What are the key elements of cognitive analytic therapy for psychosis? A Delphi study.

OBJECTIVE: There has been growing interest in the use of cognitive analytic therapy (CAT) with those facing experiences of psychosis. However, there is little research on how CAT is best applied to working with psychosis. This study aimed to identify what the key aspects of CAT for psychosis are or whether this approach requires adaptation when applied to those with experiences of psychosis, drawing on expert opinion.
METHOD: An adapted Delphi methodology was used. Items were generated during an initial workshop (N = 24) and then rated for agreement or importance via an online survey by a sample of experts with experience of CAT and working clinically with psychosis (N = 14).
RESULTS: Following two rounds of ratings, consensus was reached on most items. Additional comments emphasized the need to be flexible with regard to the varying needs of individual clients.
CONCLUSIONS: Results highlight the specific relational understanding of psychosis provided by CAT as one of the key elements of this approach. Responses emphasized the need for some level of adaptation to work with psychosis, including greater flexibility with regard to the treatment frame. PRACTITIONER POINTS: When working with experiences of psychosis, aspects of the CAT model, such as session length, pacing, and duration of therapy, are open to change and may require modification. When working with experiences of psychosis, narrative reformulation letters and sequential diagrammatic reformulation (SDR) remain essential to the therapy. This Delphi methodology study essentially relies on opinion. Further empirical research could test assumptions about the most important or therapeutically effective components of CAT in psychosis. CAT is still not widely used in the context of psychosis limiting the pool of experts available for the current sample.

Background

Cognitive analytic therapy (CAT) is an integrative, psychological therapy drawing on object relations theory and social developmental theory (Denman, 2001; Ryle & Kerr, 2002). CAT was developed in the context of affective and personality‐related difficulties, but has since been considered for psychosis (Taylor, Perry, Hutton, Seddon, & Tan, 2015). Cognitive behavioural therapy (CBT) is currently the most widely supported and recommended therapy for psychosis (NICE, 2014). The evidence base largely supports the efficacy of CBT for psychosis (Sarin, Wallin, & Widerlov, 2011; Turner, van der Gaag, Karyotaki, & Cuijpers, 2014; Wykes, Steel, Everitt, & Tarrier, 2008). However, CBT does not work for everyone, and CAT may be viable alternative for psychosis, particularly where CBT has been ineffective, where comorbid interpersonal or personality‐related difficulties exist (e.g., Gleeson et al., 2012), or where clients have a preference towards a more analytically orientated approach (Kerr, Crowley, & Beard, 2006). If CAT is to be used in the context of psychosis, a question remains around whether this approach needs adapting or modifying for this population. This study uses a Delphi methodology (Hsu & Sandford, 2007) to seek a consensus amongst practitioners on how CAT is applied to psychosis.

CAT is a relational approach, focussing on clients’ patterns of relating to others and themselves (Denman, 2001; Ryle & Kerr, 2002). These patterns of relating, internalized from ongoing interpersonal interaction, are referred to as ‘reciprocal roles’. Psychological problems are seen to arise where particular maladaptive reciprocal roles become dominant, and are reinforced by patterns of thinking, feeling, and acting referred to as ‘problem procedures’ (Leiman, 1997; Ryle, 1997). The Multiple Self States Model (MSSM; Pollock, Broadbent, Clarke, Dorrian, & Ryle, 2001; Ryle, 1997) provides a framework for understanding more complex difficulties within CAT. MSSM specifies how fluid and appropriate transition between RRs can break down in some instances, including psychosis, resulting in the emergence of specific ‘self‐states’, particularly exaggerated or dominant RRs which may feel cut‐off or fragmented from other RRs in a person's repertoire (Kerr, Birkett, & Chanen, 2003). These disconnected self‐states may result in phenomena that can ultimately appear external to the self (e.g., hallucinations). MSSM also specifies how self‐reflection may become impaired, preventing a change in unhelpful patterns. Early trauma is seen as an experience that contributes to these disturbances in self‐states (Ryle & Fawkes, 2007), consistent with the evidence that trauma increases the risk of psychosis (Varese et al., 2012). MSSM provides one way to conceptualize psychosis within CAT, but it is also possible to understand the content of psychotic experiences within CAT without this model (e.g., focussing primarily on reciprocal role procedures).

Many psychotic experiences are interpersonal in nature (Berry, Wearden, Barrowclough, & Liversidge, 2006). Individuals can experience a relationship with their voices (Pérez‐Álvarez, García‐Montes, Perona‐Garcelán, & Vallina‐Fernández, 2008), and paranoia involves a concern about interpersonal threat (Bentall, Corcoran, Howard, Blackwood, & Kinderman, 2001). A relational model like CAT may therefore be well suited to helping those with psychosis. We provide further details around specific CAT tools and concepts in Table 1.

