Feeling heard: an uncontrolled feasibility study evaluating a novel intervention pathway for psychosis patients distressed by hearing voices

ABSTRACT

Background

Cognitive Behaviour Therapy for psychosis (CBTp) has been the psychological intervention recommended for the treatment of psychosis for the past two decades. However, benefits from and access to CBTp are limited. Responses to these issues have included the targeting of CBTp at single symptoms and the delivery of CBT-informed interventions by briefly trained therapists. We sought to combine these approaches by drawing upon a wider workforce of therapists to deliver a pathway of interventions for distressing voices.

Methods

This study was an uncontrolled feasibility study evaluating a pathway of CBT-informed interventions offered over a maximum of 20 sessions. The evaluation of outcomes focused on recruitment, retention, potential for effectiveness and safety.

Results

Of the 14 consenting participants, 11 completed the intervention pathway. There was a large, significant pre-post reduction in voice related distress for the per-protocol sample (Cohen’s d = 1.51 [95% CI: 0.61 to 2.37]. No serious adverse events related to the intervention pathway occurred.

Conclusion

It is feasible to conduct an evaluation of the Feeling Heard intervention pathway. Randomised Controlled Trials are required to robustly assess its efficacy.

KEYWORDS:

• Psychosis
• auditory hallucinations
• voice hearing
• cognitive behavioural therapy
• increasing access

Introduction

Cognitive Behaviour Therapy for psychosis (CBTp) has been the psychological intervention most stongly recommended in the UK for the treatment of psychosis for the past two decades (National Institute for Health and Care Excellence [NICE], 2014). However, there have been at least two barriers to progress during this time. Firstly, CBTp targeting a range of symptoms has generated only limited benefits for psychosis patients (small-medium effect sizes at end-of-treatment for general symptoms, positive symptoms and functioning; Berendsen et al., 2024). A recent response to this barrier has been to target CBTp at one positive symptom at a time – the single-symptom approach; and results have been encouraging with enhanced benefits reported (Lincoln & Peters, 2019). Secondly, there has been limited access to CBTp with only a minority of people who experience psychosis being offered treatment in the UK (Royal College of Psychiatrists [RCP], 2018). One of the reasons for limited access is a lack of resources as CBTp is typically delivered by highly trained therapists (e.g. Clinical Psychologists) who are in short supply (Ince et al., 2016).

Within the Sussex Voices Clinic (a specialist psychological interventions service within Sussex Partnership NHS Foundation Trust - https://www.sussexpartnership.nhs.uk/sussex-voices-clinic), we have developed a suite of CBT-informed interventions targeted at the single symptom of distressing voice hearing experiences. Each of these interventions has been evaluated within Randomised Controlled Trials (RCT) and has generated effect sizes for voice-related distress at end-of-treatment ranging from medium to very large, relative to treatment-as-usual (TAU): 1) Person Based Cognitive Therapy (PBCT; d = 0.50; Chadwick et al., 2016); 2) Relating Therapy (RT; d = 1.30; Hayward et al., 2017); and 3) Guided self-help CBT (the GiVE intervention; d= 1.78; Hazell, Hayward, Cavanagh, et al., 2018). Whilst these interventions aimed to address the need for greater patient benefit, they were all delivered by highly trained therapists (typically Clinical Psychologists) who, despite national initiatives to increase the size of the profession in the UK, are currently in short supply (Psychological Professions Network, 2018). Consequently, our most recent RCT explored the contribution that briefly trained therapists (in the form of Psychology Graduates, also known as Assistant Psychologists) could make to increasing access to CBT-informed interventions for psychosis patients. We evaluated the delivery of the GiVE intervention by briefly trained therapists to psychosis patients and found a between-groups effect size in the medium range, relative to TAU (Hayward et al., 2021).

