Mental health recovery and creative writing groups: A systematic review

Background: Many studies have described how creative writing may support recovery from mental disorders; however, this has rarely been reviewed systematically in research studies. Purpose: This study aimed to gain an overview of the current evidence of the effects of facilitated, group-based creative writing interventions on mental health-related clinical and personal recovery. Methods: A systematic review of the literature was conducted in February–March 2019 using the following databases: PubMed, Web of Science, Cochrane Library, PsycINFO, Scopus


Introduction
Globally, there is an increasing interest in arts-based interventions which have explicitly been linked to health and wellbeing (Fancourt & Finn, 2019).Writing interventions in health care include various methods such as expressive writing, poetry writing, narrative therapy, and creative writing (Bolton, 2011).Creative writing has long been used in art academies (Gardner, 1983), but only in the last 15-20 years has it emerged as a distinct practice for enhancing health and wellbeing, drawing a line between personal development and therapy (Bolton, 2011).Creative writing is considered a medium for expression of emotion and thoughts, with a focus on artistic form and facilitated by professional writers with or without the attendance of health care professionals (King et al., 2013).It can re-establish a feeling of competency often compromised amongst those experiencing mental illness, and the possible benefits of creative writing for mental health have long been of interest to researchers (Bolton, 2011).A large number of publications describe how writing, both as an activity and an intervention, may support recovery from mental illness (Atkinson, 2011;Bundesen et al., 2020;Holm-Hadulla, 2013;Iszáj & Demetrovics, 2011;Jess-Cooke, 2015;Kaufman, 2001).However, the benefits of creative writing interventions on recovery from mental illness have not been reviewed systematically.
A distinction is made between non-facilitated and facilitated writing interventions.A systematic review by the National Health Service (NHS) in the UK identified 59 quantitative studies of non-facilitated writing interventions, mostly using the expressive-writing paradigm (Pennebaker, 1997) for participants with long-term somatic and psychiatric disabilities, and found little or no effect on psychiatric measures (Nyssen et al., 2016).This is supported by two meta-analyses of non-facilitated expressive writing interventions for individuals with psychiatric or stress-related symptoms (Frisina et al., 2004;Reinhold et al., 2018).Facilitated writing interventions are usually group interventions facilitated by professional authors or mental health professionals with writing experience who provide participants with feedback and guidance on their work (Nyssen et al., 2016).The NHS report identified five quantitative studies of facilitated therapeutic writing interventions, all including a creative writing activity, but the interventions varied greatly both regarding the type of writing intervention and the level of facilitation.The report concluded that evidence was insufficient and encouraged further research.
A creative writing activity can be facilitated as a group activity with peers, where a focal point is on participation and process engagement.Participatory creative writing workshops have been conceptualized as hybrids between art workshops and group therapeutic interventions which may support recovery from mental disorders, encompassing both individual creative writing and sharing of material with a group (Bundesen et al., 2020).Studies of other participatory arts activities identify mental health outcomes related to personal recovery (Damsgaard & Jensen, 2021;Stickley et al., 2018) such as empowerment, social inclusion (Hacking et al., 2008), and psychological wellbeing (Saavedra et al., 2018).Participatory art as facilitated creative writing group interventions have also shown potential for reduction of depression in older adults, but more research is needed to understand if creative writing can promote recovery, under which circumstances, and what the active mechanisms are (Toma et al., 2014).
Based on the observations above, we conducted a systematic review investigating the current evidence of the effects of facilitated, group-based creative writing interventions on recovery from mental disorders.We adapted a recovery-oriented framework (Davidson et al., 2005;Slade, 2010) where recovery from mental disorder is not simply considered as symptom reduction, but rather as "a way of living a satisfying, hopeful, and contributing life even within the limitations caused by illness" (Anthony, 1993, p. 527).Recovery is the overarching term, covering both clinical (symptom reduction) and personal recovery as conceptualized in the CHIME framework (Leamy et al., 2011).The CHIME framework includes five core components of personal recovery: connectedness, hope and optimism, identity, meaning in life, and empowerment.As personal recovery is a subjective, personal experience (Davidson et al., 2005;Slade, 2010), we have included qualitative studies.

