Hostname: page-component-6bf8c574d5-m789k Total loading time: 0 Render date: 2025-03-09T21:30:59.776Z Has data issue: false hasContentIssue false
  • English
  • Français

The role of trauma, attachment, and voice-hearer’s appraisals: a latent profile analysis in the AVATAR2 trial

Published online by Cambridge University Press:  27 February 2025

Julia Marotti Open the ORCID record for Julia Marotti [Opens in a new window] ,
Rob Saunders ,
Alice Montague  and
Miriam Fornells-Ambrojo
Julia Marotti*
Affiliation:
Research Department of Clinical, Educational and Health Psychology, University College London, London, UK
Rob Saunders
Affiliation:
CORE Data Lab, Centre for Outcomes Research and Effectiveness (CORE), Research Department of Clinical, Educational and Health Psychology, University College London, London, UK
Alice Montague
Affiliation:
Research Department of Clinical, Educational and Health Psychology, University College London, London, UK North East London NHS Foundation Trust, London, UK
Miriam Fornells-Ambrojo
Affiliation:
Research Department of Clinical, Educational and Health Psychology, University College London, London, UK North East London NHS Foundation Trust, London, UK
*
Corresponding author: Julia Marotti; Email: zcjtjma@ucl.ac.uk
Rights & Permissions [Opens in a new window]

Abstract

Background

There is evidence that attachment, trauma, and voice appraisals individually impact voice hearing in psychosis, but their intersectional relationship has not been examined. The aim of this study was to identify subgroups of individuals from the intersectional relationship between these factors and examine differences between subgroups on clinical outcomes.

Methods

A latent profile analysis was conducted on baseline data from the AVATAR2 trial (n = 345), to identify statistically distinct subgroups of individuals with psychosis who hear distressing voices based on co-occurring patterns of trauma, fearful attachment, and voice appraisals. The association between profile membership and demographic characteristics, voice severity, posttraumatic stress disorder symptoms, emotional distress, and difficulties with motivation and pleasure was then examined. Experts by experience were consulted throughout the process.

Results

Four profiles were identified: ‘adverse voices and relational trauma’, ‘low malevolent and omnipotent voices’, ‘adverse voices yet low relational trauma’, and ‘high benevolent voices’. Negative voice appraisals occurred in the presence of high and low trauma and attachment adversities. The first profile was associated with being female and/or other non-male genders and had worse voice severity and emotional distress. High adversities and worse emotional distress occurred in the presence of voice benevolence and engagement. Black and South Asian ethnicities were not associated with specific profiles.

Conclusions

The identified profiles had negative and positive voice appraisals associated with higher and lower occurrence of adversities, and different clinical outcomes. These profiles could inform detailed case formulations that could tailor interventions for voice hearers.

Keywords

Auditory verbal hallucinationPsychosisTraumaAttachmentVoice appraisalsLatent Profile AnalysisAVATAR therapy
Type
Original Article
Information
Psychological Medicine , Volume 55 , 2025 , e65
Check for updates
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2025. Published by Cambridge University Press

Introduction

Auditory verbal hallucination or voice hearing in the absence of a corresponding external stimuli, referred to as “voices” henceforth, is understood to be on a continuum (Linscott & van Os, Reference Linscott and van Os2010). It has a lifetime prevalence rate of up to one in ten individuals in the general population (Maijer et al., Reference Maijer, Begemann, Palmen, Leucht and Sommer2018) associated with reduced distress (Taylor & Murray, Reference Taylor and Murray2012), and it occurs in a range of mental health disorders including psychosis, where voices are usually reported as more severe and distressing and linked with increased clinical needs (Larøi, Reference Larøi2012) and need for care (Johns et al., Reference Johns, Kompus, Connell, Humpston, Lincoln, Longden and Larøi2014). Voice distress in psychosis has been linked with increased anxiety, depression, and a higher degree of negative voice content (Scott, Rossell, Toh, & Thomas, Reference Scott, Rossell, Toh and Thomas2020b), and the presence of posttraumatic stress disorder (PTSD; de Bont et al., Reference de Bont, van den Berg, van der Vleugel, de Roos, de Jongh, van der Gaag and van Minnen2015).

Voices in psychosis have been associated with a history of different types of trauma (Bailey et al., Reference Bailey, Alvarez-Jimenez, Garcia-Sanchez, Hulbert, Barlow and Bendall2018; Grindey & Bradshaw, Reference Grindey and Bradshaw2022), with voice content at times directly reflecting trauma content (e.g., hearing the voice of a past abuser; van den Berg et al., Reference van den Berg, Tolmeijer, Jongeneel, Staring, Palstra, van der Gaag and Hardy2023). Traumatic stress-induced changes (i.e., traumagenic neurodevelopmental model; Pruessner et al., Reference Pruessner, Cullen, Aas and Walker2017) are implicated in emotional memories being encoded without contextual information (Brewin & Burgess, Reference Brewin and Burgess2014). Trauma memories might be experienced on a continuum of contextualized autobiographical memories and fragmented sensory experiences, such as voices, that are appraised as externally sourced when re-experienced (Hardy, Reference Hardy2017).

Disrupted care experiences during childhood, including neglect, abuse, and early losses, are involved in the development of fearful attachment (van Ijzendoorn et al., Reference van Ijzendoorn, Schuengel and Bakermans–Kranenburg1999), one of three types of insecure attachment patterns developed during early caregiving relationships (Ainsworth & Bell, Reference Ainsworth, Bell and Steinberg1981). A higher prevalence of fearful attachment has been observed in individuals with psychosis (Carr et al., Reference Carr, Hardy and Fornells-Ambrojo2018a), with significantly higher levels of hallucinations and voice distress (Bucci et al., Reference Bucci, Emsley and Berry2017).

Cognitive models of psychosis propose that childhood relational trauma and insecure attachment styles renders individuals more vulnerable to negative interpretations of self and others (Garety et al., Reference Garety, Kuipers, Fowler, Freeman and Bebbington2001; Scott, Rossell, Meyer, et al., Reference Scott, Rossell, Meyer, Toh and Thomas2020a). These interpretations, especially when originating from experiences of subordination and marginalization, influence hearers’ appraisals of voices as more malevolent, persecuting, powerful, critical and are related to greater voice distress (Birchwood et al., Reference Birchwood, Meaden, Trower, Gilbert and Plaistow2000; Larøi et al., Reference Larøi, Thomas, Aleman, Fernyhough, Wilkinson, Deamer and McCarthy-Jones2019). In non-clinical hearers, voices are reported as predominantly benevolent and less distressing (Daalman et al., Reference Daalman, Boks, Diederen, de Weijer, Blom, Kahn and Sommer2011).

Relational theories also implicate interpersonal frameworks and social world relationships in how the hearer responds to voices, influencing distress (Birchwood et al., Reference Birchwood, Gilbert, Gilbert, Trower, Meaden, Hay and Miles2004; Thomas et al., Reference Thomas, McLeod and Brewin2009). It is then important to consider social factors and their effect on voices experiences (e.g., culture-wide/culture−specific views of voices; Luhrmann et al., Reference Luhrmann, Padmavati, Tharoor and Osei2015), accounting also for how social factors’ associations to voices are attenuated by age, gender, and ethnicity (Bonoldi et al., Reference Bonoldi, Simeone, Rocchetti, Codjoe, Rossi, Gambi and Fusar-Poli2013). Thus, relational therapies aim to target distress by focusing on such relationships, enabling the individual to state their needs and gain awareness (Craig et al., Reference Craig, Rus-Calafell, Ward, Leff, Huckvale, Howarth and Garety2018; Hayward et al., Reference Hayward, Berry and Ashton2011).

Most research to date has investigated the interplay between trauma, attachment, and voice appraisals in isolation or paired associations, with debates over the specificity (i.e., one process leads to voices) and equifinality (i.e., voices may occur via different processes from a variety of different initial conditions) of the multiple mechanisms and factors involved (Gibson et al., Reference Gibson, Alloy and Ellman2016). Such traditional variable-centered studies provide estimated parameters indicating how factors are related in all individuals, which are assumed to be drawn from a single population, without considering how such factors may have different interactions which could differ across subpopulations of individuals (Morin et al., Reference Morin, Boudrias, Marsh, McInerney, Dagenais-Desmarais, Madore and Litalien2017). Conversely, person-centered analyses are data-driven (e.g., Begemann et al., Reference Begemann, Sommer, Brand, Oomen, Jongeneel, Berkhout and Bell2022), relaxing this assumption, and aimed to unearth multiple specific combinations of factors in cross-sectional data based on how they differ or are similar in traits and dimensions of interest (Saunders et al., Reference Saunders, Buckman and Pilling2020).

This study aimed to (1) use latent profile analysis to identify statistically distinct groups of individuals who hear distressing voices, based on the interplay between experiences of trauma, fearful attachment style and voice appraisals, and (2) explore whether identified profiles are differentially associated with demographic characteristics and clinical outcomes. Involvement of Experts by Experience was incorporated at all stages of the analysis, in line with recommendations in research (i.e., Corstens et al., Reference Corstens, Longden, McCarthy-Jones, Waddingham and Thomas2014; National Institute for Health and Care Research [NIHR], 2012) and with the aim that chosen measures and interpretation of findings had more meaningful and real-world clinical applications.

