What Are the Components of CBT for Psychosis? A Delphi Study

Abstract

There is strong evidence supporting the implementation of cognitive behavior therapy (CBT) for people with psychosis. However, there are a variety of approaches to the delivery and conceptual underpinnings within different research groups, and the degree of consensus or disagreement regarding what are the intrinsic components has not been explored. This study uses the Delphi method to try to establish what a group of experts in CBT for psychosis view as important. Experts were invited to participate in 3 rounds of producing and rating statements that addressed areas such as principles, assessment, models, formulation, change strategies, homework, and therapists’ assumptions in order to consolidate consensus of opinion. Seventy-seven items were endorsed as important or essential for CBT for psychosis by >80% of the panel. These recommendations should ensure greater fidelity in clinical practice, allow greater evaluation of adherence within clinical trials, facilitate the development of competency frameworks, and be of value in relation to training and dissemination of CBT for psychosis.

Introduction

In recent years, the generic cognitive model¹ has been applied to our understanding and treatment of psychosis. This model suggests that the way that we interpret events will have consequences for how we feel and behave and that such interpretations are often maintained by unhelpful thinking biases and behavioral responses. It also suggests that these interpretations are influenced by our core beliefs, which are formed as a result of life experience. There have been several cognitive models of psychosis and psychotic symptoms or experiences outlined,^2–4 which suggest that it is the way that people interpret psychotic phenomena that account for distress and disability, rather than the psychotic experiences themselves. There are several comprehensive treatment manuals that describe the application of such models in greater detail.^5–9

Recent meta-analyses have concluded that cognitive behavioral treatment of psychosis is effective, demonstrating acceptable effect sizes,^10–12 and that it should be delivered routinely as part of the treatment package that is offered to people with a diagnosis of schizophrenia.¹³ As a result of the wealth of evidence in support of cognitive behavior therapy (CBT) for psychosis, recent guidelines in several countries have suggested that it should be routinely provided to people with psychosis/schizophrenia.^14,15

This consensus regarding the desirability of providing CBT for psychosis has resulted in challenges regarding dissemination of the required skills, as well as some debate regarding which components are viewed as essential. In order to be confident that people with psychosis are receiving appropriate CBT, the content and delivery need clarification, and there should be reliable, standardized, and competency-based ways of operationalizing such interventions. By doing so, this would also ensure that CBT for psychosis is comparable with the competency framework guidelines the Department of Health recommends for CBT with other disorders.¹⁶

Debates concerning what elements truly comprise CBT for psychosis have rightfully arisen, with different research groups and treatment manuals placing different emphasis on aspects such as engagement, formulation, problems and goals, homework, active change strategies, and relapse prevention. There is also some debate regarding whether CBT for psychosis is identical to, shares similarities with, or is something quite distinct from CBT for other disorders. Furthermore, experts in this field have also formed their own views on what elements are essential and should comprise CBT for psychosis, as evidenced by the various adherence measures and competency scales that have been developed and utilized within treatment trials (eg, Cognitive Therapy Scale [CTS],¹⁷ Cognitive Therapy for At Risk of Psychosis Adherence Scale [CTARPAS],¹⁸ Revised Cognitive Therapy Scale [CTS-R],¹⁹ Cognitive Therapy for Psychosis Adherence Scale [CTPAS],²⁰ Cognitive Therapy Scale for Psychosis²¹). The extent to which experts agree or disagree has yet to be explored, so this study aims to investigate, using a standardized methodology, whether a consensus regarding the components of CBT for psychosis does exist. The present study used a technique known as the Delphi process²² to elicit and quantify the opinions of a group of expert CBT clinicians working in United Kingdom. The “Delphi method” is a systematic, interactive method that relies on a panel of independent experts answering questionnaires in 2 or more rounds, with feedback from each round provided to help achieve consensus; the process is stopped after a predefined stop criterion is reached, such as number of rounds. Similar Delphi exercises have been used to clarify the concept of relapse in schizophrenia,²³ identify the key components of schizophrenia care,²⁴ delineate the practice model of an early intervention for psychosis service,²⁵ and to identify mental health first aid strategies for early psychosis.²⁶ The aim of the study was to determine the extent of expert consensus on the essential principles and structural and functional elements of CBT for psychosis.

