How do recorded mental health recovery narratives create connection and improve hopefulness?

Abstract Background Mental health recovery narratives are an active ingredient of recovery-oriented interventions such as peer support. Recovery narratives can create connection and hope, but there is limited evidence on the predictors of impact. Aims The aim of this study was to identify characteristics of the narrator, narrative content and participant which predict the short-term impact of recovery narratives on participants. Method Independent studies were conducted in an experimental (n = 40) and a clinical setting (n = 13). In both studies, participants with mental health problems received recorded recovery narratives and rated impact on hopefulness and connection. Predictive characteristics were identified using multi-level modelling. Results The experimental study found that narratives portraying a narrator as living well with mental health problems that is intermediate between no and full recovery, generated higher self-rated levels of hopefulness. Participants from ethnic minority backgrounds had lower levels of connection with narrators compared to participants from a white background, potentially due to reduced visibility of a narrator’s diversity characteristics. Conclusions Narratives describing partial but not complete recovery and matching on ethnicity may lead to a higher impact. Having access to narratives portraying a range of narrator characteristics to maximise the possibility of a beneficial impact on connection and hopefulness.


Introduction
Contemporary conceptualisations of recovery extend beyond the amelioration of symptoms, where individuals are recognised as experts based on their own lived experience (Treichler et al., 2019). In line with the international movement towards recovery-oriented practice (Salkeld et al., 2013), new interventions which utilise the personal narratives of persons in recovery have appeared in clinical practice (Thomas et al., 2016). Clinical approaches include narrative enhancement cognitive therapy, which is a groupbased manualised intervention using narrative enhancement exercises for personal story-telling within cognitive therapy . Giving personal narratives of recovery is also a central feature of the peer support worker role (Charles et al., 2021;Stratford et al., 2019), and an emerging approach used by mental health workers with lived experience (Byrne et al., 2016). Recorded narratives are also being used in online interventions, such as providing people with experience of psychosis access to a range of lived experience videos alongside peer support . Beyond direct clinical care, recovery narratives have also been introduced to clinical training where they can enhance communication skills and empathy (Repper & Breeze, 2007). They also underpin public health campaigns to reduce stigma, such as Time to Change in England and Opening Minds in Canada, which are based on facilitating social contact between people with and without experiences of mental health problems (London & Evans-Lacko, 2010;Stuart et al., 2014) with the personal narratives of individuals as a key active ingredient Pinfold et al., 2005).
Overall, narratives describing mental health recovery are in widespread use. However, whilst the impact of working on one's own narrative has been examined, for example in building a more positive self-identity (Yanos et al., 2011), limited quantitative investigation of the impact of receiving another person's mental health recovery narrative has been conducted.
Evidence on the impact of mental health recovery narratives has emerged. A systematic review of five studies identified recovery narratives could impact on people in six different ways; feelings of connectedness, understanding of recovery, reduction in stigma, validation of personal experiences, affective and behavioural responses . In a qualitative study of long-term impact, narratives were identified to have a helpful or harmful effect on the participant, with feelings of hopefulness and connection most reported . Connection from receiving narratives was identified to occur through three mechanisms; comparison, learning and empathy . Identifying factors that predict connection and hope will assist clinicians to provide personalised recommendations for narratives that may help, and not harm, their clients. It is unclear what narratives should be used in clinical practice and in national campaigns. At present, there is insufficient evidence about the participant or narrative characteristics that predict an increased sense of connection and hope because of receiving a mental health recovery narrative.
The aim of this study was to identify characteristics that predict the short-term impact of recorded mental health recovery narratives, as assessed by rating hopefulness and connection after receiving a narrative. The objectives were to investigate whether specific characteristics of participants (Objective 1) or narrative/narrator characteristics (Objective 2) influence short-term hope and connection, and whether narratives that mirror personal characteristics of participants are associated with higher narrative feedback ratings (Objective 3).

Methods
This study was conducted as part of the Narrative Experiences Online (NEON) Programme (reserchintorecovery.com/neon), which is investigating whether receiving recorded mental health recovery narratives improves the quality of life in people who experience mental health issues. The NEON Intervention is being evaluated (Slade et al., in press) in three clinical trials (ISRCTN11152837; ISRCTN63197153; ISRCTN76355273). Ethical committee approval was obtained for the collection of participant data and the development of the NEON Collection (London-West London REC and GTAC 18/LO/0991) and all participants gave written or online informed consent.