Having made progress regarding benefit from, and increased access to, CBT-informed interventions, we sought to combine these approaches to maximise both benefits and access. Consistent with the recent modularisation of CBT-informed interventions targeted at specific symptoms of psychosis (e.g. paranoid delusions; Freeman et al., 2021 [large effect size at end-of-treatment, relative to befriending]), we aimed to evaluate the benefits generated by offering a pathway of interventions for distressing voices delivered by therapists with varying levels of training to maximise benefits whilst minimising the use of therapist resources. The pathway, named Feeling Heard, contained the GiVE intervention delivered by Assistant Psychologists, and PBCT or RT delivered by Clinical Psychologists. The intervention pathway was also consistent with a move towards interventions targeting specific maintenance processes (Newman-Taylor & Bentall, 2024). Within the Feeling Heard pathway, we targeted the following mechanisms that have been demonstrated to maintain voice-related distress: beliefs about self (Fannon et al., 2009); beliefs about voices (Tsang et al., 2021); negative relating to voices and other people (Hayward et al., 2020); and mindfulness (Strauss et al., 2015).

As a first step, this uncontrolled feasibility study aimed to evaluate: (1) recruitment, (2) retention, (3) potential for effectiveness, (4) willingness to complete assessments and (5) safety of the Feeling Heard intervention pathway.

Method

Design

This study was inspired and informed by the work of Freeman et al. (2016, 2021) on the Feeling Safe Programme. We conducted an uncontrolled feasibility study of the Feeling Heard intervention pathway offered over a maximum of 20 sessions. The pathway consisted of 8 sessions of GiVE delivered by Assistant Psychologists, followed by either 12 sessions of RT or 12 sessions of PBCT delivered by Clinical Psychologists. Feasibility outcomes were: recruitment, retention, potential effectiveness and safety.

Clinical data were collected at: 1) baseline (pre-intervention), (2) monthly assessments during the intervention pathway, and (3) post-intervention. Consistent with recommendations that adverse effects of psychological therapies be monitored and reported (Parry et al. (2016), adverse events and serious adverse events were reported throughout an individual’s participation.

For the purposes of determining the feasibility of evaluating the Feeling Heard intervention pathway, the following criteria were examined:

1. Recruitment – was defined as recruiting at least 75% of the targeted sample size.

2. Retention – was defined as attending at least 14 of the 20 intervention sessions (6/8 for GiVE and 8/12 for RT/PBCT) with the aim of at least 75% of participants completing the intervention.

3. Willingness to complete assessments (an addition to the pre-specified analysis plan) - was defined as the number of assessment time points completed by participants. A threshold was not set for this as it was added post-hoc following data analysis.

4. Potential for Effectiveness – was defined as the Minimum Clinically Important Difference (MCID; the minimal score considered relevant by patients and clinicians) on the candidate primary outcome being included in the baseline to end-of-intervention 95% confidence interval for the per-protocol sample. The candidate primary outcome was voice-related distress, measured within this study by the Emotional Impact Scale of the Hamilton Programme for Schizophrenia Voices Questionnaire (HPSVQ; Van Lieshout & Goldberg, 2007. The MCID for the Emotional Impact Scale of the HPSVQ is 2 points (Hayward et al., 2021).

5. Intervention Safety - was defined as none of the participants experiencing a Serious Related Event, that is, a life-threatening or life-changing event that is due or is likely due to the intervention.

The study was registered with the ISRCTN Registry (number 16,003,237) on 4 March 2022 (https://doi.org/10.1186/ISRCTN16003237).

Participants

This study aimed to recruit 14 participants from the Secondary Care Mental Health Services within Sussex Partnership NHS Foundation Trust to provide an indication of feasibility. Allowing for attrition, this sample size was consistent with the recommendation of Julious (2005) that a group within a feasibility/pilot study should contain 12 participants. Potential participants were identified from consecutive referrals to the Sussex Voices Clinic who met the inclusion criteria.

Inclusion criteria

1. 18 years or older

2. under the care of Secondary Care Adult Services within Sussex Partnership NHS Foundation Trust

3. had a clinician-reported diagnosis of psychosis

4. experiencing current voice hearing – a score of at least 1 on item 1 (How frequently did you hear a voice or voices?) on the HPSVQ (Van Lieshout & Goldberg, 2007) – indicating that the participant has experienced at least one episode of voice hearing in the past week

5. distressed by hearing voices – scoring at least 8 out of 16 on the Emotional Impact Scale of the HPSVQ

6. willing and able to provide written/verbal, informed consent.

