Effects of visual arts-based interventions on physical and psychosocial outcomes of people with stroke: A systematic review and meta-analysis

Abstract

Background

Stroke is the leading cause of disability worldwide. People with stroke face a variety of physical and psychosocial issues as a result of their disability. Visual arts-based interventions refer to providing participants with structured visual arts-based activities and non-judgemental and safe environments to improve their health-related outcomes. Recent evidence showed that visual arts-based interventions had the potential to promote the holistic well-being of people with stroke. However, evidence of the effects of visual arts-based interventions on physical and psychosocial outcomes of people with stroke is limited.

Objective

(1) To investigate the effects of visual arts-based interventions on physical and psychological outcomes in people with stroke, and (2) to identify the effective regimen of visual arts-based interventions to improve physical and psychosocial outcomes in people with stroke.

Design

A systematic review and meta-analysis.

Setting

Not applicable.

Participants

483 people with stroke were included in this review.

Methods

Fourteen English and five Chinese databases were searched from the date of their inception to February 2022 using keywords based on the Population, Intervention, Comparison, Outcome, and Study framework. Two reviewers independently assessed trial eligibility and risk of bias and extracted data. Methodological quality was assessed using the revised Cochrane risk of bias tool for randomised trials. Meta-analysis was conducted using Review Manager 5.4. Narrative synthesis was performed when meta-analysis was inappropriate to conduct. The Grading of Recommendations, Assessment, Development, and Evaluation Approach was used to assess the certainty of evidence.

Results

Seven randomised controlled trials were included. A meta-analysis reported statistically significant effects of visual arts-based interventions on depressive symptoms (Standardized mean difference [SMD]: −1.14, 95% confidence interval [CI]: −1.67 to −0.61; three studies; moderate quality of evidence), activities of daily living (SMD: 0.96, 95% CI: 0.24 to 1.69; four studies; low quality of evidence), and upper limb function (SMD: 0.83, 95% CI: 0.42 to 1.24; two studies; low quality of evidence).

Conclusions

Visual arts-based interventions have favourable effects on depressive symptoms, activities of daily living, and upper limb function of people with stroke. However, the quality of evidence ranged from very low to moderate. Limited evidence suggested the effective regimen of visual arts-based interventions. Further rigorous randomised controlled trials should be developed to strengthen the relevant evidence.

Registration

Registered at the International Prospective Register of Systematic Reviews on 11 July 2022 (Number: CRD42022334646).

1. Contribution of the paper

What is already known

• •Stroke is a critical chronic disease and causes physical and psychosocial issues for survivors.
• •Visual arts-based interventions have the potential to promote holistic well-being in people with stroke.
• •Little is known about the effects of visual arts-based interventions on physical and psychosocial outcomes in people with stroke.

What this paper adds

• •Visual arts-based interventions have statistically significant effects on improving depressive symptoms, activities of daily living, and upper limb function in people with stroke.
• •A small number of included studies provided limited evidence of the effects of visual arts-based interventions for stroke recovery.
• •More rigorous randomised controlled trials on visual arts-based intervention for people with stroke are warranted.

2. Introduction

Stroke is the second-leading cause of death and the major cause of disability worldwide (Feigin et al., 2021). Every year, over 12.2 million people globally sustain a new stroke, and around 5.5 million deaths and 116.4 million disability-adjusted life years are caused by stroke worldwide (GBD 2016 Stroke Collaborators, 2019). With the high prevalence of stroke and the high mortality and disability caused by stroke, stroke has become a severe public health issue in the world.

A sudden stroke brings a complex disability; causes body function issues, such as functional, cognitive, visual, or speech impairments (Cramer et al., 2017); and influences survivors’ post-stroke activities and participation, such as work, leisure activities, or interactions with environments (Carmo et al., 2015; Edwards et al., 2018; Lin et al., 2021; Zhang et al., 2018), which further lead to physical or psychosocial issues for survivors (Garrett-Jones et al., 2019; Skoglund et al., 2019). Physical issues refer to the problems relating to body function, physical functioning or performance, or functional capacity (Tomey and Sowers, 2009). People with stroke commonly experience physical issues, such as difficulties with regaining optimal movement ability and independence in activities of daily living (Raghavan, 2015; Yoo et al., 2020). Psychosocial issues indicate the problems regarding mind, feeling, emotion, or involvement in activity that interacts with others in society or the community (Ro and Clark, 2009), such as depressive symptoms, anxiety, low self-efficacy, and low social participation, which also commonly occur across the acute to chronic stroke stages (McCurley et al., 2019; Nicholson et al., 2013; Poltawski et al., 2016; White et al., 2014). Evidence supports that those post-stroke physical and psychosocial issues adversely affect survivors’ adherence to rehabilitation services (Almhdawi et al., 2021; Ferro et al., 2016), functional outcomes (Haagsma et al., 2015), interpersonal relationships (Bosma et al., 2018; van der Zee et al., 2013), and reintegration into the community (Palstam et al., 2019; Wray and Clarke, 2017). A large body of evidence has further shown that physical and psychosocial issues compromise health-related quality of life (Kwon et al., 2018; Paek et al., 2014) and increase mortality after stroke (Otto, 2018; Venna and McCullough, 2015). To reduce the negative consequences caused by stroke, stroke rehabilitation programmes are necessary to support survivors to address their physical and psychosocial issues (Rudd et al., 2017; Teasell et al., 2020; Zhang et al., 2020).

Visual arts-based interventions refer to providing non-judgemental and safe environments, enhancing positive stimulations (Fraser and Sayah, 2011; Lo et al., 2018), and using structured creative visual arts-based activities to improve participants’ health-related outcomes (Davies et al., 2012; Fancourt and Finn, 2019; Januchta-Szostak, 2010). With visual arts-based activities, participants are required to use their upper limbs to perform specific movements with gross or fine motor skills to complete the activities, such as hand stretching and picking up or controlling an object with a specific grasp (Schaechter et al., 2002). Those performances further improve participants’ movement ability and contribute to their eye–hand coordination (Lusebrink, 2004; Schaechter et al., 2002). Additionally, the process of visual arts-based activities provides a period of time for participants to focus on their verbal or non-verbal self-expressions and encourages them to share their personal experiences to others in groups (Regev and Cohen-Yatziv, 2018), which further results in promoting their mental health and social functions (Caddy et al., 2012). Masika et al. (2020) also found that visual arts-based activities provided participants with positive brain-related stimulations, which contribute to their cognitive functions. Given the benefits of structured visual arts-based activities, visual arts-based interventions have been suggested as adjunctive interventions in healthcare contexts (Kim and Lor, 2022).