Table 1

Summary of key CAT concepts and tools

Theoretical concepts	Description
Countertransference	The therapist's emotional response and elicited actions to the client's transference
Dialogical Sequence Analysis (DSA)	A technique for identifying dialogical or relational patterns in clients experiences (Leiman & Stiles, 2001)
Multiple Self States Model (MSSM)	A model of identity disturbance within CAT which encompasses the idea of self‐states (see below; Ryle, 1997)
Reciprocal Role (RR)	A named pattern of relating, originating in childhood with an actual or internalized other
Reciprocal Role Procedures (RRPs)	The feelings, actions and beliefs resulting from being at one end of a reciprocal role
Reciprocal Role analysis of enactments	The enactment of reciprocal roles is continually monitored through the therapy, aided by the therapy map (SDR)
Rupture and repair sequences	Ruptures occur when problematic reciprocal role patterns take place in the therapy relationship. Rupture repair is achieved by the therapist not colluding and collaboratively using the opportunity to develop client awareness and exits around the pattern
Self‐states	Mental states that have become dissociated from one another, seen as involving a specific RR. An individual may shift in and out of this RR in an extreme or dramatic fashion
Target Problem Procedure (TPP) Transference	The problematic reciprocal role procedures that are the focus of the therapy. The unconscious redirection of inappropriate feelings in the present, of a relationship that was important in the past. This is informative in identifying RRs and RRPs
Zone of proximal development (ZPD)	This is a Vygotskian term. The ZPD refers to the client's potential space for change, with the help of another, namely the therapist

Further details can be found in Ryle and Kerr (2002); tools are presented in the order they usually appear in therapy.

The therapy itself is time‐limited, traditionally consisting of 16 sessions, or 24 in the context of complex cases, and involves three stages: reformulation (involving the identification of key relational patterns and associated ways of acting or coping); recognition (developing the client's ability to recognize unhelpful patterns); revision (developing interventions for or alternatives to the unhelpful patterns that have been identified; Ryle & Kerr, 2002). During the reformulation stage of therapy, the emphasis is on collaboratively building a shared narrative with clients concerning their difficulties, which often features a number of specific reciprocal roles at its core (Ryle & Kerr, 2002). These reciprocal roles provide a flexible means of mapping clients’ experiences, concerning patterns of relating to others (e.g., seeing others as hostile and threatening) and themselves (e.g., self‐criticism or punishing) and can be readily applied to psychotic phenomena such auditory hallucinations or persecutory delusions. A visual map detailing the problematic patterns is developed to use as a tool to understand and monitor the enactment of the reciprocal roles (the sequential diagrammatic reformulation; SDR). In the recognition stage, the goal is the development of clients’ capacity to identify and observe these processes. Clients are actively encouraged both within and in‐between therapy sessions to develop their awareness through monitoring for occurrences of the unhelpful patterns. CAT then works towards collaboratively identifying alternative ways of relating or ‘exits’ from the unhelpful patterns outlined in the reformulation (revision stage). Practitioners may draw upon other therapeutic approaches when identifying suitable exits, as long as these are consistent with the underlying reformulation. For example, a therapist may make use of compassion‐focussed techniques for someone with a self‐critical style of relating to themselves (McCormick, 2011). Management of endings is important in CAT, and the end of therapy is often a focus within sessions. The use of transference and countertransference is central to the application of CAT (Ryle, 1995). An explicit focus on the therapeutic relationship occurs at all stages of therapy (Bennett, Parry, & Ryle, 2006; Ryle & Kerr, 2002). When the unhelpful pattern takes place in the therapy, this can lead to a rupture in the therapy relationship. Repair takes place by the therapist not colluding with the reciprocal role pattern and using the experience to develop awareness and exits (Daly, Llewelyn, McDougall, & Chanen, 2009).

A number of studies have started to provide evidence of the acceptability and feasibility of CAT for those with psychosis. Pilot randomized trials of CAT for bipolar disorder (Evans & Kellett, 2014) and a multimodal, CAT‐informed intervention for comorbid early psychosis and personality disorder difficulties (Gleeson et al., 2012) have been undertaken in the UK and Australia, respectively. Case studies and small‐scale case series of CAT for psychosis also exist (Graham & Thavasotby, 1995; Kerr, 2001). Evidence of the efficacy of CAT for psychosis compared to standard interventions is still lacking, but in order to conduct such trials, a clear idea of how CAT is best applied to the context of psychosis is needed. It is possible that the standard CAT model required modification for work with psychosis, similarly to CBT for psychosis (Morrison & Barratt, 2010). This study aimed to explore how CAT should be applied to psychosis using a Delphi methodology.

Here, we view psychosis as an umbrella term capturing a variety of particular experiences (e.g., hearing voices, delusional thinking) which themselves exist on continua (e.g., van Os, Linscott, Myin‐Germeys, Delespaul, & Krabbendam, 2009). Within this frame, experiences associated with bipolar disorder could also be considered. We avoided defining psychosis in terms of particular diagnostic groups (e.g., schizophrenia) as this does not reflect the way psychotic experiences would typically be understood within the CAT model. Psychosis can present a number of potential challenges for therapists. Co‐occurring difficulties with substance use, mood disorder and self‐injury (Buckley, Miller, Lehrer, & Castle, 2009; Palmer, Pankratz, & Bostwick, 2005; Taylor, Hutton, & Wood, 2015) are not uncommon, and therapeutic engagement can be low in some settings (Doyle et al., 2014; Falchi, 2007). Individuals may present with complex histories (Falchi, 2007) including experiences of trauma (Varese et al., 2012), some of which may relate to psychiatric treatment and intervention itself (Tarrier, Khan, Cater, & Picken, 2007). Whilst CAT may be well suited in some respects to managing such complexity (Taylor, Perry, et al., 2015), it may also require modification to best suit this population and service context.