Study rationale
An exploratory analysis was conducted of data collected through an Experimental Study, and its ecological validity was then validated using data collected from a real-world clinical setting which consisted of the participants from the NEON feasibility study (known as the Clinical Study hereafter). Thus, the two studies had different definitions of outcomes and predictors.

Participants
For both studies, all participants were recruited from one mental health service in England. Inclusion criteria were current self-reported mental health condition; current user of statutory mental health services; aged over 18 years; fluent in English. Exclusion criteria were the self-reported current experience of mental health crises. To avoid bias, participants in the Clinical Study were independent of participants in the Experimental Study.

Measures
Narratives were characterised using the Inventory of Characteristics of Recovery Stories (INCRESE) . INCRESE is a 77-item inventory of manifest and latent characteristics of recovery narratives. Categories in INCRESE comprise narrative eligibility (Does the narrative meet the definition of a mental health recovery narrative as used in the NEON Study?), narrative mode (What format is the narrative presented in?), narrator characteristics (What are the personal characteristics of the narrator?), narrative characteristics (What genre, positioning, tone, relationship with recovery, trajectory or metaphor use does the narrative encompass), and narrative content (What topics does the narrative cover?).
Hope was assessed using the Herth Hope Index (HHI), a 12-item measure rated on a four-point scale, with a total score ranging from 12 (low hope) to 48. The HHI has good psychometric properties with high internal consistency (Cronbach alpha ¼ 0.97), reliability (0.91) and content validity (Herth, 1992). The HHI was only recorded in the experimental study.
Narrative feedback was provided through responses to three items presented after each narrative: How connected to the story did you feel? How connected to the narrator did you feel? How hopeful did the story make you feel? In the Experimental Study, each item was rated on an 11-point scale ranging from À5 (extremely disconnected or pessimistic) to þ5 (extremely connected or hopeful). In the Clinical Study, a 3-point scale from 1 (lower connection) to 3 for connection with the narrative/narrator, and from À1 (less hopeful than before), 0 (no change), 1 (a bit more hopeful), and 2 (much more hopeful) for hopefulness. Changes to the scale were made as a result of findings from a feasibility study of the website prototype  which indicated difficulties with the usability of the narrative feedback scales on mobile phones due to the size of the scale, leading to a decision to reduce the number of anchor points in order to improve engagement.

Procedures
Participants were recruited through posters in health services, by clinical research officers liaising with mental health teams, and by contacting participants from previous NEON studies who had provided consent for contact Ng et al., 2019;.
The NEON Collection comprises a managed set of mental health recovery narratives  for which permission for use has been provided. Narratives comprise a mixture of modalities (text, video, image, and audio), and narrators represented different age groups, genders, and ethnicities (see Online Supplement 1). Narratives also discussed trans-diagnostic mental health concerns and included narrators from different backgrounds and explanatory models. All narratives were characterised using INCRESE by one coder and a second coded for narrative eligibility and content warnings. Content warnings are presented prior to the presentation of a story and notify the participant of potential distress a narrative may cause due to the topics portrayed (Bellet et al., 2018). Content warnings included; Abuse or sexual violence, loss of life or endangerment to life, self-harm, violence or aggression, and injustice, prejudice and discrimination.
Participants in both studies gave information about their socio-demographic (including age, gender, ethnicity, education) and clinical characteristics (including diagnosis, contact with mental health services). In the Experimental Study, participants completed the HHI before receiving any narratives and self-rated their recovery status using predefined categories (working on recovery, thinking about recovery, not yet thinking about recovery, prefer not to say). In the Clinical Study, participants self-rated their recovery using predefined categories (I am recovered, I am living well, I am making progress, I am surviving day to day).
In the Experimental study, a collection of 30 narratives were selected from the NEON Collection. Narratives were selected if they required less than 10 min of engagement.
Participants were asked about their preferences for narratives in relation to narrator gender, age, and modality (text, video, and audio), however, preferences for ethnicity were not recorded. Participants were shown up to 10 randomly ordered narratives chosen to take account of their stated preferences such as preferring video to text and stopping when requested (e.g. due to participant fatigue). Narrative feedback was provided after each narrative received.
In the Clinical study, a collection of 100 narratives from the NEON Collection were selected, comprising 72 texts, 24 videos, 1 audio, 3 static images. Narratives were selected if they required less than 10 min of engagement. Narrative characteristics are presented in Online Supplement 1. Narratives were hosted on a website prototype for the NEON trial (Rennick-Egglestone et al., 2020) which offers participants the opportunity to self-select narratives based on tagged characteristics or to receive a randomly selected narrative. Tagged characteristics were based on the narrator and narrative characteristics as rated on the INCRESE inventory. The aim of this approach was to maximise the ecological validity of the findings of impact, to inform the NEON trials. Therefore, this study was conducted within a real-world context, as part of the NEON feasibility study. Participants had one face-to-face session with a researcher to complete registration processes, introduce the online portal, and practice accessing narratives. Narrative feedback ratings were obtained after each narrative and participants were given unrestricted access to the online platform for 1 month, with no recommended usage requirements such as a maximum or a minimum number of narratives to receive.

