used as a part of intervention delivery in six studies [29,34,45,46,48,49]. Phone calls were used for data collection [29,36,45] and as a part of an intervention to provide support or coaching sessions in [32,45]. E-health was integrated into SM to support the home rehabilitation programme [48,49]. The STARFISH intervention [40] was delivered through a mobile app which was also used to track and record physical activity outcomes. The TASK programme was a video-guided exercise programme [44] with two phases, 4 weeks with and 4 weeks without supervision.

4. Effects of SM on functional mobility and walking

Overall, SM seems to have a positive impact on functional mobility and walking outcomes. The Time Up and Go (TUG) was improved significantly in five studies [32,37,43,47,50]. The mean changes in the TUG result across these studies were −3.29 s (p = 0.00), −3.8 s (p < 0.01), −5.7 s (p = 0.03), −13.26 s (p = 0.01), and −2.21 s (p = 0.00). The TUG improved in 81% of the participants in [50] and subjectively a patient participant indicated “One of them there, his walking from the first day to the last day improved 100%”. In [34] there was a significant difference in the Berg Balance Scores between the intervention group and the control group of the study (mean difference −4.27, 95%CI: −6.66 to −1.87).

Walking speed had significantly improved in five studies. Using the 10-meter walking speed and the 5-meter walk test, the mean change in walking speed in [28,31,32,42,44] studies ranged between 0.11 and 0.32 m/s (p ≤ 0.02). In [47] there was a significant improvement in the intervention group as to the time parameter of the 10MWT (p = 0.008), but no significant improvement was found in the number of steps taken. In another study [39], the 6-min walking speed significantly increased in both groups post-intervention (0.09 ± 0.02 m/s in the unsupervised group, 0.06 ±  0.02 m/min in the supervised group) and remained significantly improved by 1 year. The improvement in walking speed was accompanied by a significant reduction in the sedentary time in a different study [32].

Walking distance significantly increased in six studies. The mean changes in the 6MWT were 20 ± 6 m (p = 0.02), 15.36 m (p = 0.02), 47.73 m (p < 0.001) and 85 ± 64 m post-intervention [31,41,46,47]. In [42] the 6MWT increased by 43 m (95% CI 10–76) at three months and by 61 m (95% CI 27–96) at six months post-intervention. In [37] the 6MWT was significantly increased (p = 0.03) in both groups of the study. One study showed trends toward improvement in the 6MWT in five out of eleven participants post-intervention (mean change 47.56 m) [45]. However, the authors reported a trend towards improved overall in all participants (−25.73 ± 87.75 m). The improvement in the walking distance was associated with a significant improvement in the lower body strength (2.51 chair stands, p = 0.01) [41]. The number of steps per day increased significantly in the two studies. The number increased by an average of 2895.5 steps/day (p < 0.001) in one study [35] and by 39.3% (4158–5791 steps/day) in the other one [40]. The walking time was also increased on an average of 20 min/day in this study.

Global measures of mobility

Participants had improved their ability to perform life tasks including daily activities at home and management of social tasks that involve mobility [29]. This was evident by the significant improvements in the basic lower extremity score within the function subscale of the LLFDI (increased from 59.0 to 64.9) and in four of the six LLFDI disability subscales. Participants who had received the intervention at home demonstrated a trend toward improved mobility at the end of intervention (the SIS mobility domain improved by 82.3 (95% CI (74.3–90.3) and the difference between control and intervention scores change was −9.8 (95% CI −20.1–0.4; p = 0.06) in a study [48]. Participants completed 27 min/day (95% CI 4–49) at 3 months more moderate physical activity than at baseline [42].

These findings were confirmed by participants perspectives regarding the positive changes in their mobility and walking ability after participation in SM programmes as a patient-reported “I think I can walk easier and go farther now” [30,41]. The positive change was perceived by a gradual improvement in their physical fitness for mobility (improvement in speed, balance, and strength) and in their walking ability and endurance. The benefits due to incorporation of SM into exercise programmes were recognised by some participants who showed their persistence and desirability to continue exercising after study completion indicated by a patient as “I had a lot of positive physical changes that gave me confidence to keep coming here” [31,41,50].

Grade of evidence

The quality of evidence had a very low grade for the change in walking speed, endurance, balance, and physical activity (steps/day), a low grade for (SIS-mobility domain), and a moderate grade for (Rivermead Mobility Index) (see Supplementary Appendix C for details).

5. Effect of SM on secondary outcomes

Multiple studies [32,33,35,42] showed a significant improvement in participants’ self-efficacy, exercise efficacy, and self-efficacy for physical activity (Table 1) except in two studies study [42,44] where the improvement was not statistically significant. In qualitative studies, participants perceived that the SM programme was a useful strategy for increasing physical activity and for improving confidence and skills related to participating in regular physical activity [42]. Positive change was also suggested by participants to express their improvement in confidence, motivation, self-efficacy, feeling of self-management and self-control that enhanced their recovery, level of independence and to resume their life after stroke [34,50]. The motivation was considered by participants as a facilitator that helped them to achieve positive changes. They also appreciated the presence of motivating peers and the encouragement provided by professionals and other stroke survivors [50].