Methods

Setting

The baseline assessment data from the AVATAR2 multisite (London, Manchester, Glasgow) randomized controlled trial (RCT; see Garety et al., Reference Garety, Edwards, Ward, Emsley, Huckvale, McCrone and Craig2021 for detailed description) testing the efficacy of two versions of AVATAR therapy (Craig et al., Reference Craig, Rus-Calafell, Ward, Leff, Huckvale, Howarth and Garety2018; Leff et al., Reference Leff, Williams, Huckvale, Arbuthnot and Leff2014) against treatment-as-usual was used for this study. The study was approved by London-Camberwell St Giles Research Ethics Committee (REC/HRA ethical approval 20/LO/0657) which included consent for participation in ancillary/additional studies that utilize anonymized data collected, such as the current study. The recruitment, data collection of assessment meetings, and data storage were carried out according to the AVATAR2 trial protocol (please see Garety et al., Reference Garety, Edwards, Ward, Emsley, Huckvale, McCrone and Craig2021, Reference Garety, Edwards, Jafari, Emsley, Huckvale, Rus-Calafell and Ward2024 for further information). The baseline assessment meetings completed asked participants about their experiences of their voices, overall mood, and wellbeing via multiple measures.

Participants

The 345 participants included in this study met the inclusion and exclusion criteria of the AVATAR2 RCT (for full details, see Garety et al. (Reference Garety, Edwards, Ward, Emsley, Huckvale, McCrone and Craig2021) and Supplementary Material A).

Measures

Table 1 displays measures, their descriptions, and psychometric information, for the indicator variables included in the LPA, and the distal variables in post-LPA analyses, which are demographic characteristics and clinical presentation outcomes. For further details on measures and participant characteristics, see Table 1 and Supplementary Material B.

Table 1. Table with indicator and distal variables included in the latent profile analysis and further analyses and their measure description and properties (see Supplementary Material B for further details)

Analysis

Latent profile analysis

LPA is a mixture modeling approach extending latent class analysis (Vermunt & Magidson, Reference Vermunt, Magidson, Hagenaars and McCutcheon2002) to include both continuous and categorical variables (Gibson, Reference Gibson1959). LPA was conducted using Mplus version 8 (Muthén & Muthén, Reference Muthén and Muthén2017) evaluating whether there are distinct groups (latent profiles; LPs) of voice hearing individuals based on the observed individual response patterns on the indicator variables of appraisal of voices (as malevolent, omnipotent, benevolent, and voice engagement or resistance), the presence of fearful attachment style, traumatic experiences, and association of trauma-voice content. Further information can be found in Supplementary Material C about the selection of indicator variables and negotiating with model convergence constraints.

Model fit of the LPA models was compared using The Vuong-Lo-Medell-Rubin likelihood ratio test (VLMR-LRT; Lo et al., Reference Lo, Mendell and Rubin2001), and Bootstrap likelihood ratio difference test (B-LRT; Nylund et al., Reference Nylund, Asparouhov and Muthén2007) alongside Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and entropy-based criterion values (Geiser, Reference Geiser2013). FIML is used for systematically missing data in LPA. The full information about the model selection process can be found in Supplementary Material D.

Associations between profiles and both demographic information and clinical outcomes

Following the LPA, a series of analyses were performed to explore the associations between a) LPs and demographic variables (age, gender, ethnicity) and b) LPs and clinical presentation outcomes (voice severity, PTSD symptoms, level of global distress and motivation and pleasure difficulties).

To account for misclassification accuracy in LPA (Bakk et al., Reference Bakk, Tekle and Vermunt2013; Clark & Muthén, Reference Clark and Muthén2009), multinomial logistic regression analyses via the bias-adjusted R3STEP method (Asparouhov & Muthén, Reference Asparouhov and Muthén2014) was used, with the reference profile for associations in a) being the largest sample size. Specifically, the recommended Bolck-Croon-Hagenaars (BCH) procedure (Bolck et al., Reference Bolck, Croon and Hagenaars2004) was employed, with demographic variables (age, gender, ethnicity) included as confounders in models in b) examining differences in clinical outcomes across profiles (McLarnon & O’Neill, Reference McLarnon and O’Neill2018; Nylund-Gibson et al., Reference Nylund-Gibson, Grimm and Masyn2019). The profile-specific regression intercepts yielded were compared in omnibus Wald Chi-square tests and pairwise z-tests that suggest how belonging to one profile is differentially associated with a clinical outcome beyond the demographics and other profiles (Clark & Muthén, Reference Clark and Muthén2009). Missing data on demographic and clinical variables were handled using multiple imputation in Mplus (Section 11.1 of Asparouhov & Muthén, Reference Asparouhov and Muthén2021). For detailed information about these post-LPA analyses, see Supplementary Material E.

Patient and public involvement

In line with the Hearing Voices Movement (Corstens et al., Reference Corstens, Longden, McCarthy-Jones, Waddingham and Thomas2014) and the NIHR (2012) recommendations to involve people with lived experiences in research, patient and public involvement (PPI) has played a key role at all stages for co-production of the AVATAR2 trial, including design, recruitment of staff and participants, data collection, analysis, and dissemination. An active and creative group of people was established, comprising 20 members across all four sites, from diverse backgrounds, with lived experience of psychosis and recovery, and including carers. PPI co-production was continued in this study for decision-making processes and to ensure that interpretations of the complex findings were clinically useful and with real-world reflections.

First, one-to-one consultations were conducted with five PPI consultants at the pre-analysis stage. They provided feedback about the importance of the chosen indicator variables, the retention of measures (e.g., trauma-voice content association (i.e., TVAQ; Woods et al., Reference Woods, Jones, Alderson-Day, Callard and Fernyhough2015), and the voice appraisals measure (i.e., BAVQ-R (Chadwick et al., Reference Chadwick, Lees and Birchwood2000) and not the Voices Acceptance and Action Scale (Shawyer et al., Reference Shawyer, Ratcliff, Mackinnon, Farhall, Hayes and Copolov2007), or Voice Power Differential Scale (Birchwood et al., Reference Birchwood, Meaden, Trower, Gilbert and Plaistow2000) and informed the meaningful inclusion of ethnicity and gender demographics (i.e., exploring their association with the LPA model).

Consultations were also completed post-LPA-analysis with four PPI consultants. This explored their insights and understanding of the identified LPA groupings and their indicator variables’ distributions. A qualitative analysis of the rich feedback was beyond the scope of this study; thus, PPI reflections were incorporated when discussing the results. For further detailed information and summary of consultations, see Supplementary Material F.

Results

Descriptive statistics

The full sample of 345 participants had an average age of 39.6 years old, with a higher percentage of male gender (61.4%), ‘White’ ethnicity (59.7%), followed by ‘Black or mixed Black’ ethnicity (20.9%), and the majority had a Schizophrenia or Schizoaffective diagnosis (87.2%). The full sample results, from the RSQ and Mini-TALE, found a high rate of reported interpersonal adversities of fearful attachment and trauma (only 5.8% reported no trauma) alongside negative voice appraisals (i.e., omnipotence, malevolence) and resistance to voices, with lower occurrence of benevolent voices and engagement. Such indicator variables were mostly weakly to moderately associated. As expected, participants had high voice severity and emotional distress, high levels of motivation and pleasure difficulties (albeit just below cutoff score indicative of negative symptoms; Li et al., Reference Li, Li, Zou, Yang, Xie, Yang and Chan2018), and a PTSD dimensional score indicative of a middle level of symptom severity (M = 10.6 out of a total of 24). See full details of the full sample statistics in Supplementary Material G.

Latent profile analysis

The VLMRT-LRT and B-LRT yielded significant p-values (p < 0.05), alongside decreases in the AIC and BIC values, when comparing successive models from a two-profile to a four-profile solution. At the five-profile model, the VLMR-LRT p-value increased to above 0.05 (p = 0.198), and the entropy value for the four-profile versus the five-profile model was slightly higher, therefore the four-profile solution was considered optimal for the data (see Supplementary Material D for model comparison statistics).

Latent profiles descriptions

Descriptive statistics for the full sample and distributions for each latent profile (LP) are displayed in Table 2 (presented graphically in Figure 4 in Supplementary Material H), with the description of each profile provided here:

  1. 1. LP1 (largest; approximately 44.5%, n = 157) is described as ‘Adverse voices and relational trauma’ – In comparison with other profiles, individuals in this profile have the highest scores, which are higher as compared to the full sample, of fearful attachment style, trauma, and are more likely to believe that trauma and voices are related. Compared to the full sample, they report moderately higher on beliefs of voices being malevolent, omnipotent, and resisting voices, the highest scores comparative to other profiles. It has the lowest scores of benevolent voice appraisal and engagement, akin to LP3.

