This electronic thesis or dissertation has been downloaded from the King’s Research Portal at https://kclpure.kcl.ac.uk/portal/

Social influences on dissociative processes in psychosis

Heriot-Maitland, Charlie

Awarding institution: King's College London

The copyright of this thesis rests with the author and no quotation from it or information derived from it may be published without proper acknowledgement.

END USER LICENCE AGREEMENT

Unless another licence is stated on the immediately following page this work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ You are free to copy, distribute and transmit the work Under the following conditions:  Attribution: You must attribute the work in the manner specified by the author (but not in any way that suggests that they endorse you or your use of the work).  Non Commercial: You may not use this work for commercial purposes.  No Derivative Works - You may not alter, transform, or build upon this work.

Any of these conditions can be waived if you receive permission from the author. Your fair dealings and other rights are in no way affected by the above.

Take down policy

If you believe that this document breaches copyright please contact [email protected] providing details, and we will remove access to the work immediately and investigate your claim.

Download date: 07. Oct. 2021

Social influences on dissociative processes in psychosis

Charles Heriot-Maitland

Department of Psychology Institute of Psychiatry, Psychology & Neuroscience King’s College London

This thesis is submitted to the University of London for the degree of Doctor of Philosophy. 2019

Abstract

Previous research has investigated socially based threats, such stigmatisation and shame, among people with psychosis. However, studies have typically considered these social experiences in terms of being the consequences, rather than causes, of psychosis. There have been some studies looking at stigma and shame as causal mechanisms in emotional symptoms (e.g. anxiety and depression) in the context of psychosis, and of post-psychotic trauma, but not of psychosis itself. One possible reason for why this may have been overlooked is that stigma and shame have not traditionally been regarded in ‘threat’ terms, with traumatic-like, perhaps dissociative, properties of their own. Social rank theory (Gilbert, 1992) offers an understanding of these socially-based experiences within their evolutionary context, with links to developing threat-monitoring, -processing, and -protective strategies. Social rank theory therefore provides a theoretical rationale within which to examine stigma and shame through the lens of threat and threat-protection mechanisms. These social rank threats can be investigated for links to traumatic, dissociative processes, and hence as potential contributors to causal pathways in psychosis. This thesis investigates the interaction of these social and dissociative processes in psychosis (Section A: Studies 1, 2, and 3). It also develops and tests the acceptability of a new psychological therapy that specifically targets social mechanisms, e.g. reducing shame and increasing social safeness, among people with psychosis (Section B: Study 4).

The first study designed and tested a measure of psychotic-like experiences (PLEs) that could be used for the other studies. Study 1, a questionnaire design study, analysed data from online participants for a) item selection, and b) psychometric evaluation. Scores were provided by 532 participants, and a new scale, the Transpersonal Experiences Questionnaire (TEQ), was developed. The 19-item TEQ was found to be a unidimensional scale with satisfactory internal consistency (0.85), good utility for longitudinal hypothesis-testing, and particular reliability for people endorsing a high number of PLE items. Study 2 investigated the longitudinal relationship between dissociation and PLEs, and the moderation effects of social variables, shame and social safeness, on this relationship. Participants (n=314) provided online data at two timepoints (0 and 6 months) and the analysis demonstrated a significant interaction between dissociation and shame on PLEs, despite the absence of a direct association between dissociation and PLEs. Social safeness significantly moderated this interaction effect, but not in the expected direction: higher social safeness was associated with more pronounced dissociation-shame interaction effects. In Study 3, participants (n=30) were recruited from a ‘non- clinical-in-crisis’ population; i.e. those seeking support for PLEs, but from spiritual organisations rather than mainstream services. This sample had significant trauma histories and high levels of dissociation.

2

Endorsing a higher number of PLE items was related to more distress; however, once total endorsement scores were controlled for, PLEs were rated as more positive (‘enriching’, ‘safe’ and ‘pleasant’) than negative (‘distressing’ and ‘disruptive’). The positive PLE dimensions were correlated with spirituality, post-experience growth, and social safeness. Negative PLE dimensions were correlated with depression.

The second section of the thesis reports the development and initial piloting of a new clinical intervention called Compassion Focused Therapy for Psychosis (CFTp). CFTp specifically targets building social safeness experience and compassion motivation systems with the aim of regulating threat-based dissociative processes in psychosis. Study 4 was a case series study, in which NHS participants with distressing experiences were randomised to multiple baseline periods of 2, 4 and 6 weeks, before a 26-session individual therapy. Seven out of eight participants completed the full 26 sessions. At the single case level of analysis, over half the completers showed clinically reliable improvements in depression (5/7), stress (5/7), general wellbeing (5/7), anxiety (4/7), and voices (3/5). One showed a deterioration in anxiety (1/7) and dissociation (1/7). At the group level of analysis (n=7), there were significant improvements in group mean scores of depression, stress, wellbeing, voices and delusions, and at 6-8 week follow-up, the wellbeing, voices and delusions improvements remained significant.

Further investigations are required of the theoretical relationships surfacing in this thesis, as well as the evaluation of CFTp. Specifically, the future modelling and testing of three-way interacting mechanisms (dissociation x social rank x attachment) in pathways to PLEs is recommended. For CFTp to be considered as a clinical treatment, it requires further testing in a pilot Randomised Controlled Trial, with different therapists, and with researchers who are unaware of the group allocation.

3

Acknowledgements

This thesis was supported by an MRC Clinical Research Training Fellowship, which paid me a part-time salary for five years (from June 2014) to study for a PhD. Without this financial input, it would not have been possible to conduct a project of this size, and to write a thesis, whilst also supporting my young family of five. Much gratitude therefore to the MRC for funding this work and to the Department of Psychology at IoPPN, King’s College London for helping me put together a successful grant application.

I would like to acknowledge the 609 participants who shared their personal experiences with me across the various studies, and particularly the seven of them who talked with me at great length and detail for the therapy study (at least 30 hours each) whilst courageously engaging with the most distressing parts of their experience. I feel deeply honoured to have been given access to this vast pool of wisdom and insight, and I sincerely hope that I have done some justice to this in my reporting.

I am very grateful to the support and guidance from my PhD supervisors, Dr Emmanuelle Peters, Prof Dame Til Wykes, and Prof Andrew Gumley. I have huge admiration for Emmanuelle’s remarkable attention to detail. Nothing sneaks past Emmanuelle. Ever. Although this can be daunting at times, it is also incredibly validating to experience someone with such knowledge and expertise so generously giving their time and concentration to your work. Til’s wonderful ability to get to the point and elicit action is second to none, and Andrew’s caring presence and intellectual guidance over the final two- year stretch in Glasgow was hugely valuable. Detail, action, and care: that’s what I call a dream team.