In the last decade, an increasing number of researchers have examined the benefits of visual arts-based interventions on physical or psychosocial outcomes in people with stroke. Previous researchers provided evidence that visual arts-based interventions contributed to improvements in limb functions (Baumann et al., 2013; Morris et al., 2016), social interactions (Sit et al., 2017; Morris et al., 2016), mental health (Beesley et al., 2011), and self-efficacy in people with stroke (Baumann et al., 2013; Symons et al., 2011). A recent critical review gathered quantitative and qualitative findings from ten studies involving qualitative studies, case studies, randomised controlled trials, and quasi-experimental studies on visual arts-based interventions for people with stroke (Pang et al., 2021). These reviewers suggested that visual arts-based interventions potentially contribute to holistic well-being in people with stroke (Pang et al., 2021). However, in this review, the latest experimental studies written in English or Chinese were not included, and the effects of interventions on physical and psychosocial outcomes and the effective regimen of interventions were not identified due to insufficient quantitative data (Pang et al., 2021). To better understand the effects of visual arts-based interventions for people with stroke, an updated review of the literature with meta-analysis is needed.

This systematic review and meta-analysis aimed to summarise the current evidence on the effects of visual arts-based interventions for people with stroke. The specific objectives were (1) to investigate the effects of visual arts-based interventions on physical and psychosocial outcomes of people with stroke and (2) to identify an effective regimen of visual arts-based interventions to improve physical and psychosocial outcomes of people with stroke.

3. Methods

This review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Page et al., 2021). It was registered in the International Prospective Register of Systematic Reviews on 11 July 2022 with registration number: CRD42022334646.

3.1. Search strategy

The reviewers used the three-stage approach as the search strategy to identify published and unpublished relevant studies written in English or Chinese. Firstly, we identified the initial keywords, index terms, and matched subject headings to search in electronic bibliographic databases. The search strategies specific to each database were developed in consultation with a university librarian. Secondly, we extensively searched the target databases with developed search strategies to identify the potentially relevant studies that met the eligibility criteria. Thirdly, we screened the reference list of retrieved studies to identify relevant studies that were not searched by the search strategies (Aromataris and Munn, 2020).

Fourteen English databases (MEDLINE, CINAHL Complete, EMBASE, PubMed, Scopus, Academic Search Ultimate, Arts and Humanities Citation Index at Web of Science, British Nursing Index, Cochrane Library, ERIC, Global Health, JBI Connect+, PsycINFO, and Wiley Online Library) and five Chinese databases (Airiti Library, China Academic Journal Network Publishing Database, Wanfang database, Chinese Social Sciences Citation Index, and VIP Chinese Science Journals Database) were searched to identify published relevant studies. The search keywords with three different languages were designed based on the Population, Intervention, Comparison, Outcomes, and Study framework. The English keywords, such as ‘stroke’, ‘visual art’, ‘creative art’, ‘psychosocial’, ‘physical’ and ‘function’; the Traditional Chinese keywords, such as ‘腦中風’, ‘視覺藝術’, ‘創作藝術’, ‘社會心理’, ‘生理’, and ‘功能’; and the Simplified Chinese keywords, such as ‘脑卒中’, ‘视觉艺术’, ‘社会’, 心理’, ‘生理’, and ‘功能’, were adopted. The full lists of search terms and strategies are shown in Supplemental Materials 1 to 3.

Other online resources were searched to identify unpublished relevant studies. Two grey literature resources, including Open Grey and Grey Literature Report, were searched to identify potentially relevant grey literature. Six trial registries, including CenterWatch, ClinicalTrials.gov, the University of Hong Kong Clinical Trials Registry, the United States National Institutes of Health, the World Health Organisation, and the International Clinical Trials Registry Platforms, were searched to identify ongoing studies or clinical trial registries. Ten dissertation and thesis sources, including Airiti Theses, China Doctoral and Masters Dissertations Full-text Database, Digital Access to Research Theses - Europe E-Theses Portal, National Digital Library of Theses and Dissertations in Taiwan, Networked Digital Library of Theses and Dissertations, Electronic Theses and Dissertations Collection of the Chinese University of Hong Kong, Electronic Theses of the Hong Kong Polytechnic University, Digital Dissertations Consortium of The University of Hong Kong, Edinburgh Research Archive, and ProQuest Dissertations and Theses, were also searched to identify relevant dissertations or theses.

3.2. Eligibility criteria

The eligibility criteria were performed in accordance with the Population, Intervention, Comparison, Outcomes, and Study framework.

3.2.1. Population

Regardless of gender and ethnicity, participants who were 18 years old or older and had either first-ever or recurrent ischaemic or haemorrhagic stroke were included in this review. Study participants could be recruited from either inpatient or community settings.

3.2.2. Intervention

Interventions consisting of one or more than one of the following structured visual arts-based activities as the key components of interventions were included in this review: (1) traditional fine art activity, such as drawing, painting, printmaking, photography, or sculpture; (2) media art activity, such as film, graphic communications, or animation; and (3) folk art activity, such as ceramic, fibre, jewellery, or work in wood, paper, or clay (Davies et al., 2012; Fancourt and Finn, 2019; Fraser and Sayah, 2011; Januchta-Szostak, 2010). In addition, the included studies were not limited by their delivering the intervention in different formats (individually, in group, or hybrid), modes (in person, online, or hybrid), number of sessions (at least one), settings (inpatient, community, or hybrid), or leadership by either trained or non-trained facilitators, such as professionals, peers, or volunteers.

3.2.3. Comparison

The comparisons included waitlist, usual stroke care, or non-visual arts-based interventions.

3.2.4. Outcomes of interest

Studies had to report at least one of the following outcomes:

Psychological outcomes: We defined psychological outcomes as the outcomes relating to mind, feeling, or emotion, such as depressive symptoms, anxiety, or self-efficacy.

Social outcomes: We defined social outcomes as the outcomes relating to involvement in activity that interacts with others in society or the community, such as social participation.

Physical outcomes: We defined physical outcomes as the outcomes relating to body function, physical functioning or performance, or functional capacity, such as upper-limb function, hand function, or activities of daily living.

General health outcomes: We defined general health outcomes as the outcomes relating to the complete physical, mental, and social health status, such as health-related quality of life.

Other outcomes: Adverse event.

3.2.5. Study design

Randomised controlled trials were included.

3.2.6. Language

Studies written in English, Traditional Chinese, or Simplified Chinese were included.

3.3. Exclusion criteria

Participants with multiple diagnoses that caused pre-existing physical disability, mental health or social problems, dysphasia, or severe cognitive impairment were excluded. In addition, we excluded brief reports, clinical guidelines or recommendations, conference abstracts, letters, editorials, study protocols, non-experimental studies, quasi-experimental studies, case studies, cross-sectional studies, cohort studies, qualitative studies, and reviews.