The aim of this study was to identify areas of consensus amongst clinicians regarding the key features of CAT when working with those struggling with psychosis, and what, if any, modifications are required, using a Delphi methodology. This approach has been used effectively in similar contexts, such as establishing the components of CBT for psychosis (Morrison & Barratt, 2010), and provides a useful starting point in considering how a therapy might be adapted to a new context.

Method

Delphi method

Within this study, we adopted a Delphi approach (Hsu & Sandford, 2007; Norcorss, Hedges, & Prochaska, 2003) to establish a consensus between individuals with expertise concerning the key elements of CAT for psychosis. The Delphi approach is based on the idea that consensus judgements between panel members tend to outperform individual expert opinion (Clayton, 1997). Sample sizes in Delphi studies depend on the area and available pool of experts, but samples of 10–15 are suitable for specialist areas of inquiry (Day & Bobeva, 2005; Godfrey, Haddock, Fisher, & Lund, 2006; Tersine & Riggs, 1976). The Delphi method traditionally involves three rounds (although sometimes a fourth round is included; Hsu & Sandford, 2007): an initial item generation stage following by two further rounds where participants rate items (e.g., rate their agreement; Linstone & Turoff, 1975). We revised this approach by having the initial item generation stage take place during a workshop, as opposed to some remote medium (e.g., mail, email). This approach was adopted as it meant initial items could emerge through face‐to‐face discussion so that the process of consensus formation was already taking place. The potential risks posed by the lack of anonymity in this first stage (e.g., some individuals dominating the process) were offset by the anonymity imposed in the subsequent rounds of the study. Ethical approval for this study was obtained.

Participants

Item generation workshop

For the item generation, workshop clinicians were invited who (1) had experience of undertaking CAT or CAT‐informed work, or had a working knowledge of CAT; and (2) had experience of working therapeutically with individuals with psychosis. Invitations were distributed nationally via CAT mailing lists, local therapy services, to corresponding authors of relevant research, and social media linked to CAT interest groups. Individuals were also encouraged to share information about the event. An expert in relational approaches to understanding psychosis and an individual with first‐hand experience of CAT and psychosis were also invited and attended the event.

Attendance forms indicate that 24 individuals attended the 1‐day workshop. The majority (n = 14) were qualified CAT practitioners, the remainder undergoing CAT training (n = 3), undertaking CAT‐informed work CAT (n = 3), or having a working knowledge of CAT but no current use of this approach in their clinical work (n = 3). The final individual was not a clinician but a service‐user with first‐hand experience of psychosis and CAT. All but three of the clinicians (n = 20) reported current clinical work with psychosis, the remainder (n = 3) reporting recent work with individuals with experiences of psychosis (past 6 months).

Consensus development

For the subsequent consensus development phase of the study, the following inclusion criteria were adopted: (1) be a qualified health practitioner (e.g., clinical psychologist, psychiatrist); (2) uses CAT as part of their work with clients; (3) has experience of working therapeutically with individuals with experiences of psychosis (including bipolar disorder and schizophrenia spectrum disorders). Participants were recruited via three methods: (1) clinicians who had published existing work on CAT and psychosis were identified via their publications; (2) clinicians who attended the 1‐day workshop on CAT and psychosis (see above), and who have agreed to being re‐contacted regarding future work in this area; (3) clinicians who were on a mailing list maintained by the first author concerning an interest in CAT and psychosis. Invitation emails provided a link to the Delphi survey.

Fourteen participants responded to the first round of the online survey (M age  = 43.67 years; SD = 7.74; three males). The majority (n = 12; 85.71%) were clinical psychologists, the remainder being a forensic psychologist and a social worker. Eight participants had completed the post‐qualification CAT training, the remainder either currently undertaking post‐qualification CAT training (n = 6). All participants reported working with individuals with psychosis as part of their current clinical work, except for one participant who reported past experience of clinical work with psychosis. As the latter individual was an expert in CAT who had published research on this subject, it was deemed appropriate to include them within the study. Of these fourteen, 12 agreed to participate in the second round of the survey (M age  = 44.20 years; SD = 8.43; three male). Two clinical psychologists did not continue with the study (one CAT trained and one currently undergoing post‐qualification training).