  2. 2. LP2 (second largest; approximately 25.2%, n = 84) is described as ‘Low malevolent and omnipotent voices’– Individuals in this profile are set apart by their scoring the lowest on all subscales relating to negative voice appraisals of omnipotence, malevolence and resistance to voices, compared to other profiles and the full sample. Compared to the full sample, individuals in this group have a marginally lower number of traumas experienced, traumas’ relatedness to voices and fearful attachment. Benevolent voice appraisal and engagement scores were low, similar to the full sample, however not as low as LP1 and LP3.

  3. 3. LP3 (second smallest; approximately 16.5%, n = 57) is described as ‘Adverse voices yet low relational trauma’ – Individuals in this profile are reporting the lowest scores across profiles for relational traumas. Compared to the full sample they have much lower experiences of fearful attachment, moderately a lower number of traumas reported, and where such traumas are not believed to be associated to voices heard. Although, such individuals are also scoring the second highest on subscales relating to omnipotent and malevolent voice appraisals, which is marginally higher than average full sample. Resistance to voices is similar to the high score in the full sample. Additionally, akin to LP1, they have a lower score of benevolent voice appraisals and engagement.

  4. 4. LP4 (smallest; approximately 13.8%, n = 47) is described as ‘High benevolent voices’ – Individuals in this profile are set apart by their scoring very high as compared to the full sample and other profiles on benevolent voice appraisal and engagement with voices. Although, they are similar to the full sample, with a higher number of traumas, beliefs of voices being related to traumas and experiencing fearful attachment and higher scores of omnipotent voice appraisals. Scores of malevolence voice appraisals and resistance to voices were low, lower than the full sample scores.

Table 2. Full sample and latent profiles indicator variable distribution

a see Supplementary Material B for full measures information.

Exploring profiles’ association with demographic information and clinical presentation outcomes

The next step was to analyse whether (a) LPs identified were associated with demographics (age, gender, ethnicity) and (b) LPs were differentially associated with outcome variables which represented participants’ clinical presentations (severity of voices, PTSD symptoms, global distress and motivation and pleasure difficulties). See Supplementary Material I for full statistical results of these analyses, with summaries presented later.

a) Demographic factors associated with profiles

The descriptive statistics for demographics are shown in Table 3, where there is a higher proportion of male gender across all profiles. LP1 ‘Adverse voices and relational trauma’ has the highest proportions of female and other non-male genders as compared to other profiles. The multinomial logistic regressions considering the association of demographics and LPs showed a significant association between individuals being female and other non-male genders and an increased likelihood of belonging to LP1 ‘Adverse voices and relational trauma’ as compared to LP2 ‘Low malevolent and omnipotent voices’ (OR [95% CI] = 0.472 [0.247; 0.901]). For ethnicity, only individuals within the ‘Any other’ ethnicity category, which has the lowest percentage, as compared to ‘White’, were significantly more likely to belong to LP4 ‘High benevolent voices’ as compared to LP1 ‘Adverse voices and relational trauma’ (reference profile; OR [95% CI] = 8.78 [2.75; 28.03]).

Table 3. Distribution of demographic covariates across profiles and full sample

a Please note that when considering statistical testing of the demographic covariate associations, due to sample size (i.e., only 4 other non-male respondents), female and other non-male gender categories were collapsed together, thus the analysis includes a binary variable of ‘male’ and ‘female and other non-male genders’.

b Please see Table 2 in Supplementary Material B for full details of ethnicity for each of the categories in line with the larger AVATAR2 RCT.

b) Clinical presentation outcomes associated with profiles

Table 4 shows the distribution of clinical outcomes in the full sample and across LPs. The profile-specific intercept regressions showed that all profiles have a unique influence on the distal outcomes independent of the influence of the demographic covariates accounted for in the model (treated as binary: male/female and other non-male genders), and white/not-white (i.e., all other ethnicities)). When comparing these profile-specific intercepts in equivalence omnibus chi-square tests, these indicated that LPs were only significantly different in their associations with voice severity (X 2(3) = 10.4, p = 0.015) and global distress outcomes (depression: X 2(3) = 8.4, p = 0.038*; anxiety: X 2(3) = 10.2, p = 0.017; stress: X 2(3) = 10.3, p = 0.016), yet not with PTSD (X 2(3) = 7.2, p = 0.066) and motivation and pleasure difficulties (X 2(3) = 6.7, p = 0.083) outcomes, after controlling for covariates’ influences.

Table 4. Latent profiles and associated clinical outcome distal variables

a see further information about measures in Supplementary Material B.

b DASS-21 is scored via doubling summed scores for each subscale given it is the short form for the scale.

c Missing cases are represented here for the descriptive statistics however multiple imputation was utilized for missing values on distal outcomes for the statistical analyses.

Thus, considering the pairwise z-tests for the significant outcomes, LP1 ‘Adverse voices and relational trauma’ closely followed by LP3 ‘Adverse voices yet low relational trauma’ (not statistically different from each other) had the highest voice severity scores, relative to the full sample, which were both statistically different in their influence of such outcomes when compared to LP2 ‘Low malevolent and omnipotent voices’ and LP4 ‘High benevolent voices’ (not significantly different between themselves) which were associated with lower voice severity scores. Additionally, LP1 ‘Adverse voices and relational trauma’ was significantly associated with individuals having the worst emotional distress (ranges: severe depression and stress, extremely severe anxiety) as compared to better outcomes in LP2 ‘Low malevolent and omnipotent voices’ (ranges: moderate depression and anxiety, mild stress). Specifically for depression, LP1 ‘Adverse voices and relational trauma’ and LP4 ‘High benevolent voices’ (second highest mean score of severe depression) are significantly different in their association with depression outcomes. See Figure 5 in Supplementary Material I for a graphical representation of these results.

Table 5 provides a visual summary of all findings.

Table 5. Summary of Latent Profile results and clinical presentation outcome with traffic light system representing severity*

Note: The color scheme represents a red-amber-green traffic light system to visually indicate severity of the clinical presentation outcomes across profiles (red: highest severity to green: lowest severity, with amber in between).

Discussion

This study identified four statistically different subgroups of participants based on their experiences of fearful attachment, trauma, beliefs about trauma-voice relatedness, and voice appraisals: LP1 ‘Adverse voices and relational trauma’, LP2 ‘Low malevolent and omnipotent voices’, LP3 ‘Adverse voices yet low relational trauma’, and LP4 ‘High benevolent voices’. Differential profile membership was significantly associated with outcomes of voice severity and global distress but not for other clinical outcomes (internal experience of motivation and pleasure and related behaviors and PTSD). LP1 and LP2 reflect the two ends of the severity spectrum of adversities and voice relationships theorized in various models of voices, whereas the latter two revealed profiles (LP3, LP4) suggest a novel intersection of voice experiences and adversity. There were no clear patterns of demographic effects across profiles, with only gender association with LP1 and no differences across the more represented minority groups of Black and Asian backgrounds. The reflections from PPI consultations are incorporated throughout the discussion of results.

Interpersonal adversity and impact on voices: support for existing models

In line with the existing models associating trauma and insecure attachment with voice appraisals and severity (e.g., Bailey et al., Reference Bailey, Alvarez-Jimenez, Garcia-Sanchez, Hulbert, Barlow and Bendall2018; Berry et al., Reference Berry, Varese and Bucci2017; Scott, Rossell, Meyer, et al., Reference Scott, Rossell, Meyer, Toh and Thomas2020a), the larger LP1 ‘Adverse voices and relational trauma’ had higher interpersonal adversities linked to negative voice appraisals with worst voice severity and emotional distress outcomes. LP2 ‘Low malevolent and omnipotent voices’ had better scores on clinical outcomes alongside less adverse factors linked to substantially lower negative voice appraisals. PPI consultants discussed how the experience of cumulative trauma has a role in voice appraisals of powerlessness and persecution, in line with relational models which posit that interpersonal adversities can be internalized, individuals see themselves lower in power, mirrored in their voice relating (Hayward, Reference Hayward2003; Thomas et al., Reference Thomas, McLeod and Brewin2009) and more distressed responses to voices (Pilton et al., Reference Pilton, Bucci, McManus, Hayward, Emsley and Berry2016). In this profile, trauma memories or dissociative trauma mechanisms, which can be predisposed by fearful attachment (Puckett et al., Reference Puckett, Sood and Newman-Taylor2023), could result in trauma-related memories encoded in fragmented ways, re-experienced as negative voices (Hardy, Reference Hardy2017; Pilton et al., Reference Pilton, Varese, Berry and Bucci2015).

The likelihood of belonging to the first more severe profile was associated with individuals being more likely to be female or other non-male genders as compared to the second profile. Increased severity and voice hearing distress has been reported in females in some studies (Murphy et al., Reference Murphy, Shevlin, Adamson and Houston2010; Suessenbacher-Kessler et al., Reference Suessenbacher-Kessler, Gmeiner, Diendorfer, Schrank, Unger and Amering2021; Toh et al., Reference Toh, Gurvich, Thomas, Tan, Neill and Van Rheenen2020), compared to males (see Barajas, Ochoa, Obiols, & Lalucat-Jo, Reference Barajas, Ochoa, Obiols and Lalucat-Jo2015 for less favorable outcomes in males), with passive relating to voices explaining this association (Schlier et al., Reference Schlier, Sitara, Strauss, Rammou, Lincoln and Hayward2021) and male gender norms promoting the minimization of reporting voices (Goldstein & Lewine, Reference Goldstein, Lewine, Castle, McGrath and Kulkarni2000), powerlessness, or distress (Parent, Hammer, Bradstreet, Schwartz, & Jobe, Reference Parent, Hammer, Bradstreet, Schwartz and Jobe2018).

