Effectiveness of arts interventions to reduce mental-health-related stigma among youth: a systematic review and meta-analysis

Background Educational interventions engage youth using visual, literary and performing arts to combat stigma associated with mental health problems. However, it remains unknown whether arts interventions are effective in reducing mental-health-related stigma among youth and if so, then which specific art forms, duration and stigma-related components in content are successful. Methods We searched 13 databases, including PubMed, Medline, Global Health, EMBASE, ADOLEC, Social Policy and Practice, Database of Promoting Health Effectiveness Reviews (DoPHER), Trials Register of Promoting Health Interventions (TRoPHI), EPPI-Centre database of health promotion research (Bibliomap), Web of Science, PsycINFO, Cochrane and Scopus for studies involving arts interventions aimed at reducing any or all components of mental-health-related stigma among youth (10–24-year-olds). Risk of bias was assessed using the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool for Quantitative Studies. Data were extracted into tables and analysed using RevMan 5.3.5. Results Fifty-seven studies met our inclusion criteria (n = 41,621). Interventions using multiple art forms are effective in improving behaviour towards people with mental health problems to a small effect (effect size = 0.28, 95%CI 0.08–0.48; p = 0.007) No studies reported negative outcomes or unintended harms. Among studies using specific art forms, we observed high heterogeneity among intervention studies using theatre, multiple art forms, film and role play. Data in this review are inconclusive about the use of single versus multiple sessions and whether including all stigma components of knowledge, attitude and behaviour as intervention content are more effective relative to studies focused on these stigma components, individually. Common challenges faced by school-based arts interventions included lack of buy-in from school administrators and low engagement. No studies were reported from low- and middle-income countries. Conclusion Arts interventions are effective in reducing mental-health-related stigma to a small effect. Interventions that employ multiple art forms together compared to studies employing film, theatre or role play are likely more effective in reducing mental-health-related stigma. Supplementary Information The online version contains supplementary material available at 10.1186/s12888-021-03350-8.