3.4. Study selection

All articles were checked for duplicates. Two reviewers (MWMC and AT) independently screened the initial inspection of titles, keywords, and abstracts based on the eligibility criteria. A full-text review was performed on the remaining articles to confirm their eligibility. Any discrepancies in study selection between two reviewers were discussed and resolved with the third reviewer (SHSL).

3.5. Data extraction

Two reviewers (MWMC and AT) independently extracted details of included studies using a self-developed form (see Supplemental Material 4). The extracted data included publication information (authors, publication year, and publication country), study characteristics (study design and setting), participant characteristics (sample size, gender, mean age, type of stroke, duration since stroke,  and stroke severity), intervention characteristics (conceptual framework or model application, component, delivery format and mode, duration and frequency, facilitator, and types and contents of control group), outcome measurement (instrument and time point of measurement), and findings (effect, adverse event, and attrition rate). Authors of the included studies were contacted to obtain additional data if necessary.

3.6. Methodological quality assessment

Two reviewers independently assessed the methodological quality of included studies (MWMC and AT). Any discrepancies between two reviews were discussed and resolved with the third reviewer (SHSL) to achieve consensus. The revised Cochrane risk of bias tool for randomised trials was used to assess the risks of bias in five domains: randomisation process, deviations from intended interventions, missing outcome data, outcome measurement, and outcome selection (Higgins et al., 2021; Sterne et al., 2019). The judgements of each domain and overall risk of bias were inferred as low risk of bias, some concerns, or high risk of bias (Higgins et al., 2021). The methodological quality results were presented as a traffic light plot with the online Risk-Of-Bias VISualization tool (McGuinness and Higgins, 2021).

3.7. Certainty of evidence appraisal

The Grading of Recommendations Assessment, Development, and Evaluation method was used to assess the certainty of evidence. The starting point is a high quality rating for randomised clinical trials. The quality rating can be downgraded based on five aspects: risk of bias, inconsistency, indirectness, imprecision, and publication bias (GRADEpro GDT, 2021). A single downgrade corresponds to one quality level change in the ranking of high, moderate, low, and very low (Balshem et al., 2011; Schünemann et al., 2013).

3.8. Data synthesis

A meta-analysis was conducted using Review Manager 5.4 software (Review Manager, 2020). A minimum of two studies with similar outcome measurements was involved in each meta-analysis (Higgins et al., 2021; Ryan and Hill, 2016). The standardised mean difference (SMD) and the 95% confidence interval (CI) were calculated to synthesise the pooled effects of continuous data generated by similar outcome measurements across the included studies. The effect size was interpreted as small (0.2=SMD<0.5), moderate (0.5=SMD<0.8), and large (SMD≧0.8) (Cohen, 1988). I-squared (I²) statistics were used to evaluate the heterogeneity across the included studies in each meta-analysis. I²≥50% was considered substantial heterogeneity (Higgins et al., 2002; Higgins et al., 2021). Given the consideration of clinical and methodological heterogeneity with different modes, formats, duration, frequency, and sessions, random-effect models were more appropriate than fixed-effect models for the use of meta-analysis (Borenstein et al., 2010; Riley et al., 2011). All significance levels were set at p<0.05. Additionally, sensitivity analysis was conducted to determine the robustness of specific outcomes by removing one included study at one time if substantial heterogeneity was observed (Cohen, 1988; Higgins et al., 2021). If statistical pooling was not possible, the results were presented by narrative synthesis for further data synthesis and discussion.

4. Results

4.1. Search results

A total of 1391 articles from English and Chinese electronic databases and other online resources and seven articles from screening the reference list of included studies were found. Due to duplication, 193 articles were removed. After screening the remaining articles, a total of 1171 articles were excluded because of irrelevant titles, keywords, or abstracts. After a full-text review of the remaining 34 articles, 27 were excluded due to irrelevant intervention (n = 9), non-randomised controlled trial (n = 14), study protocol (n = 3), or abstract (n = 1). Finally, seven full-text articles were included (see Fig. 1). In addition, the first reviewer (MWMC) contacted the authors of one included study (Liu and Lu, 2018) for further details.

Fig. 1.

Flow chart diagram presenting the study selection with Preferred Reporting Items for Systematic Reviews and Meta-analyses guideline.

Note. RCT= Randomised controlled trial.

4.2. Methodological quality assessment

A summary of methodological quality is shown in Fig. 2(a) and 2(b). Three included studies were assessed to have low risk of bias (Ellis-Hill et al., 2019; Kongkasuwan et al., 2016; Morris et al., 2019), two had some concerns (Huang et al., 2020; Wu et al., 2011), and two had high risk of bias (Huang et al., 2016; Liu and Lu, 2018).

Fig. 2.

Risk of bias: Overall (a) and assessment of individual studies (b).

All included studies were found to have low risk of bias in outcome measurement (Ellis-Hill et al., 2019; Huang et al., 2016, 2020; Kongkasuwan et al., 2016; Liu and Lu, 2018; Morris et al., 2019; Wu et al., 2011). Four studies presented some concerns in selecting the reported result because of absence of information about the pre-specified analysis plan (Huang et al., 2016, 2020; Liu and Lu, 2018; Wu et al., 2011). Some concerns were identified in the randomisation process because four studies did not mention the information about allocation concealment (Huang et al., 2016, 2020; Liu and Lu, 2018; Wu et al., 2011). Two studies were assessed to have a high risk of bias in missing outcome data because of absence of information about the attrition rate, which might be associated with the true values of the missing outcome data (Huang et al., 2016; Liu and Lu, 2018). In addition, two studies were rated as having high risk of bias in deviations from the intended intervention because of absence of information about the protocol and blinding (Huang et al., 2016; Liu and Lu, 2018).

4.3. Study characteristics

Seven included studies were published between 2011 and 2020. All studies were conducted as randomised controlled trials and were conducted in Mainland China (n = 4) (Huang et al., 2016, 2020; Liu and Lu, 2018; Wu et al., 2011), the United Kingdom (n = 2) (Ellis-Hill et al., 2019; Morris et al., 2019), and Thailand (n = 1) (Kongkasuwan et al., 2016). Five studies were conducted in stroke rehabilitation units (Huang et al., 2016; Kongkasuwan et al., 2016; Liu and Lu, 2018; Morris et al., 2019; Wu et al., 2011), one study was conducted in community centres (Ellis-Hill et al., 2019), and the remaining one was conducted in a stroke rehabilitation unit and at home (Huang et al., 2020) (see Table 1).

Table 1.

Summary of study and participant characteristics.