My acknowledgements and gratitude also to the clinical supervisors on my PhD, Prof Paul Gilbert and Dr Chris Irons. I feel very fortunate to have learnt Compassion Focused Therapy from the founder and world experts, and the generous input from both on this project has taken many forms beyond their clinical supervisory roles. My wonderful comrade, Dr Eleanor Longden, who has heavily shaped my learning about how to help people with psychosis; our collaboration on the PhD has produced writings, trainings, resources, and friendship. Kate Anderson, the talented artist who brought ‘Compassion for Voices’ to life through animation so that a central message of my PhD could be shared with the world. My ‘CAVEtalk’ peer supervision group, in particular, Elisabeth Svanholmer and Rufus May. Karlijn Hermans, my excellent MSc Erasmus student, who was a great ally in data collection and grappling with statistics. My fellow PhD students with whom I’ve shared offices, especially Pamela Jacobsen, who knows how to create an ideal study environment with wall charts and keeping plants alive, and Stephanie Allan, who always seems to know the exact right time to send a supportive message.

4

I am grateful to the NHS sites that supported recruitment for two of my studies: South London and Maudsley NHS Foundation Trust (London) and NHS Greater Glasgow and Clyde (Glasgow). In London, I am also grateful for the support of the NIHR/Wellcome Trust King’s Clinical Research Facility at King’s College Hospital. This facility hosted my therapy study, providing both a therapy room for me to use and nursing staff to take measurements of heart rate variability. I am also grateful to the Glasgow Clinical Research Facility, who provided nurses to collect these measurements at the Glasgow site.

My own family, Lexi and our three boys. I started the PhD in June 2014 when our youngest son was only 4 months old. He doesn’t yet know what life is like without a PhD in it, and he’s nearly six. Lexi, you have been so patient and well, quite frankly, heroic in keeping our home and family life in one piece and meeting all the twists and turns of this PhD Fellowship with your love and smiles and care.

5

Table of Contents

ABSTRACT ………………………………………………………………………………………………………………………. 2 ACKNOWLEDGEMENTS ………………………………………………………………………………………………...... 4 TABLE OF CONTENTS …………………………………………………………………………………………………………… 6 LIST OF TABLES ……………………………………………………………………………………………………………………….. 10 LIST OF FIGURES …………………………………………………………………………………………………………… 11 LIST OF APPENDICES …………………………………………………………………………………………………………… 12

INTRODUCTION Chapter 1. Introduction to thesis ………………………………………………………………………………….. 13 1.1. Psychological approaches to psychosis ……………...... 13 1.1.1. A (recent) history of psychological approaches…….………………………………………………. 13 1.2. Overarching thesis aims and objectives ……………………………………………………………………….. 15 1.2.1. Understanding pathways to psychosis: threat and dissociation …………………….. 15 1.2.2. Improving interventions for psychosis: Compassion Focused Therapy (CFT) ………… 17 1.3. Thesis structure and plan for addressing these aims ………………………………………………. 18 1.3.1. Section A ……………………………………………………………………………………………………………… 18 1.3.2. Section B ……………………………………………………………………………………………………………… 19

SECTION A Chapter 2. Introduction to Section A: Trauma and social pathways to psychosis, and where the two paths meet ……………………………………………………………………………….… 20 2.1. Trauma / adversity pathways to psychosis …………………………………………………………………… 20 2.1.1. Empirical and theoretical associations of trauma and psychosis …..………………… 20 2.1.2. Specific trauma pathways to specific symptoms? ………………………………………………. 21 2.2. Social pathways to psychosis ……………………………………………………………………………………. 23 2.2.1. Social factors in risk of psychosis …………………………………………………….…………………. 23 2.2.2. Social context of anomalous experiences …………………………………………………………… 25 2.2.3. Social-rank mechanisms ……………………………………………………………………………………. 26 2.2.4. Attachment mechanisms ……………………………………………………………………………………. 29 2.3. Where the two paths meet: Interaction of trauma and social pathways …………………….. 31 2.3.1. The traumatic/dissociative properties of shame ………………………………………………. 31 2.3.2. Summary of the pathways and their areas of interaction ………………………………….. 33 2.3.3. Aims and objectives of Section A studies …………………………………………………………… 34 Chapter 3. Study 1) – Detecting anomalous experiences in the community: the Transpersonal Experiences Questionnaire (TEQ) ….…………………………….………….. 36 3.1. Abstract ……………………………………………………………………………………………………………… 36 3.2. Introduction …………………………………………………………………………………………………………….. 37 3.3. Methods …………………………………………………………………………………………………………….. 40 3.3.1. Sample and design ……………………………………………………………………………………. 40 3.3.2. Measures ………………………………………………………………………………………………… 40 3.3.3. Procedure ………………………………………………………………………………………………… 41 3.3.4. Analysis ………………………………………………………………………………………………… 41 3.4. Results …………………………………………………………………………………………………………….. 41 3.4.1. Characteristics of the samples ……………………………………………………………………….. 42 3.4.2. Part A: Item selection ……………………………………………………………………………………. 43 3.4.2.1. Classical test theory ……………………………………………………………………….. 43 3.4.2.2. Item response theory ……………………………………………………………………….. 43 3.4.3. Part B: Psychometric evaluation of TEQ …………………………………………………………… 45