Conclusion: Arts interventions are effective in reducing mental-health-related stigma to a small effect. Interventions that employ multiple art forms together compared to studies employing film, theatre or role play are likely more effective in reducing mental-health-related stigma.
Keywords: Mental health, Youth, Stigma, Art, Systematic review, Meta-analysis, Performing, Film, Role-play, Theatre Background Stigma or a negative disposition towards mental ill-health and people with mental health problems is a widely recognized barrier in help-seeking for mental health problems [1]. Public stigma, consists of 'problems of knowledge (ignorance), problems of attitude (prejudice), and problems of behaviour (discrimination).' [2] Such stigma especially inhibits help-seeking by youth due to their inability to recognize mental health problems, difficulty in talking about their problems for fear of peer pressure and a negative perception of people with mental health problems as dependent, which clashes with their desire to be self-reliant [3]. Thus, although an estimated 10-20% of youth aged 10-24 years suffer from mental health problems, [4] 63-86% of all mental health problems that require a diagnosis generally go undetected [5]. Therefore, interventions targeting non-clinical youth groups to reduce mental-healthrelated stigma may promote youth help-seeking behaviour and ultimately address unattended mental health needs.
Most anti-stigma interventions and/or campaigns have been conceptualized using knowledge-attitude-behavior paradigm [6]. Knowledge is defined as information an individual perceives about mental health as a function of memory and stereotyping (related to, for e.g., treatment efficacy, symptom recognition, help-seeking, and employment), attitude is defined as perceptions or views towards people with mental disorders or about mental disorders (related to negative attitudes, for e.g., desiring social control and social distance), and behavior as intended or actual discriminatory actions towards people with mental health problems (related to, for e.g. social exclusion, which may contribute to status loss or human rights violations of someone living with a mental health problem) [2,[7][8][9]. Further, effective strategies in antistigma interventions include education, social contact (interaction with a person who suffers from a mental health problem) and protest [10]. In addition, effective interventions are often locally tailored, perceived as credible and of a longer duration [11]. In school-based settings, experiential learning (learning through reflection on doing), empathy building, interactive and prolonged exposure to anti-stigma content is likely effective [12,13]. Overall, systematic reviews of anti-stigma intervention studies report that in the long term and among youth, educational interventions are likely more effective than social contact interventions in reducing stigma with moderate effect [14][15][16]. Among the approaches used, educational interventions have employed a variety of visual, literary and performing arts to improve relatability, interactivity and engagement.
Art is broadly defined as any means for expression of individual and social values, through concrete and artistic activities and processes [17]. Further per Dewey's conceptualisation of art, arts interventions may communicate moral purpose or education [18] or explain experiences of one's daily emotional and rational world [19]. The evidence for arts-based educational interventions is generally limited, despite its documented emotional and visceral effects [20]. Despite multiple, relevant systematic reviews, uncertainties remain regarding the overall effectiveness of arts-based interventions in reducing mental-health-related stigma and relative effectiveness of interventions employing different art forms, varying durations and conceptualizations of stigma. A review of 22 studies evaluating the impact of mass media interventions including film, photographs, radio and comics attributed reduced prejudice (attitude) for mental health problems to creative and artistic content [21]. The majority (86%) of studies in this review focused on student populations. Other reviews of studies among 11-18 year olds using creative activities such as music, dance, singing, drama and visual arts [22] and performing arts, [23] indicate that arts-based interventions improve knowledge, another component of mental-health-related stigma. As some reviews are focused on educational versus social contact-based interventions, [14][15][16] they do not focus on the distinguishing role of arts-based elements in achieving impact nor suggest the relative impact from using role play, theatre, film compared to other art forms.
Previous studies show that arts-based interventions have the potential to reduce mental-health-related stigma as they have improved individual components of such stigma, i.e. attitude and knowledge. However, little is known about the effectiveness of arts-based interventions in reducing overall mental-health-related stigma among youth, and whether interventions using specific art forms, duration and content on all stigma components of knowledge, attitude and behavior are more effective in reducing such stigma compared to individual components. The objectives of this study are to: (a) assess the effectiveness of arts-based interventions to reduce stigma associated with mental health among youth; (b) assess effectiveness of arts-based interventions by their duration; (c) assess whether a comprehensive approach to stigma is more effective than a focus on individual stigma components; and (d) identify barriers and facilitators in implementation of arts-based interventions and the role of implementation in building participant engagement and ultimately influencing how effective such interventions are in reducing stigma.

Eligibility criteria
Studies will be included in the review if they contain: -Interventions using any form of art or creative expression or storytelling as a key method were included. Such art forms include (1) using words in literary art (in stories, poetry, creative writing, essays and other forms), and through creation of physical objects and experiences, through (2) visual art (drawing, painting, sculpture, crafts, pottery, installation), and (3) performing art (theatre or dramatic improvisation or role-play, dance, puppetry, music, stand-up comedy, folk dance-drama).
In this review, participants in included arts interventions should either be exposed to art (e.g., as an active observer/audience interpreting and responding to scenarios in a theatre production) or create their own art (e.g., as generating thought, meaning, aesthetic or object/s). -Interventions delivered to youth aged 10-24 years.
-Outcomes related to at least one component of mental-health-related public stigma (three components outlined by Thornicroft  Mixed methods studies were defined as studies which involved "sequential or simultaneous use of both qualitative and quantitative data collection and/ or data analysis techniques." [24] Studies were excluded from the review if they met one of the following criteria: -Target clinical, high-risk or at-risk populations (youth with mental disorders, including outpatients, in schools for special needs, in prisons, foster homes/ shelters and conflict zones or exposed to violence) or caregivers as these groups have unique personal experiences that might distinguish them from the general population. -Use mass media (newspapers, television and radio programmes, advertising, popular culture, cinema and songs, social media, blogs and other Internet or mobile phone). -Combine art with other strategies, where the effect of art is not separately reported. -Focus on intimate partner violence, sexual violence and gender-based violence, cyberbullying and domestic abuse.

Search strategy
The broad categories of terms used included art; mental health disorders/conditions; youth; and stigma (see Supplementary Table 1  . If two or more articles on the same intervention and target population were found, the most relevant article was retained for analysis. The Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA) guidelines were used to report updated study findings (see Supplementary Table 2 for checklist) [25].

