‘ Someone on my level ’ : A Meta-Ethnographic Review of Therapeutic Relationships in Cognitive Behavioural Therapy for Psychosis

Objectives: Cognitive Behavioural Therapy for Psychosis (CBTp) demonstrates variable and at times mild to moderate effect sizes; thus, its therapeutic processes are important to explore. Establishing a secure therapeutic relationship is one such key process where barriers may exist, including those related to psychotic symptoms and associated stigma. This review synthesizes the available qualitative research pertaining to the experience of the therapeutic relationship from the perspective of those experiencing psychosis. Methods: A systematic review was undertaken using PRISMA guidelines. Search terms included variants of ‘ psychosis ’ , ‘ therapy ’ and ‘ qualitative ’ . PsycInfo, CINAHL, EmBase, MedLine and Web of Science were searched, and reference lists were hand-scanned. Yardley's quality appraisal tool was utilized and Noblit and Hare's seven-stage process for conducting a meta-ethnographic review. A line-of-argument synthesis is presented. Results: Fourteen papers were identified using inclusion and exclusion criteria. Twelve papers were deemed to have satisfactory quality. The line-of-argument synthesis used attachment theory to propose four semi-distinct stages to establishing a therapeutic relationship: beginning; safety, hope and trust; the practicalities of therapy; and branching out. Findings suggest that the therapist's persona and use of CBTp techniques such as collaboration and shared agency over the process were important in establishing for the patient a sense of self as normal, equal and worthwhile. Conclusions: Attachment security may be an important strand of CBTp and warrants further research and clinical investigation as a process and an outcome. Future research can benefit from increased transparency regarding researcher positionality as a potential source of bias.


| INTRODUCTION
Cognitive Behavioural Therapy for Psychosis (CBTp) has historically demonstrated variable and at times limited effect sizes in producing a range of outcomes in psychosis (Jones et al., 2018;Laws et al., 2018;Mehl et al., 2015). Recent trends in CBTp research have focussed on symptom-specific approaches to identifying and addressing maintenance factors, such as addressing the role of worry processes and sleep in delusional beliefs (Freeman et al., 2015;Lincoln & Peters, 2019), the role of trauma in both hallucinations and delusions (Folk et al., 2019;Larkin & Read, 2008) and the role of attachment in voice hearing and paranoia (Berry et al., 2008;Gumley et al., 2014). A recent meta-analysis reported improvements in CBTp for addressing delusions over time, suggesting that the shift towards symptomfocussed CBTp approaches offers improved focus and efficacy in addressing specific difficulties compared to broader packages (Sitko et al., 2020). As such, the processes underlying the efficacy of CBTp are important to explore further.
One strand of research enquiry is considering the role of attachment in psychosis, which may have twofold importance as a developmental risk and/or maintenance factor in psychotic experiences and as a determinant for the development of a secure therapeutic relationship (Berry et al., 2019). Quantitative research has long suggested that the development of a secure therapeutic relationship is a predictor of positive therapeutic outcomes (Ardito & Rabellino, 2011), with research replicating this finding in therapy for psychosis (Browne et al., 2019;Priebe et al., 2011). However, the experience of psychosis is known to relate to potential difficulties engaging in therapy, such as paranoia impairing one's ability to relate to a therapist and hallucinations negatively impacting active engagement (Allott et al., 2018;Gottlieb et al., 2011). Research has also found elevated rates of insecure attachment in people experiencing psychosis Gumley et al., 2014), and this is known to limit the development of a secure therapeutic relationship Mikulincer & Shaver, 2012). Further research has considered that insecure attachment styles may contribute to negative underlying beliefs, such as a basic mistrust of others (Berry & Bucci, 2015;Debbané et al., 2016), and maladaptive coping strategies such as avoidance or dissociation , which may in part maintain distressing voice hearing and paranoia. Some studies show associations between insecure attachment styles and worse positive symptoms .
Emotion regulation difficulties, one of the consequences of insecure attachment styles, have been linked to poorer quality therapeutic relationships in people experiencing psychosis (Mehl et al., 2020).
Prior research therefore suggests that attachment style may play a role in both the development of a secure therapeutic relationship and overall symptom severity, which may in turn influence therapy engagement and overall outcome (Berry et al., 2019). The present review synthesizes the existing qualitative research regarding what people experiencing psychosis find important in establishing a therapeutic relationship. A theoretical abstraction of the themes identified from the data is presented.