6

3.4.3.1. Exploratory factor analysis of TEQ ………………………………………………. 45 3.4.3.2. Confirmatory factor analysis of TEQ ………………………………………………. 45 3.4.3.3. Psychometric properties (reliability and validity) of TEQ ……………………… 46 3.4.3.4. Psychometric properties (difficulty and discrimination) of individual TEQ items …………………………………………………………………………………………………. 47 3.4.3.5. Demographic characteristics …………………………………………………………… 48 3.5. Discussion ……………………………………………………………………………………………………………… 48 3.5.1. Summary of results ……………………………………………………………………………………. 48 3.5.2. Psychometric profile and potential uses …………………………………………………………… 49 3.5.3. Individual item profiles ……………………………………………………………………………………. 49 3.5.4. Demographic profiles ……………………………………………………………………………………. 50 3.5.5. Limitations ………………………………………………………………………………………………… 51 3.5.6. Conclusions ………………………………………………………………………………………………… 51 Chapter 4. Study 2) – Social influences on the relationship between dissociation and psychotic-like experiences ………………………………………………………….………….. 52 4.1. Abstract .……………………………………………………………………………………………………………… 52 4.2. Introduction ………………………………………………………………………………………………………………. 53 4.3. Methods ………………………………………………………………………………………………………………. 54 4.3.1. Design and participants …………………………………………………………………………………….. 54 4.3.2. Measures …………………………………………………………………………………………………. 55 4.3.3. Procedure …………………………………………………………………………………………………. 56 4.3.4. Data analysis …………………………………………………………………………………………………. 56 4.4. Results ……………………………………………………………………………………………………………… 57 4.4.1. Longitudinal sample selection ……………………………………………………………………….. 57 4.4.2. Demographics ………………………………………………………………………………………………… 58 4.4.3. Correlations ………………………………………………………………………………………………… 58 4.4.4. Model A: Direct effects ……………………………………………………………………………………. 58 4.4.5. Model B: Simple moderation ……………………………………………………………………….. 59 4.4.6. Model C: Moderated moderation ……………………………………………………………………….. 60 4.5. Discussion ……………………………………………………………………………………………………………… 62 4.5.1. No direct effect of dissociation on PLEs …………………………………………………………… 62 4.5.2. The interaction effect of dissociation-shame on PLEs ………………………………….. 63 4.5.3. Social safeness influences on the interaction effect of dissociation-shame on PLEs 64 4.5.4. Research and clinical implications …………………………………………………………… 66 4.5.5. Strengths, limitations, and future research ………………………………………………. 67 4.5.6. Conclusions …………………………………………………………………………………………………. 69 Chapter 5. Study 3) – Seeking spiritual support for psychotic-like experiences: characteristics of individuals, contexts, and experience ..………………..…………….. 70 5.1. Abstract ….…………………………………………………………………………………………………………… 70 5.2. Introduction ………………………………………………………………………………………………………………. 71 5.3. Methods ………………………………………………………………………………………………………………. 73 5.3.1. Design and participants …….………………………………………………………………………………. 73 5.3.2. Measures …………………………………………………………………………………………………. 74 5.3.3. Procedure …………………………………………………………………………………………………. 76 5.3.4. Data analysis …………………………………………………………………………………………………. 77 5.4. Results ………………………………………………………………………………………………………………. 77 5.4.1. Demographic, clinical, and support-seeking characteristics …………………………………… 77 5.4.2. Psychotic-like experience (PLE) characteristics ……………………………………………….. 79 5.4.3. Relationships between social, psychological, and spiritual variables and PLE- dimensions ………………………………………………………………………………………………….. 81 5.5. Discussion ………………………………………………………………………………………………………………. 83

7

5.5.1. Demographic, clinical, and support-seeking characteristics …………………………………… 83 5.5.2. Psychotic-like experience (PLE) characteristics ……………………………………………….. 85 5.5.3. Social, psychological, and spiritual predictors of PLE-outcome ……………………… 86 5.5.4. Limitations …………………………………………………………………………………………………. 87 5.5.5. Conclusions …………………………………………………………………………………………………. 88

SECTION B Chapter 6. Introduction to Section B: Compassion Focused Therapy (CFT) for psychosis 89 6.1. Chapter information and overview .….…….……………………………………………………………… 89 6.2. Background to CFT …………………………………………………………………………………………………. 90 6.3. CFT and psychosis: trauma, attachment and dissociation ………………………………….. 90 6.4. CFT for psychosis: key elements ………………………………………………………………………… 92 6.4.1. Establish safeness and connection ……………………………………………………………. 93 6.4.2. Psycho-education about evolved brains, emotion systems, and multiple selves 94 6.4.3. Formulation not diagnosis ………………………………………………………………………… 95 6.4.4. Cultivate / deepen the compassionate self ……………………………………………….. 97 6.4.5. Direct compassion to others, self, dissociated parts/voices ……………………… 98 6.5. Previous studies of CFT with people with experiences of psychosis ……………………… 101 6.6. Summary ……………………………………………………………………………………………………………… 101 Chapter 7. Developing the CFT for psychosis intervention ………………...….……………………….. 103 7.1. Abstract ….…………………………………………………………………………………………………………… 103 7.2. Introduction ………………………………………………………………………………………………………………. 104 7.2.1. Intervention development studies ……………………………………………………………. 104 7.2.2. User-involved development ………………………………………………………………………… 104 7.2.3. Expertise by experience and user-led developments in psychosis ……………………… 105 7.2.4. Overall aims of intervention development ……………………………………………………………. 105 7.3. PHASE 1 - Developing the initial CFTp manual (pre- Dec 2014) …………………………………… 106 7.3.1. Phase 1 aims ………………………………………………………………………………………………….. 106 7.3.2. Phase 1 methods …………………………………………………………………………………….. 106 7.3.3. Phase 1 results …………………………………………………………………………………………………. 108 7.4. PHASE 2 - Further developing the CFTp manual (2015 – 2018) ………………………………….. 109 7.4.1. Phase 2 aims ………………………………………………………………………………………………….. 109 7.4.2. Phase 2 methods …………………………………………………………………………………….. 109 7.4.3. Phase 2 results …………………………………………………………………………………………………. 112 7.5. Discussion ……………………………………………………………………………………………………………… 118 7.5.1. Informing acceptability ……………………………………………………………………………………. 118 7.5.2. Informing intervention development and theory of intervention …………………….. 120 7.5.3. Limitations ………………………………………………………………………………………………… 121 7.5.4. Conclusions ………………………………………………………………………………………………… 123 Chapter 8. Study 4) – A case-series study of Compassion Focused Therapy for distressing experiences in psychosis ……………………...….………………………………………….. 124 8.1. Abstract .……………………………………………………………………………………………………………… 124 8.2. Introduction ………………………………………………………………………………………………………………. 125 8.2.1. Background ………………………………………………………………………………………………….. 125 8.2.2. Summary and aims .……………………………………………………………………………………. 126 8.2.3. Case series methodology …………………………………………………………………………………….. 126 8.2.4. Measuring processes of change ………………………………………………………………………… 126 8.2.5. Measuring outcomes …………………………………………………………………………………….. 128 8.2.6. Research questions and hypotheses ……………………………………………………………. 128 8.3. Methods ………………………………………………………………………………………………………………. 129 8.3.1. Design and participants …………………………………………………………………………………….. 129

