Introduction

Historically, a biomedical approach to psychosis interventions has been prioritised, mainly focusing on the use of antipsychotic medications to reduce positive symptoms such as delusions and voice-hearing; however, research has consistently highlighted the many psychosocial factors associated with experiences of psychosis—such as developmental trauma, poor social support, poor quality of life, loneliness, internalized stigma, racism and discrimination, inequality, and low self-esteem (Anglin, 2023; Bloomfield et al., 2021; Lim et al., 2018; Reininghaus & Priebe, 2012)—and there is a widespread consensus that psychosocial interventions are essential for rehabilitation and longer-term psychosis care (Bellack, 2001). Furthermore, best-practice guidelines in Germany (Gaebel et al., 2020), India (Grover et al., 2017), the United Kingdom (Birchwood et al., 2016) and the United States (Keepers et al., 2020) recommend various evidence-based psychosocial interventions (e.g., psychological therapies, family interventions) as frontline treatments, alongside medication.

The present review focuses on psychological interventions for psychosis, which have proven difficult to access despite being recommended within best practice guidelines. Moreover, a 2023 meta-analysis based on data from 9 different countries within 3 continents (Europe, North America, Australia) found that the average rate of receipt of Cognitive Behavioural Therapy for psychosis (CBTp) was just 24% (Burgess-Barr et al., 2023). Published accessibility data is not available for other psychological therapies but given that CBTp is the probably the most widely researched and recommended (within best-practice guidelines) psychological therapy, the rate of receipt for them is likely to be lower. There are significant barriers to the provision of CBTp and other psychological therapies for psychosis outlined in the literature, such as workforce capacity, service structure, capacity to engage, and limited collaboration and communication between service providers and potential referrers (Switzer & Harper, 2019).

Alongside the above implementation difficulties, many service users with psychosis do not engage in offered psychological interventions, with symptom severity, and levels of insight and self-esteem suggested as important uptake predictors (Gaynor et al., 2014). Another potentially important factor—and the focus of the present review—relates to the limited attention given to service user preferences surrounding the nature of these interventions (Haddock et al., 2018). While there has been considerable research into service user experiences of psychological therapies for psychosis post-intervention, far less is known in relation to preferences surrounding the nature of these interventions, which may underlie engagement issues, especially for those who have not previously completed psychological therapy (Byrne et al., 2010; Holding et al., 2016). There has also been notable research around service user preferences for the overall mental health service treatment of psychosis (Brabban et al., 2017; Byrne et al., 2010)—but far less research into specific psychological intervention preferences, which are likely to be more implementable owing to the focus on a standalone intervention rather than a multicomponent (overall service) one.

It is important to consider the psychological intervention preferences of service users with psychosis for several reasons: First, if service providers do not consider service user preferences within intervention design and delivery, they run the risk of incorrectly leveraging suboptimal values that ultimately fail to engage many service users (Poremski et al., 2016). Second, unclear shared goals for psychological interventions and the related absence of a collaborative therapeutic relationship have been reported by service users with psychosis as important barriers to their engagement (Switzer & Harper, 2019). Third, recent findings suggest that service users whose preferences for psychological interventions are not met are less likely to report meaningful therapeutic gains, and this is likely to impact upon their future engagement (Williams et al., 2016). Fourth, clear service user preferences are likely to exist and are thus worthy of exploration. Indeed, previous reviews highlighted preferences in relation to the overall mental health service treatment of psychosis, in terms of outcome priorities and delivery mode (Brabban et al., 2017; Byrne et al., 2010).

Overall, despite the previous welcome investigations of both service user post-intervention experiences of psychological interventions and their preferences for overall mental health service treatment, uncertainty remains surrounding the specific psychological intervention preferences of service users with psychosis, particularly for those who have not previously engaged. Regarding intervention outcomes, for example, it would be useful to know the degree to which service users want to focus on outcomes relating to psychosis experiences versus broader psychosocial outcomes—and to gain greater precision around preferred psychosocial outcomes. Regarding delivery mode, there are important service user preferences to consider in relation to duration, content, delivery (e.g., using technology) and the acceptability of group versus individual interventions. There is also the question of which psychological therapy approaches would be preferred: Despite the best-practice recommendation of CBTp in the United Kingdom and Germany (Birchwood et al., 2016; Gaebel et al., 2020), there are other viable options such as Compassion Focused Therapy for psychosis, which aims to foster psychological safeness and compassion towards experiences of psychosis (CFTp) (Heriot-Maitland et al., 2023) and Acceptance and Commitment Therapy for psychosis (ACTp), which attempts to alter one’s relationship to symptoms of psychosis through concepts such as ‘acceptance’ and psychological flexibility (Morris et al., 2023).