| Aims and research question
The present systematic review synthesizes the available qualitative literature pertaining to the following research question: what is important in the experience of a therapeutic relationship for people presenting with psychosis? It focusses specifically on CBTp and Cognitive Behavioural Therapy (CBT) interventions for co-morbidities within a psychosis population. As the review aimed to offer an abstract understanding of a healthcare phenomenon, meta-ethnography was deemed the appropriate synthesis methodology (Noblit & Hare, 1988). It offers an overarching account grounded in the available data and an emergent line-of-argument synthesis to inform practitioners and researchers working in this field. Relevant theory was T A B L E 1 A table summarizing the seven-stage process of metaethnography (Noblit & Hare, 1988)  Expressing the synthesis Producing a written synthesis as one of many possible forms.

Key Practitioner Message
• Individuals experiencing psychosis may present to therapy with insecurities manifesting in an initial negative experience of therapy and the therapist.
• Core therapeutic competencies and CBT techniques can instill a sense of safety and trust.
• Attachment theory may provide a framework to consider and address the relational elements of CBTp, including how one experiences oneself and the therapist in the therapy context. considered in light of the emergent themes, rather than being preimposed. The review also appraised the quality of the body of literature. Noblit and Hare's (1988) guidance for conducting meta-ethnography studies was followed and reporting adhered to published guidance (France et al., 2019). The process is summarized in the below table (see Table 1). Additionally, the seven most cited authors in the field were emailed to inquire about any additional or upcoming publications that had been missed. The PRISMA diagram can be found below (see Figure 1)  Phase of translation process Description of process Credibility checking 1) Data extraction and quality appraisal • A paper was read in full twice.

|
• Data extraction and quality appraisal tools were completed. • Paper was read in full again, and key first and second-order concepts from the entire paper relating to the therapeutic relationship were highlighted in NVivo. • Third-order interpretations were noted alongside the codes. • Repeated for each paper • Decision made with both supervisors to group all the papers together due to conceptual overlaps • Five papers in total were co-coded, one each by the two supervisors and three individually by peer doctoral trainees to monitor consistency in identifying relevant themes, data extraction and quality appraisal. • Bracketing interview and reflective diary were used to note potential bias in thirdorder interpretations.
2) Translating meanings within papers • Codes were listed in a table • Higher level thematic units of meaning were given to each code to reflect its overarching meaning; for example, a firstorder code about the importance of therapist's body language was given 'therapist's approach'. • Higher level units of meaning were grouped based on reciprocal or refutational elements. • A summary of overarching themes and codes was produced. • Process repeated for each paper • Re-iterative approach of referring to the original data once higher-level units of meaning were ascribed to consider 'best fit' • Ongoing use of supervision to review the translation process • Continued use of reflective diary to monitor research bias and risk of reducing the original meanings 3) Synthesizing translations across papers • A table containing all of the papers themes was produced. • Related themes were grouped into superordinate categories, maintaining reference to the original data and meanings from the original codes. • A re-iterative process to consider the original codes and data, reciprocal and refutational elements within themes and a 'best fit'.
• Both supervisors read one random individual paper each to track development of translations across the entire process.
4) Expressing the synthesis • A visual concept map was produced to reflect how the translations fit together.
• A table mapping original quotes and codes was completed and monitored by both supervisors.
concepts were defined as direct quotes from participants; secondorder concepts were interpretations made by the primary researchers; and third-order concepts were the lead researcher's overall interpretation of the first and second-order codes and form the final synthesis (Noblit & Hare, 1988).

| Quality appraisal
Yardley's four-criterion quality appraisal tool (Yardley, 2000(Yardley, , 2017 was used. This spans four areas: sensitivity to context; commitment and rigour; coherence and transparency; and impact and importance. A 4-point rating system was utilized to consider inter-rater reliability: criteria absent (0), less than half criteria present but limited in overall depth or description (1), more than half the criteria present and satisfactory depth to description enabling replicability (2) to all criteria present and of detailed description ( 2.5 | Synthesizing translations and expressing the synthesis

| Synthesis process
Supporting information is available upon request regarding the synthesis process. A visual concept map was drawn to express the lineof-argument synthesis (see Figure 2), and the papers referencing each overarching theme were tracked to enable an auditable process overseen by the research team.