Procedure

Item generation

The first stage of the Delphi study involved the initial generation of items concerning the use of CAT in the context of psychosis. These items were generated at a 1‐day consensus‐forming workshop taking place in January 2015. During the workshop, attendees were invited to discuss in pairs the most important or valuable aspects of CAT when working with psychosis, and what (if any) adaptions to the traditional CAT model should be considered (see Appendix B for full questions). These starting questions were kept broad to allow a wide range of ideas to emerge, rather than directing the conversations towards specific ideas. Attendees were asked to record responses on sticky notes, one response per page, as their discussion progressed. Following this activity in pairs, a group discussion was then facilitated by the first author. Within this discussion, one of the notes was obtained by the facilitator who then asked for other notes that related to this theme (e.g., flexibility about session length). Additional notes and comments related to this theme were then gathered. This process was repeated with a different note until multiple themes had developed. The themes were fed back to the group, and further discussion was undertaken to develop the ideas in the theme. Thirty themes were initially formed through this process.

These were reviewed by the research team. Three themes encompassed several specific suggestions and so were split into multiple items. For example, a general theme around flexibility in delivering sessions was divided into three separate items (concerning session number, length, and pacing). A further six themes generated at this stage were excluded from the study, either because they were overly broad (e.g., theme concerning importance of the zone of proximal development in CAT; ZPD; see Table 1), they overlapped with other themes (e.g., themes about how psychosis was represented in SDR) or were not clearly specific CAT (e.g., theme around consent and confidentiality reflecting good clinical practice more generally). The remaining themes formed the initial items used in the Dephi study proper. Due to the interactive nature of this process, it was not possible to attribute items to any particular individual.

Due to time constraints, it was not possible to include every note or idea within the discussions on the day. Therefore, all the notes completed by attendees at the workshop were reviewed by the research team (n = 90) to identify any further, specific suggestions that were not captured by existing items. This led to a further 18 themes being identified. In total, 47 items were developed through this process.

A Delphi methodology was used to form a consensus regarding the identified items. An electronic survey was created, with each item framed as a separate statement concerning the use of CAT within the context of psychosis. Items were rated on a five‐point scale (1 = ‘Essential’, 2 = ‘Desirable’, 3 = ‘Not so important’, 4 = ‘Unimportant’, 5 = ‘Detrimental’) concerning their importance (e.g., the need for more sessions than with traditional CAT when working within the context of psychosis). Some items (e.g., therapists should aim to use the same therapy space consistently throughout therapy) had a different response format based on agreement with the statement (1 = ‘Strongly agree’, 2 = ‘Agree’, 3 = ‘Disagree’, 4 = ‘Strongly Disagree’). The option of providing typed comments to support one's rating was also available. The level of agreement was calculated for each item. Items where there was either (1) a lack of agreement between respondents or (2) there was a wide range in scores were retained for a second round of the survey. In the second round, participants were provided with the responses, in an anonymous format, from the first round (percentage endorsing each response option for each item and all written comments for each item) and asked to re‐complete the survey with regard to the retained items.

An a priori criterion of ≥75% agreement in responses for a particular theme was taken to indicate consensus, based on prior research (Hsu & Sandford, 2007; Kizawa et al., 2012). A consensus supporting an item was deemed present if ≥75% of participants rated an item as either ‘Essential’ or ‘Desirable’, whilst a consensus against an item was present if ≥75% rated an item as ‘Not so important’, ‘Unimportant’, or ‘Detrimental’. For themes where the level of agreement, as opposed to the importance, was rated, ≥75% needed to rate the item as ‘Strongly agree’ or ‘Agree’ for a consensus supporting this notion, or ≥75% needed to rate the item as ‘Strongly disagree’ or ‘Disagree’ for a consensus against this item. In addition to looking at the level of agreement, items where there was a wide range in responses, operationalized as a range of ≥3, were retained for the second round of the survey. This range is equivalent to responses of both ‘Desirable’ and ‘Detrimental’ being given by different participants as responses to a particular item, or ‘Strongly agree’ and ‘Strongly disagree’. It was judged that such opposing views required further consideration.

Qualitative comments

Any open‐ended typed comments arising from the two consensus development rounds were subjected to a basic thematic analysis (Braun & Clarke, 2006) undertaken by the first author (PJT). All comments for a particular item were reviewed and codes added to represent different underlying ideas. These codes were then reviewed and commonalities identified, leading to the formation of basic themes. In line with the goals of the Delphi methodology, identification of a theme required a number of participants to suggest a common or related idea. These themes were then drawn upon to provide additional context surrounding the quantitative ratings.

Results

In Table 2, we present the items where a consensus endorsing or supporting the item was present (i.e., responses of ‘Essential’, ‘Desirable’, ‘Strongly agree’ or ‘agree’). In Table 3, we then present the items where there was either a consensus against the item (i.e., responses of ‘Not so Important’, ‘Unimportant’, ‘Detrimental’, ‘Strongly disagree’, or ‘disagree’). The items have been organized under superordinate headings. Table 4 presents a summary of qualitative themes.