Method

Participants

Participants who took part in the Delphi method were experts in the field of CBT for psychosis (ie, clinical psychologists, cognitive therapists, psychiatrists, and mental health professionals) who work or have worked in a research capacity implementing CBT for psychosis. They were identified from a list of clinical academics and trial therapists who have been delegates at the International CBT for Psychosis conferences, have written treatment manuals, and/or who have been trial therapists on randomized control trials (RCTs) of CBT for psychosis. All participants were from the United Kingdom; this is because the original treatment manuals were developed in the United Kingdom, and the majority of clinical trials have also been United Kingdom based. Each participant was invited to take part in the study via e-mail. Initially, 60 participants meeting the above criteria were invited to take part; 28 participants actually responded to the survey (a 47% response rate). The sample appeared representative, demonstrating a good geographical spread (5, Birmingham; 5, London; 4, Manchester; 4, Newcastle; 4, Salford; 2, Glasgow; 1, Lancaster; 1, Norwich; 1, Reading; 1, Southampton), as well as a good representation of different levels of expertise (7 manual authors, 5 principal investigators, 5 trial supervisors, and 11 trial therapists), both of which were proportionately representative of the original invitation list.

Procedure

The Delphi process consisted of 3 stages, which were based on the stages outlined by Langlands et al²⁶; the approach to analysis was also identical to Langlands et al.²⁶ In “stage 1,” any elements that were identified as being pertinent to CBT or CBT for psychosis from the CTPAS, the CTARPAS, the Manual of the CTS-R, and the treatment manuals, together with Department of Health competency framework guidance of Roth and Pilling (2008) were extracted by S.B. and reviewed by A.P.M. Once a final list of relevant statements was compiled (n = 90 statements), the list was sent via e-mail to a panel of experts who were invited to take part in this preliminary phase of the study (these were predominantly the authors of the existing UK treatment manuals; n = 12). Participants who agreed to take part (n = 7 at this phase) were asked to add and respond to the current list of statements. Amendments to the wording of some of the statements and any new items suggested were added to the questionnaire except for duplications. This resulted in 1 statement being removed, and an additional 45 statements were suggested.

In “stage 2,” the finalized questionnaire containing 134 items identified in stage 1 was constructed and formatted to an online web page. Sixty participants were e-mailed a link to the online version and asked to complete the questionnaire by rating the importance of each item, with regard to CBT for psychosis, on a Likert rating scale of 1–5 (1, essential; 2, important; 3, do not know/depends; 4, unimportant; and 5, should not be included). The results from the questionnaire were automatically entered into an anonymous database. Items to be included as a standard were then extracted so that the questionnaire could be amended for use in stage 3. Survey responses were recorded in an anonymous database and analyzed by obtaining group percentages. The following cutoff points were used in relation to inclusion, rerating, and exclusion criteria for the items.

1. If at least 80% or above panel members rated an item as essential or important as an ingredient of CBT for psychosis, it was included as a standard.

2. If 70%–79% of panel members rated an item essential or very important, we asked all panel members to rerate that item.

3. Any statements that did not meet the above 2 conditions were excluded.

This resulted in the inclusion of 69 items as standards; the exclusion of 41 items and 24 items required rerating in phase 3.

In “stage 3,” the 28 participants from phase 2 were sent a questionnaire of the same format as above but incorporating only those items that 70%–79% of panel members had rated as essential or important in the previous stage; we asked all panel members to rerate these items (n = 24 statements). Twenty-three participants responded at this stage, which resulted in an additional 10 statements being included and 14 excluded.

At each stage, 2 reminders were sent if potential participants had not responded to the original invitation.

Results

A sum total of 77 items from the 3 rounds were rated as essential or important by >80% of the panel members, with 2 items achieving consensus as not to be included (>80% rated as unimportant or should not be included) and 55 items being excluded. The authors grouped items of similar content under headings to make the recommendations more comprehensible and have indicated whether these components are viewed as specific to CBT or more generic factors that would be included in many different psychotherapies (see table 1 for those items with consensus for inclusion and table 2 for the items that achieved consensus as should not be included).

Table 1.