8

8.3.2. Measures …………………………………………………………………………………………………. 130 8.3.3. Procedure …………………………………………………………………………………………………. 132 8.3.4. Intervention description …………………………………………………………………………………….. 133 8.3.5. Data analysis …………………………………………………………………………………………………. 134 8.4. Results ……………………………………………………………………………………………………………… 136 8.4.1. Feasibility and acceptability ……………………………………………………………………….. 136 8.4.2. Single-case level analysis ……………………………………………………………………………………. 138 8.4.3. Group-level analysis ……………………………………………………………………………………. 147 8.5. Discussion …………………………………………………………………………………………………………….. 150 8.5.1. Summary of results ……………………………………………………………………………………. 150 8.5.2. Acceptability of CFTp ……………………………………………………………………………………. 151 8.5.3. Single case variability in results ……………………………………………………………………….. 152 8.5.4. Limitations …………………………………………………………………………………………………. 154 8.5.5. Conclusions and implications for future research ………………………………………………. 156

DISCUSSION Chapter 9. Discussion …………..………………………………………………………………………………….. 158 9.1. Thesis summary …………………………………………………………………………………………………. 158 9.1.1. Summary of aims and objectives …………..……………………………………………………………. 158 9.1.2. Summary of main findings ………………………………………………………………………… 158 9.2. Strengths and limitations ……………………………………………………………………………………. 159 9.2.1. Strengths …………………………………………………………………………………………………. 159 9.2.2. Limitations …………………………………………………………………………………………………. 161 9.3. Implications for theory of psychosis ………………………………………………………………………… 165 9.3.1. Targeted approaches, but targeting what? A symptom, a mechanism, or a function? ……………………..………………………………………………………………………….. 165 9.3.2. A three-way interaction approach …………………………………………………………… 168 9.4. Implications for psychosis interventions …………………………………………………………… 170 9.5. Concluding remarks …………………………………………………………………………………………………. 172

REFERENCES ……………………………………………………………………………………………………………………….. 174 APPENDICES ……………………………………………………………………………………………………………………….. 190

9

List of Tables

TABLE 3.1 Participant numbers for each part of the study, and demographics …...…… 42 TABLE 3.2 Loadings for the one, two, and bifactor two factor models …………..………… 46 TABLE 3.3 EFA/CFA loadings, IRT parameters, and reliability indices for TEQ items ………… 47 TABLE 4.1 Means, standard deviations, medians, interquartile ranges, and Spearman’s inter-correlations of variables at baseline and 6 months (n=314) .…………………….. 58 TABLE 4.2 Regression analysis results (n=314): Testing Models A, B & C, with TEQ6m as dependent variable …………………………………………………………………………………….. 59 TABLE 5.1 Demographic and clinical characteristics (n=30) ……………………………………………….. 78 TABLE 5.2 Means, standard deviations, and Spearman’s inter-correlations of PLE dimensions ……………………………………………………………………………………....……….. 80 TABLE 5.3 Means and standard deviations of social, psychological, spiritual characteristics, their Spearman’s correlations with TEQ, and their partial correlations with TEQ- dimension …………………………………………………………………………..…………………….. 82 TABLE 7.1 Summary of changes made and the feedback themes (and sources) that led to changes ………………………………………………………………………………………………….. 117 TABLE 8.1 Template for intervention description and replication (TIDieR) checklist …………. 133 TABLE 8.2 Clinical and demographic profiles of participants (n=7) ……………………….…………. 137 TABLE 8.3 Directions of reliable changes in process and outcome measures, showing only those that were significant against a Reliable Change Index (RCI) .…………………….. 138 TABLE 8.4 Summary of Tau analysis for sessional process measure (SSPS) ……………………… 147 TABLE 8.5 Summary of Tau analysis for sessional outcome measure (Dissociation) …………. 147 TABLE 8.6 Means and standard deviations for outcome and process measures at five assessment points ………………..…………………………………………………………………… 148 TABLE 8.7 Wilcoxon Signed Rank Tests for group-level changes across phases …………. 150

10

List of Figures

FIGURE 2.1 Social influences on the trauma pathway to psychosis …………..……………………… 34 FIGURE 3.1 IRT item information curves for Factor 1 …………..…………..……………………… 44 FIGURE 3.2 IRT item information curves for Factor 2 …………..…………..……………………… 44 FIGURE 3.3 Scree plot for exploratory factor analysis …………..…………..……………………… 45 FIGURE 3.4 Item characteristic and item information curves for TEQ items (n=532) …….…… 48 FIGURE 4.1 Models tested …………………………………………………………………………………………………. 54 FIGURE 4.2 Plots of the interaction effects (DES x OAS) on TEQ6m ………………………………….. 60 FIGURE 4.3 Plots of the interaction effects in the context of different levels of SSPS ..……….. 61 FIGURE 5.1 Characteristics of individual psychotic-like experiences (n=30) ………………….….. 80 FIGURE 6.1a CFT formulation (development) – combination of ↑ threat and ↓ safeness 96 FIGURE 6.1b CFT formulation (maintenance) ………………………………………………………………………… 96 FIGURE 8.1 Flow diagram of participant recruitment ..……………………………………………… 136 FIGURE 8.2 Changes in distressing voices and distressing delusions for each case …………. 143 FIGURE 8.3 Plots of session-by-session measures ……………………………………………………………. 144 FIGURE 9.1 A model of dissociation and social mentality (social-rank x attachment) interactions in the pathway to psychotic-like experiences (PLEs) ….………………….. 169

11

List of Appendices

APPENDIX 3.1 Additional information about analysis methodologies ………………………..……….. 190 APPENDIX 3.2 Reliability indices for the initial pool of 57 items (N=283) ……………………… 192 APPENDIX 3.3 Transpersonal Experiences Questionnaire (TEQ) …………..……………………… 194 APPENDIX 4.1 KCL ethics approval – Studies 1 & 2 ………………………………..………………..……….. 196 APPENDIX 4.2 Participant information sheet – Studies 1 & 2 ……………………….……………………… 198 APPENDIX 4.3 Comparison of completers versus non-completers ………………………………….. 201 APPENDIX 5.1 KCL ethics approval – Study 3 ……………………………………………………………………….. 202 APPENDIX 5.2 Participant information sheet – Study 3 …………………………..……………………………….. 203 APPENDIX 5.3 The TEQ-dimensions measure administration and scoring …………………….. 205 APPENDIX 5.4 The ‘other’ sources of support ………....…………………………..……………………………….. 207 APPENDIX 5.5 TEQ endorsement and rankings in spiritual support sample (Study 3, Chapter 5) and general population sample (Study 1, Chapter 3) .……………..………………….. 208 APPENDIX 7.1. Theoretical paper published in Frontiers in Psychology (Heriot-Maitland, McCarthy-Jones, Longden, & Gilbert, 2019) ..……………………. 209 APPENDIX 7.2 Feedback from the 20 voice-hearers who engaged with ‘Compassion for Voices’ 220 APPENDIX 8.1 Random allocation of multiple baselines (2/4/6 weeks) …………………………………… 222 APPENDIX 8.2 REC ethics approval – Study 4 ………………………………………………………………………… 223 APPENDIX 8.3 Participant information sheet – Study 4 …………………………………………………………… 229 APPENDIX 8.4 NHS SLAM R&D approval – Study 4 …………………………………………………………… 232 APPENDIX 8.5 NHS GG&C R&D approval – Study 4 …………………………………………………………… 234 APPENDIX 8.6 CFTp manual v2 Dec 2018 ……………………………………………………………………….. 236 APPENDIX 8.7 CFTp-ACM v2 Dec 2018 ……………………………………………………………………………………. 237 APPENDIX 8.8 Expert rating for manual adherence and therapist competence ……………………… 239 APPENDIX 8.9 Normative data used for establishing reliable change criteria ……………………… 240 APPENDIX 8.10 Reliable Change Indices for single-case-level changes across phases – full table 241