Analysis
All objectives were addressed using a univariate multi-level (participant and narrative), mixed effect statistical model. This statistical model accounts for the variability between and within participants and between and within narratives. Models were implemented in R 3.6.3 (R Core Team, 2019). Participants and narratives in the studies were given random intercepts, with other effects kept as fixed effects for hypothesis testing. The dependent variables were the narrative feedback ratings on the mechanism (hopefulness, connection to narrative, connection to narrator) relevant to each objective. The predictor variables were participant characteristics, narrative/narrator characteristics or matched participant and narrative/narrator characteristics.
Significant predictors from the univariate multi-level model from the Experimental Study and Clinical Study were compared for significance. Model fit was compared using Maximum Likelihood and ANOVA for the fit of the univariate predictor model compared to the fit of the intercept model to obtain an overall p-value. Each predictor variable was then further tested for significance using Random Effects Maximum Likelihood. b coefficients of predictor variables were not compared across the experimental and clinical studies due to differences in the scales used on the dependent variables. The reference category for comparison of ordinal data was selected based on being the smallest or least category (for example relationship with recovery, people who were surviving day-to-day was the reference category, whilst the reference category for comparison for nominal data was selected alphabetically (for example, for gender, the female category was the reference category) or the largest category to preserve power (i.e. white background in ethnicity), smaller categories of ethnicity were combined to ensure power to test for differences associated with being from ethnic minority backgrounds.
The research team consisted of a diverse range of academic and clinical perspectives including mental health services research, survivor research; critical qualitative health research, qualitative research within recovery; statistics; and clinical psychology. Some members of the research team also identify with having lived experience of mental health problems or mental health distress. Consultations with the NEON Lived Experience Advisory Panel (LEAP) occurred in three phases. First, consultations were held in the development of the NEON Collection , where LEAP members were consulted on the ethical principles of curating narratives, in the identification of narratives for inclusion in the NEON Collection, and when disagreement occurred amongst the research team with regards to the inclusion of a narrative in the NEON Collection. Second, LEAP members were consulted in the development of INCRESE, particularly in the categorisation of content warnings. Third, consultations were held with three members of the LEAP to support the interpretation of the findings. This ensured that there was a balance between academic and lived experience perspectives.

Results
The characteristics of participants in the Experimental study (n ¼ 40) and the Clinical study (n ¼ 13) are shown in Table 1.
The characteristics of the narratives received during the two studies are shown in Table 2.

Experimental study
Participants received a mean of 7 narratives (Range 3-10). Univariate predictors for the multi-level model are shown in Online Supplement 2, significant predictors from the multilevel model for both model fit, and prediction are summarised in Table 3.
Higher baseline levels of hope, as measured using the HHI, predicted higher hopefulness ratings (b ¼ 0.06, S.E ¼ 0.03, p ¼ 0.02). Narratives were rated as more hopeful for narratives depicting a narrator living well with mental health problems, that is, intermediate between making progress and being recovered (b ¼ 1.16, S.E ¼ 0.50, p ¼ 0.03). A participant's connection with the narrator was higher when the narrator's gender was the same as the participant preference for narrator gender (b ¼ 0.99, S.E.¼0.51, p ¼ 0.05). Less connection with narrators was also detected for participants from ethnic minority backgrounds compared to participants from a white background (b¼ À1.04, S.E.¼0.41, p ¼ 0.01)

