Misconception 1: CBTp doesn’t work

The UK NICE guideline (2002; 2009; 2014) recommends CBTp be offered to everyone with psychosis, regardless of the phase of illness. Despite meta-analyses supporting this recommendation, and a recent umbrella review (Solmi et al., Reference Solmi, Croatto, Piva, Rosson, Fusar-Poli, Rubio, Carvalho, Vieta, Arango, DeTore, Eberlin, Mueser and Correll2023), there is a persistent distrust of the guidance and a perception that the evidence has been overstated (e.g. Jauhar et al., Reference Jauhar, McKenna, Radua, Fung, Salvador and Laws2014). This scepticism is likely to be due in large part to the variation in effect sizes across meta-analyses conducted over the last two decades. With effects ranging from very small or negligible to moderate, some have interpreted this variability as indicating that the evidence is unreliable or exaggerated. This has been discussed at length (e.g. Birchwood et al., Reference Birchwood, Shiers and Smith2014b; Peters, Reference Peters2014), with proponents of CBTp attributing the range of outcomes to heterogeneity of (1) symptoms that have been grouped together (e.g. hallucinations, delusions, and negative symptoms), (2) therapies (e.g. first generation CBTpFootnote ¹ , avatar therapy, and problem-specific CBTp), and (3) primary outcomes – do we focus on symptom reduction, as is typical in meta-analyses, or associated distress and personal recovery, which are targeted in psychological therapies? Additionally, we’d argue that psychological therapies should be evaluated in addition to medication effects as most people receiving CBTp also take anti-psychotics.

Taking account of these issues yields more nuanced insights. For example, CBTp tends to be more effective for delusions than hallucinations or negative symptoms (Sitko et al., Reference Sitko, Bewick, Owens and Masterson2020), when interventions target mechanisms underlying specific symptoms (e.g. Freeman et al., Reference Freeman, Emsley, Diamond, Collett, Bold and Chadwick2021), and when assessing emotional and behavioural outcomes (e.g. Birchwood et al., Reference Birchwood, Michail, Meaden, Tarrier, Lewis and Wykes2014a). For example, an early trial of CBT for command hallucinations showed that while participants continued to hear voices, compliance reduced considerably; at 18 months, just 28% of the CBTp group still fully complied with their voices, compared with 46% of the control group (OR 0.45, 95% CI 0.23–0.88, p=0.021). In further support of the cognitive model, this key behaviour change was mediated by a reduction in strength of belief about the power of the voice (Birchwood et al., Reference Birchwood, Michail, Meaden, Tarrier, Lewis and Wykes2014a).

A meta-analyses completed in 2020 showed that CBTp was an effective treatment with a small, pooled effect size (e.g. Cohen’s d=0.26 for hallucinations, 0.36 for delusions; Sitko et al., Reference Sitko, Bewick, Owens and Masterson2020), comparable to some anti-psychotics (Huhn et al., Reference Huhn, Nikolakopoulou, Schneider-Thoma, Krause, Samara, Peter, Arndt, Bäckers, Rothe, Cipriani, Davis, Salanti and Leucht2019). While this may seem unimpressive compared with CBT for other presentations such as social anxiety and PTSD which yield large effects (NICE, 2005; 2013), CBTp is still a relatively young intervention and will evolve as our understanding of psychosis deepens. For example, recent research aimed at reducing delusional conviction and distress by targeting specific mechanisms contributing to persecutory delusions yielded large effects at end of treatment (Cohen’s d=0.86 and 1.20, respectively) compared with befriending (Freeman et al., Reference Freeman, Emsley, Diamond, Collett, Bold and Chadwick2021).Footnote ²

Targeting distinct causal mechanisms for different psychotic experiences holds much promise. Developments in the broader psychotherapy outcome research could also improve CBTp treatment effects. Analyses of large datasets reveal heterogenous outcomes in therapies for depression and anxiety based on pre-therapy factors (e.g. Lorenzo-Luaces et al., Reference Lorenzo-Luaces, DeRubeis, van Straten and Tiemens2017). Personalised therapy decisions informed by machine learning could clarify what works best for whom (cf. Paul, Reference Paul1967), and may dramatically change psychotherapy outcome research in the near future.

Misconception 2: CBTp is about ‘thinking positively’ or correcting people’s thinking

CBTp involves working closely with a person to develop an understanding of the cognitive, behavioural, and affective processes contributing to the development and maintenance of their psychosis and linked distress. An individualised formulation forms the basis for reflection on early learning and self-perpetuating cycles of thinking, feeling, and responding that maintain distress and disability. For example, recognising how a derogatory voice triggers and reflects self-critical thoughts and beliefs learnt early in life, which affect mood and willingness to do what’s important to the individual, leading to further self-criticism; or how anxiety drives expectations of threat from others and avoidance, leading to further anxiety when next in the feared situation. Based on an individualised formulation, we can support people to make changes in line with personal goals, through a process of guided discovery (Padesky, Reference Padesky1993; Padesky and Kennerley, Reference Padesky and Kennerley2023); for example, re-evaluating self-critical appraisals, or gradually building confidence in feared situations even when feeling anxious.