Aspects of CBT for Psychosis Included as a Standard Recommendation

Recommended Element of CBT for Psychosis	Generic/Specific	Round Included
Engagement
Interventions should be informed by client feedback	G	2
Normalizing of psychotic symptoms should be used to reduce stigma and improve engagement	G	2
The client should be allowed and encouraged to express positive and negative reactions regarding therapy	G	2
Collaborative feedback should be used to engage the client	G	2
CBT should require consistent collaboration throughout the sessions	G	2
CBT should be implemented using a collaborative approach	G	2
The client should be engaged in the therapeutic relationship	G	2
CBT should take into account the clients’ perspective and “world view”	G	2
Account always needs to be taken of presenting symptomatology, past experiences of services, and cultural/family issues in engagement	G	2
The rationale of CBT should be explained and demonstrated to the client	S	2
Structure and principles
CBT should aim to reduce distress and improve quality of life	G	2
Summaries and feedback should be used to structure the session	G	2
CBT sessions should always be accommodated to the client's needs and speed of learning	G	2
CBT should aim to reduce distress and prevent future distress	G	2
CBT should aim to elicit hope in recovery	G	2
CBT should consult the client regarding the terminology used to explain their experience	G	2
CBT should end in a planned manner and plan for long-term maintenance of gains after treatment	G	2
Session structure and content should be decided jointly between client and therapist	G	2
The client should be given a chance to explain his or her own model first	G	2
The client should make choices and take appropriate responsibility for the CBT sessions	G	2
CBT should assist the maintenance of a client's capacity to make informed decisions about their lives	G	3
The client and therapist should jointly agree a problem list	S	2
Appropriate flexibility needs to be given in constructing agendas, targets, and problem lists according to client's capacities, inclinations, and motivations	S	2
Guided discovery and Socratic questioning should be used to elicit key cognitions/images	S	2
Guided discovery should be used to help the client gain understanding	S	2
Agreed short- and long-term goals should underpin the intervention	S	2
Items on the agenda must be appropriate, clear and discrete, consistent with the formulation, and conceived to take therapy forward	S	2
Goals should be SMART (specific, measurable, achievable, Realistic and Time limited)	S	2
CBT should help the client consider a range of perspectives regarding his/her experience	S	2
CBT for psychosis should be founded upon the principles of evidence-based practice and value-based practice	S	3
CBT should help the client develop hypotheses regarding his/her current situation and to generate potential solutions for him/herself	S	2
The client should be encouraged to prioritize the items on the agenda	S	2
Socratic questioning, diaries, and monitoring procedures should help the client reflect upon and explore new meanings about their thinking, behavior, and context	S	2
Major brief summaries should occur at the beginning and end of each session	S	2
Over the course of therapy a client should work toward becoming their own therapist	S	2
Formulation
A good collaborative relationship must be formed to help develop a comprehensive formulation	G	2
A balanced conceptualization should highlight the client's strengths	G	2
The therapist must avoid overcomplex ‘kitchen sink’ formulation and intervention	G	2
CBT should develop a formulation of the client's difficulties and use psychological mechanisms to understand the processes that are controllable in relapse	G	2
A formulation should be developed and used to outline a treatment plan	G	2
Conceptualization should draw together current concerns, vulnerabilities, strengths, precipitating and perpetuating factors	S	2
A cognitive behavioral maintenance cycle should be devised and used to set targets for intervention	S	2
In order for effective CBT to occur, the conceptualization must be appropriate and shared	S	2
Assessment and model
CBT must identify the needs of the client and competency of the therapist before undertaking in-depth therapeutic work	G	3
CBT should be idiosyncratic to the individual client	S	2
CBT should examine the role that behaviors have in triggering and maintaining the clients difficulties	S	2
CBT should help a client to identify and elicit those thoughts, images, and beliefs that are fundamental to their distress (ie, the key cognitions)	S	2
CBT should elicit any behavioral features that contribute to the maintenance of the client's problems	S	2
CBT ought to elicit and examine behavioral patterns such as “safety seeking behaviors” in relation to the relevant emotions associated with them	S	2
CBT should elicit and assess the intensity of emotions associated with a particular situation or cognition	S	3
The therapist should make a full mental health assessment with attention to key cognitive, behavioral, and emotional issues	S	3
The role of safety-seeking behaviors should be demonstrated	S	2
CBT should identify emotional issues that interfere with effective change (eg, hostility, anxiety, excessive anger)	S	2
Homework
“Homework” should be a “standing” item on the agenda	S	2
Homework assignments ought to act as a bridge between therapy and the real world	S	3
Practical plans (ie, practical homework) should be developed with the client to facilitate effective change	S	3
Practice assignments (“homework”) should be planned and reviewed	S	2
Clients should be provided with an organic summary of the main ‘findings’ from the therapy to review at home	S	2
If clients fail to complete agreed between-session activities, the therapist should take at least equal responsibility for this happening and renegotiate	S	2
Change strategies
Therapists should use elements of self-disclosure to help normalize clients’ psychotic symptoms	G	2
Beliefs in omniscience and omnipotence of command hallucinations should be explored and debated	S	2
The therapist should work directly with content of voices to explore its relationship to life experiences and beliefs about the self	S	2
CBT should help a client modify core beliefs/schemas and associated behavior	S	2
Beliefs about hallucinatory mechanisms should be systematically explored and normalized, eg, reasons for external attribution and individual nature of experience	S	2
Logging pros and cons should be used to explore voices	S	2
Self-report measures and self-monitoring ought to guide therapy and monitor outcome	S	2
The client should learn to monitor features of his/her experience and gain further insight through the data-gathering process	S	3
With CBT, the client should be supported to explore alternative explanations of experiences that may be more adaptive and less distressing	S	2
CBT should identify and work with safety behaviors	S	2
Experiments should be devised to test beliefs by modifying safety behaviors	S	2
CBT should recognize and manage obstacles that a client brings to therapy	S	2
Therapist assumptions
Therapists should believe that many people experience psychotic-like symptoms without feeling distressed by them	G	2
Therapists should have a good understanding of recovery from psychosis	G	2
Therapists ought to believe that delusions can be quite understandable	G	2
Therapists should believe that it is not the hallucination or the delusion per se that is clinically relevant but the amount of distress or disability associated with it	G	2
Therapists ought to believe that hallucinations or thought disorder can happen to anyone if they are very stressed	G	2
Therapists ought to view most symptoms of psychosis as quite common in the normal population	G	3