Clinical study
Thirteen participants took part in the Clinical Study. Of the 100 narratives available to view in the Clinical Study, only 38 narratives were received by participants. On average, participants received six narratives (range 2-14). Univariate predictors for the multi-level model of the three dependent variables (hopefulness, connection with the narrative/narrator) are presented as Online Supplement 3, the significant predictors for both model fit and prediction are summarised in Table 4.

Discussion
This is the first paper to examine the predictors of impact in recorded mental health recovery narratives. Receiving narratives describing living well with mental health problems generated higher hopefulness than narratives describing surviving day-to-day. A match between the ethnic background of participants and narrators was associated with higher levels of hopefulness, yet participants from ethnic minority backgrounds rated their connection with the narrative and narrator significantly lower than participants from white backgrounds overall.

Relationship to previous studies
Narratives describing living well with mental health problems generated more hopefulness than narratives describing surviving day-to-day or being fully recovered. Concerns over the curative connotations of "being recovered" have been raised, as these may not encapsulate the experiences of many individuals (Piat et al., 2009). Therefore, narratives describing living well with mental health problems may provide insight into experiences that may be perceived to be more achievable or realistic. Qualitative studies indicate that the perceived authenticity of narratives is known to be an important mediator of the impact of recorded mental health recovery narratives . The authenticity of narratives contributes to their perceived genuineness and allows for a greater understanding of the narrator's lived experience.
Participants' demographic variables such as ethnicity, age and gender, were also predictive of hope and connection. For age and gender, a match between the participant and the narrator did not necessarily influence how connected or hopeful people felt after receiving a narrative. Ethnicity influenced ratings of connection and hopefulness, such that in the Experimental Study, participants from ethnic minority backgrounds rated their connection with the narrative and the narrator as significantly lower than participants from a white background. This may be attributable to there being fewer narrators from ethnic minority backgrounds in the NEON collection at present. However, it should be acknowledged that in both studies, participants from ethnic minority backgrounds only represented a small proportion (17.5% and 15%) which may affect the generalisability of the findings.
The findings in the Clinical Study indicate that a match between a participant and the narrator's ethnicity (for example when a recipient and narrator are both classified as being from ethnic minority backgrounds) can lead to higher levels of hope. Role-modelling hope is a core component of a peer support worker's role (Bradstreet & Pratt, 2010), and the ability to see oneself within a narrative may contribute to a sense of belonging. Additionally, people from ethnic minority backgrounds can experience significant disparities in their ability to access mental health services (Vahdaninia et al., 2020), with stigma further deterring help-seeking (Linney et al., 2020). Receiving mental health recovery narratives may be a more discreet opportunity for people from ethnic minority backgrounds to learn about recovery and may act as a form of social contact.