12

Chapter 1. Introduction to thesis

1.1. Psychological approaches to psychosis

Psychotic conditions (i.e. experiencing symptoms such as hearing voices or delusions) have some of the most severe and enduring mental health outcomes, such as high suicide rates, disengagement from society, loss of economic productivity, and poverty (Bentall & Morrison, 2002). Psychosis-related diagnoses, particularly schizophrenia, also carry severe social stigma, and many of those diagnosed will internalise this stigma to experience shame (i.e. developing negative attitudes towards themselves). Studying the psychological and social processes that aid or hinder recovery from psychosis has the potential to contribute to treatment and service developments with both social and economic impacts. Psychological approaches to psychosis are typically aligned to a continuum model framework, which is the well-evidenced theory that psychotic-like experiences exist on a continuum throughout the population (van Os, Linscott, Myin-Germeys, Delespaul, & Krabbendam, 2009). An important implication of the continuum model is that it provides a rationale for psychosis research to take place within general population samples, meaning larger and better-powered studies, which has greatly advanced our scientific study and understanding of psychological mechanisms in psychosis.

1.1.1. A (recent) history of psychological approaches

Over the last three decades, there has been a surge in the development and delivery of psychological interventions for people with psychosis. Towards the end of the last century, psychological approaches for psychosis were still barely registering on the awareness of mainstream mental health services as a psychosis treatment option. The biomedical approach to treating psychosis still carried all the weight, and in the first UK national guidelines for treating schizophrenia (NICE, 2002), antipsychotic treatment was recommended as the front-line intervention for acute psychotic episodes, with alternative (non-medical) treatments assigned secondary, optional status, and only deemed appropriate for later stages. However, by the time the guidelines were updated seven years later, there had been a shift towards a more equal recommendations for both biomedical and psychological treatments across all stages (NICE, 2009).

The people responsible for putting talking therapy on the psychosis map are the early pioneers of Cognitive Behavioural Therapy approaches for Psychosis (CBTp), such as Kingdon, Turkington, Garety,

13 and Morrison (Garety, Kuipers, Fowler, Freeman, & Bebbington, 2001; Kingdon & Turkington, 1994; Morrison, 2001), who adapted Beck’s (1979) original cognitive model of emotional disorders within psychosis settings, leading to testable hypotheses and therapy. CBTp is premised on the same cognitive principle that our interpretation of events will have consequences for how we feel and behave, and as such CBTp is focused on generating less distressing explanations for psychotic experiences. Garety et al.’s (2001) influential cognitive model postulates two routes to the development of positive psychotic symptoms: one in which cognitive changes give rise to an ‘anomalous experience’ (accompanied by an emotional response), which is then appraised in a particular way; and a second in which affective changes alone lead to a particular appraisal. So, it is not the anomalous experience itself which directly leads to psychosis, but rather the appraisal, which is influenced by various factors such as pre-existing cognitive biases and expectations (Garety et al., 2001). Morrison (2001) similarly identifies the misinterpretation of experiences in the development of psychosis but draws attention to the cultural acceptability of interpretations as to whether or not they are deemed psychotic. In Morrison’s (2001) model, the experiences themselves are considered to be essentially ‘normal’ intrusions into awareness, and their interpretations, which may or may not be culturally acceptable, are determined by prior experiences, beliefs and knowledge. The idea that perceptual experiences and intrusions are themselves not pathological is consistent with the aforementioned continuum model of psychosis.

By 2004, a review reported that 20 randomised controlled trials of CBTp had been published, concluding that “overall there is good evidence for the efficacy and effectiveness of CBTp in the treatment of schizophrenia” (p. 1377, Tarrier and Wykes (2004)). In the years that followed, new psychological interventions emerged. In a review of the main developments, Tai and Turkington (2009) identified ‘3rd wave’ therapies that had started being applied to psychosis: Mindfulness-Based Approaches, Acceptance and Commitment Therapy, Compassionate Mind Training, Meta-cognitive therapy, and The Method of Levels. Since that review, there have been additional developments, with key areas of development and interest being therapies that help voice-hearers to change their relationship with voices, namely COMMAND (Birchwood et al., 2018), Relating Therapy (Hayward, Jones, Bogen-Johnston, Thomas, & Strauss, 2017), Avatar Therapy (T. K. Craig, 2019; T. K. Craig et al., 2018), and Talking with Voices (Corstens, Longden, McCarthy-Jones, Waddingham, & Thomas, 2014; Longden, Read, & Dillon, 2018). The competitive memory training (COMET) approach of van der Gaag, van Oosterhout, Daalman, Sommer, and Korrelboom (2012) has also been investigated as a targeted approach for voice-hearers. In addition, there have been promising developments in Cognitive Remediation Therapy for psychosis (Reeder et al., 2016) and testing is underway to evaluate Open

14

Dialogue (Seikkula et al., 2006) as a whole team approach in NHS mental health services (Peer- Supported Open Dialogue, Stockmann et al. (2019)).