In addition to formulating, we seek to understand how an individual knows themselves and the world – their epistemic style (Royce et al., Reference Royce and Smith1964; Royce et al., Reference Royce1978). Moorey (Reference Moorey2024) describes three epistemic styles linked to change in cognitive behavioural therapies – people with a rational/empirical style prioritise logical consistency and objective validity (e.g. evidence for and against), people with a more pragmatic style benefit from considering impact and utility (e.g. ‘does it help you to …?’), and those with a constructivist style benefit from generating alternative perspectives (e.g. ‘how would you like to be in this situation?’). Guided discovery is likely to be most effective when drawing on the person’s preferred epistemic style(s). For example, someone with a rational/empirical style who is fearful of the power of a voice might benefit from testing whether the voice can cause harm (e.g. by gradually postponing compliance). If they had a pragmatic style, we might encourage them to consider the impact of spending hours ruminating on the voice, given their goal to spend more time on their artwork. If they had a more constructivist style, we might support them to develop a coherent narrative of past traumas to help make sense of core self- and other-beliefs reflected in the voice. In practice, we draw on all these approaches; the skilful CBTp therapist recognises people’s epistemic style(s) and adapts therapy accordingly.

The mistaken notion that CBTp is about ‘thinking positively’ or correcting people’s thinking is sometimes advanced by people who have not been trained in CBTp. A more subtle version of this argument sees CBT as paternalistic (doing to), or simplistic (correcting rather than guideing discovery). It is not our role to decide what people should or should not think, or to educate people in ‘correct thought’. Indeed, we are likely to work with many people whose opinions and views we disagree with, and may sometimes find objectionable. Our focus is on the problems for which the person is seeking help, and supporting their goals for personal recovery and autonomy (Hutton and Morrison, Reference Hutton and Morrison2013). Our role is to create the conditions for change Footnote ³ that enable a person to step back and re-evaluate self-defeating appraisals, try out different behaviours, and develop ways to manage difficult emotions. The fallacy of Orwellian ‘thought police’ is also undermined by the fact that changing what are typically over-learnt and emotionally driven responses to voices, paranoia and other psychotic experiences do not simply follow corrective information or encouragement to ‘think positively’.

CBTp is not about thinking in any particular way. It is about creating the conditions for people to make changes that are important to them, if they so choose. Skilful CBTp involves an awareness of individuals’ epistemic styles (and our own) to support therapeutic change in line with the person’s formulation and personal goals for recovery and autonomy.

Misconception 3: CBTp doesn’t recognise the therapeutic relationship (or it’s all about the therapeutic relationship)

The therapeutic relationship is the basis for all CBT; ‘The aspiring cognitive therapist must be, first, a good psychotherapist’ (Beck, Reference Beck1979; p. 22), and the most common mistake made by those new to or unfamiliar with CBT is ‘[s]lighting the therapeutic relationship’ (p. 32). Despite this clear emphasis from the earliest CBT texts, many criticise CBT and therefore CBTp for being mechanistic in delivery and effecting superficial change, and (paradoxically) that any benefit is due solely to relationship factors (cf. Proctor, Reference Proctor2003). The suggestion that CBTp does not recognise the therapeutic relationship is a violation of the core principles and practice of CBT (e.g. Brabban et al., Reference Brabban, Byrne, Longden and Morrison2017). The suggestion that the impact of CBTp is entirely due to the relationship is inconsistent with the evidence (e.g. Freeman et al., Reference Freeman, Emsley, Diamond, Collett, Bold and Chadwick2021).

In CBTp, as in CBT, we articulate and evaluate the therapeutic relationship in terms of interpersonal effectiveness and collaboration (Blackburn et al., Reference Blackburn, James, Milne, Baker, Standart, Garland and Reichelt2001). Following Rogers (Reference Rogers1957), interpersonal effectiveness describes the ability to communicate genuine regard, empathy, and warmth. Collaboration describes the ability to engage a person in the active task of therapeutic discovery and change. If we fail to attend to the therapeutic relationship, we are not delivering CBT. Like any therapy, CBTp done poorly could cause harm (cf. Parry et al., Reference Parry, Crawford and Duggan2016; Proctor, Reference Proctor2008).