| Reflexivity statement
The lead author is a white, British male completing doctoral training in Clinical Psychology. Having spent 3 years working in an Early Intervention in Psychosis service prior to doctoral training, he developed interests in user engagement in mental health care where barriers seemed to exist and how services responded to this. He has theoretical interests in relational and attachment-based approaches to mental health. A bracketing interview was conducted to identify key personal and professional sources of resonance and potential bias, which was monitored by the research team throughout the analytic process.

| Outcome of systematic search
The PRISMA diagram below summarizes the results of the systematic search (see Figure 1). The systematic search found 1,323 articles.
Following title and abstract screening, 1,278 were removed.
Forty-five papers were then screened in full using the inclusion and exclusion criteria, resulting in 14 papers being included in the study.

| Characteristics of included studies
Full study characteristics can be found in Table 3.

| Quality appraisal
Twelve papers were rated as having satisfactory quality overall based on the information provided, although the majority of these had sample sizes of below 10 (n = 6). Two papers were deemed to have limited quality, one of these (Halpin et al., 2016) had the lowest overall rating of sensitivity to context having provided limited description of the sample and recruitment strategy and having a sample size of only three. Notable further issues in the two limited quality papers were absence of user involvement, limited reports of data saturation, limited credibility checking beyond the supervisor or immediate  (Halpin et al., 2016;Waller et al., 2015).
These omissions may be due to limited space in publishing rather than the absence of consideration. Messari and Hallam's (2003) and Kilbride et al.'s (2013)

| Outcome of synthesis
The second-order translations were organized into four themes reflecting the meanings ascribed to the therapeutic relationship throughout the process of therapy; beginning; safety, hope and trust; practicalities; and branching out. Two additional themes reflected facilitators and barriers to relating. Themes will be outlined before the

| The practicalities of therapy
This theme describes individual's experience of therapeutic tasks and techniques. Twelve papers endorsed this theme. Dilks et al.'s (2013) grounded theory paper proposed a model of therapeutic processes: The central therapy activity identified was conceptually summarized as a process of building bridges to observational perspectives. This was defined as a jointly negoti-  Bjornestad et al.'s (2018, p.  Authors interpreted that the experience of a positive therapeutic relationship offered a juxtaposition or an 'antidote' to negative ways of relating pertaining to paranoia (Lawlor et al., 2015), trauma symptoms (Tong et al., 2019), submissive relating styles associated with voice hearing (Hazell et al., 2017) or internalized self-stigma (Wood et al., 2016).

| Barriers and facilitators to engagement
Eleven papers reported barriers and facilitators to engaging in a therapeutic relationship. These are summarized in the below table (see Table 4). The factors were nonlinear and largely unrelated to each other; rather, they were embedded within the experience of therapy.
Third-order concepts are presented using titles reflecting the language of the original papers.

| Line of argument synthesis
As the third-order interpretations ascribed to each theme appeared to relate to participants' sense of self and others in relation to each 'stage' of the therapeutic relationship, attachment theory (Bowlby, 1969) was considered an appropriate theory to provide an overarching synthesis (see Figure 2). The therapeutic relationship and core CBT techniques were experienced by many participants as reflecting the self as acceptable and worthwhile: she makes me feel like I'm someone special, I'm worthy (Dilks et al., 2013, p. 220).
It is therefore proposed that attachment security presents a cog in a machine which moves people through the therapeutic process; movement through stages may reflect shifts in one's sense of self and others. There are also opposing facilitator and barrier cogs influencing its flow, containing factors influencing the development of a secure relationship across the stages. Papers imbued the latter stages of therapy, propelled on by the relationship, with various affective and relational gains: the fog lifted a little, I had not had that experience before.
I found the missing words for these thoughts and feelings. (Bjornestad et al., 2018, p. 4).
This may reflect an established secure attachment relationship, whereby a child internalizes a sense of 'good enough' and safety to explore the world around them (Bowlby, 1969): Therapists' use of self and CBT techniques were both highlighted as important in facilitating this security and a range of facilitators and barriers were identified (see Table 3). Findings will next be considered in light of existing research.