Exclusion criteria

1. established organic cause for voice hearing (e.g. brain disease or injury)

2. primary diagnosis of substance misuse

Measures

Impact of voices

• Hamilton Programme for Schizophrenia Voices Questionnaire (HPSVQ; Van Lieshout & Goldberg, 2007) – is a 9-item self-report measure assessing physical characteristics of voices (5 items) and the emotional impact of voices (4 items) across two scales. The HPSVQ has demonstrated internal consistency (Cronbach’s alpha = 0.74, 0.83 and 0.85 at baseline, 1 and 11 weeks, respectively; Van Lieshout & Goldberg, 2007) and a two-factor structure has recently been corroborated (Berry et al., 2021).

• Voice Impact Scale (VIS; Strauss et al., 2024) – is a 24-item self-report questionnaire measuring the “negative impact of voices”, the “positive impact of voices” and “living well with voices”. The three scales were confirmed by factor analysis. The VIS has been found to have high internal consistency (Cronbach’s alpha = 0.91), concurrent validity and is able to capture unique outcomes.

Recovery

• CHoice of Outcome In Cbt for psychosEs (CHOICE) – is a self-report measure assessing patient-defined recovery. The psychometric properties of the 12-item short form have recently been evaluated and results demonstrated high internal consistency (Cronbach’s alpha = 0.93), high levels of sensitivity to change and good construct validity (Webb et al., 2020).

Mechanisms of action

• The Brief Core Schema Scale (BCSS; Fowler et al., 2006) – a self-report measure demonstrating good internal consistency (Cronbach’s alpha = 0.75), test-retest reliability, and validity within a psychosis population (Fowler et al., 2006). This study used the 12-item scales measuring “negative self-schema” and “positive self-schema”.

• Beliefs About Voices Questionnaire – Revised (BAVQ-R) – a self-report measure of persecutory beliefs (Cronbach’s alpha = 0.88) and benevolent beliefs (Cronbach’s alpha = 0.87) over 14 items, recently found to have a 2-factor structure (Strauss et al., 2018).

• Approve-Voices and Approve-Social (Hayward et al., 2020) – are separate 15-item measures of relating to voices and relating to people within the social environment. Factor analysis revealed three subscales for each measure: passive, assertive and aggressive. The measures have been found to have robust psychometric properties regarding internal consistency (mean Cronbach’s alpha = 0.84), discriminant and criterion validity.

• Five-Facet Mindfulness Questionnaire − 15 (FFMQ-15; Gu et al., 2016) – a 15-item self-report scale assessing five mindfulness factors, with three items each: observing, describing, acting with awareness, non-judgement and non-reacting. The Observe subscale has poor validity in novice meditator samples and can be excluded when calculating a total score (Gu et al., 2016). The short form without the Observe subscale has adequate indices of internal consistency (Cronbach’s alpha = 0.73) and sensitivity to change.

Intervention protocol

Following the baseline assessment, all participants were offered the GiVE intervention delivered by an Assistant Psychologist (HA, HF or EV). Following the completion of the GiVE intervention, an expert consultation was offered by a Clinical Psychologist (MH) and supported the participant to select their next step within the intervention pathway (either leave the pathway or RT or PBCT). If selected, the second intervention was delivered by Clinical Psychologist (MH or CS). The intervention pathway was offered over a maximum of 20 sessions.

GiVE intervention

GiVE is an 8-session intervention that follows a workbook (Hazell, Hayward, Strauss, et al., 2018) which is based upon the Overcoming Distressing Voices self-help book (Hayward, Strauss, et al., 2018). Participants also had the opportunity to access the “CHOICES” mobile phone application (an “app” containing interactive pages which are structured in the same way as the workbook).

After an introductory session on coping, the intervention covered three core modules: (1) beliefs about the self, (2) beliefs about voices and (3) relationships. Modules (1) and (2) drew upon psychoeducation and cognitive behavioural strategies to help participants to re-evaluate their negative or unhelpful beliefs related to the self and voices. Module (3) involved work on how to relate to others and voices more assertively.