Data extraction
All titles and abstracts were assessed by a single reviewer (SMG). A second reviewer (SU) assessed 10% of all titles and abstracts to confirm accuracy of inclusion. The updated search was conducted by the first author and 895 additional articles were retrieved. Using the Quality Assessment Tool for Quantitative Studies developed by the Effective Public Health Practice Project (EPHPP), [26] a framework for data extraction was developed. The framework captured additional data on intervention characteristics and study design, related to review objectives. Full-text articles were independently assessed as per the EPHPP framework and data were entered in to tables by a single reviewer (SMG). The second reviewer assessed all full-text articles and cross-checked data in the framework. Discussion between reviewers compared quality ratings and key findings. Where consensus was not reached, a third reviewer (MP) was consulted.

Summary measures
The main study outcome was mental-health-related public stigma, which is composed of three components: knowledge, attitude and behaviour. Measures of these components include means and standard deviations, difference between means and level of significance (p-value).

Synthesis and reporting of results
Demographic information of participants and qualitative themes were compiled in a narrative form. Firstly, means, standard deviations and sample sizes were pooled for each stigma-component/outcome for all studies, followed by art form or intervention type, to assess whether the type of intervention was responsible for a difference in outcomes. Change in stigma was plotted by pooling study-wise difference of means and standard deviations per component of stigma among studies with a study design rating of 1 or 2 per the EPHPP component ratings. As an illustration, for the behavior component of stigma we pooled mean differences from the Social Distance scale (a common proxy measure for behavioral intent) [27] and the Reported and Intended Behavior Scale [28]. If a study reported on multiple items within each stigma component, then the item with the lowest (stigmatising) mean score change was included. We calculated the mean score change from data available in the study text and tables, wherever available. Change in outcomes related to stigma (knowledge, attitude and actual or intended behaviour) were pooled by intervention type or art form, i.e. multiple art forms, film, theatre and role play. Next, a post-hoc sub-group analysis of data was conducted by intervention duration (single versus multiple sessions). Finally, studies which took a comprehensive approach to stigma (measured knowledge, attitude and behaviour components, together and likely also included content addressing each of these components) were pooled for their the impact on individual stigma components. These analyses were presented alongside pooled studies measuring individual stigma components such as knowledge or behaviour alone. This was done in order to assess whether a comprehensive approach leads to better outcomes within each stigma component. Meta-analysis, where appropriate, was conducted using Review Manager software (Version 5.3.5) [29]. Heterogeneity of studies was assessed through I 2 values> 0, and random-effects models were generated to calculate the effect size on stigma. A random-effects model with standardized mean differences was preferred as study populations and locations, recruitment processes, points of time for implementation and assessment measures varied. Narrative synthesis was used to collate findings regarding barriers and facilitators in reducing stigma.

Results
The search produced 19,892 articles, of which 187 articles were identified for full-text review (Fig. 1). Of these, 132 were either contextual articles without an arts intervention, epidemiological studies assessing impact from exposure to media, reviews on related aspects of stigma or youth or studies that did not meet the inclusion criteria. Finally, 57 studies (53 full-text articles and four conference abstracts) were included in this systematic review.

Qualitative studies
Of six qualitative studies, one did not clearly define the method of qualitative research, [73] two used in-person and/or telephonic semi-structured interviews, [75,77] a study used focus group discussions (FGD), [76] another used ethnographic procedures, [74] a study used students' reflective essays, short films, and course evaluations, [78] and two used field notes and observation [74,76].