In line with the well-established argument for the role of common interpersonal factors across psychological therapies (Rosenzweig, Reference Rosenzweig1936; Wampold, Reference Wampold2001; Wampold and Imel, Reference Wampold and Imel2015), meta-analyses consistently indicate a modest effect of relationship quality on clinical outcomes, irrespective of modality and presenting problem (e.g. Martin et al., Reference Martin, Garske and Davis2000). Meta-analyses also evidence this for psychosis specifically, showing that relationship quality has a small to moderate impact on therapy engagement and outcomes, including in CBTp (Bourke et al., Reference Bourke, Barker and Fornells-Ambrojo2021; Browne et al., Reference Browne, Wright, Berry, Mueser, Cather, Penn and Kurtz2021). Furthermore, in a secondary analysis of a CBTp trial, Goldsmith et al. (Reference Goldsmith, Lewis, Dunn and Bentall2015) showed that the therapeutic relationship is likely to have a causal effect on outcomes; in the context of a strong relationship, more sessions led to better outcomes (β=−2.91), whereas with a poor relationship, more sessions had a detrimental effect (β=7.74).

The consistent finding that the relationship contributes to therapeutic outcomes is particularly important in CBTp given that people with psychosis typically delay accessing services for 1−2 years (Birchwood et al., Reference Birchwood, Connor, Lester, Patterson, Freemantle, Marshall, Fowler, Lewis, Jones, Amos, Everard and Singh2013), and about a third then disengage following initial contact (Doyle et al., Reference Doyle, Turner, Fanning, Brennan, Renwick, Lawlor and Clarke2014; Kreyenbuhl et al., Reference Kreyenbuhl, Nossel and Dixon2009) including from CBTp (though estimates vary, e.g. Peters et al., Reference Peters, Godfrey, Khalil, McInerney, Parker and Soares2015; Richardson et al., Reference Richardson, Dasyam, Courtney, White, Tedbury, Butt and Newman-Taylor2019). The quality of the relationship early in CBTp may therefore determine whether someone remains engaged and receives ‘full dose’ therapy.

In CBTp, we assume that the quality of the relationship is necessary but not sufficient to effect therapeutic change (cf. Beck, Reference Beck2020). Meta-analyses of first generation CBTp indicate small effects compared with (increasingly improved) treatment as usual, and active therapeutic controls such as supportive counselling (Solmi et al., Reference Solmi, Croatto, Piva, Rosson, Fusar-Poli, Rubio, Carvalho, Vieta, Arango, DeTore, Eberlin, Mueser and Correll2023). More recent trials of problem-specific CBTp show large effects compared with befriending (e.g. Freeman et al., Reference Freeman, Emsley, Diamond, Collett, Bold and Chadwick2021). Both supportive counselling and befriending focus primarily on fostering a strong and supportive therapeutic relationship, and the outcomes are not as good as for CBTp. Collectively, the evidence indicates that CBTp is likely to be most beneficial when based on a sound therapeutic relationship, and targeting the psychological mechanisms maintaining specific psychotic experiences.

Misconception 4: CBTp focuses only on psychotic symptoms

Rarely is someone referred for CBTp with a single, discrete problem or diagnosis. Most people have a history of adversity or trauma, multiple co-existing symptoms, and a range of social needs. CBTp often faces criticism for being too narrowly focused to benefit people who have complex presentations, who may not be asking for help with discrete problems such as voices or paranoia, or who are more focused on finding a job or coping with the effects of early childhood abuse.

While CBTp is most developed for voices and paranoia, the assumption that therapy focuses only on these symptoms, disregarding other problems and goals that the individual wishes to address, is incorrect. The misunderstanding that CBTp is unsuitable for people with additional needs discourages services from offering an evidence-based, potentially transformative intervention to people who could benefit from the opportunity to address a range of psychological and social challenges.

During CBTp assessment and formulation, individuals are encouraged to reflect on what is most problematic for them, how they would like things to be different, and what they hope to achieve in therapy. These objectives typically extend beyond symptom reduction to aspects of personal recovery and autonomy, such as improving relationships, being ready to find meaningful employment, and self-acceptance. As part of the assessment, we collaborate with the person to agree personally meaningful therapy goals that align with what CBTp can offer. Addressing these goals typically involves focusing on a range of presenting problems, including psychosis, trauma symptoms, mood and sleep disturbance, social withdrawal, and relationship issues.

Some people prioritise understanding the past or managing fears about the future. Where people want to make sense of complex and confusing experiences, past and present, a developmental formulation can give context and coherence to a complex personal history and ongoing impact on current experience and relationships. For many, past episodes of psychosis and linked treatments have themselves been traumatic, prompting Gumley and Schwannauer (Reference Gumley and Schwannauer2006) to develop a framework to formulate and guide treatment for fear of relapse and linked iatrogenic harms.