Please see Hazell et al. (2020) for a qualitative exploration of participants’ experience of the GiVE intervention.

Relating therapy

Relating Therapy (RT) is a 16-session intervention which supports patients to respond assertively within difficult relationships with voices and other people. Therapy is conducted over three phases:

• – Phase 1: socialization to RT and its implications for the interrelating between the hearer and the voice/other people. Consideration of the typical ways of responding to negative relating. Introduction of the possibility of relating differently to voices/other people.

• – Phase 2: exploration of themes within the relational history of the hearer and their experience of relationships with voices, and interpersonal relating within the family and social environment. Development of connections across all forms of relating.

• – Phase 3: exploration and development of assertive approaches to relating to voices/other people. Selection of a relationship to be the focus of treatment. Exploration of a current conversation within chosen relationship, responses within this conversation, identifying responses as passive, aggressive, or assertive and the generation of assertive responses. Experiential role-plays are used extensively within this phase to explore and practice relating in an assertive manner.

As RT was introduced to participants during Module 3 of the GiVE intervention, it was offered over a shorter 12-session format within the pathway.

Please see Hayward, Bogen-Johnston, et al. (2018) for a qualitative exploration of participants’ experience of Relating Therapy.

Person based cognitive therapy

Person Based Cognitive Therapy (PBCT) is a 12-session group intervention which draws upon cognitive behavioural strategies and mindfulness principles to facilitate acceptance of distressing voices and the re-evaluation of beliefs about self and voices.

All sessions begin with mindfulness practice and discussion. Mindfulness practice in PBCT is brief (10 min) with frequent guidance that includes reference to voice hearing experiences and combines focussed attention on body and breath with open awareness. Sessions 1–3 Socratically draw out participants’ voice hearing experiences (onset, impact, meaning, distress and coping) and frame them using the ABC cognitive model. Sessions 4–6 explore personal control, Socratically weakening voice omnipotence and enhancing autonomy. Sessions 7–12 add focus on identifying and decentring from negative schemata and building positive schematic beliefs (including using experiential chairwork) alongside recognition that the self is complex and changing. Participants are encouraged to practice mindfulness daily at home, using a supplied 10 minute recording.

PBCT is typically delivered within a group format. However, this format is not ideal for a pathway of interventions, as participants may be kept waiting for other participants to be ready to commence the group. Consequently, PBCT was adapted and offered individually over 12 sessions within this study.

Therapy sessions were conducted either face-to-face (at a mental health centre) or remotely (by phone or video call) in order to respect both patient choice and any restrictions imposed due to the COVID-19 pandemic.

Please see May et al. (2014) for a qualitative exploration of participants’ experience of PBCT.

Treatment-as-usual

Treatment as usual (TAU) was delivered according to national and local service protocols and guidelines and mainly consisted of antipsychotic medication and support and monitoring from the local clinical team. All participants were encouraged to engage in and continue with existing treatments throughout the duration of the study.

Training and supervision of therapists

The Assistant Psychologists attended two half-day workshops on the delivery of the GiVE intervention within the Sussex Voices Clinic. They received monthly supervision from a Clinical Psychologist (MH) during the study. The Clinical Psychologists (MH or CS) had extensive experience of delivering all of the interventions prior to the study. They offered each other regular peer supervision during the study.

Data analysis

Data were cleaned and checked for errors.

Descriptive statistics were calculated for all variables at all time points. Thereafter, analysis of the intention-to-treat and per-protocol samples were restricted, due to the varying number of monthly assessments across participants and for ease of interpretation, to baseline (pre-intervention), the monthly assessment following completion of the GiVE intervention (different for each participant – either month 2, 3 or 4) and the final assessment (post-intervention) following the completion of PBCT/RT. Baseline to final assessment, baseline to post-GiVE and post-GiVE to final assessment Cohen’s d effect sizes and their respective 95% confidence intervals were calculated.