Methods
This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines (Moher et al., 2016), and was preregistered at Prospero (ID:CRD42019126031) in February 2019.The protocol was updated in June 2020 because the initial protocol did not specify the study population to be individuals with mental illness.

Inclusion criteria
The inclusion criteria were:

Quality assessment
The Standard Quality Assessment Criteria for Evaluating Primary Research Papers From a Variety of Fields (Kmet et al., 2004) was applied to assess the quality of included studies.Quantitative studies were rated on a three-point scale (2 = yes, 1 = partial, 0 = no) indicating the extent to which 14 key methodological aspects of quantitative studies, e.g.study design, sample size, control for confounding were addressed in the paper.Qualitative studies were rated on an identical scale, covering 10 core measures important for the quality of qualitative studies, e.g.appropriateness of study design, sampling strategy, data analysis.Scores for each applicable item were summarized and divided by the total possible score.Scores could range from 0.00-1.00,higher numbers indicating higher quality.Quality was assessed independently by two reviewers (SM & BK).Disagreements were solved by discussing studies with a third assessor (LH) until consensus was reached.

Data analysis
Two authors (SM & BK) independently extracted the main findings of the studies with the intention to conduct meta-analyses, but this was not possible due to the limited number of quantitative studies identified.
A narrative synthesis of qualitative and quantitative studies was articulated under two overarching themes: 1) personal recovery and 2) clinical recovery.This was done following the Guidance on the Conduct of Narrative Synthesis in Systematic Reviews (Popay et al., 2006).
One author (SM) coded and organized extracted findings according to the CHIME framework (Leamy et al., 2011).Findings could encompass several CHIME components.The authors also examined if any results were specific for interventions with certain characteristics, e.g., if facilitator background was associated with specific findings such as connectedness or hope and optimism.Due to the novelty of the field, findings from all studies were weighed equally regardless of quality ratings.However, the quality assessment is an important aspect of the interpretation of the overall strength of evidence.
Only one study addressed symptomatic recovery (King, 2018), hence results on clinical recovery were solely summarized descriptively.

Results
The search resulted in 7743 records that were uploaded to the online software program Covidence (Covidence -Accelerate Your Systematic Review, n.d.).Two reviewers (SM & BK) independently screened title abstracts, leaving 506 records to be assessed for eligibility at full-text level with the assistance of a third reviewer (LE).Finally, six studies were included (see figure 1): two quantitative and four qualitative studies (tables 1 and 2).

Participant characteristics
Studies included veterans with PTSD, depression, and anxiety (King, 2018), individuals with chronic mental health conditions (Williams et al., 2018), individuals with serious mental health disorders in rehabilitation programs (King et al., 2013), users of mental health services (Jensen & Blair, 1997), and in-patients with eating disorders (Padfield et al., 2017).

Intervention characteristics
All interventions were conducted in high-income countries such as England (2), the United States (1), Australia (2), and Scotland (1) in both inpatient and outpatient mental health care (Cooper, 2013;King, 2018;Padfield et al., 2017), and community-based mental health care settings (Jensen & Blair, 1997;King et al., 2013;Williams et al., 2018).Four studies offered writing groups as interventions in mental health care settings (Cooper, 2013;King et al., 2013;King, 2018;Padfield et al., 2017), and in two studies as an activity in a community setting (Jensen & Blair, 1997;Williams et al., 2018).Facilitators were either professional writers (King et al., 2013;Williams et al., 2018), health care professionals with a degree in English literature (Padfield et al., 2017), mental health service users (Cooper, 2013), or clinical psychology interns (King, 2018).One study did not specify the facilitator's background (Jensen & Blair, 1997).The writing groups varied in structure and format: two were ongoing and patients could flexibly join and exit (Cooper, 2013;Padfield et al., 2017), but most groups had a fixed length ranging from three to ten sessions (Jensen & Blair, 1997;King et al., 2013;King, 2018;Williams et al., 2018).
Each session usually consisted of three phases: 1) facilitator presentation of a topic or exercise, 2) individual writing, and 3) sharing with group.Most groups had themed writing exercises that could be more or less related to participants themselves (Cooper, 2013;Jensen & Blair, 1997;Padfield et al., 2017;Williams et al., 2018), while others were instructed to write stories based on their lives (King et al., 2013;King, 2018).Some interventions encouraged creative free writing while others focused on writing techniques.In all groups, participants were encouraged to share their work by reading it aloud and discussing it with  facilitators and group members.In one study, participants could e-mail their work to the facilitator for technical feedback (King et al., 2013).In two of the groups, homework was assigned between meetings (Cooper, 2013;Williams et al., 2018).Participants were often encouraged to share their stories in front of an audience at the end of the program (Jensen & Blair, 1997;Williams et al., 2018), or material could be published in a participating organization's magazine (Cooper, 2013).