Table 2

Items with a consensus endorsing the item (responses of ‘Essential’, ‘Desirable’, ‘Strongly agree’ or ‘Agree’) with associated ratings

Items	Percentage of ratings endorsing the item
	Round 1	Round 2
Treatment frame and flexibility
1. The number of sessions may need to be extended beyond the traditional 16 sessions	92.86	–
2. The pacing of sessions may need to be slower, with more time spent on each component (reformulation, recognition, revision)	85.71	–
3. Sessions may need to be shorter to accommodate those with cognitive impairments or struggling with medication side effects	78.57	–
4. It is acceptable to use elements of CAT rather than the full CAT model with clients with experiences of psychosis	92.86	–
5. Therapists should aim to use the same therapy space consistently throughout therapy	100	–
6. An assertive outreach approach is helpful in managing engaging, missed sessions and where sessions take place	92.86	–
CAT tools
7. A narrative reformulation letter is used	92.86	–
8. Reformulation letter may come later on in therapy, as opposed to being introduced around session four	92.86a	91.67
9. Symptoms (e.g., hearing voices, delusional thinking, mood swings) are explicitly featured on the map or SDR	92.86	–
10. Maps or SDRs should be used to help clients understand the functionality of particular experiences or symptoms (e.g., paranoia as a defence against low self‐worth)	92.86	–
11. Initially mapping out client's experiences is helpful in building engagement early in therapy	100	–
12. The use of memory aids or audio recordings of sessions or letters should be considered	71.43	75.00
13. Therapists should typically aim to work towards a single map or SDR encompassing the client's difficulties and exits	92.86	–
Methods and process within therapy
14. Metaphors and imagery are used to help individual's make sense of their experiences	78.57	–
15. Creative strategies (e.g., using art or visual mediums) are used to help individual's make sense of their experiences	78.57	–
16. It is important the therapist helps the client to feel heard through the use of narrative reflections during therapy	100	–
17. Experiences are normalized by linking these to an individual's past experiences or wider context	100	–
18. A non‐blaming approach whereby symptoms are not seen as rooted within the individual is emphasized	100	–
19. A joint narrative is developed with the client with the purpose of forming a shared understanding the client's experiences and bringing coherence to these	92.86	–
20. Triggers, warning signs, and cycles related to the psychosis (or bipolar disorder) are mapped out with the individual as part of therapy	100	–
21. Therapy should focus on what the client brings and initially work with their understanding of their experiences and difficulties	100	–
22. The target of the therapy is the distress an individual experiences rather than the psychotic symptoms themselves	85.71	–
23. The experience of hearing voices is understood from a relational perspective (e.g., self–self and other–self relating) and may mirror past relationships	100	–
24. Unusual beliefs are understood in terms of underlying reciprocal roles and related procedures	92.86	–
25. Therapists should work towards clients understanding their experiences and difficulties in terms of Reciprocal Roles and associated procedures	92.86	–
26. Therapists need to be mindful not to ‘over‐psychologize’ and assume every symptom or experience can be linked to a particular enactment or Reciprocal Role	92.86	–
27. In some cases, it is more appropriate to adopt a ‘here & now’ focus and focus less on past history	85.71	–
Therapeutic relationships
28. Therapists need to have an awareness of whether they are being drawn into particular dynamics (or reciprocal roles) with clients, related to their difficulties	100	–
29. The therapeutic relationship with the client should be at the centre of the therapy	100	–
30. Psychotic experiences including voices and paranoia can impact upon the therapeutic relationship	100	–
Considering broader systems
31. Reformulation is shared with the clinical team (or Multidisciplinary Team)	92.86	–
32. The therapy refers to the wider social, cultural, and political context surrounding the client	92.86	–
33. Therapists need to have an awareness of, and work with, problematic dynamics between clients and the broader systems they inhabit (e.g., other clinical staff, staff teams, family)	100	–
Contra‐indications and challenges
34. In some stages or phases of psychosis and bipolar disorder, CAT is not helpful and cannot be readily applied	71.43a	75.00
35. Therapists need to be mindful of how previous involvement with psychiatric services has impacted on the client	100	–
36. Therapists need to be mindful of how medication side effects can impact on the client	92.86	–
Exits
37. Dialectical behaviour therapy (DBT) strategies can be used as exits, particularly in helping individuals to regulate their emotions	85.71	–
38. Cognitive behavioural therapy (CBT) coping techniques (e.g., distraction) can be used in helping with experiences such as voices	92.86	–
39. Mindfulness can be used as an exit	85.71	–
40. Compassionate mind‐based approaches can provide exits, particularly when working with persecutory voices or relating	92.86	–

Item retained for second round as range of response in first round ≥3.