The current study encompasses—to our knowledge—the first systematic review on the specific psychological intervention preferences of service users with psychosis. We aim to increase understating of these preferences so that they can be more easily accommodated into the design and delivery of psychological interventions specifically (as opposed to broader services), to ultimately contribute to improved engagement and service user experiences.

Methods

Eligibility Criteria for Study Selection

The following inclusion criteria were applied:

  • Participants: Service users with current or previous experiences of psychosis, with or without formal psychiatric diagnoses. This criterion was kept broad to reflect the wide spectrum of psychosis experiences, and the increasingly common transdiagnostic psychological treatment of psychosis.

  • Interventions: findings had to be directly applicable to psychological interventions (including psychotherapy and psychoeducation) for psychosis and/or related psychosocial outcomes delivered by a clinical/counselling psychologist or psychotherapist (e.g. psychotherapy, counselling, psychoeducation).

  • Outcome: Service users’ self-report about their intervention preferences.

  • Studies: only primary studies (not reviews or commentaries); quantitative, qualitative; mixed-methods studies included. Only studies published in scientific journals.

The following exclusion criteria were applied:

  • Studies investigating previous experiences of psychological interventions (e.g. feedback about a recently completed intervention) rather than preferences for psychological interventions more generally.

  • Studies focusing on the broad concept of “recovery”, rather than psychological interventions. Relatedly, findings applicable to overall mental health services but not directly applicable to psychological interventions.

  • Studies investigating dementia-related psychosis, due to the potential overlap of symptoms with dementia. Neurologically-related psychosis would likely not be treated in the same way as other forms of psychosis.

  • Studies investigating interventions for substance use and other non-psychosis mental health disorders in psychosis populations.

  • Studies investigating preferences in relation to the overall mental health service treatment of psychosis.

Literature Search and Data Extraction

Search terms relating to psychosis (i.e., psychosis OR psychotic OR “voice hear*” OR “hear* voice*” OR hallucinat* OR paranoi* OR delusion* OR schizo*) were combined with terms relating to service user preferences (i.e., “patient preference*” OR “client preference*” OR “service user preference*” OR “patient perspective” OR “client perspective*” OR “service user perspective*” OR “patient priorit*” OR “client priorit*” OR “service user priorit*”) and psychological interventions (i.e., psychosocial OR therapy OR psychotherapy OR psycholog* OR psychoeducation OR counsel* OR intervention OR recovery OR outcome*). PsycINFO, MEDLINE, CINAHL, and Academic Search Complete comprised the databases, searched on April 1st, 2023 with an updated search performed on January 1st 2024. Additional records were identified through hand-searching reference lists of included studies and a Google search.

Both authors independently conducted all steps of the screening and data extraction process (i.e. record search, screening, full-text, review and data extraction), and collaboratively reviewed progress after each step. Upon disagreement regarding the screening outcome of an abstract, it was included in screening at “full-text” level. When we disagreed at “full-text” level we discussed the reasons for inclusion and exclusion and made a joint decision. Data were managed using the COVIDENCE programme (https://app.covidence.org/) and word processing software. Extracted data covered study setting, participant characteristics, how preferences were assessed, and service user preferences organised by category.

Study Quality Assessment

We anticipated that most—if not all—the studies included in this review would consist of surveys, which are often methodologically simplistic, and usually involve convenience sampling methods. Therefore, study quality was expected to be relatively low. To assess study quality, we considered survey-specific quality assessment tools but could not identify any pre-existing measures with adequate supportive evidence. We therefore decided to use 3 items—highly relevant to survey studies—from the widely-used Newcastle–Ottawa Scale (NOS) cross-sectional studies version (Wells et al., 2013). Although we considered using more comprehensive non-validated measures for study quality assessment, we reasoned that focusing on three directly applicable items from the NOS would be preferable over including a longer measure with items that we adjudged to be less clearcut in terms of their applicability. The three NOS items related to representativeness of the sample (compared with the total population), sample size justification (including sample size calculation), and the use of a validated measure of psychosis. Quality ratings for each item were performed on a present/absent basis derived from what was reported in the paper; the maximum scores for each item and each study were 1 and 3, respectively. The assessment was conducted independently by both authors, with no discrepant ratings arising.