Unexpected adversity and voice profiles

LP3 ‘Adverse voices yet low relational trauma’ and LP4 ‘High benevolent voices’ had an interesting combination of psychosocial factors, which may have been overlooked in the literature perhaps given previous studies’ different methodologies (e.g., isolated/paired psychosocial associations; Saunders et al., Reference Saunders, Buckman and Pilling2020). PPI consultants suggested that, where much lower fearful attachment and traumas in the LP3 group were reported, other ongoing environmental adversities (e.g., urbanicity, living in deprived areas) or difficult experiences (e.g., being from a minority group and experiencing subordination, discrimination, bullying), that were not captured in our demographic variables, can be impacting the individual. Where a continuing sense of threat results in hypervigilance which has been implicated in making negative information more salient shaping more negative voice-content (Larøi et al., Reference Larøi, Thomas, Aleman, Fernyhough, Wilkinson, Deamer and McCarthy-Jones2019). As such, reflected in LP3 voices being highly malign and omnipotent, with the second-worst voice severity outcome.

Alternative non-trauma routes to voices in LP3 could be a genetic/family history (van Winkel et al., Reference van Winkel, Van Nierop, Myin-Germeys and van Os2013), cognitive vulnerability, such as, strategies using punishment and worry for controlling unwanted thoughts, which are implicated with more psychological dysfunction and distress (Morrison & Wells, Reference Morrison and Wells2000), and low self-esteem (Williams et al., Reference Williams, Bucci, Berry and Varese2018). This group’s lower emotional distress could be linked with the lower fearful attachment suggesting more robust attachment styles (e.g., secure) may contribute to this group’s better emotion regulation and coping (e.g., lower emotional hyperactivity linked to positive psychosis symptom vulnerability; Berry et al., Reference Berry, Varese and Bucci2017). It may also be indicative of aspects not captured in this study, such as voice acceptance lowering negative distress from the lesser emotional resistance to voices, impacting voice distress (Varese et al., Reference Varese, Morrison, Beck, Heffernan, Law and Bentall2016) and psychological flexibility’s positive impact on general emotional wellbeing (Morris et al., Reference Morris, Garety and Peters2014).

LP4 ‘High benevolent voices’ has a substantially higher level of benevolent voice appraisals, highly engaged with, alongside reported high fearful attachment, trauma, and omnipotent voices, similar to the first profile. With less supportive attachment relationships to provide a buffer to traumas reported, PPI consultants and previous literature suggest that other experiences from an upbringing in cultures and religious contexts, with helpful positive beliefs in relation to voices, could provide a soothing or accepting stance for voice hearing (Larøi et al., Reference Larøi, Luhrmann, Bell, Christian, Deshpande, Fernyhough and Woods2014). Voices then can be experienced as powerful, as seen in this group, yet this power is reported as more benevolent (e.g., Cottam et al., Reference Cottam, Paul, Doughty, Carpenter, Al-Mousawi, Karvounis and Done2011). Of note, being of ‘Any other’ (and not other core minority Black and Asian backgrounds) rather than ‘White’ ethnicity was associated with this profile (as compared to LP1) and further information would be needed to interpret this finding.

In line with other studies showing a positive relationship between benevolent voices and lower voice distress (Sanjuan et al., Reference Sanjuan, Gonzalez, Aguilar, Leal and Van Os2004), the greater voice engagement may help lower the intensity and severity of voices in LP4 (Sayer et al., Reference Sayer, Ritter and Gournay2000), potentially due to a voice dialogue where the hearer is alongside the voice, rather than a victim, as suggested by PPIs. However, when considering LP4’s greater emotional distress, in line with psychological flexibility models, an appraisal of voices as omnipotent in this profile (regardless of it being a benevolent intention) can imply greater judgment toward these experiences and be responded to literally in subordinate ways which can impact emotional wellbeing (Gilbert et al., Reference Gilbert, Birchwood, Gilbert, Trower, Hay, Murray and Miles2001). Additionally, experiencing voices as friendly in this profile, especially in the context of potentially impoverished social interactions, may be associated with retreating to the voice relationship for comfort/companionship (Miller et al., Reference Miller, O’Connor and DiPasquale1993), impacting social functioning and seeking treatment support (Favrod et al., Reference Favrod, Grasset, Spreng, Grossenbacher and Hodé2004). Of importance, the current study’s participants are seeking treatment of a relational kind. Yet, the social withdrawal (LP4’s average motivation and pleasure difficulties being close to cutoff indicative of negative symptoms) may have associations with higher chronicity of symptoms, previously proposed to be that with more turbulent patients they are prone to more positive relationship with voices, and poorer outcomes including emotional distress (Favrod et al., Reference Favrod, Grasset, Spreng, Grossenbacher and Hodé2004).

Limitations

This is a cross-sectional study where LPA identifies subgroups dependent on the variables included which was limited to the available sample size (Tein et al., Reference Tein, Coxe and Cham2013; see Supplementary Material C). LPA is recognized as not sufficient to prove that profiles found will exist as tangible groups of people in other data samples, they provide guides for clinical consideration, but are not suggested as reified profiles (Williams & Kibowski, Reference Williams, Kibowski, Jason and Glenwick2016). It thus should be noted that this study sample is specific individuals distressed by persistent voices who view a relational AVATAR treatment approach as relevant to them which can influence findings, since other voice hearers might not conceptualize their voice experience as a relationship (e.g., Chin et al., Reference Chin, Hayward and Drinnan2009). Thus, this is not representative of all voice hearers, some of whom might not engage with services.

The trauma measure (i.e., TALE) used, although a clinically useful measure, requires further better-quality evidence (Airey et al., Reference Airey, Taylor, Vikram and Berry2023) which incurs limitations for the reliability of interpreting trauma associations in this study. The selection of the shorter (Mini-TALE) measure in the AVATAR2 trial assessment protocol (Garety et al., Reference Garety, Edwards, Ward, Emsley, Huckvale, McCrone and Craig2021, Reference Garety, Edwards, Jafari, Emsley, Huckvale, Rus-Calafell and Ward2024) was intended to minimize participant burden from using longer intrusive questionnaires (Fornells-Ambrojo et al., Reference Fornells-Ambrojo, Johns, Onwumere, Garety, Milosh, Iredale and Jolley2017). This measure captures only the cumulative elements of trauma, without differentiating childhood and adulthood trauma. Given evidence about the specific role of childhood trauma (Stanton et al., Reference Stanton, Denietolis, Goodwin and Dvir2020) and the cumulative or shared impact of adversity across different life stages (Pastore et al., Reference Pastore, de Girolamo, Tafuri, Tomasicchio and Margari2020; Trauelsen et al., Reference Trauelsen, Bendall, Jansen, Nielsen, Pedersen, Trier and Simonsen2015) in psychosis, the findings of the current study require replication with a more robust trauma measure that distinguishes trauma across the lifespan and includes other trauma types, which PPI feedback for all LPs highlighted as important, such as, poverty, discrimination, and urbanicity factors (Fett et al., Reference Fett, Lemmers-Jansen and Krabbendam2019; Varchmin et al., Reference Varchmin, Montag, Treusch, Kaminski and Heinz2021). There are inconsistent findings related to associations of trauma type and hallucination modality (c.f. Barnes et al., Reference Barnes, Emsley, Garety and Hardy2023). Thus, due to statistical constraints related to model convergence (see Supplementary Materials C for further consideration), this study was not able to separate the types of interpersonal traumas (c.f. Begemann et al., Reference Begemann, Sommer, Brand, Oomen, Jongeneel, Berkhout and Bell2022) since this would complicate the model and increase error where with the current sample size this would yield meaningless profiles. Non-trauma-related factors such as comorbid difficulties (e.g., anxiety and depression, autism spectrum disorder) could also have a mediating role for all profiles, perhaps impacting cognitive functioning and schemas of self-other, as such, consideration of such further factors are important and suggested by PPIs in future studies.

Interpretability of variables in a clinical utility sense were continually negotiated, however, for analysis ethnicity categories needed to be collapsed into mixed groups and binary codes making findings harder to interpret (i.e., association of ‘Any other’ ethnicity and increased likelihood of membership to LP4 ‘High benevolent voice’). This highlights the limitations in investigating the role of ethnicity within studies not specifically powered for this, as is the case of the current study.

Clinical and research implications

The current findings highlight the importance of thorough assessments of adverse experiences, alongside careful formulations of the meaning traumas have in reference to voices and how they relate to ways the voice hearer does/does not and has/has not been able to form safe attachments with others. Given profiles from a sample of individuals with distressing voices have both negative and positive voices, asking patients for detailed descriptions and interpretations of their voice could aid clinicians to not miss information, such as, benevolent appraisals and engagement with voices valued by individuals and where fewer interpersonal adversities co-occur with distressing negative appraisals. Further exploration of other factors with stronger influences, including other adversities and comorbid difficulties not measured in this study, should be explored.