As this was a feasibility study, data analysis focused on intervention completers (i.e. the per-protocol sample), however findings for the candidate primary outcome were also calculated for the intention-to-treat sample. Whilst inferential statistics are reported, namely repeated measures ANOVAs with three timepoints (baseline, post-GiVE and final assessment), the statistical significance of these tests are not foregrounded given the design did not include a power calculation and the study was not intended to determine the statistical significance of effects.

Results

Recruitment

Fifteen patients were referred to the study between 16/02/22 and 04/11/22 and 14 consented (one patient declined to participate as the intervention sessions could not be offered within their preferred format). Of the 14 participants, 7 identified as female and 7 as male. Their mean age was 42 years (SD = 1.41) and all identified as White. Most of the participants (N = 8) reported being in a long-term relationship. Diagnoses were primarily Schizophrenia Spectrum (N = 12; 2 participants were given a diagnosis of Bipolar Disorder) and the mean duration of voice hearing experiences was 14.3 years (SD = 4.24).

Retention

Of the 14 participants, 11 were intervention completers (79%), meeting the feasibility criterion of at least 75% of participants completing the pathway. Of the 3 who did not complete: one completed the GiVE intervention over an extended time period and was not offered an expert consultation due to the time constraints of the study; one completed the GiVE intervention, attended the expert consultation and chose not to receive a second intervention; and one completed the GiVE intervention and 6 sessions of RT before withdrawing from the study due to competing treatment priorities. See Figure 1 for the flow of participants through the trial.

Figure 1. Trial/Recruitment flowchart.

Of the 11 participants within the per-protocol sample, 7 chose RT as their second intervention and 4 chose PBCT. Each of the intervention completers attended 8 GiVE sessions and a mean of 10.8 of the 12 PBCT/RT sessions (SD = 2.83). Table 1 shows descriptive statistics for the 11 completers at baseline, post-GiVE and following the second intervention. The descriptive statistics for all time points are available from the corresponding author upon request.

Table 1. Descriptive statistics at baseline, post-GiVE intervention and at the final assessment session (N = 11 per protocol sample).

	Baseline mean (SD)	Post-GiVE mean (SD)	Final assessment mean (SD)	Cohen’s d – baseline to Final assessment (95% CI)*
Emotional Impact of Voices (HPSVQ)	11.73 (2.33)	7.82 (3.60)	4.45 (4.23)	1.51 (0.61 to 2.37)
Physical Characteristics of Voices (HPSVQ)	13.18 (2.23)	11.55 (2.11)	7.55 (6.22)	0.75 (0.06 to 1.41)
Voice Impact (VIS)	72.55 (41.91)	94.82 (31.00)	118.46 (24.45)	-0.92 (−1.62 to −0.19)
Recovery (CHOICE)	4.54 (1.80)	5.68 (1.60)	6.65 (1.76)	-0.74 (−1.39 to −0.05)
Negative self-schema (BCSS)	9.36 (6.39)	6.64 (6.17)	3.00 (3.35)	1.08 (0.31 to 1.82)
Positive self-schema (BCSS)	7.09 (5.58)	11.27 (4.80)	13.64 (4.97)	-0.84 (−1.51 to −0.13)
Persecutory beliefs about voices (BAVQ-R)	1.56 (0.91)	1.17 (0.85)	0.99 (1.02)	1.14 (0.35 to 1.89)
Benevolent beliefs about voices (BAVQ-R)	0.58 (0.79)	0.36 (0.69)	0.25 (0.72)	0.76 (0.07 to 1.42)
Assertive social relating (APPROVE-Social)	4.67 (2.33)	5.58 (2.18)	6.78 (1.72)	-0.78 (−1.44 to −0.08)
Aggressive social relating (APPROVE-Social)	2.35 (1.85)	1.65 (2.01)	1.18 (1.03)	0.62 (−0.04 to 1.26)
Passive social relating (APPROVE-Social)	5.00 (2.21)	4.38 (2.18)	2.98 (1.56)	0.87 (0.15 to 1.56)
Assertive relating to voices (APPROVE-Voices)	4.49 (2.71)	5.25 (2.45)	6.67 (2.11)	-0.58 (−1.21 to 0.07)
Aggressive relating to voices (APPROVE-Voices)	3.20 (2.06)	1.85 (1.82)	1.84 (1.62)	0.90 (0.17 to 1.59)
Passive relating to voices (APPROVE-Voices)	5.29 (2.40)	3.67 (2.80)	2.96 (2.34)	1.05 (0.29 to 1.78)
Mindfulness (FFMQ-15, minus Observe)	32.45 (3.83)	35.00 (8.83)	36.45 (6.86)	-0.64 (−1.28 to 0.03)