Research designs
Most qualitative studies aimed to evaluate (Cooper, 2013;Jensen & Blair, 1997;Padfield et al., 2017) or describe (Jensen & Blair, 1997;King et al., 2013) a participatory creative writing intervention.Data sources varied (table 2), including a combination of observations of group activities, telephone interviews with participants and informal staff feedback (King et al., 2013), written evaluation forms for patients and staff interviews (Jensen & Blair, 1997), or solely written evaluation forms filled out by patients (Padfield et al., 2017).None of the studies included face-to-face interviews with participants.The timing of the follow-up interviews was sometimes unspecified.In a qualitative study with a comparator group (Cooper, 2013), the evaluations were conducted after the sixth session to match the length of the comparator group, even though the creative writing group was ongoing.
The two quantitative studies evaluated the writing groups in non-randomized trials using self-report surveys to measure their primary outcomes.One was a single group preand post-intervention while the other had a choir singing group as an active comparator (King, 2018).The choir group had a one-year duration whereas the creative writing group comprised only ten sessions.Participants could freely choose between the two groups.Both groups were assessed at three time points: baseline data were collected within three weeks of group onset, the two subsequent measurements were conducted with one to two-month intervals for the creative writing group, and five to six-month intervals for the choir group (Williams et al., 2018).
The quantitative studies had scores ranging from 0.65-0.68(M = 0.67) (Appendix 2b).Two studies (King, 2018;Williams et al., 2018) did not provide a rationale for sample size, did not control for confounding or report detailed results.The study with a comparator group did not use blinding of investigators or randomization of participants (Williams et al., 2018).

Clinical recovery
The only study addressing clinical recovery was a quantitative study which found that depression scores improved significantly from pre-to post-intervention; on average, participants went from moderate depression at pre-intervention (mean = 12, SD = 6) to mild depression post-intervention (mean = 8, SD = 4) (King, 2018).