‘Negative symptoms’ describe reductions in usual affect, cognition, motivation, social interaction, and behavioural expression. Whether these are due to primary neurocognitive changes or secondary to other factors (such as social deprivation, trauma, substance misuse, and medication side-effects), cognitive models assume that repeated setbacks and failures foster low expectations about performance (e.g. ‘I can’t do it so there’s no point trying’), relationships (‘I’m not normal so no one will want me’), pleasure and success (e.g. ‘If I get a job I won’t enjoy it’), and resources (e.g. ‘my memory is terrible’). These maintain disengagement and avoidance, functioning to protect the person from expected pain and rejection (Beck et al., Reference Beck, Himelstein, Bredemeier, Silverstein and Grant2018; Beck et al., Reference Beck, Grant, Inverso, Brinen and Perivoliotis2020; Perivoliotis and Cather, Reference Perivoliotis and Cather2009). CBTp for negative symptoms draws on the theory and practice of CBT for depression, targeting these appraisals, and utilising behavioural activation by working on small steps towards personal goals, while accommodating neurocognitive impairments and engagement difficulties (e.g. in planning, concentration, and memory; Beck et al., Reference Beck, Himelstein, Bredemeier, Silverstein and Grant2018; Beck et al., Reference Beck, Grant, Inverso, Brinen and Perivoliotis2020; Perivoliotis and Cather, Reference Perivoliotis and Cather2009). CBTp is less well developed in this area, there are far fewer trials, and reviews indicate small pooled effects, usually as secondary outcomes (Aleman et al., Reference Aleman, Lincoln, Bruggeman, Melle, Arends, Arango and Knegtering2017; Lutgens et al., Reference Lutgens, Gariepy and Malla2017; Riehle et al., Reference Riehle, Böhl, Pillny and Lincoln2020). It will be interesting to see whether, like CBTp for voices and paranoia, future research adopts a more symptom specific approach, targeting likely maintenance mechanisms, and if this leads to improved outcomes. Broader psychosocial interventions such as exercise, music, dance, and cognitive remediation also show promise in this area (Cella et al., Reference Cella, Roberts, Pillny, Riehle, O’Donoghue, Lyne and Preti2023; Lutgens et al., Reference Lutgens, Gariepy and Malla2017).

The breadth of people’s needs requires therapists to be skilled in delivering CBT for a wide range of issues including but not limited to ‘positive symptoms’ of psychosis. The UK CBTp training curriculum, based on the UCL competence framework (Roth and Pilling, Reference Roth and Pilling2013), is designed to reflect this, equipping therapists with the knowledge and skills to address common affective presentations, including depression, social anxiety and PTSD, as well as psychosis. Trainees’ competencies are rigorously assessed throughout their training, so that when qualified, therapists are able to provide high quality, evidence-based treatment for a range of presenting problems.

Misconception 5: Collaboration means never being direct

A common issue we encounter when training clinicians to deliver CBTp is novice therapists equating collaboration with passivity. The Cognitive Therapy Scale-Revised (CTS-R) describes collaboration as encouraging active engagement in therapy, leading to productive teamwork (Blackburn et al., Reference Blackburn, James, Milne, Baker, Standart, Garland and Reichelt2001). Skilful collaboration involves the therapist facilitating a shared responsibility for the therapeutic process through the use of non-verbal prompts, as well as joint decision making and problem solving. While a questioning approach is important, therapist passivity is highlighted as an obstacle to effective collaboration.

As an example, CBTp is a goal-oriented therapy and so at assessment we discuss what the person would like to get out of the sessions in order to agree personalised therapy goals. This does not mean that the therapist should only ask the individual about their goals, without also explaining what CBTp can offer and making suggestions, given their training and experience, as to what might be helpful. People with psychosis often say they would like to get rid of their voices or the images they see, or request help reporting neighbours or others interfering in their lives to the police. Others may struggle to articulate goals or have such low expectations that they cannot imagine how therapy could be beneficial. We spend time supporting people to think through possible goals and provide information about how we may be able to help make sense of unshared experiences, cope with them more effectively, and address key maintenance processes (e.g. worry and sleep in the case of persecutory delusions). We would also want to provide information about other evidence-based treatments where these align better with the person’s needs and goals, for example family intervention for psychosis to optimise caregiving when day-to-day arguments make things even more difficult for the person, and exacerbate their distress further (see Jolley and Grice, Reference Jolley and Grice2024).

For any of us, when we go to see a healthcare professional, unless we have specific training in that field, we don’t know what we don’t know. We expect the clinician to share with us information about the nature of the problem, evidence-based treatment options, and likely outcomes, in order to make informed treatment decisions. Similarly, collaboration in CBTp includes being explicit about what we can and cannot offer and what, from our perspective, may be helpful, in order that the person can make an informed decision about whether to pursue therapy, and if so, the goals they would like to prioritise.