| Therapeutic relationships in CBTp
The  (Linnington, 2019). The present findings suggest that this in itself may be a change process, as well as facilitating engagement with other treatment strategies. This is salient given the limited effectiveness of CBTp in research trials; this review proposes that attachment security sets the context for therapy to occur effectively, as well as providing a key intervention strategy itself. It may be that attention afforded to attachment and relational issues directly influences treatment efficacy, which requires further research. Participant experiences in this respect suggest that CBTp can produce quality of life improvements and reductions in distress as well as symptom reductions although these outcomes are not always reflected in research trials (Jones et al., 2018;Laws et al., 2018). Attachment as an active therapeutic process in CBTp warrants further investigation. For example, given research suggesting that relationally orientated therapies can influence positive change in attachment security (Travis et al., 2001), it would be interesting to investigate how attachment security changes over the course of CBTp and affects outcome, and underlying beliefs about oneself, others and the world. Some research has explored selfconcept transformation as a process and outcome of psychotherapy for psychosis (Lysaker et al., 2003(Lysaker et al., , 2015; attachment theory may offer a way to conceptualize and measure this in further research within CBTp. The present findings are largely in line with previous reviews of user perspectives of CBTp in that the therapeutic relationship is an important aspect of therapy (Berry & Hayward, 2011;Wood et al., 2015). CBTp strategies such as collaboration may be experienced as novel aspects of therapy and go some way to alleviate the relational sequelae of psychosis, such as passivity (Messari & Hallam, 2003). However, cultural issues such as ethnic background and gender were not wholly reported or accounted for in the studies.
Some research suggests that cultural beliefs inform beliefs about psychosis, therapy and mental health services more broadly (Dutta et al., 2019;Naeem et al., 2015;Rathod et al., 2010).

| Social relationships and psychosis
The role of attachment in psychosis is relevant to wider considerations of the social world of those experiencing psychosis, where symptoms, internalized self-stigma and negative self-concepts can impair the ability to relate (Brabban et al., 2017;Lim et al., 2018).
Research considering attachment and psychosis has begun to inform work in various settings, including inpatient wards (Campbell et al., 2014). There is also growing research into the benefits of affirmative spaces such as the charity The Hearing Voices Network that report similar themes regarding the freedom to be oneself, connect with others and make sense of one's experiences (Payne et al., 2017).
It may be that individual therapy provides one opportunity for a secure relationship to ameliorate these issues and support individuals to branch out to other positive social experiences. However, alternatives are also important to consider. The role of the self and identity has been considered in research in psychosis (Braehler & Schwannauer, 2012;Lysaker et al., 2003;Meehan & MacLachlan, 2008). Attachment theory could provide a framework for interventions targeting such factors via clinical design and research (Gumley et al., 2014). security. It is acknowledged that this is but one frame from which to consider these dynamic processes. Being qualitative in nature, the review also offers no clarity as to causation and only represents the experiences of some.

| Limitations
As discussed earlier, the use of a quality appraisal tool in qualitative reviews is under debate.

| Clinical implications
The findings here suggest a number of clinical implications. Attachment security as a concept could be formulated and inform intervention for people experiencing psychosis as an important factor in both therapy engagement and outcomes. Individuals with pre-existing attachment security may engage with therapy processes more easily than insecurely attached people (Mallinckrodt & Jeong, 2015), which may have particular salience for those experiencing psychosis where symptoms and stigma pose additional barriers to treatment engagement. CBTp strategies such as collaboration, shared agency over process and flexibility may go some way to model a secure attachment relationship.
CBTp treatment strategies may have limited effectiveness and may even be appraised as threatening in the absence of consideration of attachment and pre-existing relational beliefs. This resonates with wider suggestions for a focus on engagement, normalization and person-centred, relational engagement (Brabban et al., 2017).
This review also suggests that facilitating clients 'branching out' is an important aspect of clinical care, so facilitating engagement with safe, affirming social spaces may be of benefit. Hearing Voices Network groups, vocational resources and groups or activities affording social connection may maintain gains made from a positive experience of therapy (Brabban et al., 2017;Cooke, 2017). It may be that individuals need therapeutic support to make these connections and test their helpfulness.
Therapist use of self and therapist beliefs were identified as impor-