Mixed methods studies
Mixed methods were used to supplement studies, which were overtly quantitative: a RCT, [86] a controlled   [80,[83][84][85] and two qualitative studies, included surveys [81,82]. These studies reported the use of observations [79,82,86] open-ended questionnaires, [79,[82][83][84]86] semistructured interviews, [80] group interviews, reflective groups and FGDs, [81,82] drawing and explanatory writing, [84] and field notes [86]. Among eight RCTs, varied combinations of stigma components improved in a majority of studies, except a study using role play which reported no significant change in any aspect of stigma [72]. Only one RCT measured and improved all components of mental health stigma (KAB) using film [34]. Other RCTs improved: attitude and behaviour using film [39]; knowledge and attitude using multiple art forms [53] and film [31]; attitude using film [43,45]; and behavior using film [41,52,56]. Intervention content included facts on incidence, causes, symptoms and warning signs of mental illness, broadly and about specific disorders, including bipolar disorder, schizophrenia, depression, substance abuse and suicidal behaviours. Social contact was used to describe negative experiences of stigmatisation using filmed or dramatised interventions. Even single session interventions included multiple art forms [30,48,63].

Qualitative studies
From the six qualitative studies, three studies employed theatre interventions, [74,75,77] two used multiple arts forms, [73,78] such as music, radio, documentary and visual arts or students' reflective essays and short films [78] and another used dance [76]. In four out of six qualitative studies, participatory or collaborative approaches involved students as performers of art or as collaborators in co-creating art with persons living with mental health problems [73,75,76,78]. Only one intervention study comprised a single session [74] and other the other five studies involved multiple sessions, with intervention duration ranging from 2 weeks [73] to 8 months [75].
The study reporting a positive improvement in all components of mental health stigma (KAB) used a professional play, followed by role-play [74]. Among other studies, two using theatre, [75,77] one using music and visual arts [73] and one using dance [76] suggest positive gains in knowledge and attitude (KA) related to mental health problems and drug use, and reducing awkwardness and increasing empathy. The study involving music and creation of short films on dementia reported that 27% of participants continued to volunteer in dementia care settings after the intervention was completed [78].

Mixed methods studies
Six out of eight studies involved professionally created art, [79,81,[83][84][85][86] while the other studies helped youth create their own rap songs [82] and participate in a choir [80]. Three interventions used multiple art forms (film, theatre, rap songs, role play and educational materials) [81,82,86] and one each used children's books, [79] song lyrics, [81] film [83] and interactive theatre [84]. Collaborative art or co-created art was evaluated using a post-only survey and interviews [82]. Only one study used an intervention that lasted a single day [84] and other interventions ranged from between 4 weeks [86] to 10 weeks [80,83]. In another study follow-up material was mailed to participants for 12 months after the intervention [85]. Only one mixed-methods study stated that youth participation was voluntary [82].
Three studies report changes in all components of mental health stigma (KAB), one using theatre, [84] another using film [85] and another other using multiple art forms [86]. These studies include knowledge about drugs, mental health awareness and self-recognition; attitudinal change that anyone could be affected by mental health problems and behaviours such as reduced negative words and desire to help those in need (intended behaviour). A study each improved acceptance and bystander responses (AB), [79] and knowledge about substances (K) [81]. and two studies improved knowledge and attitudes related to substances and dementia (KA) [82,80].
Overall, studies using film had good quality, studies using theatre had moderate quality and studies using multiple art forms and role play had weak study designs. Confounders were better addressed by studies using theatre and multiple art forms, compared to role play and film. Valid and reliable data collection instruments were used by studies using theatre and film, followed by multiple art forms and lastly, role play.

Qualitative studies
From six qualitative studies, a study lacked quotations to assess validity of conclusions, [73] and one study presented quotations as a response to questionnaires [74] [87]. One study reported full participation, [75] rate of participation varied from 10 to 88% in three studies [74,76] and was not specified in the remaining three studies [73,77,78].
Out of six qualitative studies one reported positive changes in KAB, [74] three studies improved knowledge and attitudes (KA) associated within mental health [75,77,78] and the remaining two studies improved knowledge (K) by way of recall and the level of awareness about mental health problems [73,76]. Three studies focused on reducing stigma associated with drugrelated issues, [73,74,76], one study on dementia, [78] one study on depression, anxiety, panic and stress, [75] and one on trauma [77]. Studies focused on the process of using art as a stimulus for discussion and narratives focused on achieving attitudinal change, positive selfesteem, purpose and satisfaction from participation. Some of these studies highlight the emotional impact of art as a tool to relate stories and personal experiences, [77,78] changes in how youth use labels and describe interactions with people living with mental health problems, [75] and one reportedly led to substantial increases in requests for counselling [74].
From five mixed methods studies, three studies reported positive results on KAB, one study reported positive impact on attitude and behaviour (AB), [79] another two studies reported positive changes in knowledge and attitude (KA), [82,86] one study improved specific attitudes only [83] and lastly, one study improved knowledge [81].