Study	Country	Study design	Setting	Participant
				Sample size	Mean age	Gender	Type of stroke	Time since stroke	Stroke severity
Ellis-Hill et al. (2019)	The United Kingdom	Feasibility RCT (parallel-arm)	Community	Total: 56 IG: 29/CG: 27	70±12.1	M: 32 F: 24	Ischaemic stroke: IG: 21/CG: 20	≦2 years	Mild to moderate
Huang et al. (2016)	Mainland China	RCT (two-arm)	Stroke rehabilitation unit	Total: 48 IG: 24/CG: 24	IG: 50.26±17.88 CG: 52.67±14.74	M: 21 F: 27	Ischaemic stroke: IG: 16/CG: 15	>30 days to 12 months	Brunnstrom IV-VI stage
Huang et al. (2020)	Mainland China	RCT (two-arm)	Stroke rehabilitation unit + home	Total: 60 IG:30/CG:30	62.03±8.16	M: 40 F: 20	Ischaemic stroke: IG: 28/CG: 29	>2 weeks	Brunnstrom IV-V stage
Liu and Lu (2018)	Mainland China	RCT (two-arm)	Stroke rehabilitation unit	Total: 60 IG: 30/CG: 30	60	M: 36 F: 24	Not stated	>2 weeks	NR
Kongkasuwan et al. (2016)	Thailand	RCT (two-arm)	Stroke rehabilitation unit	Total: 118 IG:59/CG:59	IG: 67.1 ± 9.2 CG: 65.5 ± 9.9	M:55 F:63	Ischaemic stroke: IG: 46/CG: 43	NR	NR
Morris et al. (2019)	The United Kingdom	Feasibility RCT (two-arm)	Stroke rehabilitation unit	Total: 81 IG:41/CG:40	IG: 77 CG: 75.6	M:36 F:45	Ischaemic stroke: IG: 36/CG: 35	≦2 weeks	Mild to moderate
Wu et al. (2011)	Mainland China	RCT (two-arm)	Stroke rehabilitation unit	Total: 60 IG: 30/CG: 30	49±6.3	M: 38 F: 22	Ischaemic stroke: IG: 17/CG: 14	13.7 ± 7.5 months	NR

Notes. NR=Not Reported; IG=Intervention Group; CG=Control Group; M=Male; F=Female; RCT=Randomised Controlled Trial.

4.4. Participant characteristics

A total of 483 people with stroke (intervention group [n = 243]; control group [n = 240]) were included in this review. The sample size varied from 48 to 118. The mean age of participants ranged from 49±6.3 to 70±12.1 years old. The majority of participants suffered an ischaemic stroke (n = 320, 66.25%). Most participants were male (n = 257, 53.4%). At the time of recruitment, the duration since stroke ranged from 2 weeks to less than 2 years. Only two included studies reported levels of stroke severity or disability in participants (Ellis-Hill et al., 2019; Morris et al., 2019). A total of 134 participants experienced mild to moderate stroke severity or disability at the time of recruitment (see Table 1).

4.5. Interventional characteristics

4.5.1. Use of conceptual framework or model

One included study used an intervention developed by underpinning a conceptual framework (Ellis-Hill et al., 2019). This study used the Life Thread Model (Ellis-Hill et al., 2008) and the Self-Discrepancy Theory (Gracey et al., 2008, 2009) to provide a structured intervention design for each session to re-establish a positive sense of self and confidence. In this study, meaningful narrative, non-verbal expression, exploration of sense, and emotional support by the facilitator and group sessions were adopted to approach the personal construction of self within this conceptual framework (Ellis-Hill et al., 2019) (see Table 2).

Table 2.

Summary of intervention and outcome characteristics.

Study	Details of intervention		Outcome measurements	Findings
	IG (Intervention type, conceptual framework, components, strategies, facilitator, duration and frequency, delivery mode and format)	CG
Ellis-Hill et al. (2019)	heART of Stroke (HoS) – community art-based and health group intervention Conceptual framework: The life thread model and the self-discrepancy theory Components: Multi-visual arts-based activities (painting, drawing, clay, textiles, or mixed-media) Strategies: Sharing and support Facilitator: Art and health practitioner Duration and frequency: 2-hour session with total 10 sessions over 14 weeks (Total 20 h). The First 7 sessions weekly, session 8–9 fortnightly, session 10 at 3 weeks later Delivery mode and format: Face-to-face Format: Group	Post-stroke outpatient follow-up: Stroke consultants, day hospital service, additional therapy, or support	At baseline and 1 month post-intervention (18 weeks): Anxiety: HADS Depressive symptoms: HADS HRQOL: SF-36	Attrition rate: 16.1% Adverse events: NR Effects: No significant differences on anxiety, depressive symptoms, and HRQOL between groups at 1 month post-intervention
Huang et al. (2016)	Handicraft training programme Conceptual framework: NA Components: Handicraft-making (plastic flowers making and collages) Strategies: Task setting Facilitator: Handicraft therapist Duration and frequency: 30-minute session with 6 times per week for 4 weeks (Total 12 h) Delivery mode: Face-to-face Format: Individual and group	Usual stroke care in inpatient rehabilitation: Medication and rehabilitation nursing care	At baseline and immediately post-intervention (4 weeks): ADL: MBI Upper limb function: FMA-UE Hand function: STEF	Attrition rate: NR Adverse events: NR Effects: Compare with the CG, participants in IG had: Statistically significant improvement in upper limb function Statistically significant improvement in hand function Statistically significant improvement in ADL Immediately post-intervention
Huang et al. (2020)	Ultra-light clay manual training programme Conceptual framework: NA Components: Handicraft-making (Making ultra-light clay) Strategies: Provision of information, sharing, feedback, and support Facilitator: Therapist, physician, and nurse Duration and frequency: Around 30-minute session twice per day for 2 months (Total around 60 h) Delivery mode: Face-to-face Format: Individual (early stage) and group (later stage)	Usual stroke care in inpatient rehabilitation: Medication, exercise therapy, physiotherapy, and acupuncture	At baseline, 1 month, and immediately post-intervention (2 months): ADL: BI Upper limb function: FMA-UE	Attrition rate: 0% Adverse events: NR Effects: Compare with the CG, participants in IG had: Statistically significant improvement in upper limb function Statistically significant improvement in ADL At 1 month and immediately post-intervention
Liu and Lu (2018)	Mandala drawing therapy Conceptual framework: NA Components: Drawing (drawing and colouring Mandala pictures) Strategies: Sharing Facilitator: NR Duration and frequency: 1-hour session with 3 times per week (Not specified the duration) Delivery mode: Face-to-face Format: NR	Usual stroke care in inpatient rehabilitation: Medication, exercise therapy, physiotherapy, and cognitive speech therapy	At baseline, 2 weeks, 4 weeks, 6 weeks, and immediately post-intervention (8 weeks): Depressive symptoms: HAMD Anxiety: HAMA	Attrition rate: NR Adverse events: NR Effects: Compare with the CG, participants in IG had: Statistically significant improvement in depressive symptoms Statistically significant improvement in anxiety At 2 weeks, 4 weeks, 6 weeks, and immediately post-intervention
Kongkasuwan et al. (2016)	Creative art therapy Conceptual framework: NA Components: Multi-visual arts-based activities (drawing, claying, and handicrafts-making) Strategies: Sharing and support Facilitator: Physical therapist and creative art therapist Duration and frequency: 90–120-minute session with twice a week for 4 weeks (Total 12–16 h) Delivery mode: Face-to-face Format: Group	Physiotherapy in inpatient rehabilitation: 60–120 min session with 5 times per week for 4 weeks (Total 20–40 h)	At baseline and immediately post-intervention: Anxiety: HAMA Depressive symptoms: HAMD Physical performance: MBI HRQOL: The pictorial Thai Quality of Life questionnaire	Attrition rate: 4.23% Adverse events: NR Effects: Compare with the CG, participants in IG had: Statistically significant improvement in depressive symptoms Statistically significant improvement in physical function Statistically significant improvement in HRQOL No significant differences on anxiety Immediately post-intervention
Morris et al. (2019)	Visual art participation programme - the Tayside Creative Engagement Intervention (TCEI) Conceptual framework: NA Components: Multi-visual arts-based activities (drawing, collage, printing, painting, or mixed-media techniques) Strategies: Task setting and sharing Facilitator: Visual artists Intensity: 60-minute (individual)/ 90-minute (group) session with maximum of 8 sessions for 3 to 5 weeks (4 individual sessions, 4 group sessions) (Total 10 h) Delivery mode: Face-to-face Format: Individual and group (a maximum of 5 participants per group)	Usual stroke care in inpatient rehabilitation: Physiotherapy, occupational therapy, and as necessary, speech, and language therapy with 90 min session with 5 times per week (mostly) (Not specified the duration)	At baseline, immediately, and 3 months post-intervention: HRQOL: SIS Social participation: SIS Hand function: SIS Self-efficacy: GSES	Attrition rate: 23% Adverse events: NR Effects: Compare with the CG, participants in IG had: Improvement in social participation immediately post-intervention No significant differences on social participation at 3 months post-intervention No significant differences on hand function and self-efficacy immediately and 3 months post-intervention
Wu et al. (2011)	Handicraft training programme Conceptual framework: NA Components: Handicraft-making (embroidery, paper cutting, and collage) Strategies: Provision of information, task setting, sharing, feedback, and support Facilitator: Therapist Duration and frequency: 1-hour session once a day for 6 months (Total 168 h) Delivery mode: Face-to-face Format: Group (10 patients per group, 2 patients per team)	Usual stroke care in inpatient rehabilitation: Medication and rehabilitation nursing care	At baseline, 2 months, and immediately post-intervention (6 months): Depressive symptoms: HAMD ADL: BI	Attrition rate: 0% Adverse events: NR Effects: Compare with the CG, participants in IG had: Statistically significant improvement in depressive symptoms Statistically significant improvement in ADL At 2 months and immediately post-intervention