Table 3

Items with a consensus against the theme (responses of ‘Not so Important’, ‘Unimportant’, ‘Detrimental’, ‘Strongly disagree’, or ‘Disagree’) or no consensus reached with associated ratings

Items	Percentage of ratings against the item
	Round 1	Round 2
Treatment frame and flexibility
1. Psychoeducation about what psychosis (or bipolar disorder) is and what it involves is introduced prior to starting CAT proper	57.14	83.33
2. Breaks from therapy are introduced	71.43	83.33
3. Flexibility is required around when therapy ends as this may come earlier than planned/expected	42.86	58.33
CAT tools
4. The reformulation letter may be simpler or shorter for clients with experiences of psychosis	35.71	50.00
5. The psychotherapy file should be used within therapy	78.57	–
Methods and process within therapy
6. Initially, a therapist might be cautious around introducing the concept of Reciprocal Roles and may start by focussing more on current or core states, or target problem procedures	42.86	41.67
Contra‐indications and challenges
7. If a positive therapeutic relationship does not seem to have developed in the first few sessions of therapy a therapist should consider ending the therapy	50.00	83.33

Table 4

Summary of qualitative themes

Item	Qualitative theme
Treatment frame and flexibility
The number of sessions may need to be extended beyond the traditional 16 sessions	Depends on client‐related factors
The pacing of sessions may need to be slower, with more time spent on each component (reformulation, recognition, revision)	Depends on client‐related factors
	Sessions may need to be longer earlier in therapy
Sessions may need to be shorter to accommodate those with cognitive impairments or struggling with medication side effects	If needed or required for some individuals
Psychoeducation about what psychosis (or bipolar disorder) is and what it involves is introduced prior to starting CAT proper	Depends on client‐related factors
	Psycho‐education can be incorporated into CAT (e.g., as part of reformulation)
Breaks from therapy are introduced	If needed or required for some individuals
	Depends on client‐related factors
	Important to maintain momentum in therapy (Breaks a threat to this)
	This theme is not specific to CAT
Flexibility is required around when therapy ends as this may come earlier than planned/expected	Endings are important in CAT
	Important to have a ‘good’ or planned ending to therapy
CAT tools
A narrative reformulation letter is used	Only if this is tolerable to the client
Reformulation letter may come later on in therapy, as opposed to being introduced around session four	May be as late as session six or eight
	Should not be left too late
	Depends on client's readiness for letter
	May come later if more time needed for engagement
The use of memory aids or audio recordings of sessions or letters should be considered	Depends on client‐related factors
The reformulation letter may be simpler or shorter for clients with experiences of psychosis	Issue is not specific to psychosis
	Depends on client‐related factors
The psychotherapy file should be used within therapy	Helpful to use aspects of the psychotherapy file
Methods and process within therapy
Experiences are normalized by linking these to an individual's past experiences or wider context	Depends on client's ZPD or tolerance
The target of the therapy is the distress an individual experiences rather than the psychotic symptoms themselves	Can work with both the distress and the symptoms/experiences as both are related anyway
Therapists should work towards clients understanding their experiences and difficulties in terms of Reciprocal Roles and associated procedures	Depends on client‐related factors/ZPD
Initially a therapist might be cautious around introducing the concept of Reciprocal Roles and may start by focussing more on current or core states, or target problem procedures	Depends on client‐related factors/ZPD
Considering broader systems
Reformulation is shared with the clinical team (or Multidisciplinary Team)	Depends to an extent on the nature of the team or service
	Client consent important here
Contra‐indications and challenges
In some stages or phases of psychosis and bipolar disorder, CAT is not helpful and cannot be readily applied	CAT still applicable to some extent in these situations/CAT‐informed work still possible
	CAT‐informed consultation or indirect working is an option still
If a positive therapeutic relationship does not seem to have developed in the first few sessions of therapy a therapist should consider ending the therapy	Working with such problems in the therapeutic relationship is the focus of CAT
	More time might be needed to build appositive relationship
	Client's choice is important here

Items not included where no related themes emerged. ZPD = zone of proximal development.

Treatment frame and flexibility

Results supported the notion of added flexibility when applying CAT for psychosis, for example around session number, length (e.g., longer sessions recommended earlier in therapy), and pacing. Using elements of CAT rather than the full model was also seen as acceptable for this population. Participants suggested that such flexibility was dependent upon the needs of the client (‘depends on complexity/stability of person’). However, the introduction of breaks in therapy was seen as unnecessary, suggesting this aspect of the model may be unchanged when working with experiences of psychosis. Participants recognized the potential risks of therapy becoming disjointed as a result of such breaks (‘breaks generally not desirable/keep momentum of therapy’). There was no consensus around the importance of allowing flexibility regarding the end of therapy. Comments highlighted the importance of endings, especially of having a planned ending, within CAT for psychosis (‘endings should always be attended to in CAT’, ‘would work towards planned ending wherever possible’). No consensus was also reached around the introduction of psychoeducation about psychosis at the start of therapeutic work. It was noted that psychoeducation could be incorporated into CAT, as opposed to occurring pre‐therapy, for example forming part of the reformulation (‘CAT formulation may be an ideal opportunity to help to develop a psychologically informed understanding of psychotic experience’).

There was 100% agreement on the need to make consistent use of a single therapy space but also agreement on the value of an assertive outreach approach (including around where sessions take place). There is clearly a tension between these two positions, and this perhaps reflects a conflict between encouraging engagement with a population where this can be a challenge (Falchi, 2007), and maintaining a consistent treatment frame.

In summary, whilst some aspects of the CAT treatment frame were viewed as being flexible when working with those with psychosis, others were seen as less adaptable and possibly more central to the model.