Meta-analysis Effectiveness of art in reducing components of stigma
There was no significant difference in whether arts interventions improved behaviour towards people with mental health problems compared to a control group (effect size = 0.12, 95%CI -0.01-0.25; p = 0.07) (Fig. 4), and moderate heterogeneity was reported across studies (I 2 = 47%). High heterogeneity of studies on knowledge and attitude outcomes made meta-analysis inappropriate (88-94%).

Effectiveness of different art forms
The largest positive effect on knowledge may be attributed to interventions using multiple art forms (effect size = 1.47, 95%CI -0.19-3.13; p = 0.08), followed by film (effect size = 0.14, 95%CI -0.21-0.50; p = 0.42) (Fig. 5). However, the I 2 value for pooled studies in this metaanalysis, reporting knowledge-related outcomes, was between 84 and 98%. No data were available for theatre, role play or other studies with respect to knowledge. Similarly, the impact of interventions using theatre, film, multiple art forms and role play on changing attitudes was not significant. Studies pooled by each of these art forms had heterogeneity, ranging from I 2 = 80-94%. Interventions using multiple art forms were the only ones that significantly reduced stigmatising, practised or intended behaviours (effect size = 0.28, 95%CI 0.08-0.48; p = 0.007) (Fig. 5). Theatre-based interventions pooled by behavioural outcomes showed low heterogeneity (I 2 = 20%) and film-based studies pooled by behavioural outcomes showed moderately high heterogeneity (I 2 = 67%). No data were available from studies using role play and other art forms, due to lack of precise measurement or poor quality of reporting.

Effectiveness of interventions by duration
Studies pooled by duration, i.e., whether single session interventions or multi-session interventions, displayed moderate to high heterogeneity (I 2 = 51-99%) and did not show any significant effect on knowledge, attitude or behavior (Fig. 6).

Comprehensive approach versus focus on individual stigma domains
Studies that took a comprehensive view of stigma showed no significant improvement in behavior (effect size = 0.12, 95%CI -0.03-0.27; p = 0.11). The I 2 = 0% indicates that pooled studies had low heterogeneity, and therefore that the meta-analytic approach was appropriate (see Supplementary material 4: figure). These studies focused on all components of knowledge, attitude and behaviour (KAB) in measurement and possibly also in intervention content. For knowledge-and attituderelated outcomes in studies reporting KAB, there appeared to be a positive effect (effect size = 0.09-0.25), but there was high heterogeneity among pooled studies (I 2 = 84%) and these results were not significant. There were no studies that focused on knowledge, attitude and behavior components of stigma alone and that met study design quality benchmarks for meta-analyses.