Notes. NA=Not Applicable; NR=Not Reported; IG=Intervention Group; CG=Control Group; BI=Barthel Index; FMA-UE=Fugl-Meyer Assessment for Upper Extremity; GSES= General Self-Efficacy Scale; HADS=Hospital Anxiety and Depression Scale; HAMA=Hamilton Anxiety Scale; HAMD=Hamilton Depression Rating Scale; MBI=Modified Barthel Index; SIS=Stroke Impact Scale; STEF=Simple Test for Evaluating Hand Function; SIS=Stroke Impact Scale; ADL=Activities of Daily Living; QOL=Quality of life; HRQOL=Health-Related Quality of Life.

4.5.2. Duration and frequency

The duration of interventions ranged from 4 weeks to 6 months (Huang et al., 2016; Kongkasuwan et al., 2016; Morris et al., 2019; Wu et al., 2011). The frequency was between twice weekly and twice daily (Huang et al., 2020; Kongkasuwan et al., 2016). The number of sessions ranged from 8 to 168 (Kongkasuwan et al., 2016; Morris et al., 2019; Wu et al., 2011), and the length of each session varied from 25 min to 120 min (Ellis-Hill et al., 2019; Huang et al., 2020; Kongkasuwan et al., 2016) (see Table 2).

4.5.3. Delivery mode and format

All interventions were delivered in person. Three interventions were conducted in groups with two to 10 participants (Ellis-Hill et al., 2019; Kongkasuwan et al., 2016; Wu et al., 2011). Three interventions were conducted as hybrid of individual and group-based interventions (Huang et al., 2016, 2020; Morris et al., 2019). As for the hybrid interventions, the one-to-one session was implemented for beneficial engagement with the facilitator (Huang et al., 2016; Morris et al., 2019) and the early training in visual arts-based activities (Huang et al., 2020); the group session was performed for the benefit of social interaction (Huang et al., 2016, 2020; Morris et al., 2019). Liu and Lu (2018) did not report the delivery format (see Table 2).

4.5.4. Components

Visual arts-based activity was identified as the key intervention component in included studies, which included plastic handicraft-making, such as flower making, embroidery, collage, pottery, and textiles, as well as drawing, such as colouring, printmaking, and painting. Two included studies selected specific visual arts-based activities as the components based on the participants’ interests or condition of physical impairments (Ellis-Hill et al., 2019; Morris et al., 2019). One intervention adopted multi-visual arts-based activities, which included drawing and handicraft-making (Kongkasuwan et al., 2016) (see Table 2).

In addition, we also identified that all included studies adopted at least one of five motivation strategies to assist the intervention delivery. Five motivation strategies included provision of information about stroke and self-care skills, setting goals about performing visual arts-based activities, sharing experiences about visual arts-based activities with participants or facilitators in group, providing positive feedback, and receiving support for visual arts-based activities or emotional support from facilitators or participants in group (see Table 3).

Table 3.

Summary of five motivation strategies.

Study	Components of the interventions
	Provision of information	Task setting	Sharing	Feedback	Support
Ellis-Hill et al. (2019)	NA	NA	Sharing art pieces or arts-based processes with peers or facilitators	NA	Support by facilitators and peers
Huang et al. (2016)	NA	Setting tasks about arts-based activity as goals to match and train participants’ limb function	NA	NA	NA
Huang et al. (2020)	Provision of information about stroke and self-care	NA	Sharing art pieces or arts-based processes with peers	Positive feedback by facilitators	Support by facilitators and peers
Liu & Lu (2018)	NA	NA	Sharing art pieces with facilitators	NA	NA
Kongkasuwan et al. (2016)	NA	NA	Sharing art pieces or arts-based processes with peers or facilitators	NA	Support by facilitators
Morris et al. (2019)	NA	Setting tasks about arts-based activity as goals to match participants’ interests and physical function	Sharing art pieces or arts-based processes with peers or facilitators	NA	NA
Wu et al. (2011)	Provision of information about health and stroke care	Setting tasks about arts-based activity as goals to match and train participants’ limb function	Setting role models to share arts-based process with peers	Positive feedback by facilitators	Support by facilitators and peers

Note. NA=Not Applicable.