Quality of life measures was significantly improved in five studies. The SIS and SSQoL scores of different domains were improved (p < 0.05) in [32,36,43,48]. Also, there were some trends toward improvement in the SIS (mean change = 1.45 ± 4.66 ) in [45]. The improvement of the physical dimension score of the SIS was associated with increasing in the Barthel Index (mean difference = 10 and 95% CI= −16.98, −3.02) in [43]. The SAQoL significantly improved (mean = 0.25, p = 0.01) and the improvement remained for three months post-intervention [50]. The result of the EuroQual-5D showed some trends toward improvement (15 out of 100 points (95% CI 4–26) greater than at baseline) [42]. Although the findings of the reviewed studies showed changes in the self-efficacy and quality of life to different extents, the quality of evidence had a very low grade due to methodological issues (see Supplementary Appendix C for details).

Other physical secondary outcomes included participation, extended ADLs, fatigue, pain, and impairment. The included studies showed a significant improvement in the Chedoke-McMaster Stroke Assessment score (p < 0.01) for the leg impairment [32], Nottingham Extended ADLs (p = 0.01) [48], the pain of the knee and back [28], and fatigue (p = 0.003) in the intervention group [40]. In the PROPEL trial, the intervention group had higher participation in physical activity measured by PASIPD (Hedge’s g ≥ 0.5), had fewer barriers to physical activity, and showed higher outcome expectations of their exercise than the comparison group (effects sizes ≥0.5 for SSOE) [38]. The SF 36 physical component was significantly improved (p < 0.01) in the supervised group who also had trends towards the improvement of HAP results [39]. One study showed a moderate improvement in activity level in both groups of the study with no between-group differences [33]. Also, the SF36 physical and general health domains showed trends towards increased positive change following the intervention, but these did not reach significant levels. Physical activity levels did not change in one study [44].

The psychological outcomes reported were mood, depression and anxiety, self-esteem, psychological general well-being index, patient confidence and satisfaction. The reviewed studies showed significant improvements in patient’s anxiety and caregiver’s depression (p = 0.023 and p = 0.003) in [49], mood (p = 0.04) in [46], and cognition (p < 0.01) in [32]. In [47] the score for the self-esteem within the DUKE Health Profile was significantly higher in the intervention group than that in the controls. Implementing of SM intervention for inpatients led to a significantly shorter stay at the hospital for the intervention group (p = 0.04) and fewer readmissions over 12 months of their discharge (p < 0.05) [48]. Qualitative findings indicate that SM interventions such as video conferencing reduce isolation [34]

Stroke knowledge was significantly increased in one study with 95% of participants improving their awareness about stroke and exercise tolerance (p = 0.00) [50]. Qualitative findings [30,41] showed that increasing knowledge was considered by participants as an important component of SM interventions, explained by one participant as “that was absolutely important, because it made sense of why you are doing” [30]. However, it was not measured appropriately in most of the studies. Also, goal setting which is deemed to be central for SM programmes was not commonly assessed. Only in [46], 66% of participants achieved their long-term goals and they met about 68% of their short-term weekly goals. The qualitative findings of one study indicated that most of the participants had not achieved all of their goals [50].

Discussion

This study is one of the first that looked at the role of SM interventions to improve mobility as a specific outcome of stroke rehabilitation. Regardless of limitations in some studies, mainly due to methodological weaknesses, the overall findings of the reviewed literature indicate that SM interventions such as patient education, goal setting, action planning, self-monitoring, and professionals/carers support can help to improve functional mobility and walking. Additionally, other physical activity outcomes, self-efficacy, and quality of life of survivors with mild to moderate impairments post-stroke also showed improvement. The qualitative studies reviewed highlighted the survivors’ perceptions that indicated the importance of involving survivors in their care programmes and facilitating their independence using self-management. Across all studies, SM interventions were found to be feasible, acceptable, and appreciated by participants and their caregivers. SM interventions were not associated with any serious adverse event and can be applied safely in different contexts. Only three studies reported adverse events such as episodes of muscle soreness, fatigue, trips, or non-injurious falls [30,40,42].

Our findings confirm the short-term efficacy of self-management interventions for mobility rehabilitation post-stroke which is consistent with other reviews on the effectiveness of SM programmes for improving functional ability, self-efficacy, participation, and quality of life post-stroke [11,14,16]. Sustaining of the positive effect of SM interventions for the long term is important for stroke rehabilitation; yet most of the studies did not assess the long-term effect of the interventions.