This may all seem common sense. However, many trainees worry that providing information, guiding a conversation, making suggestions, and sometimes interrupting people, will be invalidating and disrupt the therapeutic relationship. Behavioural experiments in (or agreed in) supervision to experience being given information, guided in a discussion, and respectfully interrupted (and observing patients’ responses to similar) can be very effective in addressing these unhelpful appraisals. When delivering training in the Feeling Safe approach (Freeman et al., Reference Freeman, Emsley, Diamond, Collett, Bold and Chadwick2021) the trainers (including L.I.) are sometimes asked whether offering therapy with the explicit goal of ‘feeling safer, happier, and getting back to doing more of the things you want to do’ is too directive. Clearly, insisting that the person must engage in a specific approach would be highly uncollaborative, but providing information that allows people to decide if this is something they would like to pursue is consistent with therapeutic collaboration.

Finally, it is important to note that CBTp often requires a degree of assertive outreach when making an offer of therapy. Some people may not initially recognise that they have a difficulty as described in typical psychiatric language (e.g. schizoaffective disorder), may not know what psychological therapy involves, or may not see that therapy has anything to offer, and so not seek referral. As clinicians, we therefore sometimes need to be proactive in making an offer of therapy and ensure we remain up-to-date with the evidence base so we can give clear and accurate information to support genuine collaboration and informed decision making.

Misconception 6: CBTp isn’t trauma informed

Early adversity, particularly childhood interpersonal trauma, is associated with the development, severity and persistence of psychosis (Bailey et al., Reference Bailey, Alvarez-Jimenez, Garcia-Sanchez, Hulbert, Barlow and Bendall2018; De Vries and Goggin, Reference De Vries and Goggin2018; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechbauer, Read, van Os and Bentall2012), and shapes voices and other hallucinatory experiences phenomenologically (Peach et al., Reference Peach, Alvarez-Jimenes, Cropper, Sun, Halpin, O’Connell and Bendall2021; van den Berg et al., Reference van den Berg, Magaraggia, Schreiber, Hillhouse and Porter2022). People with psychosis report higher levels of trauma and PTSD than the general population, including psychosis-related PTSD (Fornells-Ambrojo et al., Reference Fornells-Ambrojo, Gracie, Brewin and Hardy2016). In a large Dutch sample, de Bont et al. (Reference de Bont, van den Berg, van der Vleugel, de Roos, de Jongh, van der Gaag and van Minnen2015) found that over 78% of people with psychosis reported at least one traumatic event in their lives, and 16% were likely to meet criteria for PTSD (in contrast to 0.5% reported in clinical notes). The literature indicates that psychosis is likely to be both a response to and cause of trauma (Gumley and MacBeth, Reference Gumley and MacBeth2007; Hardy, Reference Hardy2017; Larkin and Read, Reference Larkin and Read2008; Morrison et al., Reference Morrison, Frame and Larkin2003), which is often not recognised, and more often not recorded (de Bont et al., Reference de Bont, van den Berg, van der Vleugel, de Roos, de Jongh, van der Gaag and van Minnen2015). CBTp therapists must therefore be willing and able to assess and treat trauma and the consequences of trauma, drawing on CBTp as a trauma-informed therapy and directly with trauma-focused CBTp (Hardy et al., Reference Hardy, Keen, van den Berg, Varese, Longden, Ward and Brand2023).

Over the last decade, trauma-informed care has gained widespread recognition as a foundational component of recovery focused mental health services. Trauma-informed care incorporates the realisation of how trauma can affect individuals and communities, a recognition of the signs of trauma, organisational policy and procedural responses, and avoidance of iatrogenic re-traumatisation (e.g., Substance Abuse and Mental Health Services Administration, 2014). To be consistent with these principles, CBTp assessment must involve asking sensitively about early and subsequent adversity and trauma (see Read (Reference Read2007) for guidance on asking people with psychosis about past trauma and responding effectively). Formulation then involves linking current psychotic experiences and associated appraisals, emotions, and behavioural responses, with beliefs and assumptions that have typically been learnt early in life, to help make sense of the development and maintenance of distressing psychosis. For many, this is a validating process, recognising, often for the first time, that psychosis and responses to psychosis can be understood in the context of earlier life experiences.