Barriers and facilitators in implementation and reducing stigma
Overall, multiple mechanisms and contingencies were reported to influence implementation and participant engagement, especially attendance and quality of delivery.
Fluctuating intervention attendance, [82] awkwardness and scepticism, [57,59,86] language-related issues, [74,82] group and gatekeeper dynamics [57,66,75] and logistical issues [70] influenced implementation in several studies. In a large number of studies females were overrepresented [32,38,42,45,50,51,67,75,76]. Unintended consequences were reported in a study using multiple art forms (professional theatre, quiz and games) with school children, where mental health problems were normalised to the extent that participants felt that these problems did not have much of an adverse impact [37]. Active ingredients that facilitate successful delivery of arts-based interventions include institutional endorsement for the initiative at educational institutions [39,75] and scheduling sessions during class times [31,38,41,49]. Clear content, [55,68] a diversity of views from presenters [34,66] and involving people with mental health problems for embedded social contact were perceived to reduce stigma [31,34,42,50]. Visual stimuli and expressive artsbased techniques were useful tools to facilitate participation [64,86]. Further, high quality, emotionally powerful art performed may help achieve a stronger, anti-stigma stance among participants. Thus, several studies highlight the value of involving professional artists [37,60,73,82,84]. In one study that involved youth in performing scripted theatre, [54] the authors observed that even deeper participation was needed to reinforce key ideas. A study which enabled youth to act in scripted plays highlighted the importance of public reinforcement of messages through performance, however to align content to participant experiences, it recommended that participants write their own scripts [54].
Youth arts projects meet social needs of young people to engage in a popular programme [76]. They involve people external to educational institutions, which studies felt youth appreciate [65]. Additionally, youth projects accord an equal status among participants, [39] which according to Fernandez et al. is ideal for 'cooperational education,' where students learn and evaluate key programme messages collaboratively. Such interventions simultaneously use skills-building and educational appeals, rather than purely emotional ones, an approach which has been suggested as longer lasting. The Studio 3 Arts project among 13-21 year olds in the United Kingdom created participatory music, radio, documentary and visual arts for drugs-related awareness [73]. However, findings were inadequately reported from the perspective of effective intervention components. The project was reported in a brief, non-technical, magazine style which described the process and provided limited participant quotations or summaries of their experience as support. A pilot mixed methods study of VoxBox, co-creating rap music with high school students in Australia showed non-significant positive changes in knowledge, attitude and intended behaviour related to alcohol users [82]. Twardzicki et al. conducted a study in the UK in 2008 which generated theatre productions through discussion between people with mental health problems and college students [65]. Rowe et al. conducted a similar study in 2013 with students belonging to a theatre major, who co-created art on the theme of mental health with users of mental health services [75]. Although this study had a small sample size, its authors suggest that 'shared, theatre-making may create an environment that challenges stereotypes and reduces prejudice.' [75] Studies that used participatory, cocreation of art in this review, predominantly used qualitative and mixed methods for evaluation. These studies also demonstrate the impact of youth participatory arts projects focused on mental-health-related public stigma as a theme on critical thinking, problem-solving and building team spirit. Arts interventions are generally effective when they use multiple art forms, but with a small effect. This study also demonstrates that we do not affirmatively know whether interventions with multiple sessions had a greater effect on stigma, relative to single day interventions. Further, it remains inconclusive whether a comprehensive approach to stigma (including all stigma components of knowledge, attitude and behaviour in an intervention study), translates to significant improvements in knowledge, attitude and/or behaviour relative to studies focused on changing each of these individual stigma components. Common challenges faced by interventions related to buy-in from school or college stakeholders and youth engagement. No studies were reported from low-and lower-middle-income countries, and this highlights the need to develop, and report results from arts-based interventions in those contexts. No studies reported negative outcomes or unintended harms.
This review does not provide evidence to support conclusions from reviews by Schachter [13] and Mellor [12] on school-based interventions to reduce mental health stigma, that use of multiple art forms may coincide with multiple exposures and a more intensive engagement. The use of multiple art forms may have attracted and engaged participants with varied interests to reinforce concepts related to the theme of mental-health-related stigma. Overall, the most commonly reported underlying theory is Bandura's social learning theory, where youth are likely to emulate [89] less stigmatizing behaviour if they observe stereotypes or are able to concretize their experiences through art. Film-based studies were too heterogenous, likely due to varying educational content including filmed theatre or social contact or documentary; varying duration of films and varying complementary activities such as discussion or role-play.
Multiple art forms are potentially more impactful than other art forms in lowering stigma as a combination of art forms likely aims for a more intense experiences compared to use of a single art form [90]. These programs have the potential for greater interactivity and longer duration as well as the possibility of attracting youth who may be interested in using or engaging with at least one art form among several deployed. Findings related to the effect of theatre and role play in this review, are supported by Joronen's review on school-based drama, which showed short-term effects on health-related knowledge and behaviour [91]. Our findings related to implementation barriers such as inconsistent participant attendance may be overcome by recommendations by authors of included studies to use participatory student arts-based projects that involve direct youth engagement. Given that including voluntary role play as 20% of an intervention on mental illness led to changes in youth knowledge and attitudes in a recent study, [92] one may expect a positive response and increased acceptability in studies where participantcreated art is a complementary component. However, most intervention studies in this review involved mandatory attendance of professionally created art.
In this review, collaborative art or co-created art involving students was evaluated using mixed methods (post-test only for one group) or qualitative research. Other recent studies place the responsibility of creating art directly in the hands of young people through a variety of art forms: photo-voice; scripting, filming, and editing a public service announcement targeted to peers, and words and messages in response to a participatory public art project on mental health [93][94][95][96]. Study outcomes relate to enabling participants to describe their perceptions in relation to mental health, share personal experiences of stigmatized topics and the ability to participate in a project that validates that mental illness is real and acknowledges the need for shame-free mental health awareness [95]. As more rigorous evaluations of these participatory interventions are conducted, and an expanded range of outcomes are studied, their effectiveness in changing participants' knowledge, attitudes and behavior associated with people with mental health problems will become clearer.
Most arts-based interventions target health professionals in-training. College students from other backgrounds should justifiably have access to age-appropriate interventions on mental health stigma for prevention, early detection and acceptance of people with mental health problems. Further, three studies in this review observed that their interventions were likely more effective for older adolescents compare to children, [62,69,79] potentially because older adolescents have the confidence to communicate and skills to analyse complex, social and individual emotional responses.