CAT tools

It was suggested that experiences of psychosis might be explicitly captured within CAT on the SDR. This tool can also be used to highlight how such symptoms may serve a function or provide a means of coping. Mapping clients’ experiences in this way was seen as a means of building engagement. It was agreed that a narrative reformulation letter was still important in this population, as long as this was judged to be tolerable to the client (‘not if overwhelming – diagrammatic representation(s) may be more digestible’), and that a single SDR, encompassing client's difficulties, was preferable to using multiple SDRs. Participants noted the added simplicity or clarity of a single SDR. No consensus was reached that the reformulation letter requires simplifying and this may depend on the client (e.g., ZPD; see Table 1) rather than being a decision specific to psychosis (‘depends on the person not diagnosis…I don't see this as something specific to psychosis’).

The psychotherapy file (see Table 1) was not viewed as important or as relevant for those with psychosis, although some sections were still seen as helpful (‘sometimes I might only use the states section’). It was also agreed that for those with psychosis, the reformulation letter may be introduced later, suggesting that greater time spent on engagement may sometimes be needed for this population (‘exception might be if earlier sessions have been focused on engaging the client’). However, comments also echoed the need to not leave this letter too late (‘yes but don't leave it too long otherwise much less useful’). Participants agreed that using memory aids (e.g., audio recordings) could be helpful.

Methods and process within therapy

There was consensus that various non‐specific aspects of the CAT approach (allowing clients to feel heard, normalising their experiences, adopting a non‐blaming approach, initially working with client's own understandings of their experiences and developing a joint narrative of experiences) were also important when working with psychosis. Results indicated that psychotic experiences including voices and delusions should be understood in terms of relational patterns and procedures. Participants suggested that trigger and warning signs related to psychotic experiences should form part of the mapping of clients’ experiences within therapy. However, there was agreement that therapists should not to try to fit every experience or symptom to the CAT model and also that a more ‘here & now approach’ could be useful in some cases. Participants agreed that the target of therapy should be clients’ distress as opposed to specific psychotic symptoms. There was agreement around the importance of using metaphors, imagery, and creative strategies, and developing a joint narrative of clients’ difficulties. Across these items, participants often stated that these answers still depended on the client and their ZPD (‘within limits of ZPD & what is manageable for client’). No consensus was reached for the suggestion that therapists may initially be cautious about introducing the concept of reciprocal roles for those with psychosis, and several commented that this, again, depended on the client and ZPD (‘therapist has to judge what the client can work with and stay within the zone of proximal development’).

Therapeutic relationships

Consistent with the standard CAT model, there was agreement that therapists should be aware of being drawn into enactments and dynamics within sessions (‘one of the most important aspects and what sets CAT apart from some other therapies’) and that the therapeutic relationship should be at the centre of CAT for psychosis. It was agreed that psychotic experiences can impact upon the therapeutic relationship, suggesting a possible complication to therapy in this population.

Considering broader systems

Participants agreed upon the importance of sharing reformulations with clinical teams, dependent on client consent and the nature of the service (‘pending agreement & consent by client’). This is relevant as many UK services that work with individuals struggling with psychosis adopt a multidisciplinary team approach. It was also deemed important to consider the broader social, cultural, and political context and problematic dynamics between the client and these broader systems.

Contra‐indications and challenges

Participants agreed on the importance of being mindful of how psychosis medication side effects and involvement with psychiatric services could impact upon clients and agreed that in some states of psychosis (i.e., more acute, florid), CAT may not be helpful. However, it was suggested that some CAT‐informed work may still be possible (‘it could be helpful as a consultation approach for staff working with the client’), including more indirect or consultancy‐based approaches. Participants disagreed with the idea that if a positive therapeutic relationship does not emerge early on in therapy, a therapist should consider ending the therapy. It was noted that working with ruptures and difficulties in the therapeutic relationship was a major focus of CAT and that developing a positive therapeutic relationship may take time in this population (‘it may take longer to develop a therapeutic relationship and may require creative/flexible approaches’).

Exits

Participants agreed on the importance or value of a variety of different exit strategies drawn from other models when working with experiences of psychosis (e.g., dialectical behaviour therapy).

Discussion

The aim of the current study was to establish a consensus regarding the use of CAT for psychosis, the key elements of this approach, and any modifications that are required. The study allows for separating out what are considered, by experts in the field, as central features for CAT in cases of psychosis and those more peripheral or open to change. The results indicated several areas where clinicians using CAT with individuals with experiences of psychosis, felt flexibility, or adaptation were important. These included numerous aspects of the treatment frame, including therapy length, pacing, and length. Flexibility in the use of some CAT‐specific tools was also suggested, including the reformulation letter coming later, and the psychotherapy file potentially being excluded altogether or just used in part. The need for this added flexibility may be a reflection of the added challenges and complexity of work with individuals struggling with experiences of psychosis. Aspects of the CAT model were deemed less open to change included the use of narrative reformulation letter, the use of a single SDR (as opposed to multiple separate maps), and the introduction of breaks (participants were against this). The results also highlight how clinicians have applied the CAT relational model to understanding experiences like hallucinations and delusions and the capture of these within sessions through CAT tools like the SDR. The inherently interpersonal nature of many psychotic experiences may make then particularly amenable being understood in this way (e.g., Berry et al., 2006; Pérez‐Álvarez et al., 2008). This particular relationship approach to understanding psychological difficulties is the hallmark of CAT, including its application to psychosis (Kerr et al., 2003, 2006; Ryle & Kerr, 2002).