*Please note that effect sizes are all in the hypothesized direction. Positive effect sizes show a decrease in scores on measures where higher scores indicate greater distress. Negative effect sizes show an increase in scores on measures where higher scores indicates less distress.

Willingness to complete assessments

The cumulative number of assessment timepoints completed by participants were 4 (n = 14), 5 (n = 12), 6 (n = 11), 7 (n = 10), 8 (n = 8), 9 (n = 5), 10 (n = 2) and 11 (n = 1). This suggests that the number of timepoints included in the study was not feasible. Including 6 timepoints in future research seems feasible as 11 of the 14 participants (79%) completed assessments on at least 6 occasions. Where an assessment timepoint was attended, there was virtually no missing data suggesting that participants were willing to complete the battery of included measures when they attended an assessment.

Potential for effectiveness

Candidate primary outcome

An intention-to-treat (i.e. including all 14 participants) repeated measures one-way ANOVA comparing baseline scores on the HPSVQ Emotional Impact subscale (M = 12.07, SD = 2.23) with scores following the GiVE intervention (M = 8.93, SD = 3.87) and in the final assessment session following the second intervention (M = 6.14, SD = 5.01) showed a significant effect of time, F = 10.78, p < .001, with a large baseline to final assessment effect size, Cohen’s d = 1.18 [95% CI: 0.47 to 1.85]. Follow-up repeated contrasts showed a large and significant effect in improvement from baseline to following the GiVE intervention (F = 9.09, p = .010, d = 0.81 [95% CI: 0.19 to 1.40]) with a further non-significant improvement from post-GiVE to the final assessment with a medium effect size (F = 3.52, p = .070, d = 0.53 [95% CI: −0.04 to 1.08]).

A repeated measures one-way ANOVA on the per-protocol sample (N = 11) on the HPSVQ Emotional Impact subscale also showed a significant effect of time, F = 4.84, p = .019, with a large baseline to final assessment effect size, Cohen’s d = 1.51 [95% CI: 0.61 to 2.37]. Repeated contrasts showed a large and significant effect in improvement from baseline to following the GiVE intervention (F = 3.85, p = .078, d = 0.96 [95% CI: 0.22 to 1.66]) with a further non-significant improvement from post-GiVE to the final assessment with a medium effect size (F = 3.52, p = .090, d = 0.58 [95% CI: −0.08 to 1.20]). The MCID of 2 points was included in the 95% confidence interval for baseline-final assessment improvement, meeting the feasibility criteria for potential effectiveness.

Secondary clinical outcomes

For the per-protocol sample, effects on secondary clinical outcomes were all medium to large in size and in the direction of benefit. Cohen’s d effect sizes (with minus signs removed for clarity) for changes from baseline to the final assessment ranged from 0.74–0.92 (see Table 1 for descriptive statistics).

Mechanisms of action

For the per-protocol sample, effects on proposed mechanisms of action were all medium to large in size and in the direction of benefit. Cohen’s effect sizes (with minus signs removed for clarity) for changes from baseline to the final assessment ranged from 0.58–1.14 (see Table 1 for descriptive statistics).

Further details of the analysis of secondary outcomes and mechanisms of action are available from the corresponding author upon reasonable request.

Intervention safety

During the study, there were six Serious Adverse Events (SAEs) reported for four participants (3 × overdose, 2 × hospital admission for deterioration of physical health and 1 × hospital admission due to increased distress). Four of the six SAEs were classified by an independent medical monitor to be unrelated to the study and the remaining two were classified as unlikely to be related. This meets the feasibility criteria of none of the participants experiencing an SAE that is related to, or likely to be related to, the study or intervention.