Study strengths
This review is unique because it collates evidence on pragmatic dilemmas of mental health promotion faced by policy-makers, researchers, practitioners and communicators/educators. It is also unique in its comprehensiveness, as it explores the effectiveness of arts-based interventions across a range of mental health stigmarelated outcomes, study designs, art forms and intervention durations. This review takes a broad view of art and mental-health-related stigma. Other systematic reviews of interventions in mental health prevention include creative, artistic or entertainment techniques, and also do not acknowledge them as 'art.' [97][98][99][100][101][102][103][104][105][106][107][108][109][110][111][112][113] Many studies in this review use arts-based interventions, but do not explicitly recognise or state that they use art, expand on the purpose of art or define a clear pathway to change or theory of change through arts interventions. This review included all such studies in addition to including a wide range of arts-based techniques and mental health conditions. This review examines the theoretical understanding that comprehensively addressing all components of stigma is likely to impact intended behaviour towards people with mental health problems and towards helpseeking, rather than focusing on knowledge or attitude alone. The sub-group analysis by stigma components, extends the approach in a systematic review by Hanisch et al. in 2016, where they assessed and plotted successful impact on knowledge, attitude and behaviour outcomes from workplace interventions [114]. While duration of follow-up has been a subject of investigation for many systematic reviews, the impact of intervention duration (single vs multiple sessions) is explored by this review, although we observed inconclusive results.

Limitations
As per the EPHPP tool, [26] none of the quantitative studies received a strong overall rating. Studies scored poorly in terms of blinding of researcher awareness to intervention allocation, and selection bias due to convenience sampling and participant self-selection, which is common in researcher-led communication or public engagement interventions. Our search returned few randomized and/or controlled trials assessing the effectiveness of arts-based interventions on mental-healthrelated stigma overall, and its components of knowledge, attitude and behavior. Since interventions are continuously being designed and developed, this review sought to analyse all available evidence to inform stigmareduction initiatives amongst young people. Thus, we have included all studies (including quasi-experimental studies) of generally high quality in our meta-analyses, to identify a direction of impact, no impact or negative impact rather than focus on estimates of expected change in outcomes. Readers are encouraged to review confidence intervals and heterogeneity to gauge the level of certainty of expected outcomes when implementing a study using arts interventions.
Specific subgroup analyses were affected by high heterogeneity (I 2 values). In addition, several studies provided inadequate data and therefore, were not pooled. For the sub-group analysis by duration, varying timepoints for follow-up and lack of follow-up implied that studies could not be pooled and that only short-term effects at post-test (up to 1 month) could be feasibly calculated. Sub-group analysis by middle school, high school and university was not conducted due to fewer pooled studies. Other aspects that may have led to general heterogeneity include complementary components such as social contact [10,13,115,116] and differences in measuring stigma. Finally, the concept of art, relationship of participants with observing and creating different art forms and therefore the relative effectiveness of interventions based on arts, are likely influenced by the cultural context in which such art interventions are applied. It was not feasible for this study to factor in cultural differences in how the impact of arts interventions vary across cultures.
Studies measured different combinations of mentalhealth-related stigma components. The most common methodological issue cited by nearly all studies was the extent to which participant responses were affected by social desirability. Several studies used intended behaviour as a reasonable measure of actual behaviour, since measuring actual stigma-related behaviour is challenging [31,33,34,36,47,55,79]. A study argued that intended behaviour consisted of beliefs, self-efficacy to act on those beliefs and perceived benefit from behaviour [62]. To address these issues, this review focused on a multi-pronged concept of stigma, which is more comprehensive (included a combination of knowledge, attitude and behaviour components) and also focused on intended behavior. If studies found that both knowledge and attitude or any combination of knowledge, attitude and behavior (as mental-health-stigma-related components) changed after an arts intervention, we found that such studies did not correlate or discuss the relationship between knowledge, attitude and behavior components. We believe these findings could be important for readers interested in implementing arts interventions who may need to understand whether incremental changes in knowledge may or may not be correlated with changes in attitude and intended behavior.