Where people report trauma symptoms, show signs of trauma-related mechanisms (e.g. dissociation), or meet criteria for PTSD, they are likely to need a trauma-focused therapy, such as trauma-focused CBTp, Imagery Rescripting, EMDR and dialogic approaches, to process trauma-related memories, beliefs and emotions directly (Hardy et al., Reference Hardy, Keen, van den Berg, Varese, Longden, Ward and Brand2023). Here, again, clinical practice falls well behind the evidence base and best practice guidelines which recommend trauma interventions for people with psychosis and PTSD (NICE, 2014). This is likely to be due in large part to clinicians’ caution about re-traumatising people, and increasing risks to self or others, despite evidence that PTSD can safely and effectively be treated in people with psychosis (Sin and Spain, Reference Sin and Spain2016). The current STAR trial incorporates CBT for PTSD and CBTp for people with psychosis (Peters et al., Reference Peters, Hardy, Dudley, Varese, Greenwood, Steel, Emsley, Keen, Bowe, Swan, Underwood, Longden, Byford, Potts, Heslin, Grey, Turkington, Fowler, Kuipers and Morrison2022), and is likely to shape the field.

Misconception 7: CBTp is simple and doesn’t require specialist or extensive training

A common mistake is to assume that interventions based on a simple premise are simple to put into practice. While CBTp is based on the arguably simple premise that people can effect emotional and behavioural change through cognitive change, in practice this requires considerable expertise.

To deliver competent CBTp, a therapist must have a critical knowledge of the relevant theoretical and research literature, and the clinical competence to translate this knowledge into practice, for a wide range of presentations (see Misconception 4). The UK CBTp training curriculum is based on the UCL competence framework (Roth and Pilling, Reference Roth and Pilling2013), which describes the knowledge, skills, and meta-competences required. These include fundamental principles of CBT theory; problem specific models and protocols for depression, anxiety presentations, trauma, bipolar and psychosis; and broader CBTp relevant constructs (e.g. how memory works). Clinical application requires the ability to develop and maintain good therapeutic relationships (including with people who find it hard to trust others and may incorporate the therapist in their psychosis); to conduct CBT assessments and develop collaborative, accessible formulations that draw on multiple models where needed; to deliver culturally competent, individualised interventions with respect for the person’s values, culture and social differences; to work within a person’s belief system as required; and to utilise standardised and idiosyncratic measures to monitor and support therapeutic progress.

The depth of knowledge, skill and experience required to deliver good quality CBTp is considerable. Almost all clinicians attending our one- and two-year postgraduate training courses in CBTp, including qualified practitioner psychologists, do not meet competency standards on therapy recordings at the start of training. This is to be expected (though often disappointing to trainees) and the purpose of CBTp training. CBTp skills can be acquired (the majority meet and exceed these standards by the end of training), but doing so requires specialist teaching, close supervision, and practise over an extended period. Competency in CBT and CBTp is not acquired by attending a few workshops, and the failure to recognise this risks patients being given poor quality interventions which are not, in fact, CBTp, but labelled as such. This in turn can result in poor outcomes, iatrogenic harms, people being less willing to engage in future therapy, and CBTp developing a poor reputation with referrers and the general public. Current UK guidance on Early Intervention in Psychosis services requires that services only record delivery of CBTp if the clinician has completed one- or two-year specialist CBTp postgraduate training, evidencing competence, or equivalent.

Poor public understanding of CBTp, and reports that some people have found (mislabelled) ‘CBTp’ to be simplistic, invalidating, or ineffective, behove our professional bodies to maintain and publicise rigorous standards of care, particularly in terms of training and ‘full dose’ therapy. In medicine, we expect clinicians to be appropriately trained and qualified, and their practice held to account by relevant governance systems. Likewise, prescription of medicines without the active ingredients, or in insufficient quantities, would be considered unethical and malpractice. While not everyone with psychosis will choose or benefit from CBTp, as is the case for most treatments, it is essential that we are clear about what CBTp is and is not, and distinguish evidence-based CBTp from widely used ‘CBT-based’ practice if we are to ensure people can make informed decisions about treating and managing their mental health.

Misconception 8: This person is too complex (too ill, too recently discharged …) for CBTp

Most of the people we see in UK secondary care services present with complex needs, including additional diagnoses, substance use, and/or risk (usually to themselves). The NICE guideline (2014) recommends CBTp for everyone with psychosis including in the acute phase.

Early suitability criteria for CBT were based on predictors of outcome following short-term therapy. These criteria emphasised awareness of thoughts and feelings, ability to engage in a therapeutic relationship, shorter-term problems, and personal responsibility for change (Safran et al., Reference Safran, Segal, Vallis, Shaw and Samstag1993). People with psychosis may initially struggle to articulate their internal experiences, be cautious about trusting a therapist, present with complex and long-standing difficulties, and be ambivalent about change. This is understandably daunting for novice therapists.