4.5.5. Facilitators

Five interventions were facilitated by different trained facilitators, such as therapist, physician, nurse, or qualified artist (Huang et al., 2016, 2020; Kongkasuwan et al., 2016; Morris et al., 2019; Wu et al., 2011). One intervention was facilitated by art and health practitioners with experience in working with various patients in art participation programmes (Ellis-Hill et al., 2019). Liu and Lu (2018) did not provide information regarding the facilitator (see Table 2).

4.6. Comparison characteristics

Six studies adopted usual stroke care in stroke rehabilitation units as the control group, which involved providing medication, nursing care, physiotherapy, exercise therapy, cognitive therapy, occupational therapy, speech and language therapy, or acupuncture (Huang et al., 2016, 2020; Kongkasuwan et al., 2016; Liu and Lu, 2018; Morris et al., 2019; Wu et al., 2011). One included study reported that the control group received outpatient follow-up services, which mainly included consultant service, day hospital service, or additional therapy if necessary (Ellis-Hill et al., 2019) (see Table 2).

4.7. Outcome characteristics

Six studies measured the outcomes immediately post-intervention (Huang et al., 2016, 2020; Kongkasuwan et al., 2016; Liu and Lu, 2018; Morris et al., 2019; Wu et al., 2011). Two studies performed further follow-up outcome assessments: one was at one month post-intervention (Ellis-Hill et al., 2019), and one was at three months post-intervention (Morris et al., 2019). Amongst the included studies, activities of daily living and upper limb function were measured immediately post-intervention; depressive symptoms and anxiety were assessed immediately and one month post-intervention; hand function, self-efficacy, social participation, and health-related quality of life were evaluated immediately and three months post-intervention.

In addition, no included study specified primary and secondary outcomes or reported adverse events. Five reported attrition rates below 25% with reasons for attrition (Ellis-Hill et al., 2019; Huang et al., 2020; Kongkasuwan et al., 2016; Morris et al., 2019; Wu et al., 2011). The remaining two studies did not provide attrition rates (Huang et al., 2016; Liu and Lu, 2018) (see Table 2).

4.7.1. Effects of visual arts-based interventions

Meta-analyses were conducted by combining specific outcome measurements immediately post-intervention from six studies (Huang et al., 2016, 2020; Kongkasuwan et al., 2016; Liu and Lu, 2018; Morris et al., 2019; Wu et al., 2011).

4.7.1.1. Psychological outcomes

Depressive symptoms: Three studies evaluated the impact of visual arts-based interventions on depressive symptoms (Kongkasuwan et al., 2016; Liu and Lu, 2018; Wu et al., 2011). This outcome was measured using the Hamilton Depression Rating Scale or the Hospital Anxiety and Depression Scale. Pooled results from meta-analysis revealed that visual arts-based interventions could significantly decrease depressive symptoms in people with stroke completing the outcome measurements immediately (SMD: −1.14, 95% CI: −1.67 to −0.61; three studies, 233 participants; moderate quality of evidence) (see Fig. 3(a) and Table 4) and showed significant heterogeneity (I²: 70%) (see Fig. 3(a)). Although there was a substantial heterogeneity, the direction of effects was consistent, and sensitivity analysis indicated that the pooled results were robust.

Fig. 3.

Forest plots of visual arts-based interventions on depressive symptoms (a), anxiety (b), activities of daily living (c), upper limb function (d), and hand function (e).

Notes. Chi²=Chi-square statistic; CI=Confidence interval; df=Degrees of freedom; I²=I-square heterogeneity statistic; IV=Weighted mean difference; p = p-value; SD=Standard deviation; Tau²=Tau-square statistic; Z = Z statistic.

Table 4.

Summary of quality of evidence appraisal.

Certainty assessment							№ of patients		Effect		Certainty	Importance
№ of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Intervention group	Control group	Relative (95% CI)	Absolute (95% CI)
Depressive symptom (assessed by HAMD & HADS)
3	Randomised  trials	Serious	Not serious	Not serious	Serious	Strong association	114	119	–	SMD 1.14 lower (1.61 lower to 0.67 lower)	⨁⨁⨁◯ Moderate	CRITICAL
Anxiety (assessed by HAMA & HADS)
2	Randomised  trials	Serious	Serious	Not serious	Serious	Strong association	84	89	–	SMD 0.80 lower (1.71 lower to 0.11 higher)	⨁⨁◯◯ Low	CRITICAL
Activities of daily living (assessed by BI & MBI)
4	Randomised  trials	Serious	Serious	Not serious	Serious	Strong association	138	143	–	SMD 0.96 higher (0.24 higher to 1.69 higher)	⨁⨁◯◯ Low	CRITICAL
Upper limb function (assessed by FMA-UE)
2	Randomised  trials	Very serious	Not serious	Not serious	Serious	Strong association	54	54	–	SMD 0.83 higher (0.42 higher to 1.24 higher)	⨁⨁◯◯ Low	CRITICAL
Hand function (assessed by STEF & SIS)
2	Randomised  trials	Serious	Serious	Not serious	Serious	None	57	62	–	SMD 0.47 SD higher (0.43 lower to 1.37 higher)	⨁◯◯◯ Very low	CRITICAL

Note. BI=Barthel Index; FMA-UE=Fugl-Meyer Assessment for Upper Extremity; HADS=Hospital Anxiety and Depression Scale; HAMA= Hamilton Anxiety Scale; HAMD=Hamilton Depression Rating Scale; MBI=Modified Barthel Index; SIS=Stroke Impact Scale; STEF= Simple Test for Evaluating Hand Function; SMD=Standardised mean difference; SD= Standard deviation.

As only one study reported depressive symptoms at one month follow-up, meta-analysis could not be conducted (Ellis-Hill et al., 2019). This study showed that the visual arts-based intervention decreased participants’ depressive symptoms at one month post-intervention (see Table 2).

Anxiety: Two studies examined anxiety using the Hamilton Anxiety Rating Scale or the Hospital Anxiety and Depression Scale (Kongkasuwan et al., 2016; Liu and Lu, 2018). Meta-analysis showed no difference in participants’ anxiety between the intervention and control groups immediately post-intervention (SMD: −0.80, 95% CI: −1.71 to 0.11; two studies, 173 participants; low quality of evidence) (see Fig. 3(b) and Table 4). Significant heterogeneity was found in the pooled results (I²: 86%) (see Fig. 3(b)). Sensitivity analysis indicated that the pooled results were robust.