Another major development in the evaluation of psychological approaches to psychosis has been a move away from ‘generic’ approaches to intervention (i.e. therapies that work with the range of symptoms and difficulties) to more targeted approaches (i.e. targeting a specific symptom or a specific psychological mechanism). A major impetus for these changes has come from several meta-analyses of generic CBTp which reveal only modest effect sizes (e.g. Bighelli et al. (2018): d=0.29 compared to inactive control, d=0.30 compared to treatment as usual, d=0.47 compared to supportive therapy, and d=0.24 compared to waitlist), whereas targeted CBTp interventions are found to have higher effect sizes (reviewed by Lincoln and Peters (2019)). In an important commentary about the field, Freeman (2011) recommended that to move CBTp forward, studies should evaluate therapies that target one putative causal factor at a time. In this vein, a number of studies have targeted specific aspects of delusions, such as reasoning biases (Moritz et al., 2011; K. Ross, Freeman, Dunn, & Garety, 2011; Waller, Freeman, Jolley, Dunn, & Garety, 2011) and worry (Foster, Startup, Potts, & Freeman, 2010), and other studies have also targeted specific aspects of voices, such as the power relationship and compliance with a commanding voice (Birchwood et al., 2018), but not all evaluations have produced positive results. A review of key developments in psychological therapies for voices advocates for more targeted studies to “better understand specific processes, therapeutic methods, and applicability for different voice hearers” (S209, Thomas et al. (2014)). As we learn more about mechanisms and pathways to different symptoms and symptoms dimensions, the more our interventions can become tailored to individuals and their experiences. In Chapter 2, there is a review of the current knowledge about key mechanisms in psychosis.

1.2. Overarching thesis aims and objectives

1.2.1. Understanding pathways to psychosis: threat and dissociation

The first aim of this thesis is to improve our understanding of mechanisms involved in psychosis by investigating the interaction of dissociation and socially based threats, in particular those linked to shame, in pathways to PLEs. The suggestion, in line with evolutionary social rank theory (Gilbert, 1992), is that experiences of shame may threaten our basic survival needs for social acceptance and connection, thereby activating threat processing systems. Within this framework, where shame is

15 understood as a ‘social rank threat’, there may be relevance in investigating the potential role of shame in interacting with, and maintaining, dissociative processes in psychosis.

Clinical observations of psychotic symptoms, such as paranoid beliefs and commanding voices, often signal underlying themes of threat, and studies show that people with psychosis have biased recall of, and attention towards, threatening information (Dorahy & Green, 2008; Savulich, Shergill, & Yiend, 2012; Underwood, Kumari, & Peters, 2016). Psychological models have recognised the importance of threat processing; for example, characterising psychotic symptoms as the manifestation of highly sensitised threat-response systems (Gumley, Braehler, Laithwaite, MacBeth, & Gilbert, 2010). Threat themes are also common to many of the well-established psychosis risk factors, such as abuse, trauma, and social adversity (Dean & Murray, 2005). Threat is prominent in cognitive theories of the development of different symptoms; for instance persecutory delusions are conceptualised as threat beliefs (Freeman & Garety, 2014), and in voice-hearing models, adverse life experiences are important factors in negative voice content (Laroi et al., 2019).

Dissociation is a psychological threat response; essentially an automatic, internal strategy to protect the mind from becoming overwhelmed by distressing information. Dissociation itself is not pathological; it can occur for a variety of reasons, generally adaptive (e.g. during trauma, drug-use, isolation or impasse). Theoretical links between psychosis and dissociation have become increasingly prominent in the literature (Heriot-Maitland, 2012; Pilton, Varese, Berry, & Bucci, 2015), and the associations between dissociative experiences and psychotic intrusions have been demonstrated in a wide range of clinical and non-clinical populations (Moskowitz, Barker-Collo, & Ellson, 2005).

Threat and dissociation have received increasing attention in the psychosis research field, with both often implicated in psychosis phenomenology and causal pathways. Threat-based experiences, such as trauma, are typically regarded as causal determinants, with dissociation as a mediator, on the pathway to psychosis in theoretical accounts (Varese, Barkus, & Bentall, 2012). Further discussion of the trauma pathways literature is given in Chapter 2.

The pathways from trauma – via dissociation – to psychosis have been thoroughly tested and evidenced, but what has received less attention has been the social pathways – via dissociation – to psychosis. Often social factors are more commonly linked to other influences, e.g., to appraisals and the creation of negative schema in cognitive models (Garety et al., 2001), or to unsupportive caregiving experiences where there is high ‘expressed emotion’ (Brown, Birley, & Wing, 1972).

16

However, evidence is now emerging that negative social rank experiences, such as being excluded or shamed, may themselves have dissociative properties (Dorahy et al., 2017), which poses intriguing questions as to how trauma pathways and social pathways might interact.

1.2.2. Improving interventions for psychosis: Compassion Focused Therapy (CFT)

The second aim of this thesis is to improve our psychological interventions for people with psychosis by developing and testing the acceptability of a novel therapy called Compassion Focussed Therapy (CFT) for Psychosis, which specifically targets shame and regulates overactive threat processing systems by building internal feelings of safeness and affiliation.

CFT is designed to help people develop their capacity for affiliative relating (in self-to-other and self- to-self relationships) through the activation and cultivation of care-giving social motives. The rationale for targeting these processes and mechanisms is based on research on the physiology of emotion regulation through social and affiliative experience (Hostinar, Sullivan, & Gunnar, 2014) and evidence that compassion-focused practices can influence different physiological processes important for emotional wellbeing (Mascaro, Rilling, Tenzin Negi, & Raison, 2013; Pace et al., 2009; Weng et al., 2013). By helping individuals to foster these processes, CFT aims to optimise their chances of compassionate engagement with their threat-based emotions and experiences, and to work with them towards therapeutic processing, resolution, and change. To inform its interventions, CFT researchers have sought to understand the neurophysiological underpinnings of these processes, particularly their links to the vagus nerve and heart rate variability (HRV) (Kirby, Doty, Petrocchi, & Gilbert, 2017). CFT is now recognised as an integrated bio-psycho-social approach, and while evidence for its effectiveness as a psychotherapy is still in its early days, the research interest is beginning to grow, as demonstrated by two recent literature reviews (Kirby, 2017; Leaviss & Uttley, 2015).

There has been very little research on CFT for people with psychosis. Laithwaite et al. (2009) and Braehler et al. (2013) both studied CFT for psychosis in group format, and Mayhew and Gilbert (2008) reported a case series of 3 participants with malevolent voices receiving CFT in individual format. Despite the paucity of evaluation studies to date, there is a strong theoretical rationale for why CFT may be a suitable psychosis intervention. This rationale has been previously outlined by Gumley et al. (2010) and is explored in depth in this thesis (Chapter 6 and Appendix 7.1).

17

1.3. Thesis structure and plan for addressing these aims

The first section of the thesis reviews the state of knowledge in trauma and social pathways to psychosis and reports three empirical studies investigating these pathways. The second section builds on this framework to advance our knowledge of one potential psychological treatment for people with psychosis through developing and testing a novel intervention, CFTp. In line with the continuum model of psychosis, the 4 studies in this thesis recruited a range of different samples from both clinical and non-clinical populations.

1.3.1. Section A

The literature review focusses on our current understanding of trauma and social pathways to psychosis and then considers the potential mechanisms and their relationships, specifically i) dissociation, ii) attachment, and iii) social rank/shame.