Note: CBT, cognitive behavior therapy; G, generic; S, specific.

Table 2.

Aspects of CBT for Psychosis With Consensus That Should Not Be Included

Element of CBT for Psychosis Agreed Not to Be Included	Generic/Specific	Round Included
Therapist assumptions
Therapists should believe that clients with psychosis are very different to clients with other mental health difficulties	G	2
Therapists should believe that there is a clear boundary between being mentally unwell and mentally healthy	G	3

Note: CBT, cognitive behavior therapy; G, generic; S, specific.

Discussion

To our knowledge, this is the first study that attempts to achieve consensus regarding the important ingredients of CBT for psychosis. A high degree of consensus was obtained for a wide range of items regarding assessment and model, engagement, structure and principles, formulation, change strategies, homework, and therapist assumptions. These recommendations as to what constitutes CBT for psychosis are distinct from the information that is provided via existing training courses and treatment manuals, in that they have been endorsed by a large number of experts.

It is to be expected that many of the elements that were identified as important/essential were more generic factors that would apply to many psychotherapies (eg, collaboration, need for formulations, a good therapeutic relationship), whereas others were more specific to CBT (eg, specific change strategies, need for a cognitive model). The assumptions and beliefs of cognitive behavior therapists are clearly likely to influence the delivery of CBT. It is interesting that there was consensus regarding attitudes that should be held by therapists as well as consensus regarding attitudes that should not be held. The views that were agreed to be undesirable were both at odds with the normalizing rationale of CBT for psychosis, which was a strongly endorsed aspect.