It is essential that senior clinicians uphold best practice guidelines, recognise when junior colleagues are hesitant to offer therapy in these contexts, and support access to training and supervision as needed. Equally, service leads should not use criteria developed before the evidence for CBTp evolved to manage scarce resources. Where we have insufficient provision, for CBTp and other evidence-based therapies, we need to be transparent about this and not suggest that any lack of offer is due to the person being too unwell, too recently discharged, or otherwise too complex. For example, a recent review found that CBT can be delivered in acute care and is likely to be beneficial (Wood et al., Reference Wood, Williams, Billings and Johnson2020).

Barton and colleagues (Reference Barton, Armstrong, Wicks, Freeman and Meyer2017, Reference Barton, Armstrong, Robinson and Bromley2023) describe problem complexity, when people have multiple problems that interact with each other, and relational complexity, when complications in the therapeutic relationship or service context interfere with effective collaboration. In CBTp, we are likely to see both problem and relational complexity. People with psychosis often present with a mix of psychosis (e.g. voices, paranoia, social withdrawal), affective (e.g. depression, anxiety) and other difficulties (e.g. substance misuse, risk). Additionally, psychosis can have a direct impact on the therapeutic relationship, e.g. voices often worsen prior to or during therapy, may criticise the person for coming to sessions, and undermine the therapist. Social withdrawal and confusion can make it hard for people to attend and engage actively in therapy. Paranoia and grandiosity can make it difficult to trust and talk to a therapist. In a qualitative study of grandiose delusions, participants emphasised how their beliefs made sense of anomalous or adverse experiences, but that these were often hard to put into words (Isham et al., Reference Isham, Griffith, Boylan, Hicks, Wilson, Byrne, Sheaves, Bentall and Freeman2021). Problem and relational complexity need to be recognised, formulated, and addressed in therapy (and supervision) for CBT to be effective (Barton et al., Reference Barton, Armstrong, Wicks, Freeman and Meyer2017; Barton et al., Reference Barton, Armstrong, Robinson and Bromley2023).

As an example, a young man with a history of disrupted attachments, substance misuse and various previous diagnoses presented with voices, paranoia, social withdrawal, anxiety, anger and some ongoing drug use (problem complexity). He did well to simply get to therapy sessions (and occasionally didn’t). When he attended, he appeared distracted by his voices and wary of the therapist (K.N.-T.), consistent with a pervasive paranoia about people in authority (relational complexity). Supervision and some flexibility in service expectations were helpful in clarifying (1) the focus for therapy in the context of problem complexity – first to make sense of his experiences (formulation), and then to address fears about seeing family, friends and working (his goals), and (2) the impact of relational complexity – the ways in which his psychosis, anxiety, and substance use were likely to effect the therapeutic relationship, sensitively anticipating this with him, and problem solving together when these issues arose.

Misconception 9: CBTp can be done entirely in the clinic room

A core aspect of CBT involves taking therapy out of the clinic room and into the real world (Kennerley et al., Reference Kennerley, Kirk and Westbrook2016). When working with someone with depression, we might go for a brisk walk together to find out what impact this has on the person’s mood. In social anxiety, we might visit a café, order a coffee, and observe whether, as the person predicts, the waitress notices and responds negatively to their anxiety. In treating OCD, we might go with someone to use a public restroom whilst refraining from compulsive hand washing afterwards, to discover whether they become physically unwell or experience an intolerable level of anxiety, as feared.

Such in vivo work offers many advantages over and above therapy conducted solely in the confines of the therapy room by:

• Allowing the therapist to provide immediate encouragement and support when the person tries a feared or avoided activity for the first time.

• Identifying key information that might not be recognised if the therapist was not present (e.g. subtle safety seeking behaviours).

• Providing an opportunity for modelling, practising, and refining new strategies, and problem-solving difficulties as they arise.

• Allowing the therapist to support the person to articulate new appraisals in the moment of discovery (e.g. ‘I can go out safely even when feeling afraid’).

• Building momentum in therapy – providing more opportunities for key learning that can be built on with between-session tasks.

• Strengthening the therapeutic alliance, by being alongside the person when commencing the difficult process of behaviour change.

These benefits are just as relevant when working with people with psychosis, and recommended in CBTp manuals and approaches (e.g. Morrison, Reference Morrison2017). Despite this, clinicians are often reluctant to get out of the clinic room. In a recent training event, more than 100 CBTp practitioners were asked what proportion of their patients they had routinely been out with as part of their therapy; 92% reported having done this with fewer than half their patients, and 33% said they never did this.