Discussion

This feasibility study evaluated outcomes related to recruitment, retention, willingness to complete assessments, effectiveness and safety of the Feeling Heard intervention pathway. Outcomes showed promise for the pathway with all feasibility criteria being met. First, the targeted sample size for the study was recruited (above the 75% feasibility threshold). Second, retention was achieved as the pathway was completed by 79% of participants (above the 75% feasibility threshold). Third, 79% of participants completed assessments on at least 6 occasions (with 100% completing assessments on at least 4 occasions) suggesting that including multiple assessment timepoints is feasible, if limited to a maximum of 6. Moreover, participants were willing to complete the full battery of included measures at each assessment timepoint they attended, with virtually no missing data. Fourth, there was a large pre-post improvement on the candidate primary outcome of voice-related distress, as the MCID on the HPSVQ Emotional Impact Scale was included in the 95% confidence interval for the pre-post effect. Fifth, no serious adverse events were deemed to be related to, or likely to be related to the study or intervention by an independent medical monitor.

Evaluating the delivery of the Feeling Heard intervention pathway for distressing voices is feasible, as participants were recruited and retained within the study above the specified threshold. The 100% completion rates for the GiVE intervention offer further evidence for the willingness of psychosis patients to receive CBT-informed interventions from briefly trained therapists (Hayward et al., 2021).

The large pre-post effect size for the overall pathway on the candidate primary outcome of voice-related distress (Cohen’s d = 1.18 in intention-to-treat sample and 1.51 in per-protocol sample) is encouraging, and further supports the suggestion that the single-symptom approach can enhance the benefits of CBT-informed interventions for psychosis patients (Lincoln & Peters, 2019). Moreover, the second intervention (PBCT/RT) produced an additional medium sized improvement on the candidate primary outcome following the large effect of the first intervention (GiVE). Although uncontrolled, these findings suggest that adding an expert-delivered CBT-informed intervention for distressing voices to the GiVE intervention has the potential to improve outcomes further.

The profile of changes for the hypothesised mechanisms suggests that the pathway interventions may be targeting variables that can play a role in the maintenance of voice-related distress. A larger and adequately powered trial is required to explore any mediating or moderating roles of the hypothesised mechanisms before any conclusions can be drawn in this respect.

The study has several limitations. Firstly, the study was conducted within a specialist service for the treatment of distressing voices and findings may not generalise to non-specialist services. Secondly, the study was uncontrolled, and benefits may be attributable to factors beyond the intervention pathway. Thirdly, the assessments were conducted by researchers who could not be blinded, creating the potential for observer bias. Fourthly, follow-up data was not collected, leaving questions unanswered about the durability of benefits. Fifthly, the adherence of therapists to the intervention protocols was not assessed. This is particularly pertinent for Assistant Psychologists who have limited experience of delivering interventions. This issue is currently being explored within a full trial of the GiVE intervention delivered to psychosis patients by Assistant Psychologists (Hayward et al., 2023). Sixthly, the sample lacked diversity with regard to ethnicity and conclusions cannot be drawn about the appropriateness of the intervention pathway to all psychosis patients. Finally, we acknowledge that alternative psychological interventions could be included in the Feeling Heard pathway, e.g. AVATAR Therapy (Craig et al., 2018) as an alternative to Relating Therapy, due to its targeting of negative relating. Each of these limitations would need to be addressed within an adequately powered Randomised Controlled Trial of the Feeling Heard pathway.

Acknowledgement

We are grateful to the participants who supported our learning within this study. Thanks are also due to the Psychosis Theme Group at Sussex Partnership NHS Foundation Trust who supported the design of the study.

Disclosure statement

MH & CS are authors on the self-help book and workbook that were used to assist the delivery of the GiVE intervention within the study. Each author receives a small royalty for each copy of the self-help book and workbook that are sold.

Additional information

Funding

This work was supported by a South east Network for Social Sciences (SeNSS) Supervisor-led Collaborative Studentship, sustained by the Economic and Social Research Council [ES/P00072X/1].