Results

Study Selection Process

After duplicates were removed, a total of 742 records were screened. There was a 94% agreement rate between authors in the process of independent abstract screenings. All discrepancies were rescreened collaboratively, and 20 studies were selected for full-text review. Both authors independently included 10 studies in the review (Fig. 1).

Fig. 1
figure 1

Literature search flow

Full size image

Study Characteristics

Table 1 outlines the characteristics of the 10 included studies, and the main findings. Publication dates ranged from 2010 to 2023, and most were conducted in public mental health services in the United Kingdom (UK). Mode of recruitment for the majority was convenience sampling within the service. Sample sizes ranged from 53 to 1809 (Combined N = 3121), and it is worth noting that one study (Freeman et al., 2019) had by far the largest n (of 1809); though all-bar-one of the other studies had ns either approaching 100 or in excess of 100. The mean ages ranged from 31 to 44 years, and most studies had slightly more male than female participants. Seven studies investigated preferences through a quantitative survey, one study used a qualitative interview, and two studies used both. Owing to basic survey methodologies deployed, study quality was low: five studies received no points (out of three) for study quality, while five scored one point.

Table 1 Study characteristics and the psychological intervention preferences of service users with psychosis
Full size table

Service User Preference Findings

Eight studies (combined n = 2936) investigated outcome-related preferences (Berry et al., 2023; Byrne & Morrison, 2014; Fornells-Ambrojo et al., 2017; Freeman et al., 2019; Greenwood et al., 2010; Longden et al., 2023; Moritz et al., 2017; Morrison et al., 2023), and three (combined n = 520) investigated delivery mode of CBTp (Berry et al., 2023; Haddock et al., 2018; Sumner et al., 2014), with the Berry et al. (2023) study investigating both outcome and delivery mode considerations. It therefore made sense to synthesise the findings under two categories: outcome priorities and delivery mode. We used both quantitative and qualitative findings in this synthesis.

Regarding outcome priorities, across, studies, service users mostly prioritised various emotion-related (e.g., worrying less), behavioural (e.g., drive and motivation, coping mechanisms), and self-reflective (e.g., self-esteem, self-confidence) outcomes and these psychosocial outcomes were generally deemed as more important than psychosis symptoms outcomes. One exception was the Longden et al. (2023) study which specifically investigated voice-hearing outcomes using a predetermined list that included a narrow range of psychosocial outcomes: here service-users were more likely to prioritise a clinically focussed view of recovery (e.g., voice cessation) over a psychosocial one (e.g., functioning while living with voices). Additionally, one study investigated attitudes towards outcome monitoring; concluding that service users were generally open to completing routine outcome measures (Fornells-Ambrojo et al., 2017).

Regarding delivery mode preferences, across studies, service users mostly indicated a preference for individual format (e.g., with therapist or peer support worker) over group format (Berry et al., 2023; Greenwood et al., 2010; Haddock et al., 2018; Sumner et al., 2014). Amongst the CBTp preference studies, no option of individual face-to-face sessions was provided; but individual telephone delivery was preferred to telephone delivery in conjunction with groupwork (Haddock et al., 2018; Sumner et al., 2014). Further, meeting a therapist face-to-face in an individual session was the highest preference recorded in Berry et al. (2023), followed by hybrid, telephone and video call. The latter study examined other questions relating to delivery mode: 9–12 sessions of therapy was preferred in terms of duration, closely followed by 13–16 sessions; weekday afternoon was the preferred start time. Finally, in two studies, service users reported preferences for collaborative therapy modalities with a task-focus on formulation in psychosis (Berry et al., 2023; Byrne & Morrison, 2014).

Discussion

The main findings of this review pointed to service user preferences for psychosocial outcomes over outcomes related to psychosis symptoms, and preferences for some degree of individualised care.