As only one study reported the level of anxiety at one month post-intervention (Ellis-Hill et al., 2019), meta-analysis could not be conducted. The result of this study was that the participants had a lower score of  anxiety in the intervention group at one month post-intervention compared to the baseline assessment (see Table 2).

Self-efficacy: One study reported general self-efficacy by the General Self-Efficacy Scale; thus, meta-analysis could not be conducted (Morris et al., 2019). This study found no difference in general self-efficacy between the intervention and control groups immediately and three months post-intervention (see Table 2). It is worth mentioning that a significant difference in general self-efficacy was found between groups at the baseline assessment (Morris et al., 2019).

4.7.1.2. Social outcomes

Social participation: Meta-analysis could not be conducted, as only one study reported the finding of social participation by the Stroke Impact Scale (Morris et al., 2019). According to this study, the visual arts-based intervention improved social participation immediately post-intervention. However, no improvement in social participation was found in the intervention group at three months follow-up (see Table 2).

4.7.1.3. Physical outcomes

Activities of daily living: Four studies reported the effects of visual arts-based interventions on activities of daily living (Huang et al., 2016, 2020; Kongkasuwan et al., 2016; Wu et al., 2011). This outcome was measured using the Barthel Index or the Modified Barthel Index. The pooled results from meta-analysis showed a statistically significant improvement on activities of daily living immediately post-intervention (SMD: 0.96, 95% CI: 0.24 to 1.69; four studies, 281 participants; low quality of evidence) (see Fig. 3(c) and Table 4). A significant heterogeneity was found (I²: 87%) (see Fig. 3(c)). Sensitivity analysis indicated that the pooled results were robust.

Upper limb function: Two studies examined upper limb function using the Fugl-Meyer Assessment for the upper extremity (Huang et al., 2016, 2020). The pooled results indicated that visual arts-based interventions had a statistically significant effect on upper limb function immediately post-intervention (SMD: 0.83, 95% CI: 0.42 to 1.24; I²: 6%; two studies, 108 participants; low quality of evidence) (see Fig. 3(d) and Table 4).

Hand function: Two studies measured hand function using the Simple Test for Evaluating Hand Function or the hand function domain of the Stroke Impact Scale (Huang et al., 2016; Morris et al., 2019). The pooled results of meta-analysis indicated no effect on hand function in the intervention group immediately post-intervention (SMD: 0.47, 95% CI: −0.43 to 1.37; two studies, 119 participants; very low quality of evidence) (see Fig. 3(e) and Table 4). Significant heterogeneity was found in the pooled results of hand function (I²: 87%) (see Fig. 3(e)). Sensitivity analysis indicated that the pooled result was robust.

As only one study reported hand function at three months post-intervention, meta-analysis could not be conducted. This single study found no improvement in hand function in the intervention group at three months follow-up (Morris et al., 2019) (see Table 2).

4.7.1.4. General health outcomes

Health-related quality of life: Two studies measured health-related quality of life using the 36-item Short Form Survey or the pictorial Thai Quality of Life questionnaire (Ellis-Hill et al., 2019; Kongkasuwan et al., 2016). However, meta-analysis could not be conducted because one included study measured health-related quality of life immediately post-intervention and one assessed this outcome at one month post-intervention. Two included studies found that participants who received the visual arts-based intervention had a better health-related quality of life immediately post-intervention (Kongkasuwan et al., 2016) and one month follow-up (Ellis-Hill et al., 2019) compared to the baseline assessment (see Table 2).

5. Discussion

In this systematic review, we critically summarised seven randomised controlled trials involving 483 participants, with the objectives of assessing the effects of visual arts-based interventions on physical and psychosocial outcomes after stroke and determining the effective regimen of visual arts-based interventions for people with stroke. The findings of this review indicated that visual arts-based interventions contributed to statistically significant improvements in depressive symptoms, activities of daily living, and upper limb function in people with stroke immediately post-intervention. However, the effects of interventions on anxiety, self-efficacy, social participation, hand function, and health-related quality of life after stroke were inconclusive due to the limited number of included studies.

After receiving visual arts-based interventions, people with stroke had statistically significant improvements in activities of daily living and upper-limb function, according to the pooled results. These results aligned with the previous critical review that visual arts-based interventions contributed to post-stroke physical outcomes (Pang et al., 2021). However, there was only one case study with one participant (Kim et al., 2008) reporting motor activity and functional independence in the previous review (Pang et al., 2021). Insufficient evidence in the previous review could not provide specific evidence about the effects of interventions on physical outcomes in people with stroke, which resulted in the effect of interventions on hand function in this review being inconsistent with previous findings (Pang et al., 2021). Additionally, it should be noted that the quality of evidence regarding physical outcomes assessed by the Grading of Recommendations Assessment, Development, and Evaluation method was very low to low due to the limitations of risk of bias, inconsistency, and imprecision. Thus, we have little confidence in the effect estimate.

We also found that visual arts-based interventions had a statistically significant effect on depressive symptoms but had no effect on anxiety. A previous review reported that visual arts-based interventions decreased depressive symptoms and anxiety in people with stroke (Pang et al., 2021), which differed from the findings about anxiety in this review. It should be noted that the findings about anxiety in the previous review were not synthesised by meta-analysis due to insufficient data. While the included individual study results in this review showed a potentially beneficial effect of interventions on anxiety, this is insufficient data from which to draw certain conclusions. Existing studies regarding visual arts-based interventions did not provide sufficient quantitative evidence to examine the effect of interventions on anxiety, which might cause inconclusive findings about anxiety. Therefore, more trials in the future are needed to investigate the effects of visual arts-based interventions on post-stroke anxiety. Additionally, the quality of evidence regarding depressive symptoms was moderate, while the quality of evidence regarding anxiety was low; therefore, we have relatively less confidence on the findings of psychological outcomes.

The included individual study results also showed that the interventions had potential benefits to self-efficacy, social participation, and health-related quality of life; however, those studies provided insufficient data. Thus, we could not draw certain conclusions on those outcomes. All interventions were delivered in person, and six of seven interventions involved group-based delivery. The face-to-face with group-based delivery format was identified as a feature of visual arts-based interventions (McCaffrey et al., 2021; Shoesmith et al., 2021) and commonly used in the included studies. This delivery format enhanced interactions amongst participants, facilitators, or peers (Blancafort Alias et al., 2021; French et al., 2014). Three previous qualitative studies suggested that art participation with six to eight participants per group could provide opportunities to facilitate social interactions in people with stroke during the processes of visual arts-based activities and group discussions (Gonen and Soroker, 2000; Le Navenec and Bridges, 2005; Morris et al., 2016), which resulted in enhancing self-efficacy, social participation, and health-related quality of life (Beesley et al., 2011; Morris et al., 2016). However, this review did not identify certain effects on those outcomes due to insufficient data for meta-analysis. Given the suggestions from previous studies and the findings of this review, subsequent testing in randomised controlled trials regarding face-to-face with group-based interventions is needed to further examine the effectiveness of interventions on self-efficacy, social participation, and health-related quality of life.