Conclusion and implications
Overall, the studies reviewed demonstrate that arts interventions have limited effects on reducing young people's discriminatory behaviour towards people living with mental health problems. The review specifically indicates that using multiple art forms in arts-based interventions likely impact youth behaviour towards people living with mental health problems. While the quality of evidence included in this review is modest, the number of interventions using arts-based methodologies and a strong direction of travel for impact on stigma indicate the scope for application of its findings.
This review identifies several opportunities to develop arts-based education to reduce mental-health-related stigma. First, the dearth of such interventions in low-and lower-middle-income countries calls for the development of new, contextual initiatives. Second, since most interventions are implemented in partnership with the education sector, school and college authorities should be sensitized to the need for mental health promotion and should consider including arts-based educational interventions as part of their curriculum. Third, interventions may focus on young adults in college and not just those who are training for healthcare-related careers. Fourth, student-led arts projects may be useful to explore mentalhealth-related stigma in an interactive format, which may then serve to reinforce social norms that are anti-stigma. Future intervention development may involve empirical development of student arts projects or participatory artsbased interventions to reduce stigma. Finally, robust, realworld evaluations are needed in the future that go beyond short-term follow-up periods.
The review suggests that conceptualization of art and content also require closer attention. For instance, the purpose of using art may be expanded beyond information-sharing to a transformative process, providing a sense of agency to participants to take supportive decisions and actions when confronted by a person with a mental health problem or attending situation. Student art projects or co-creation of art to reduce mental-health-related stigma may embody such a concept, and finds support in two theories: 1) Fisher's communication narrative theory where art is a form of communication and storytelling and storytelling has the potential to re-shape the social world [117] and 2) Goldblatt's interpretation of Dewey's theory of art as experience, which highlights the transformative role of art in removing fear and prejudice, spurring critical analysis and empowering youth to achieve social justice [18]. With regard to content, future research on stigmarelated theories may define conceptual boundaries between stigma components of knowledge, attitude and behaviour, and interrelations and possible hierarchies among these components. Such research would strengthen and guide intervention content, for example, by informing intervention planners whether a gain in knowledge about causes of mental health problems or change in attitude that people with mental health problems are to be feared could be instrumental in reducing negative behaviours, such as the use of harsh words against people with mental health problems. Such research must be based on cultural understanding and interpretations of mental health problems.
Notably, this is the first global review of arts-based interventions to reduce stigma associated with mental health problems. Practical and action-oriented findings from the review may inform anti-stigma interventions and other mental health promotion interventions using youth engagement strategies. Continuous knowledge-sharing of active ingredients in effective interventions and implementation research is needed to ensure the successful adaptation of arts-based interventions across settings.
Abbreviations PRISMA: Preferred Reporting Items for Systematic reviews and Meta-analyses; SE: Standard Error; CI: Confidence Interval; K: Knowledge; A: Attitude; B: Behaviour