In line with the outcome-related findings, previous reviews summarizing preferences surrounding the overall mental health service treatment of psychosis found that service users prioritized social, functional, and recovery-based outcomes, rather than treating and monitoring psychosis symptoms (Brabban et al., 2017; Byrne et al., 2010). A user-led Delphi study included in the present review (Byrne & Morrison, 2014) also found broader service-based preferences for psychosocial outcomes, in line with those of the present review. The one included study (Longden et al., 2023) that did not support the prioritisation of psychosocial outcomes solely focused on voice-hearing outcomes (with all questionnaire items in some way referring to voice-hearing). This contrasted with other relevant studies wherein questionnaires addressed a broader range of symptoms-based (e.g. paranoia) and psychosocial outcomes (e.g., self-esteem, self-confidence, well-being), and the narrower focus on voice hearing might explain the outlying finding here. Overall, there appears to be a growing consensus among service users with psychosis surrounding the prioritisation of psychosocial outcomes—in both psychological and broader mental health service interventions. This may come as a surprise to the many services that continue to prioritise the biomedical approach to psychosis interventions, yet it supports the growing recognition of psychosocial interventions in best-practice guidelines in Germany (Gaebel et al., 2020), India (Grover et al., 2017), the United Kingdom (Birchwood et al., 2016) and the United States (Keepers et al., 2020).

The indicated preference for some form of individualised care concords with psychosis service research highlighting the benefits of individualised treatment approaches such as shared-decision making and collaborative care during times of crisis (Farrelly et al., 2014; Thomas et al., 2021). Such individualised and collaborative approaches are well placed to address the many barriers to engagement for people with psychosis such as stigma, shame, low self-worth, and negative beliefs about treatment (Stovell et al., 2016). On the other hand, it is possible that some of the findings surrounding preferences for individualised telephone support could be related to the accessibility gains arising from remote delivery rather than preferences for one-to-one over group support.

Regarding other findings, it was interesting that there were some reported intervention preferences for collaborative formulation (Berry et al., 2023; Byrne & Morrison, 2014). This highlights previously documented misconceptions about the unsuitability of formulation-based CBT for many service users with psychosis (Switzer & Harper, 2019) and suggests that many service users have a strong interest in insight-building around experiences of psychosis and broader life experiences.

It is worth highlighting that a small percentage of service users indicated little or no desire for engagement with psychological interventions (Haddock et al., 2018; Sumner et al., 2014). Therefore, it is important to avoid assuming all service users want to engage in psychological interventions, and there are various other psychosocial interventions that can be considered such as family and exercise-based interventions, and peer support groups such as those provided by Hearing Voices Network (Longden et al., 2018). On the other hand, it seems that many service users with psychosis wish to receive some level of psychological input, and that there are likely to be various preferences around this input. Indeed, recent results from a broad scope study investigating service user preferences for psychological treatment in various mental health disorders detailed that 86% expressed a preference relating to their treatment (Williams et al., 2016).

Implications for Clinical Practice

Arguably, the highlighted psychosocial outcomes should be prioritised in the planning of psychological interventions for psychosis, though it is possible for these interventions for to concurrently target both symptoms and psychosocial outcomes, as shown by recent meta-analysis support for CBTp (Wood et al., 2020). Improved self-esteem seems to be a particularly important psychosocial outcome: it was a consistent priority in the present review, and it has been highlighted in recent research examining service user experiences of group cognitive behaviour therapy for psychosis (Lecomte et al., 2015). Of relevance, services could consider CFTp which prioritises similar psychosocial outcomes such as increased self-compassion and reduced shame (Heriot-Maitland et al., 2023).

If psychological interventions are to accommodate the delivery mode preferences highlighted in this review, they should be delivered with some degree of individualised input (even within group interventions through means such as individual review sessions and concurrent individual telephone support). As it is possible that some of the findings surrounding preferences for individualised support could be related to remote delivery, ways to increase accessibility through remote delivery options—for individualized and group-based interventions—should also be considered by services. Of relevance, a recent study involving public health service users with psychosis in the United Kingdom found that over half of the 200 participants preferred a hybrid model involving both face-to-face and remote delivery of healthcare (Kuhn et al., 2024). The Berry et al. (2023) study highlighted preferences around duration (9–12 sessions) and an afternoon start time; however, this contrasts with United Kingdom best-practice guidelines which state that CBTp should be delivered over at least 16 sessions (Birchwood et al., 2016). On a cautionary note, the above accommodations (e.g., additional telephone and individualised supports) are resource intensive in terms of finance, staff requirements, and time. This poses a significant barrier to their implementation, particularly in low and middle income countries, where inadequate funding for psychosis services is commonplace, alongside a shortage of basic services and trained professionals (Singh & Javed, 2020).