Therapists cite a number of reasons for their reluctance to leave the clinic. Some have not tried in vivo experiments before and are anxious about them ‘going wrong’. Others are nervous about managing risk away from the team base. These anxieties are understandable but likely to impede therapy. Accessing supervision to support the planning and execution of within session activities, and robust (but not excessively cautious) risk assessment and planning, should ensure these not be default reasons to avoid in vivo work. Most commonly, however, therapists say they simply do not have time to leave the clinic with patients, reporting systemic pressures, such as needing to see people ‘back-to-back’ and in sessions of no more than 50 minutes. This is particularly concerning; systematically excluding a core tenet of CBT risks diluting the intervention and reducing likely treatment effectiveness. We suggest it is unethical to tell people they are receiving a NICE recommended treatment and then giving a substandard ‘dose’. Once more, this would not be tolerated in physical health settings (e.g. providing half the recommended dosage of diabetes medication), and should not be tolerated in mental health care.

Misconception 10: Measures are too burdensome (and don’t get to what’s really important)

Routine outcome measurement is an integral part of CBT, and can benefit the person with psychosis, the therapist, and the service (Faija et al., Reference Faija, Bee, Lovell, Lidbetter, Gellatly, Ardern, Rushton, Brooks, McMillan, Armitage, Woodhouse and Barkham2022). The wider psychotherapy outcome literature shows that taking measures improves engagement and outcomes (Lambert et al., Reference Lambert, Whipple, Hawkins, Vermeersch, Nielsen and Smart2003; De Jong et al., Reference De Jong, Timman, Hakkaart-Van Roijen, Vermeulen, Kooiman, Passchier and Busschbach2014). Indeed, most people with psychosis find measures useful as a way of ‘feeling understood, valuing opportunities to reflect, expressing feelings, and tracking progress towards goals’ (Fornells-Ambrojo et al., Reference Fornells-Ambrojo, Johns, Onwumere, Garety, Milosh, Iredale and Jolley2017; p. 254).

At the start of their training, many of our CBTp trainees are using few or no regular measures. Some are uncertain what to measure or which scales to use, but most are concerned that people will find measures aversive or reductive, and that this will hamper engagement. It is of note that this assumed threat to engagement is also given as a reason for excluding other core aspects of CBT, such as setting an agenda and agreeing homework tasks. Whilst we wholeheartedly support continuous attention to, and prioritisation of, the therapeutic relationship (see Misconception 3), this does not necessitate omitting core components of CBT when the evidence shows that done well, routine outcome measurement enhances both engagement and outcomes.

How to introduce measures? Outcome measurement is introduced as part of CBTp assessment, with a clear rationale in line with the person’s goals. Potential benefits include further understanding of the problem(s) for which they are seeking help (and linked mechanisms), identifying treatment targets, and capturing progress (Kennerley et al., Reference Kennerley, Kirk and Westbrook2016). This is usually highly valued as a way to keep therapy on track in line with personal goals for recovery (cf. Waller and Turner, Reference Waller and Turner2016). Without a clear and shared rationale, measures may be seen as irrelevant, and this can certainly hinder engagement (Rao et al., Reference Rao, Hendry and Watson2010).

Which measures to use? CBTp utilises a limited combination of standardised measures, brief idiosyncratic sessional measurements (e.g. subjective units of distress), and within session tools (e.g. belief ratings in behavioural experiments and activity diaries). The choice of measures is based on the person’s presentation, needs and goals, considering:

• Is the measure useful? What will it tell us and does the person agree it will be helpful? Is this the best option?

• For a standardised measure – is it valid and reliable (psychometric properties)?

• Is the measure accessible? This takes account of the person’s literacy and belief system. Occasionally, we may change a word if likely to be confusing or unacceptable (e.g. replacing ‘delusions’ with ‘concerns others are trying to harm me’). While better not to change standardised questionnaires, minor adaptations are sometimes necessary, and preferable to omitting measures altogether.

How often and in what format? Brief measures can be used sessionally, whereas longer (typically standardised) measures are used less frequently at regular intervals. The best combination of measures varies from person to person, taking account of their problems, priorities and capacity. Some people are happy to complete measures between sessions, others prefer taking a few minutes to run through these with the therapist each week, and some need to spread just a few measures across sessions to reduce burden.

In this way, we come to a shared agreement about which measures to use, why, how often, and in what format, ensuring the combination is relevant, useful, and not burdensome. Too often, clinicians fail to take any measures, having assumed without discussion that doing so would be arduous and without benefit, when a limited number of carefully selected measures is both possible and likely to be helpful.

Routine outcome measurement is built into UK NHS Talking TherapiesFootnote ⁴ services to monitor impact at individual, service, and national levels. This is likely to play a significant role in ongoing funding decisions. While national guidelines recommend similar for people with severe mental ill-health (Royal College of Psychiatrists, 2024), adherence is very patchy. A minimum co-produced dataset could support and shape services, and provide the large scale data needed to advance personalised psychological therapies for people with psychosis (Newman-Taylor and Bentall, Reference Newman-Taylor and Bentall2024).