Many other items were excluded as they appeared to be dependent on a number of factors; this was evident both from comments made to the investigator and from many items achieving a large number of “depends/don't know” responses. This could have been expected as there was a consensus of opinion that CBT for psychosis should be idiosyncratic and that the targets for treatment should be collaboratively negotiated, based on a shared list of problems and goals, and that particular change strategies should be formulation driven. In addition, CBT for people with psychosis often involves a very broad range of problems (unlike CBT for anxiety disorders, eg), often including anxiety, low mood, self-esteem, etc, as well as psychotic experiences; this wide range of possible treatment targets and corresponding treatment strategies pose particular problems for generating a finite number of agreed items for a consensus view. It is also likely that engagement issues and the service user's own perception of the problem and readiness to change would affect the suitability of these items. Therefore, consensus was not established for many items, most of which were related to specific treatment targets (such as “CBT should focus on negative symptoms,” “CBT should reduce symptoms of psychosis,” and “CBT should focus on depression, interpersonal anxiety and unwanted intrusive memories”) and specific change strategies (such as “Thought records should be used to detect, examine and help the client reality test automatic thoughts/images,” “The therapist should suggest alternative explanations for delusional beliefs if the client unable to do so,” and “The advantages and disadvantages of beliefs classified as delusional ought to be explored honestly and collaboratively with a client before attempting further change or intervention”). It is likely, however, that such items would have been included if they had been preceded by the assumptions mentioned above (ie, if they were on the problem list and consistent with the case formulation, etc). It is also likely that some further structuring of the list (eg, into “mini-lists” that are specific to certain problems, ie, “If voices were viewed as problematic and change was desired …”) might result in greater consensus for these items and could result in consensus statements for each specific problem (eg, hearing voices, paranoid thinking, delusions of interference, flat affect, etc), which would be in keeping with the single-symptom approach that has advanced our understanding of the mechanisms involved in the development and maintenance of such difficulties.²⁷

There are several limitations to this study. The definition of expert (as being an author of a treatment manual or having been a therapist/supervisor on an RCT of CBT for psychosis) seemed to have faced validity and be relatively conservative, but it may have excluded many clinical experts working within routine services. The response rate of 47% was lower than hoped and is an obvious weakness, but the sample of responders appeared representative, with a similar profile in terms of geography and seniority/experience as the initially invited sample. The use of “depends/don't know” as an anchor made the responses to some items difficult to interpret; either an alternative anchor or different wording of some of the items may have allowed consensus to be achieved regarding more idiosyncratic or problem-specific aspects. Alternatively, it is possible that some items may require a hierarchical or clustering approach similar to a decision tree, where certain aspects would apply only if certain prerequisites were met (eg, if something was on the problem list or indicated within the formulation). This could result in a similar framework to the competency framework developed in relation to anxiety disorders by Roth and Pilling, which involves meta-competencies, generic competencies, and problem-specific competencies. It should also be noted that this study identifies what experts think are intrinsic elements of CBT for psychosis but is clearly not concerned with what constitutes the essential components in terms of efficacy, which is a different study altogether. It is also likely that the profile of agreed aspects of CBT for psychosis may differ between countries, so the exclusively UK-based sample is also a limitation. Finally, the decision regarding classification of items as generic or specific to CBT was a subjective one taken by the authors; different results may have been achieved if this question had been put to the panel.

There are many implications of these results. They may prove useful to professionals responsible for provision or coordination of training in CBT for psychosis. It could be that they form the underpinnings of a competency framework. They could also be used to inform the evaluation of adherence and fidelity to CBT for psychosis in both clinical trials and services and could help to form the basis of an adherence measure that would be based on wider consensus than that currently in existence. If this information is shared with service users and carers, then this may help them to determine whether therapy that they are receiving would be recognizable as CBT and could potentially be quite empowering. Similarly, audit against these recommended elements could provide useful information to service providers and commissioners. The categories of elements and the distinction between specific and generic elements could help with the planning and delivery of CBT for psychosis training and supervision and lead to greater standardization, which could promote more equitable service delivery in relation to current differences that may be present due to geography, professional background, resources and funding, etc. Finally, the identification of assumptions that are agreed to be important may also have implications for the overall ethos and culture of services for people with psychosis, as well as for the selection of staff for training.

There is clearly a need for further research to examine several issues; perhaps the most important of these is whether these components are related to outcome, which could potentially be examined if a measure of adherence or competence that incorporated these factors were related to clinical and social outcomes across a number of trials. Our study might serve as the basis for the development of a more formal competency framework that then could be evaluated. Other research could examine the generalizability of these results to other countries, investigate the importance of therapist assumptions, and determine whether there are important differences according to geography, professional background, and service culture or organization.