These case formulations could support clinicians and service users to discuss appropriate tailored treatment. For example, where relational therapies address past relationships with abusive caregivers via how these are represented in the voice interaction, aiming to change relating behavior with voices (O’Brien et al., Reference O’Brien, Rus-Calafell, Craig, Garety, Ward, Lister and Fornells-Ambrojo2021). Additionally, incorporating what the individual values about benevolent voices, especially respecting cultural or religious aspects, can help interventions to shape the voice relationship to how the individual wants, so they can feel safer and are meeting their own needs.

The findings in this study would benefit from further research considering generalizability of current profiles in other psychosis populations and non-clinical voice hearers, since this is a sample of individuals distressed by persistent voices who were willing to participate in the AVATAR2 RCT. A longitudinal investigation of identified profiles and their association with outcomes would be informative. Such as, AVATAR2 studies considering whether identified profiles respond differently to the AVATAR treatments received, which may inform understanding of how best to target treatments to different groups. The invaluable collaboration from PPIs in this study, in shaping variable selection and meaning making of results, is an important demonstration of how Experts by Experience can and should be incorporated in future research so that this can refine psychological support to meet the needs of voice hearers. Details of cultural background, historical wellbeing prior to voices, relationship to help, protective family experiences, occupational context, and developmental stages (stress/pressure) were raised by PPIs which should be incorporated in further research.

Conclusion

This study uses LPA to provide new insights related to the complex interplay of interpersonal adversities co-occurring with positive and negative voice appraisals that are differentially associated with voice and emotional distress. Clinical information from this LPA can inform individualized assessments including careful consideration of voice appraisals, especially benevolent voices, and how these are linked to interpersonal adversities, to support decisions around helpful interventions. Information from the identified profiles could inform services, audits, and evaluations.

Supplementary material

The supplementary material for this article can be found at http://doi.org/10.1017/S003329172500008X.

Acknowledgments

The research was funded by the Wellcome Trust (FWBC-AVATAR 098272/z/12/z). The views expressed in this research are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.

Funding statement

This study is funded as part of the AVATAR2 Randomized Controlled Trial which is funded by The Wellcome Trust Ltd., through an Innovations Project award (grant reference 215471/Z/19/Z). The funding body has no role in the design of the study or the collection, analysis and interpretation of data or the writing of the manuscript. This research was part of JM’s Doctorate in Clinical Psychology thesis which received funding from the UCL Research Department of Clinical, Educational and Health Psychology.

Competing interest

All authors have no conflicts of interest or funding.

Ethical standard

The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.