Finally, given the highlighted intervention preferences for formulation (Berry et al., 2023; Byrne & Morrison, 2014), a greater emphasis on collaborative formulation surrounding psychosis is also worthy of exploration in the development of future interventions. In particular, there is a growing recognition of psychosis-related post-traumatic stress disorder and there is a rationale for psychological interventions that facilitate trauma informed, collaborative formulation (Buswell et al., 2021).

Implications for Future Research

First, as the review findings have mainly arisen from aggregated service user preferences gathered from quantitative surveys, individual-level approaches such as shared-decision making should be considered in future research to explore individualised preferences for psychological psychosis interventions, and the feasibility of their implementation (Thomas et al., 2021).

As well as investigating service user preferences in treatment, one included study conducted a survey that rated psychosocial interventions in terms of helpfulness (Moritz et al., 2017). CBTp was rated 10th, with 20% of service users reporting that they found CBTp ‘not helpful at all’. Talk therapy, followed by psychoanalytical therapy, metacognitive training and psychoeducation with relatives were amongst the highest rated psychological interventions. Although this sample was limited in terms of representation, the indicated preference for alternative psychological approaches is worthy of further exploration considering that CBTp is the routinely recommended treatment in various best-practice guidelines (Birchwood et al., 2016; Gaebel et al., 2020). A future research study could, for example, offer service users the fully-informed choice of CBTp, and other increasingly popular and evidence-supported psychological interventions such as CFTp and ACTp and investigate the reasons for the selected intervention approach.

Future research would also benefit from the use of validated measures or clinical interviews for assessing psychosis, as such measures were absent in several studies included in the present review. The use of such measures would clarify the extent of psychosis symptomology among participants, as well as providing diagnostic clarity—overall, this information would be useful in investigating if service users’ preferences differ along the continuum of severity and/or between sub-types of psychosis (e.g., the preferences of voice hearers compared with individuals experiencing paranoia). It may also be helpful to employ more qualitative exploratory methods to capture whether preferences within different stages of recovery vary and to investigate if the prominent symptoms experienced by service users (paranoia, delusions, and hallucinations) influence preferences. Previous research has highlighted the lack of qualitative research investigating different types of preferences among service users with psychosis (Byrne et al., 2010).

A notable omission arising from reviewed studies concerns preferences around family member involvement (both initially and over time) in psychological interventions—this is important given the common involvement of family members in psychosis help-seeking, their value in terms of ongoing support and achieving clinical outcomes, and the recent finding that most service users (n = 761) requested some form of family involvement in their care within an Early Psychosis Service (Jones et al., 2021). Preferences around family involvement in psychological interventions should therefore be prioritised in future research.

Limitations

Our findings should be considered preliminary due to the limited number of studies, though there was a large combined sample size (N = 3121). Furthermore, most studies took place in public mental health settings in the United Kingdom and involved service users were mostly aged 31–44. These factors limit generalizability of the findings across geographic and age divides, and to people with psychosis not attending public mental health services. Relatedly, the limited number of studies precluded the investigation of links between preferences and the ethnic, cultural, and socioeconomic backgrounds of participants.

We excluded studies broadly investigating recovery, as there is a lack of consensus around how recovery is defined, likely due to its idiosyncratic nature as well as varying views reported between clinicians and service users (Law & Morrison, 2014). Research that exclusively focused on general mental health service preferences and past experiences with psychological interventions was excluded. This decision aimed to specifically delve into preferences for psychological interventions, enhancing the accuracy of the findings. However, this approach may have overlooked broader findings that could be relevant and recent intervention experiences that might be more salient to service users.

Our adapted version of the NOS (Wells et al., 2013) illustrated the low quality of studies involved, largely as result of convenience sampling and simplistic survey methods. However, the suitability of the NOS for the quality assessment of survey studies is unknown—we used it in the absence of survey-specific quality assessment measures.

Conclusion

Notwithstanding various methodological limitations, the above findings could be considered as a progressive preliminary step towards the increased accommodation of the specific psychological intervention preferences of service users with psychosis. The key takeaways for psychosis psychological intervention providers are to consider the prioritisation of psychosocial outcomes and—if resources allow—preferences for some level of individualised care over the sole provision of group interventions.