Personal recovery
Most studies aimed to create positive changes related to mental wellbeing.A quantitative study examined the association between participation in a creative writing group and wellbeing (Williams et al., 2018), whereas, two studies specified that mental wellbeing was not an explicit aim of the creative writing group but merely a byproduct of the creative activity (Jensen & Blair, 1997;King et al., 2013).The included studies addressed the following CHIME constructs: Connectedness, empowerment and identity, meaning in life, and hope and optimism about future.
Connectedness was the most frequently addressed component.In the qualitative studies, groups were found to facilitate the development of friendships (Jensen & Blair, 1997) and meaningful peer connections that contributed to feeling supported (Padfield et al., 2017), feeling less lonely (Padfield et al., 2017), and a sense of belonging (Cooper, 2013).Additionally, the group enabled revelation, communication, and exploration of emotions otherwise difficult to share (Padfield et al., 2017).
Despite the emphasis on connectedness, some participants experienced interpersonal issues and felt stigmatized by other group members (Jensen & Blair, 1997).Another study examining an intervention of three workshops also did not describe any findings related to connectedness (King et al., 2013).This association between shorter duration (three to four-day workshops) and lack of positive impact may indicate that shorter duration influences the peer-group impact, as it takes time to build group coherence and a trusting relationship with a facilitator.However, studies with longer interventions highlighted problems in group dynamics and that some participants felt exposed and vulnerable (Cooper, 2013;Padfield et al., 2017).Some participants reported that the demand for interpersonal resources could evoke feelings of guilt and incompetency (Padfield et al., 2017), others that pressure and competition could arise in the group (Cooper, 2013;Padfield et al., 2017).One study emphasized that a competent facilitator is necessary to avoid group dynamics that reproduce harmful narratives about mental disorder and self (Cooper, 2013).
Writing groups were found to support empowerment as the process of writing and sharing written material was described as a medium that could support being heard and finding a voice (Cooper, 2013;Jensen & Blair, 1997) and as an arena for facing and communicating difficult emotions that can establish a sense of control over such emotions (Cooper, 2013;Jensen & Blair, 1997;Padfield et al., 2017).
Throughout the studies, we found consistency in that the writing intervention gave participants a sense of personal achievement by obtaining status as a writer or an artist (Cooper, 2013;Jensen & Blair, 1997;King et al., 2013).The experience of being acknowledged as a person with abilities and as something other than someone with a mental disorder is central to personal recovery and highly related to a sense of identity.Two studies highlighted the importance of the facilitator being a professional writer as opposed to a mental health professional to establish a sense of producing art (King et al., 2013;Williams et al., 2018).
Meaning in life was rarely touched upon in the included studies, but obtaining a social status as a writer can be considered part of the process of reestablishing a purpose in life.In general, participants expressed gratitude for being taught something unrelated to the mental disorder (King et al., 2013) and highlighted the need to be part of meaningful interventions and activities outside of the psychiatric setting.Hope and optimism was the only CHIME component that was not identified in the findings.It could be argued, though, that the meaningful relations to peers described as connectedness could also reflect meaning and hope (Padfield et al., 2017).
In summary, we found indications of a potentially positive impact on constructs related to personal recovery, particularly connectedness, empowerment, and identity.However, studies are scarce and have limitations, not allowing solid conclusions.Additionally, the research on clinical recovery is too scarce to detect any indications of benefits for the patients.

Discussion
This systematic review provides an overview of the current knowledge on the effect of facilitated, group-based creative writing interventions on clinical and personal recovery.In line with our finding, five studies in the NHS report, including participants with mental disorders (Golkaramnay et al., 2007), post-tramatic stress disorder (PTSD) (Lange et al., 2003;Sloan et al., 2012), dementia (Hong & Choi, 2011), and serious physical illness (Rickett et al., 2011), found significant improvements in depressive symptoms, but also symptoms of PTSD, anxiety, somatic complaints, and sleep problems, indicating that facilitated writing interventions may impact important aspects of subjective suffering related to mental and somatic illness.However, interventions ranged from written communication in chat forums to written exposure therapy (Lange et al., 2003;Sloan et al., 2012), making them difficult to compare to a creative writing workshop.Furthermore, the studies had methodological shortcomings, e.g.lack of randomization and concealment of sequence generation, blinding, etc.
Several factors may be the active mechanisms of facilitated, group-based creative writing interventions.Studies primarily highlighted that the very act of creative writing could strengthen a sense of capability for creative writing, emotional expression, and obtaining status as a writer (Cooper, 2013;Jensen & Blair, 1997;Padfield et al., 2017).But also the peer aspect of the interventions was highlighted as a central active ingredient (Cooper, 2013;Jensen & Blair, 1997;Padfield et al., 2017), reflected in the analysis where connectedness is the most frequently identified process.The connectedness established by being part of a peer group may mediate effects in other aspects of personal recovery.For example, being able to communicate and identify emotions in the company of others with whom there is a sense of group cohesion may mediate effects on empowerment and identity building.However, interpersonal problems could also arise in the groups, creating negative group dynamics (Jensen & Blair, 1997;Padfield et al., 2017).Therefore, the authors recommend considering the frame and dynamics of creative writing groups.The active ingredients of the interventions are likely a combination of the art-based factors and the group-based processes.
This review indicates that certain factors may influence the effect of the interventions, such as the time frame of the intervention and the facilitator's background (Cooper, 2013;King et al., 2013).The studies including a facilitator with a writing background underlined this as an essential element to ensure focus on improving and developing creative writing skills and obtaining writer status (King et al., 2013;Williams et al., 2018), and noted that mental health professionals as facilitators may be directly detrimental (King et al., 2013).However, this was contradicted in a study comparing a creative writing intervention facilitated by mental health service users to a problem-solving writing group facilitated by an occupational therapist (Cooper, 2013).Good practice may entail both the presence of a mental health professional and a skilled author to ensure that both dimensions are given equal weight.To ensure reproducibility and comparison between studies, future research should include clear, and structured guidelines for the writing interventions with thorough conceptualization.
Studies discuss pros and cons of performance or publication of material produced in the group that should also be considered when designing an intervention: for some, it may be a motivating factor inducing a sense of achievement and pride in the work, while in other cases it could enhance tensions and competitiveness in groups.Therefore, it is important to consider the participants' attitudes towards performance and to be attentive towards group dynamics (Cooper, 2013;Jensen & Blair, 1997;Padfield et al., 2017).