Our results support the findings of other reviews [11,16] that incorporating more of SM strategies in an intervention can lead to better outcomes and that strategies with behaviour change targets have more impact than other strategies. In some studies, it was not easy to identify all strategies that were used as some of them might have been used implicitly as a part of programme delivery. For instance, the process of action planning might include identifying of contextual barriers and motivation of participants without being reported as an explicit part of the intervention. Moreover, the use of some SM strategies such as caregiver support, group therapy, home visit, or tele-visit was described as optional in some studies and the effect of using such strategies was not measured. To draw a clear conclusion about their efficacy, researchers are encouraged to describe all strategies that were incorporated in the SM interventions.

There was no specific therapy training that was considered to be most effective in working with SM. The variability in practice might affect the effectiveness of SM programmes in different contexts. For example, the CARE4STROKE integrated the same SM programme to two different practice guidelines (in Australia and in the Netherlands) and had a different level of improvement in the outcome measures. Hence SM strategies need to be supplementary, flexible and tailored to the patient’s context. Some inconsistent results were found regarding the perception of change following the intervention. Participants demonstrated that comorbidities, stroke sequelae and previous level of exercise might make a difference and work as a facilitator or barrier to the improvement in their physical activity post-stroke. All participants in the reviewed studies had mild to moderate impairments and people with severe impairments (physical, cognitive, or aphasia) were excluded. This confirms that this population might have less opportunities within community programmes as reported by the NICE guideline (2016) [5].

This review found some technologies such as telerehabilitation, smartphones, videoconferencing, and online resources as useful tools that help to implement the SM programmes for survivors who live away from the areas of services or cannot physically attend their sessions. They also have been used to promote participants’ independence in carrying out their training with minimal supervision of a therapist. This can provide proof of using these technologies for the delivery of care during the current situation of COVID-19 lockdown and related social distancing measures.

SM might not be suitable for every stroke survivor. A prior examination of the patient’s readiness and intention to follow the SM programme should be considered. This review found only a few studies that considered this readiness prior to participants’ enrolment in SM interventions. The Stages of Change tool within the Transtheoretical Model and Patient Activation Measure (PAM) to assess participants’ readiness to change have been recommended by the reviewed studies [38,43]. Qualitative findings of this review indicate that motivated people can do better with SM programmes which reflects the critical role that patient’s motivation can play in the successfulness of the implementation of SM interventions. Failure to address these confounding factors perhaps results in withdrawals that were reported in some studies.

Self-management was conceptualised variously by different stakeholders and in different cultures. It seems likely that the concept is ambiguous, or could have different meanings to different parties, or change over time depending on the purpose. This was evident in the participants perspectives about their experience with the SM programmes and in the rhetoric of researchers and authors of the reviewed studies. The conceptualisation of SM might change individual’s attitude towards SM. For instance, some patients might think SM is to prevent them from attending rehabilitation facilities and to let them encounter the challenges in their life after stroke alone and may feel abandoned by the services. Hence, there might be a need for explaining what SM means to the eligible participants.

There were some methodological issues reported in the reviewed study. Blinding was not possible in most studies due to the nature of self-management interventions. The number of men included within studies was notably higher than women in some studies. This does not reflect the incidence of stroke in women. Also, forming a group for interventional groups in the community with planned group sizes might have taken a long time, with possible wait times for some affecting participants’ time of entry into the study. The use of cluster RCT has been suggested for future studies that will examine the effectiveness of these interventions to minimise the risk of contamination among participants and to ensure the proper formation of study groups. Our findings also suggest the use of qualitative methods in addition to the quantitative within the effectiveness studies as that can provide valuable insights on how participants perceived the implementation of SM interventions and if there are contextual issues that can affect the effectiveness of these interventions beyond the therapy protocols. Further, the effects or influence of self-management which is a complex intervention should also be interpreted against the background of several confounding factors that are common to rehabilitation research such as stage of recovery, the intensity of therapy, co-morbidities in this population and physical activity and motivational levels prior to their stroke.

The overall certainty in the evidence for the outcomes was rated using GRADE which can include RCTs and other study designs and if a study design and or quality is weaker than an RCT, the quality of evidence can be downgraded. Also, if a non-RCT meets certain criteria they can still be upgraded. We acknowledge that GRADE is not specific to observational studies [55] and that methods currently have not been identified for a gold standard quality assessment across study designs and limitations do exist.

Conclusion

Self-management interventions seem to be an effective approach that helps stroke survivors to improve their mobility training outcomes. However, participants level of ability, their specific goals, and stage of recovery should be considered when planning for any self-management programme to engage motivation and self-efficacy vital for SM. Many of the identified studies were feasibility studies. Further research should examine the effectiveness of self-management intervention in the long term and report the specific components in order to compare their effectiveness.

Disclosure statement

The authors report no conflict of interests. The authors alone are responsible for the content and writing of this review.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

The authors confirm that they have not received any funding for this work.