The existing research on CAT for psychosis is limited, comprising a number of small‐scale case series alongside two pilot trials, one including an adapted form of CAT and one not yet published (Evans & Kellett, 2014; Graham & Thavasotby, 1995; Kerr, 2001; Gleeson et al., 2012). This research has not outlined ways in which CAT has been specifically adapted for this population (with the exception of Gleeson and colleagues where CAT was part of a broader multimodal therapy approach for comorbid psychosis and personality disorder) and as such provide limited guidance on whether or how CAT needs adapting for work with psychosis. The current results provide a possible guide for developing a psychosis‐specific CAT approach which can inform future trials of CAT for psychosis. As the existing research has largely concerned feasibility and not efficacy, it is not yet possible to ascertain whether the amendments to CAT suggested in this study lead to improved outcomes. Determining this could be an aim for future research.

It should also be noted that a number of the endorsed items, whilst be rated here specifically in the context of CAT for psychosis, reflect good clinical practice across therapies and problems. For example, allowing clients to feel heard or adopting a collaborative approach is also seen as important in CBT (Roth & Pilling, 2007). Therefore, CAT shares many aspects of good clinical practice with other therapies. The results also often reflected what would be considered common practice in CAT (e.g., focus on therapeutic relationships) rather than something specific to psychosis. This is similar to what was seen in a Delphi study concerning CBT for psychosis (Morrison & Barratt, 2010) and suggests that the similarities between using CAT with psychosis and using CAT with other difficulties appear to outweigh the differences. This could be taken as evidence against the idea that more specialized training in the use of CAT for psychosis, beyond the standard CAT post‐qualification training, is needed.

In line with the relational model at the heart of CAT, there was consensus on the importance of considering broader systems and how the wider social context interacts with clients’ difficulties. The suggestion of CAT‐informed indirect or consultancy‐based work also emerged when participants considered the application of CAT to more extreme or florid states of psychosis. This is consistent with recent work evaluating CAT‐informed consultancy approaches for those with experiences of psychosis (Kellett, Wilbram, Davis, & Hardy, 2014) and shows that CAT of psychosis need not be limited to one‐to‐one therapy.

Participants’ comments emphasized how clinical decisions with CAT depended on the individual client (i.e., their needs, experiences, and what they can tolerate). Despite this, a consensus was reached for most items, suggesting that whilst variability in practice across clients should be acknowledged, it is also possible to identify a general ‘direction’ or archetype in terms of how CAT is used in psychosis. The concept of the ZPD, drawn from social developmental theory and simply defined as the distance between what client and therapist may achieve together and what the client may achieve alone (Leiman & Stiles, 2010), was often cited in this context. This observation highlights how the use of the ZPD within the CAT model provides a valuable framework for enabling flexibility in line with clients’ needs.

Several limitations require note. The use of CAT within the context of psychosis, whilst a developing area of interest, is still not common practice for mental health services. This limited the initial pool of eligible individuals who could take part in the study. This also meant that procedures (e.g., peer nomination; Hsu & Sandford, 2007) used to select the most expert of qualified individuals for the panel could not readily be applied in this. Nonetheless, the obtained samples had face validity in terms of knowledge of CAT and clinical experience of psychosis. The initial item generation workshop included individuals who undertook CAT‐informed work and might therefore have less working knowledge of CAT than others. However, the goal of this initial stage was to develop a broad array of preliminary items for use in the Delphi study proper. The subsequent steps of the Delphi then provide a check of the relevance of these initial items. As a result, the inclusion of ‘CAT‐informed’ practitioners in the item generation stage is unlikely to produce any bias in the final results. The sample for rounds two and three were predominantly clinical psychologists. Moreover, only UK‐based participants took part. This is beneficial in that more homogenous samples are recommended for Delphi studies as they produce more stable results (Akins, Tolson, & Cole, 2005; Tersine & Riggs, 1976). However, it may mean that certain professional groups that make use of CAT (e.g., psychiatry; social workers) and clinicians from other countries were under‐represented or not represented at all, and differing views may have emerged if these groups had been present to a greater extent.

The current results help to separate out what is seen as more core to CAT for psychosis from aspects that seem more peripheral and open to flexibility and change. This information may be helpful to clinicians who use CAT with psychosis to reflect on their practice. We are also aware that many clinicians that do not use CAT still draw on CAT concepts, tools, and principals in their work. The current results may therefore also be helpful to such individuals in indicating which features of CAT are deemed more useful or important when working with psychosis. The current results are also helpful in providing a tighter definition around the core features of CAT for psychosis for the purposes of future research studies, where a level of consistency between therapists is needed.