In this review, all interventions were delivered by qualified, trained, or experienced facilitators, such as health professionals, artists, or therapists. These findings implied that qualified, trained, or experienced facilitators are important to deliver visual arts-based interventions. Additionally, visual arts-based intervention is regarded as complex (Reynolds, 2012). The guideline for the development of complex interventions suggests that they need to be implemented by trained facilitators (O'Cathain et al., 2019; Skivington et al., 2021). Given the findings from this review and the guideline for developing complex interventions, we suggest that, in the future, qualified or trained facilitators should deliver visual arts-based interventions for people with stroke.

The optimal duration and frequency of intervention were not identified, as various ranges were found in included studies. Nevertheless, we found that all statistically significant improvements in depressive symptoms, activities of daily living, and upper limb function occurred in the interventions that involved drawing or handicraft-making (Huang et al., 2016, 2020; Kongkasuwan et al., 2016; Liu and Lu, 2018; Wu et al., 2011). Previous researchers have suggested that handicraft-making allows people with stroke to use their affected upper limbs to perform repetitive and specific movements and control the handicraft equipment, which results in improving their upper limb functions and self-care abilities (Beesley et al., 2011; Morris et al., 2016). Previous researchers, on the other hand, also found that people with stroke can express their thoughts, feelings, and moods through drawing (Kim et al., 2013; Michaels, 2010; Sit et al., 2017). Drawing allows people with stroke to express themselves in a non-verbal way; thus, their inner voices can be spoken through their art pieces, which contribute to self-expression (Carmi and Mashiah, 1996; Sit et al., 2017). In addition, previous researchers further found that the process of handicraft-making and drawing could motivate people with stroke to explore alternative ways for their stroke recovery, such as achieving their goals, increasing their confidence, or adapting to and managing their post-stroke conditions by problem-solving (Beesley et al., 2011; Morris et al., 2016; Sit et al., 2017). Therefore, we suggest handicraft-making and drawing for future visual arts-based interventions to explore the benefits of those potentially valuable components for stroke recovery.

With a better understanding of the effects on the specific health-related outcomes in people with stroke, visual arts-based interventions grounded by an established theoretical framework are needed. In this review, all interventions employed at least one motivation strategy as follows: provision of information, goal setting, sharing of experience, providing appropriate feedback, or providing support, to assist participants’ engagement in visual arts-based activity for stroke recovery. The findings of this review were consistent with Green et al. (2021) and Lou et al. (2020) that those strategies are commonly used to support stroke recovery. Moreover, those five motivation strategies identified by this review are related to Bandura's Self-Efficacy Theory (Bandura, 1997). According to Self-Efficacy Theory, self-efficacy refers to the individual's beliefs in their capability to make behaviour changes. Self-efficacy can be enhanced by four sources of information: mastery experience, vicarious experience, verbal persuasion, and physiological and affective states (Bandura, 1977, 1986, 1997). Although seven included studies did not explicitly describe the four sources or the theory name regarding self-efficacy, the five motivation strategies used in included studies are related to Self-Efficacy Theory. For example, providing information about stroke and self-care skills and setting appropriate goals for visual arts-based activity contribute to a successful mastery experience (Elliot et al., 1997; Michie et al., 2014); sharing experience is a strategy to obtain successful experience from others, which is related to vicarious experience (Roberts, 2010). In addition, providing positive feedback and support resembles verbal persuasion (Peifer et al., 2020; Pfitzner-Eden, 2016); providing information about self-care skills is beneficial to controlling physiological and affective states (Butler et al., 2019). Given the potential value of the five motivation strategies on enhancing self-efficacy by four sources of information, Bandura's Self-Efficacy Theory is suggested as a potential theoretical framework to guide the development of visual arts-based intervention for people with stroke in the future.

5.1. Strengths and limitations of this review

This review was systematically conducted based on standardised framework, guideline, and assessment tools. However, there were also limitations in this review. First, including only studies written in English or Chinese might influence the number of included studies with appropriate data for meta-analysis. Second, the limited number of included studies could not provide sufficient data to conduct subgroup analysis. Third, the various characteristics of studies, participants, and interventions also caused difficulty in conducting subgroup analyses. Fourth, some methodological weaknesses were detected in included studies, which possibly influenced the quality of included studies.

5.2. Implications for practice and research

Several implications are suggested for future studies. First, this review indicated that visual arts-based interventions had statistically significant effects on depressive symptoms, activities of daily living, and upper limb function in people with stroke. Given the positive findings, visual arts-based interventions are recommended as adjuvant interventions for stroke recovery. Second, given that most of the interventions in included studies were delivered in groups and in person, we suggest using a face-to-face with group-based format in future visual arts-based interventions for people with stroke. Third, we suggest a theory-based intervention to address physical and psychosocial outcomes in people with stroke. However, possible theories have yet to be explored. Therefore, further studies should be conducted to understand the mechanisms of action of visual arts-based interventions for people with stroke, particularly concerning the theoretical framework about enhancing self-efficacy. Fourth, the effects of interventions on hand function, anxiety, self-efficacy, social participation, and health-related quality of life in people with stroke were inconclusive because of insufficient data; more examinations of these outcomes are needed in future studies. Fifth, most visual arts-based interventions were conducted in inpatient settings. Considering the long-term post-stroke needs and the post-stroke recovery process in the community, more interventions should be conducted in community settings. Sixth, this review could not identify the optimal delivery modalities, such as component, duration, and frequency of visual arts-based interventions for people with stroke. Therefore, future studies are needed to investigate the effective regimen of intervention. Finally, the included studies were assessed to have some methodological weaknesses due to absence of information about study protocol, randomisation, allocation concealment, blinding, attendance, or data analysis strategies; more randomised controlled trials with rigorous study designs are required to strengthen the quality of relevant evidence.

Conclusion

Visual arts-based interventions positively affected depressive symptoms, activities of daily living, and upper limb function in people with stroke. However, the effects of interventions on hand function, anxiety, social participation, and health-related quality of life were inconclusive. Difficulty in identifying the optimal implementation strategies for visual arts-based interventions was exacerbated due to a lack of relevant evidence. As the data on effectiveness is limited, we suggest further randomised controlled trials with rigorous study designs to identify the effects of visual arts-based interventions for people with stroke.

Funding sources

No external funding.

Declaration of Competing Interest

None.