References

Ainsworth, M. D. S., & Bell, S. M. (1981). Attachment, exploration, and separation: Illustrated by the behavior of one-year-olds in a strange situation. In Steinberg, L. D. (Ed.), The life cycle: Readings in human development (pp. 57–71). Columbia University Press.CrossRefGoogle Scholar
Airey, N. D., Taylor, C. D., Vikram, A., & Berry, K. (2023). Trauma measures for use with psychosis populations: A systematic review of psychometric properties using COSMIN. Psychiatry Research, 323(2023), 1–13. doi:10.1016/j.psychres.2023.115163CrossRefGoogle ScholarPubMed
Andrew, E. M., Gray, N. S., & Snowden, R. J. (2008). The relationship between trauma and beliefs about hearing voices: a study of psychiatric and non-psychiatric voice hearers. Psychological medicine, 38(10), 1409–1417. doi: 10.1017/S003329170700253XCrossRefGoogle ScholarPubMed
Asparouhov, T., & Muthén, B. (2014). Auxiliary variables in mixture modeling: Three-step approaches using Mplus. Mplus web notes, 15, 1–51. Available at: https://www.statmodel.com/download/webnotes/webnote15.pdfGoogle Scholar
Asparouhov, T., & Muthén, B. (2021). Auxiliary variables in mixture modeling: Using the BCH method in Mplus to estimate a distal outcome model and an arbitrary secondary model. Mplus web notes, 21, 1–80. Available at: https://www.statmodel.com/examples/webnotes/webnote21.pdfGoogle Scholar
Bailey, T., Alvarez-Jimenez, M., Garcia-Sanchez, A. M., Hulbert, C., Barlow, E., & Bendall, S. (2018). Childhood trauma is associated with severity of hallucinations and delusions in psychotic disorders: a systematic review and meta-analysis. Schizophrenia bulletin, 44(5), 1111–1122. doi:10.1093/schbul/sbx161CrossRefGoogle ScholarPubMed
Bakk, Z., Tekle, F. B., & Vermunt, J. K. (2013). Estimating the association between latent class membership and external variables using bias-adjusted three-step approaches. Sociological methodology, 43(1), 272–311. doi:10.1177/0081175012470644CrossRefGoogle Scholar
Barajas, A., Ochoa, S., Obiols, J. E., & Lalucat-Jo, L. (2015). Gender differences in individuals at high-risk of psychosis: a comprehensive literature review. The Scientific World Journal, 2015, 1–13. doi: 10.1155/2015/430735CrossRefGoogle ScholarPubMed
Barnes, G. L., Emsley, R., Garety, P., & Hardy, A. (2023). Investigating specific associations between childhood victimization profiles and positive psychosis symptoms: the mediating roles of anxiety, depression, and schema. Schizophrenia Bulletin Open, 4(1), sgad017.CrossRefGoogle ScholarPubMed
Begemann, M. J. H., Sommer, I. E., Brand, R. M., Oomen, P. P., Jongeneel, A., Berkhout, J., & Bell, I. H. (2022). Auditory verbal hallucinations and childhood trauma subtypes across the psychosis continuum: a cluster analysis. Cognitive Neuropsychiatry, 27(2–3), 150–168. doi:10.1080/13546805.2021.1925235CrossRefGoogle ScholarPubMed
Berry, K., Varese, F., & Bucci, S. (2017). Cognitive attachment model of voices: Evidence base and future implications. Frontiers in Psychiatry, 8(2017), 1–13. doi:10.3389/fpsyt.2017.00111CrossRefGoogle ScholarPubMed
Birchwood, M., Meaden, A., Trower, P., Gilbert, P., & Plaistow, J. (2000). The power and omnipotence of voices: subordination and entrapment by voices and significant others. Psychological medicine, 30(2), 337–344. doi:10.1017/s0033291799001828CrossRefGoogle ScholarPubMed
Birchwood, M. A. X., Gilbert, P., Gilbert, J., Trower, P., Meaden, A., Hay, J., & Miles, J. N. (2004). Interpersonal and role-related schema influence the relationship with the dominant ‘voice’ in schizophrenia: a comparison of three models. Psychological medicine, 34(8), 1571–1580. doi:10.1017/s0033291704002636CrossRefGoogle ScholarPubMed
Bolck, A., Croon, M., & Hagenaars, J. (2004). Estimating latent structure models with categorical variables: One-step versus three-step estimators. Political analysis, 12(1), 3–27. doi:10.1093/pan/mph001CrossRefGoogle Scholar
Bonoldi, I., Simeone, E., Rocchetti, M., Codjoe, L., Rossi, G., Gambi, F., & Fusar-Poli, P. (2013). Prevalence of self-reported childhood abuse in psychosis: a meta-analysis of retrospective studies. Psychiatry research, 210(1), 8–15. doi:10.1016/j.psychres.2013.05.003CrossRefGoogle ScholarPubMed
Brewin, C. R., & Burgess, N. (2014). Contextualisation in the revised dual representation theory of PTSD: A response to Pearson and colleagues. Journal of Behavior Therapy and Experimental Psychiatry, 45(1), 217–219. doi: 10.1016/j.jbtep.2013.07.011CrossRefGoogle ScholarPubMed
Bucci, S., Emsley, R., & Berry, K. (2017). Attachment in psychosis: A latent profile analysis of attachment styles and association with symptoms in a large psychosis cohort. Psychiatry Research, 247, 243–249. doi:10.1016/j.psychres.2016.11.036CrossRefGoogle Scholar
Carr, S., Hardy, A., & Fornells-Ambrojo, M. (2018b). The Trauma and Life Events (TALE) checklist: development of a tool for improving routine screening in people with psychosis. European journal of psychotraumatology, 9(1), 1–10. doi:10.1080/20008198.2018.1512265CrossRefGoogle ScholarPubMed
Carr, S. C., Hardy, A., & Fornells-Ambrojo, M. (2018a). Relationship between attachment style and symptom severity across the psychosis spectrum: A meta-analysis. Clinical Psychology Review, 59, 145–158. doi:10.1016/j.cpr.2017.12.001CrossRefGoogle ScholarPubMed
Chadwick, P., Lees, S., & Birchwood, M. A. X. (2000). The revised beliefs about voices questionnaire (BAVQ–R). The British Journal of Psychiatry, 177(3), 229–232. doi:10.1192/bjp.177.3.229CrossRefGoogle Scholar
Chin, J. T., Hayward, M., & Drinnan, A. (2009). Relating to voices: Exploring the relevance of this concept to people who hear voices. Psychology and Psychotherapy: Theory, Research and Practice, 82(1), 1–17.CrossRefGoogle ScholarPubMed
Clark, S. L., & Muthén, B. (2009). Relating latent class analysis results to variables not included in the analysis. Available at: https://www.statmodel.com/download/relatinglca.pdfGoogle Scholar
Cloitre, M., Hyland, P., Prins, A., & Shevlin, M. (2021). The international trauma questionnaire (ITQ) measures reliable and clinically significant treatment-related change in PTSD and complex PTSD. European Journal of Psychotraumatology, 12(1), 1–12.CrossRefGoogle ScholarPubMed
Cloitre, M., Shevlin, M., Brewin, C. R., Bisson, J. I., Roberts, N. P., Maercker, A., & Hyland, P. (2018). The International Trauma Questionnaire: Development of a self-report measure of ICD-11 PTSD and complex PTSD. Acta Psychiatrica Scandinavica, 138(6), 536–546. doi:10.1111/acps.12956CrossRefGoogle ScholarPubMed
Corstens, D., Longden, E., McCarthy-Jones, S., Waddingham, R., & Thomas, N. (2014). Emerging perspectives from the hearing voices movement: implications for research and practice. Schizophrenia bulletin, 40(Suppl_4), S285–S294. doi:10.1093/schbul/sbu007CrossRefGoogle ScholarPubMed
Cottam, S., Paul, S. N., Doughty, O. J., Carpenter, L., Al-Mousawi, A., Karvounis, S., & Done, D. J. (2011). Does religious belief enable positive interpretation of auditory hallucinations? A comparison of religious voice hearers with and without psychosis. Cognitive neuropsychiatry, 16(5), 403–421. doi:10.1080/13546805.2010.548543CrossRefGoogle ScholarPubMed
Craig, T. K., Rus-Calafell, M., Ward, T., Leff, J. P., Huckvale, M., Howarth, E., & Garety, P. A. (2018). AVATAR therapy for auditory verbal hallucinations in people with psychosis: a single-blind, randomised controlled trial. The Lancet Psychiatry, 5(1), 31–40. doi:10.1016/S2215-0366(17)30427-3CrossRefGoogle ScholarPubMed
Daalman, K., Boks, M. P., Diederen, K. M., de Weijer, A. D., Blom, J. D., Kahn, R. S., & Sommer, I. E. (2011). The same or different? A phenomenological comparison of auditory verbal hallucinations in healthy and psychotic individuals. The Journal of clinical psychiatry, 72(3), 320–325. doi:10.4088/JCP.09m05797yelCrossRefGoogle ScholarPubMed
de Bont, P. A., van den Berg, D. P., van der Vleugel, B. M., de Roos, C., de Jongh, A., van der Gaag, M., & van Minnen, A. (2015). Predictive validity of the Trauma Screening Questionnaire in detecting post-traumatic stress disorder in patients with psychotic disorders. The British Journal of Psychiatry, 206(5), 408–416. doi:10.1192/bjp.bp.114.148486CrossRefGoogle ScholarPubMed
Drake, R., Haddock, G., Tarrier, N., Bentall, R., & Lewis, S. (2007). The Psychotic Symptom Rating Scales (PSYRATS): their usefulness and properties in first episode psychosis. Schizophrenia research, 89(1–3), 119–122. doi:10.1016/j.schres.2006.04.024CrossRefGoogle ScholarPubMed
Favrod, J., Grasset, F., Spreng, S., Grossenbacher, B., & Hodé, Y. (2004). Benevolent voices are not so kind: the functional significance of auditory hallucinations. Psychopathology, 37(6), 304–308. doi:10.1159/000082269CrossRefGoogle Scholar
Fett, A. K. J., Lemmers-Jansen, I. L., & Krabbendam, L. (2019). Psychosis and urbanicity: a review of the recent literature from epidemiology to neurourbanism. Current opinion in psychiatry, 32(3), 232–241. doi:10.1097/YCO.0000000000000486CrossRefGoogle ScholarPubMed
Fornells-Ambrojo, M., Johns, L., Onwumere, J., Garety, P., Milosh, C., Iredale, C., & Jolley, S. (2017). Experiences of outcome monitoring in service users with psychosis: findings from an Improving Access to Psychological Therapies for people with Severe Mental Illness (IAPT-SMI) demonstration site. British Journal of Clinical Psychology, 56(3), 253–272. doi:10.1111/bjc.12136CrossRefGoogle ScholarPubMed
Garety, P., Edwards, C. J., Ward, T., Emsley, R., Huckvale, M., McCrone, P., & Craig, T. (2021). Optimising AVATAR therapy for people who hear distressing voices: study protocol for the AVATAR2 multi-centre randomised controlled trial. Trials, 22(1), 1–17. doi:10.1186/s13063-021-05301-wCrossRefGoogle ScholarPubMed