Strengths and limitations of the paper
There are several strengths of the paper. First, narratives were provided in multiple modalities. Although only text-based narratives generated higher ratings of connection with the narrative in the Experimental Study, the provision of different narrative modalities may cater to different learning styles. Second, this is the first study to utilise narratives discussing trans-diagnostic mental health concerns, involving narrators from diverse backgrounds, and differing explanatory models (for example mad studies, survivor, and clinical models). Third, previous studies have only examined the impact of narratives in experimental settings. The Clinical Study was the first study to examine the impact of recovery narratives within a real-world context as part of a feasibility study of the NEON trial . Finally, the provision of narratives differed between the two studies. Participants in the Experimental Study were provided randomly selected narratives, whilst participants in the Clinical Study could obtain a randomly selected or selfselected narrative. It is plausible that participants may choose narratives that reflect their personal characteristics, however, except for a match in participant and narrator ethnicity, the findings do not indicate that a match in personal characteristics leads to higher connection or hope. Despite the ability to self-select narratives, the process undertaken by a participant in choosing a narrative is unclear. Future qualitative research using a think-aloud protocol to delineate the decisions behind choosing a narrative may assist in improving the categorisation of narratives.
Several limitations can be identified. First, a small sample size across both studies was used. The generalisability of the findings may be limited, but findings provide a preliminary insight into variables that may predict the impact of recorded recovery narratives on participants. To increase generalisability and the applicability of findings, further investigation using a larger sample size in both participants and narratives within a real-world setting, beyond English speakers, is required to identify predictors of impact more definitively. Second, the study's power may be limited. Participants in the Clinical Study provided feedback ratings for 38 narratives, despite having access to narratives for a 1month period. This may be indicative of low participant engagement, which may be rectified by sending email reminders. Third, to maximise the benefits of receiving narratives, it is recommended that narratives that depict narrators living well with mental health problems be presented. However, further research is required to understand why narratives of fully recovered narrators are less helpful and whether these narratives are more helpful in other clinical or public settings (for example in-patient settings or antistigma campaigns). Fourth, ethnicity was categorised as a dichotomous variable in the quantitative analysis. This does not take into consideration that ethnicity is a contested construct given the significant heterogeneity between individuals who identify as being from ethnic minority backgrounds. It is therefore incorrect to assume that participants and narrators from ethnic minority backgrounds will have similar experiences based solely on the ethnicity construct. Moreover, an individual's ethnic identity can intersect with other confounding factors such as social class, socio-economic status, an individual's appearance, or accent, to affect how a participant connects or derives hopefulness from the narrative. The media format used to present narratives may have exacerbated perceived differences. Audio and video narratives provide additional cues which can change one's perception of a narrator. Participants from ethnic minority backgrounds identified feeling less connected with the narrator and the narrative, however other confounding variables not examined by the study may have contributed to perceived differences. Future research could qualitatively examine the specific components that participants attend to when receiving narratives and whether these components differ between individuals from different ethnic backgrounds. The inclusion of more ethnic categories, beyond the dichotomous white and ethnic minority categories in the quantitative analysis, with larger samples for each ethnic category, may provide a more in-depth understanding of the influences of the ethnicity construct on the impact of mental health recovery narratives is needed. Similarly, examining the differing effects of each narrative format may assist to establish whether there is an ideal narrative format for generating connection and hopefulness.

Implications for clinical practice and research
Results from the current study suggest clinical approaches to influencing personal recovery and point mental health service providers in a direction that further embeds recovery into service delivery. Therefore, the findings of the present study can influence clinical practice and research in several ways. First, the provision of narratives that depict a narrator as fully recovered may not promote hopefulness in participants at an earlier relationship with the recovery stage. Rather, narratives of individuals living well with mental health problems but still experiencing challenges may show a more realistic and hope-promoting portrayal of recovery. It is advisable that mental health services and systems that utilise recovery narratives as part of the promotion of services refrain from only using narratives of individuals who are fully recovered. Furthermore, national anti-stigma campaigns should make use of authentic narratives including those with elements of ongoing struggle, rather than solely using narratives in which all difficulties are fully resolved. Second, prior to recommending specific narratives, it is recommended that mental health workers have ongoing discussions with people with mental health problems about their relationship with recovery as this may affect an individual's subsequent experience of connection and hopefulness.
Third, diversity in the narratives presented to individuals with mental health problems is required. This not only maximises the likelihood of individuals connecting with a narrative or narrator and feeling hopeful, but it also allows for a greater understanding of perspectives that previously may not have been encountered. For example, in a clinical context, it may be important to include less articulate and polished narratives or ones where the narrative is not fully resolved or is expressed metaphorically rather than literally, to maximise the likelihood of impact. This also has implications for countries with first-nation peoples. For example, in Canada future research with Indigenous groups within Canada may build upon these findings. The Mental Health Commission of Canada guidelines includes a chapter on working with First Nations, Inuit and Metis, recognising that recovery principles and the Indigenous shared understanding of wellness overlap extensively (Mental Health Commission of Canada, 2015). The use of recovery narratives with this population, offered in their own languages, could prove beneficial. Culture-specific recovery narratives are a practical means of providing relevant and tailored support. Fourth, large-scale trials involving a diverse collection of recorded recovery narratives (such as the NEON Trial)  are required to robustly assess the impacts of receiving recorded recovery narratives.

Disclosure statement
No potential conflict of interest was reported by the author(s). The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.