Strengths and limitations
The strengths of this systematic review are that it is based on a comprehensive search of several databases following the PRISMA guidelines.
One of the challenges of research in recovery is the lack of clarity regarding the concept of recovery (Bird et al., 2014;Leamy et al., 2011), which may lead to ambiguous interpretations (Vogel et al., 2020).The CHIME framework defines personal recovery, which, as it is an individual and dynamic process, is a concept that is difficult to grasp and measure.The framework is validated for mental health consumers (Bird et al., 2014), and it is considered a strength of the present study that the framework was used to organize findings.The framework can be used to identify target areas of interventions and guide the choice of recovery measures and outcomes (Bird et al., 2014;Leamy et al., 2011), which would make the comparison of studies and the building of an evidence base for creative writing interventions easier.
Earlier studies have used unclear definitions of participants' diagnoses.We only included studies of participants with a diagnosis, but the studies do not specify how participants were diagnosed, which is a limitation of this review.
The included studies were heterogeneous and varied regarding target groups, interventions, settings, and facilitators which compromised the comparison of results across studies.Different clinical outcome measures made meta-analyses unattainable.Even though studies reached a moderately high quality, the quantitative studies did not consider sample size, and they either lacked a control group or did not use blinding of investigators or randomization.Qualitative studies were compromised by lack of clarity about sampling strategy, data analysis, use of verification procedures, and whether conclusions were supported by results -shortcomings that, combined, may have resulted in biased results and uncertainty about the robustness of findings.
Most studies included facilitator interviews, but only one qualitative study (King et al., 2013) included interviews with participants.This is a considerable limitation in terms of elucidating the highly individual personal recovery processes.One study highlighted the need for qualitative studies including in-depth interviews with participants (Jensen & Blair, 1997), emphasizing the participant perspective.Interviews with both facilitators and participants are important to understand the mechanisms within the groups.From a research perspective, studies should have independent assessors conducting interviews and collecting data.

Conclusion
Despite findings indicating some promising outcomes of creative writing on personal recovery from mental health disorders, this systematic review highlights the need for further research.Overall, the lack of a clear definition, in addition to differing measures of personal recovery, as well as the heterogeneity of interventions, and methodological shortcomings of the included studies may have influenced the detection of effects across the included studies.
A potentially positive impact on constructs related to personal recovery -particularly connectedness, empowerment, and identity -was identified.However, studies on clinical recovery were scarce and no conclusions could be drawn.This warrants investigating the active mechanisms behind the CHIME components related to personal recovery in order to further support the recovery of people with mental health disorders.
Future studies should include a clearly described intervention and a theory and evidence-based program theory, including hypotheses on the mechanisms of effect of key components of the intervention, e.g.how the peer aspect, the duration of interventions, the setting, and the facilitator background impact personal and clinical recovery.Furthermore, randomized designs using both quantitative and qualitative methods in the evaluation are recommended, including a systematic investigation of the participants' perspectives.

Figure 1 .
Figure 1.Flowchart of Records in the Review

Table 1 .
Characteristics of Quantitative Studies

Table 2 .
Characteristics of Qualitative Studies