Garety, P. A., Edwards, C. J., Jafari, H., Emsley, R., Huckvale, M., Rus-Calafell, M., & Ward, T. (2024). Digital AVATAR therapy for distressing voices in psychosis: the phase 2/3 AVATAR2 trial. Nature Medicine, 1–11.Google ScholarPubMed
Garety, P. A., Kuipers, E., Fowler, D., Freeman, D., & Bebbington, P. E. (2001). A cognitive model of the positive symptoms of psychosis. Psychological medicine, 31(2), 189–195. doi:10.1017/s0033291701003312CrossRefGoogle ScholarPubMed
Geiser, C. (2013). Data Analysis with Mplus. Guilford, New York.Google Scholar
Gibson, L. E., Alloy, L. B., & Ellman, L. M. (2016). Trauma and the psychosis spectrum: A review of symptom specificity and explanatory mechanisms. Clinical Psychology Review, 49, 92–105. doi:10.1016/j.cpr.2016.08.003CrossRefGoogle ScholarPubMed
Gibson, W. A. (1959). Three multivariate models: Factor analysis, latent structure analysis, and latent profile analysis. Psychometrika, 24, 229–252. doi:10.1007/BF02289845CrossRefGoogle Scholar
Gilbert, P., Birchwood, M., Gilbert, J., Trower, P., Hay, J., Murray, B., & Miles, J. N. V. (2001). An exploration of evolved mental mechanisms for dominant and subordinate behaviour in relation to auditory hallucinations in schizophrenia and critical thoughts in depression. Psychological medicine, 31(6), 1117–1127.CrossRefGoogle ScholarPubMed
Goldstein, J. & Lewine, R. (2000). Overview of the sex differences in schizophrenia. In Castle, D., McGrath, J., & Kulkarni, J. (Eds.), Women and Schizophrenia (pp. 57–68). Cambridge University Press.Google Scholar
Griffin, D. W., & Bartholomew, K. (1994). Models of the self and other: Fundamental dimensions underlying measures of adult attachment. Journal of Personality and Social Psychology, 67(3), 430–445. doi:10.1037/0022-3514.67.3.430CrossRefGoogle Scholar
Grindey, A., & Bradshaw, T. (2022). Do different adverse childhood experiences lead to specific symptoms of psychosis in adulthood? A systematic review of the current literature. International journal of mental health nursing, 31(4), 868–887. doi:10.1111/inm.12992CrossRefGoogle ScholarPubMed
Haddock, G., McCarron, J., Tarrier, N., & Faragher, E. B. (1999). Scales to measure dimensions of hallucinations and delusions: the psychotic symptom rating scales (PSYRATS). Psychological medicine, 29(4), 879–889. doi:10.1017/s0033291799008661CrossRefGoogle ScholarPubMed
Hardy, A. (2017). Pathways from trauma to psychotic experiences: a theoretically informed model of posttraumatic stress in psychosis. Frontiers in psychology, 8, 1–20. doi:10.3389/fpsyg.2017.00697CrossRefGoogle ScholarPubMed
Hayward, M. (2003). Interpersonal relating and voice hearing: to what extent does relating to the voice reflect social relating?. Psychology and Psychotherapy: Theory, Research and Practice, 76(4), 369–383.CrossRefGoogle Scholar
Hayward, M., Berry, K., & Ashton, A. (2011). Applying interpersonal theories to the understanding of and therapy for auditory hallucinations: a review of the literature and directions for further research. Clinical psychology review, 31(8), 1313–1323. doi:10.1016/j.cpr.2011.09.001CrossRefGoogle Scholar
Henry, J. D., & Crawford, J. R. (2005). The short-form version of the Depression Anxiety Stress Scales (DASS-21): Construct validity and normative data in a large non-clinical sample. British journal of clinical psychology, 44(2), 227–239. doi:10.1348/014466505X29657CrossRefGoogle Scholar
Horan, W. P., Kring, A. M., Gur, R. E., Reise, S. P., & Blanchard, J. J. (2011). Development and psychometric validation of the Clinical Assessment Interview for Negative Symptoms (CAINS). Schizophrenia research, 132(2–3), 140–145.CrossRefGoogle ScholarPubMed
Hyland, P., Shevlin, M., Brewin, C. R., Cloitre, M., Downes, A. J., Jumbe, S., & Roberts, N. P. (2017). Validation of post-traumatic stress disorder (PTSD) and complex PTSD using the International Trauma Questionnaire. Acta Psychiatrica Scandinavica, 136(3), 313–322.CrossRefGoogle ScholarPubMed
Johns, L. C., Kompus, K., Connell, M., Humpston, C., Lincoln, T. M., Longden, E., & Larøi, F. (2014). Auditory verbal hallucinations in persons with and without a need for care. Schizophrenia bulletin, 40(Suppl_4), S255–S264.CrossRefGoogle ScholarPubMed
Kring, A. M., Gur, R. E., Blanchard, J. J., Horan, W. P., & Reise, S. P. (2013). The clinical assessment interview for negative symptoms (CAINS): final development and validation. American journal of psychiatry, 170(2), 165–172. doi:10.1176/appi.ajp.2012.12010109CrossRefGoogle ScholarPubMed
Larøi, F. (2012). How do auditory verbal hallucinations in patients differ from those in non-patients?. Frontiers in human neuroscience, 6(25), 1–9. doi:10.3389/fnhum.2012.00025CrossRefGoogle ScholarPubMed
Larøi, F., Luhrmann, T. M., Bell, V., Christian, W. A. Jr, Deshpande, S., Fernyhough, C., & Woods, A. (2014). Culture and hallucinations: overview and future directions. Schizophrenia bulletin, 40(Suppl_4), S213–S220. doi:10.1093/schbul/sbu012CrossRefGoogle ScholarPubMed
Larøi, F., Thomas, N., Aleman, A., Fernyhough, C., Wilkinson, S., Deamer, F., & McCarthy-Jones, S. (2019). The ice in voices: Understanding negative content in auditory-verbal hallucinations. Clinical psychology review, 67, 1–10. doi:10.1016/j.cpr.2018.11.001CrossRefGoogle ScholarPubMed
Lee, J., Lee, E. H., & Moon, S. H. (2019). Systematic review of the measurement properties of the Depression Anxiety Stress Scales–21 by applying updated COSMIN methodology. Quality of Life Research, 28, 2325–2339.CrossRefGoogle ScholarPubMed
Leff, J., Williams, G., Huckvale, M., Arbuthnot, M., & Leff, A. P. (2014). Avatar therapy for persecutory auditory hallucinations: What is it and how does it work?. Psychosis, 6(2), 166–176. doi:10.1080/17522439.2013.773457CrossRefGoogle ScholarPubMed
Li, Y., Li, W. X., Zou, Y. M., Yang, Z. Y., Xie, D. J., Yang, Y., & Chan, R. C. (2018). Revisiting the persistent negative symptoms proxy score using the clinical assessment interview for negative symptoms. Schizophrenia research, 202, 248–253. doi:10.1016/j.schres.2018.07.005CrossRefGoogle ScholarPubMed
Linscott, R. J., & van Os, J. (2010). Systematic reviews of categorical versus continuum models in psychosis: evidence for discontinuous subpopulations underlying a psychometric continuum. Implications for DSM-V, DSM-VI, and DSM-VII. Annual review of clinical psychology, 6, 391–419. doi:10.1146/annurev.clinpsy.032408.153506CrossRefGoogle ScholarPubMed
Lo, Y., Mendell, N. R., & Rubin, D. B. (2001). Testing the number of components in a normal mixture. Biometrika, 88(3), 767–778. doi:10.1093/biomet/88.3.767CrossRefGoogle Scholar
Luhrmann, T. M., Padmavati, R., Tharoor, H., & Osei, A. (2015). Differences in voice-hearing experiences of people with psychosis in the USA, India and Ghana: interview-based study. The British Journal of Psychiatry, 206(1), 41–44. doi:10.1192/bjp.bp.113.139048CrossRefGoogle ScholarPubMed
Maijer, K., Begemann, M. J. H., Palmen, S. J. M., Leucht, S., & Sommer, I. E. C. (2018). Auditory hallucinations across the lifespan: a systematic review and meta-analysis. Psychological Medicine, 48(6), 879–888. doi:10.1017/S0033291717002367CrossRefGoogle ScholarPubMed
McLarnon, M. J., & O’Neill, T. A. (2018). Extensions of auxiliary variable approaches for the investigation of mediation, moderation, and conditional effects in mixture models. Organizational Research Methods, 21(4), 955–982. doi:10.1177/1094428118770731CrossRefGoogle Scholar
Miller, L. J., O’Connor, E., & DiPasquale, T. (1993). Patients’ attitudes toward hallucinations. American Journal of Psychiatry, 150, 584–584. doi:10.1176/ajp.150.4.584Google ScholarPubMed
Morin, A. J., Boudrias, J. S., Marsh, H. W., McInerney, D. M., Dagenais-Desmarais, V., Madore, I., & Litalien, D. (2017). Complementary variable-and person-centered approaches to the dimensionality of psychometric constructs: Application to psychological wellbeing at work. Journal of Business and Psychology, 32, 395–419. doi:10.1007/s10869-016-9448-7CrossRefGoogle Scholar
Morris, E. M. J., Garety, P., Peters, E. (2014). Psychological flexibility and nonjudgemental acceptance in voice hearers: relationships with omnipotence and distress. Australian & New Zealand Journal of Psychiatry, 48(12), 1150–1162. doi: 10.1177/0004867414535671CrossRefGoogle ScholarPubMed
Morrison, A. P., & Wells, A. (2000). Thought control strategies in schizophrenia: a comparison with non-patients. Behaviour Research and Therapy, 38(12), 1205–1209. doi: 10.1016/s0005-7967(99)00153-9CrossRefGoogle ScholarPubMed
Murphy, J., Shevlin, M., Adamson, G., & Houston, J. E. (2010). Positive psychosis symptom structure in the general population: Assessing dimensional consistency and continuity from “pathology” to “normality.” Psychosis, 2, 199–209. doi: 10.1080/17522430903437087.CrossRefGoogle Scholar
Muthén, L.K. & Muthén, B.O. (2017). Mplus user’s guide. Eighth Edition.Google Scholar
National Institute for Health Research (NIHR) (2012). Briefing notes for researchers: public involvement in NHS, public health and social care research. Available at: https://www.nihr.ac.uk/documents/briefing-notes-for-researchers-public-involvement-in-nhs-health-and-social-care-research/27371Google Scholar
Ng, F., Trauer, T., Dodd, S., Callaly, T., Campbell, S., & Berk, M. (2007). The validity of the 21-item version of the depression anxiety stress scales as a routine clinical outcome measure. Acta neuropsychiatrica, 19(5), 304–310.CrossRefGoogle ScholarPubMed
Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural equation modeling: A multidisciplinary Journal, 14(4), 535–569. doi:10.1080/10705510701575396CrossRefGoogle Scholar
Nylund-Gibson, K., Grimm, R. P., & Masyn, K. E. (2019). Prediction from latent classes: A demonstration of different approaches to include distal outcomes in mixture models. Structural equation modeling: A multidisciplinary Journal, 26(6), 967–985. doi:10.1080/10705511.2019.1590146CrossRefGoogle Scholar