The first study designs a questionnaire to measure psychotic-like experiences (PLEs) that can be used in the thesis and to support other studies in this area. The requirement for such a measure is to have utility for longitudinal, time-sensitive research with particular reliability for people with higher incidence of PLEs (the upper end of the psychosis continuum).

The second study is longitudinal and involves a general population sample with two time points (six months apart) to test a 2-way interaction model (dissociation x shame) and a 3-way interaction model (dissociation x shame x social safeness) in the pathway to PLEs. This study comprised a total sample of 314 online participants who completed measures at both timepoints.

The third study recruited individuals who have high levels of persistent PLEs and who are seeking support, but from spiritual support organisations not mainstream services. This group can potentially inform future research on the interaction of trauma/dissociative and social mechanisms as they are likely to experience different social contexts (and narratives around their PLEs) than their clinical counterparts. As this is a novel population, the aims were primarily exploratory, with a focus on characterising the population (e.g. their trauma histories and their social rank, attachment, and PLE characteristics) in order to inform the design of future (hypothesis-testing) studies in this population. The total sample was n=30.

18

1.3.2. Section B

The first of the chapters in Section B (Chapter 6) describes a Compassion Focused Therapy (CFT) approach to psychosis, including the theoretical rationale, and illustrations of how this translates (in general terms) into therapeutic processes and techniques. The CFT approach to psychosis is informed by two major theoretical strands (attachment theory and social rank theory), and because the social rank aspects are explored in Section A and Appendix 7.1, the focus of Chapter 6 is on the attachment themes. The chapter is published in a book on Attachment and Psychosis (Heriot-Maitland & Kennedy, 2019) and is co-authored with Dr Angela Kennedy, who is an experienced NHS clinician with expertise in attachment and dissociation processes in psychosis.

Chapter 7 describes the intervention development procedures, including the knowledge exchange partnerships, collaborations, and activities that informed the development of CFT for Psychosis. Intervention development took place in two phases: Phase 1 before the evaluation period (pre-Dec 2014) and Phase 2 during the evaluation period (2015 – Dec 2018). This chapter reports on the engagement activities and results during each phase that led to the 50-page CFTp manual (Dec 2018).

A fourth, clinical study (Study 4, reported in Chapter 8) tested the feasibility of a 26-session individual CFT for distressing experiences in psychosis (CFTp). A single case series design was used, with participants randomly allocated to one of three baseline periods (2, 4 and 6 weeks), before starting CFTp. Seven participants completed the intervention, all of whom were NHS mental health service users with distressing psychotic experiences. Participants provided brief measures of social safeness and dissociation every week, as well as a longer battery of process and outcome measures at five timepoints throughout the therapy (baseline, pre-, mid-, post-, and follow-up).

The final chapter summarises and integrates the main findings, considers the overarching strengths and limitations of the thesis in meeting its aims and objectives, the implications of this work for psychological theories and treatments, and suggestions for future research.

19

Chapter 2. Introduction to Section A: Trauma and social pathways to psychosis, and where the two paths meet

2.1. Trauma / adversity pathways to psychosis

2.1.1. Empirical and theoretical associations of trauma and psychosis

The strength of the empirical evidence linking trauma and psychosis is reported to be of a magnitude comparable to the causal association between smoking and lung cancer (Bentall et al., 2014). In a meta-analyses of 41 studies, Varese, Smeets, et al. (2012) found that childhood adversity and trauma increases the odds of psychosis by 2.8, and that this increased risk of psychosis held across multiple different types of trauma (odds ratios for: sexual abuse 2.38, physical abuse 2.95, emotional abuse 3.40, bullying 2.39, and neglect 2.90). The only specific trauma type not significantly associated with psychosis risk was parental death (odds ratio 1.70). Importantly, there is also evidence of a cumulative relationship between trauma and psychosis, i.e. the more trauma types experienced, the greater likelihood of psychosis; for example, Shevlin, Houston, Dorahy, and Adamson (2008) reported that experiencing three types of trauma made the odds of psychosis 18 times more likely, and experiencing five types of trauma made the odds of psychosis 193 times more likely. There is also found to be a cumulative relationship between trauma and psychotic experiences (Croft et al., 2019). With the causal link between trauma and psychosis now widely recognised, in terms of both acute trauma reactions (Steel, Fowler, & Holmes, 2005) and childhood, trait-related, trauma reactions (Read, van Os, Morrison, & Ross, 2005; Varese, Smeets, et al., 2012), the theoretical application of trauma models to psychosis has become an increasingly important guide for hypotheses and research developments.

Cognitive models of trauma and post-traumatic stress disorder (PTSD) (Brewin, Gregory, Lipton, & Burgess, 2010; Ehlers & Clark, 2000) illustrate how acute dissociative states can arise in reaction to extreme or adverse experiences. This reaction is thought to involve a de-synchronisation or dissociation of processing between the conceptual and perceptual elements of the experience, resulting in unprocessed perceptual memories that lack spatial, temporal, and conceptual context. Moreover, if exposure to such adverse experiences is prolonged or repeated, especially during childhood, dissociative states and processes can become more a part of personality formation; i.e. dissociative traits (Perry, Pollard, Blakley, Baker, & Vigilante, 1995). Both state- and trait-related

20 dissociation can create the potential for de-contextualised memories to be triggered into awareness, experienced by the individual as involuntary perceptual intrusions.

Such ‘multi-level’ trauma processing models have been applied to formulating psychotic symptoms (Fowler et al., 2006; Heriot-Maitland, 2012; Morrison, 2001; Morrison, Frame, & Larkin, 2003; Steel et al., 2005). According to Morrison et al. (2003), unusual perceptual experiences in psychosis, such as hearing voices, may be phenomenologically similar to the perceptual intrusions in PTSD, such as flashbacks; i.e. dissociated and de-contextualised perceptual memories. However, a crucial difference lies in the appraisal of these intrusions, because, in psychosis, the voice or image may not be directly attributed as a component of memory, but rather a novel experience from an alternative source. The lack of contextual information accompanying the perceptual intrusion will increase its vulnerability to conceptual misinterpretation, which, according to cognitive models, is the main factor in the development of psychosis (Garety et al., 2001; Morrison, 2001). Fowler et al. (2006) argue that “problems in contextual processing associated with vulnerability to psychosis may then have the capacity to distort or exaggerate personally significant threat” (p 116).