O’Brien, C., Rus-Calafell, M., Craig, T. K., Garety, P., Ward, T., Lister, R., & Fornells-Ambrojo, M. (2021). Relating behaviours and therapeutic actions during AVATAR therapy dialogue: An observational study. British Journal of Clinical Psychology, 60(4), 443–462. doi:10.1111/bjc.12296CrossRefGoogle ScholarPubMed
Parent, M. C., Hammer, J. H., Bradstreet, T. C., Schwartz, E. N., & Jobe, T. (2018). Men’s mental health help-seeking behaviors: An intersectional analysis. American journal of men’s health, 12(1), 64–73. doi: 10.1177/1557988315625776CrossRefGoogle ScholarPubMed
Pastore, A., de Girolamo, G., Tafuri, S., Tomasicchio, A., & Margari, F. (2020). Traumatic experiences in childhood and adolescence: a meta-analysis of prospective studies assessing risk for psychosis. European Child & Adolescent Psychiatry, 31, 215–228. doi:10.1007/s00787-020-01574-9CrossRefGoogle ScholarPubMed
Pilton, M., Bucci, S., McManus, J., Hayward, M., Emsley, R., & Berry, K. (2016). Does insecure attachment mediate the relationship between trauma and voice-hearing in psychosis?. Psychiatry research, 246, 776–782.CrossRefGoogle ScholarPubMed
Pilton, M., Varese, F., Berry, K., & Bucci, S. (2015). The relationship between dissociation and voices: A systematic literature review and meta-analysis. Clinical Psychology Review, 40, 138–155.CrossRefGoogle ScholarPubMed
Pruessner, M., Cullen, A. E., Aas, M., & Walker, E. F. (2017). The neural diathesis-stress model of schizophrenia revisited: An update on recent findings considering illness stage and neurobiological and methodological complexities. Neuroscience & Biobehavioral Reviews, 73, 191–218. doi:10.1016/j.neubiorev.2016.12.013CrossRefGoogle ScholarPubMed
Puckett, J., Sood, M., & Newman-Taylor, K. (2023). Does disorganised attachment lead to auditory hallucinations via dissociation? An experimental study with an analogue sample. Psychology and Psychotherapy: Theory, Research and Practice, 2023(00) 1–17.Google Scholar
Ratcliff, K., Farhall, J., & Shawyer, F. (2011). Auditory hallucinations: a review of assessment tools. Clinical Psychology & Psychotherapy, 18(6), 524–534. doi:10.1002/cpp.729CrossRefGoogle ScholarPubMed
Sanjuan, J., Gonzalez, J. C., Aguilar, E. J., Leal, C., & Van Os, J. (2004). Pleasurable auditory hallucinations. Acta Psychiatrica Scandinavica, 110(4), 273–278. doi:10.1111/j.1600-0447.2004.00336.xCrossRefGoogle ScholarPubMed
Saunders, R., Buckman, J. E., & Pilling, S. (2020). Latent variable mixture modelling and individual treatment prediction. Behaviour research and therapy, 124(2020), 1–11. doi:10.1016/j.brat.2019.103505CrossRefGoogle ScholarPubMed
Sayer, J., Ritter, S., & Gournay, K. (2000). Beliefs about voices and their effects on coping strategies. Journal of Advanced Nursing, 31(5), 1199–1205. doi:10.1046/j.1365-2648.2000.01375.xCrossRefGoogle ScholarPubMed
Schlier, B., Sitara, X., Strauss, C., Rammou, A., Lincoln, T. M., & Hayward, M. (2021). Can gender differences in distress due to difficult voices be explained by differences in relating?. Cognitive Therapy and Research, 45, 831–839.CrossRefGoogle Scholar
Schmitt, D. P., Alcalay, L., Allensworth, M., Allik, J., Ault, L., Austers, I., & ZupanÈiÈ, A. (2004). Patterns and universals of adult romantic attachment across 62 cultural regions: Are models of self and of other pancultural constructs?. Journal of Cross-Cultural Psychology, 35(4), 367–402. doi:10.1177/0022022104266105CrossRefGoogle Scholar
Scott, M., Rossell, S. L., Meyer, D., Toh, W. L., & Thomas, N. (2020a). Childhood trauma, attachment and negative schemas in relation to negative auditory verbal hallucination (AVH) content. Psychiatry research, 290, 1–7. doi:10.1016/j.psychres.2020.112997CrossRefGoogle ScholarPubMed
Scott, M., Rossell, S. L., Toh, W. L., & Thomas, N. (2020b). The relationship between anxiety, depression, and subtypes of negative auditory verbal hallucination (AVH) content in affective and non-affective psychosis. Psychiatry Research, 294 113500. doi: 10.1016/j.psychres.2020.113500CrossRefGoogle ScholarPubMed
Shawyer, F., Ratcliff, K., Mackinnon, A., Farhall, J., Hayes, S. C., & Copolov, D. (2007). The voices acceptance and action scale (VAAS): Pilot data. Journal of clinical psychology, 63(6), 593–606. doi:10.1002/jclp.20366CrossRefGoogle ScholarPubMed
Stanton, K. J., Denietolis, B., Goodwin, B. J., & Dvir, Y. (2020). Childhood trauma and psychosis: an updated review. Child and Adolescent Psychiatric Clinics, 29(1), 115–129. doi:10.1016/j.chc.2019.08.004CrossRefGoogle ScholarPubMed
Suessenbacher-Kessler, S., Gmeiner, A., Diendorfer, T., Schrank, B., Unger, A., & Amering, M. (2021). A relationship of sorts: gender and auditory hallucinations in schizophrenia spectrum disorders. Archives of Women’s Mental Health, 1–12.Google ScholarPubMed
Taylor, G., & Murray, C. (2012). A qualitative investigation into non-clinical voice hearing: what factors may protect against distress?. Mental Health, Religion & Culture, 15(4), 373–388.CrossRefGoogle Scholar
Tein, J. Y., Coxe, S., & Cham, H. (2013). Statistical power to detect the correct number of classes in latent profile analysis. Structural equation modeling: a multidisciplinary journal, 20(4), 640–657. doi:10.1080/10705511.2013.824781CrossRefGoogle ScholarPubMed
Thomas, N., McLeod, H. J., & Brewin, C. R. (2009). Interpersonal complementarity in responses to auditory hallucinations in psychosis. British Journal of Clinical Psychology, 48(4), 411–424. doi:10.1348/014466509X411937CrossRefGoogle ScholarPubMed
Toh, W. L., Gurvich, C., Thomas, N., Tan, E. J., Neill, E., Van Rheenen, T., et al. (2020). The infuence of gender on emotional aspects of auditory verbal hallucinations. Psychiatry Research, 284, 112642. https://doi.org/10.1016/j.psychres.2019.112642CrossRefGoogle Scholar
Tolmeijer, E., Hardy, A., Jongeneel, A., Staring, A. B., van der Gaag, M., & van den Berg, D. (2021). Voice-hearers’ beliefs about the causes of their voices. Psychiatry Research, 302(2021), 1–3. doi:10.1016/j.psychres.2021.113997CrossRefGoogle ScholarPubMed
Trauelsen, A. M., Bendall, S., Jansen, J. E., Nielsen, H. G. L., Pedersen, M. B., Trier, C. H., & Simonsen, E. (2015). Childhood adversity specificity and dose-response effect in non-affective first-episode psychosis. Schizophrenia research, 165(1), 52–59. doi:10.1016/j.schres.2015.03.014CrossRefGoogle ScholarPubMed
van den Berg, D., Tolmeijer, E., Jongeneel, A., Staring, A. B., Palstra, E., van der Gaag, M., & Hardy, A. (2023). Voice phenomenology as a mirror of the past. Psychological Medicine, 53(7), 2954–2962. doi:10.1017/S0033291721004955CrossRefGoogle ScholarPubMed
van Ijzendoorn, M. H., Schuengel, C., & Bakermans–Kranenburg, M. J. (1999). Disorganized attachment in early childhood: Meta-analysis of precursors, concomitants, and sequelae. Development and psychopathology, 11(2), 225–250. doi:10.1017/s0954579499002035CrossRefGoogle ScholarPubMed
van Winkel, R., Van Nierop, M., Myin-Germeys, I., & van Os, J. (2013). Childhood trauma as a cause of psychosis: linking genes, psychology, and biology. The Canadian Journal of Psychiatry, 58(1), 44–51. doi:10.1177/070674371305800109CrossRefGoogle ScholarPubMed
Varchmin, L., Montag, C., Treusch, Y., Kaminski, J., & Heinz, A. (2021). Traumatic events, social adversity and discrimination as risk factors for psychosis-an umbrella review. Frontiers in psychiatry, 12, 665957. doi:10.3389/fpsyt.2021.665957CrossRefGoogle Scholar
Varese, F., Morrison, A. P., Beck, R., Heffernan, S., Law, H., & Bentall, R. P. (2016). Experiential avoidance and appraisals of voices as predictors of voice-related distress. British Journal of Clinical Psychology, 55(3), 320–331. doi: 10.1111/bjc.12102CrossRefGoogle ScholarPubMed
Vermunt, J. K., & Magidson, J. (2002). Latent class cluster analysis. In Hagenaars, J., & McCutcheon, A. (Eds.), Applied latent class analysis (pp. 89–106). Cambridge University Press.CrossRefGoogle Scholar
Williams, G. A., & Kibowski, F. (2016). Latent class and latent profile analysis. In Jason, L. A., & Glenwick, D. S. (Eds.), Handbook of methodological approaches to community-based research: Qualitative, quantitative, and mixed methods (pp.). Oxford University Press.Google Scholar
Williams, J., Bucci, S., Berry, K., & Varese, F. (2018). Psychological mediators of the association between childhood adversities and psychosis: A systematic review. Clinical psychology review, 65, 175–196. doi:10.1016/j.cpr.2018.05.009CrossRefGoogle ScholarPubMed
Wongpakaran, N., DeMaranville, J., & Wongpakaran, T. (2021). Validation of the relationships questionnaire (RQ) against the experience of close relationship-revised questionnaire in a clinical psychiatric sample. Healthcare, 9(9), 1–13. doi:10.3390/healthcare9091174CrossRefGoogle Scholar
Woods, A., Jones, N., Alderson-Day, B., Callard, F., & Fernyhough, C. (2015). Experiences of hearing voices: analysis of a novel phenomenological survey. The Lancet Psychiatry, 2(4), 323–331. doi:10.1016/S2215-0366(15)00006-1CrossRefGoogle ScholarPubMed
Figure 0

Table 1. Table with indicator and distal variables included in the latent profile analysis and further analyses and their measure description and properties (see Supplementary Material B for further details)

Figure 1

Table 2. Full sample and latent profiles indicator variable distribution

Figure 2

Table 3. Distribution of demographic covariates across profiles and full sample

Figure 3

Table 4. Latent profiles and associated clinical outcome distal variables

Figure 4

Table 5. Summary of Latent Profile results and clinical presentation outcome with traffic light system representing severity*

Supplementary material: File

Marotti et al. supplementary material

Marotti et al. supplementary material
Download Marotti et al. supplementary material(File)
File 2.9 MB