2.1.2. Specific trauma pathways to specific symptoms?

Bentall et al. (2014) suggest that there may be slightly different trauma pathways for different symptoms of psychosis, citing the heterogeneity of psychosis, with evidence of at least three different clusters of symptoms of schizophrenia: positive, negative, and cognitive disorganisation (Liddle, 1987). However, Bentall et al. (2014) go further to argue for specificity of mechanisms between the different psychotic experiences themselves (i.e. within the positive symptom cluster), suggesting that the dissociation pathway (outlined above) may be more relevant to voice-hearing than to paranoia. After reviewing evidence for different trauma types as determinants for voices and paranoia, Bentall et al. (2014) argue that childhood sexual abuse is more implicated in the pathway to voice-hearing and that attachment-disrupting events are more implicated in the pathway to paranoia. However, as the authors note, most of the evidence for specific causal pathways to voices and paranoia does not control for the co-occurrence of these positive symptoms, an issue they address in a study using data from the UK 2007 Adult Psychiatric Morbidity Survey (Bentall, Wickham, Shevlin, & Varese, 2012). After controlling for co-occurring voices and paranoia, they found that childhood sexual abuse was associated with voices, institutional care (an indicator of disrupted attachment) was associated with paranoia, and physical abuse was associated with both voices and paranoia (Bentall et al., 2012).

21

Bentall et al.’s (2014) case for specificity is provided with some support from a replication using a US survey (Sitko, Bentall, Shevlin, O'Sullivan, & Sellwood, 2014) and a prison sample (Shevlin, McAnee, Bentall, & Murphy, 2015). However, closer examination of these studies does not paint a particularly clear picture. For example, Shevlin et al. (2015) found that while childhood sexual abuse did show higher odds for hallucinations (odds ratio 2.37) than paranoia (1.20), living in an institution as a child had only slightly higher odds for paranoia (1.49) than hallucinations (1.09). Bullying, a stronger predictor of paranoia (1.99), does not necessarily represent attachment disruption, but rather inter- personal threat and powerlessness (i.e. similar themes to those in sexual abuse). Unwanted sexual attention in prison was also a stronger predictor of paranoia (1.63) than living in an institution.

The specificity picture is further complicated by evidence of sexual abuse pathways to paranoia (Murphy, Shevlin, Houston, & Adamson, 2012) and attachment disruption pathways to voices (Pilton et al., 2016). In their cognitive-attachment model of voices (CAV), Berry and Bucci (2015) propose that a combination of trauma, disorganised attachment, and dissociation are all involved in the pathway to voices. The CAV model is elaborated, along with a thorough review of all the supporting evidence and implications, by Berry, Varese, and Bucci (2017). Another interesting aspect to this specificity question emerged in a study of mediating mechanisms among 112 participants with distressing psychotic experiences (Pearce et al., 2017). In this study, dissociation was found to mediate the relationship between childhood trauma and voices (as expected) and the relationship between trauma and paranoia (more surprising). This was the first time that the dissociation-paranoia connection had been made in the psychosis literature (Kilcommons & Morrison, 2005). Their interpretation of this result was to suggest that perceptual anomalies in dissociation may also contribute to the formation of paranoid beliefs, and they recommended future research into threat processing mechanisms in relation to both dissociation and paranoia (Pearce et al., 2017). In a recent commentary, Dorahy and Green (2019) suggest that the reason this connection was not made before was due to some divergence in the aims of dissociation and psychosis research. They present studies that indicate “shared neurobiological processes for dissociation and paranoia, manifesting in both increased salience of internal threat and perception of threat” (p 295).

In summary, Bentall et al.’s (2014) call for specificity is well-founded, certainly when it comes to differentiating between positive symptoms, negative symptoms and thought disorder. However, within the psychotic (‘altered reality’) domain of positive symptoms, there is evidence to suggest roles of both dissociative and attachment mechanisms along the pathways to both voices (dissociative (Varese, Barkus, et al., 2012); attachment (Pilton et al., 2016)) and paranoia (dissociative (Pearce et

22 al., 2017); attachment (Wickham, Sitko, & Bentall, 2015)). There is also evidence for more global, generalised effects of childhood adversities on different symptoms (Longden, Sampson, & Read, 2016), which is contrary to the specificity model. Perhaps, therefore, when it comes to psychotic experiences, instead of discounting mechanisms for the sake of specificity, it might be helpful for researchers to consider multiple interacting-mechanism models, e.g. in the vein of CAV (Berry et al., 2017), and to investigate the relative contributions of each of these processes to the interactions.

One potential area of overlap with trauma pathways (e.g. around social adversity risk factors and attachment processes) is social pathways to psychosis, and these are reviewed below.

2.2. Social pathways to psychosis

2.2.1. Social factors in risk of psychosis

There is considerable evidence that adverse social experiences are risk factors for psychosis. However, broader interpersonal experience, beyond traditional definitions of ‘trauma’, which can also increase the risk of psychosis, includes inner city living (Vassos, Pedersen, Murray, Collier, & Lewis, 2012), isolation/loneliness (Michalska da Rocha, Rhodes, Vasilopoulou, & Hutton, 2017), discrimination (Janssen et al., 2003), and migration (Cantor-Graae & Selten, 2005). Each will be outlined in turn.

Inner city (urban) living is believed to create a risk of psychosis due to the lack of social cohesion and neighbourhood trust typical of urban environments, combined with the higher incidence of crime (Newbury et al., 2017); factors which lead to people feeling socially unsafe. Newbury et al. (2017) found that adolescents raised in urban neighbourhoods, compared to rural neighbourhoods, were significantly more likely to have psychotic experiences (odds ratio 1.67). Interestingly, when Kirkbride et al. (2008) studied the notion of social capital, i.e. a measure of embedded community characteristics that provide support, they found a U-shaped relationship, whereby both low and high rates of social capital were associated with psychosis. They suggested that while social capital may have benefits, e.g. in buffering social stress, there may also be costs to individuals who are excluded from the social capital that is available (Kirkbride et al., 2008).

In a systematic review of loneliness in psychosis, Lim, Gleeson, Alvarez-Jimenez, and Penn (2018) define loneliness as when “one perceives their relationships to be inadequate to meet their need for belonging” (p 221). Their review highlights evidence for a significant positive relationship between

23 loneliness and psychotic symptoms (Michalska da Rocha et al., 2017); however they suggest that whether this a causal relationship is unclear due to lack of quality studies. One causal pathway from social isolation to psychosis has been proposed by Hoffman (2007) in the social deafferentation hypothesis, which suggests that when the social brain is deprived of input (through isolation), the neural networks for processing the social intentions, actions etc. of others produce ‘spurious’ social meaning in the form of voices and delusions. However, there is no empirical evidence to support the causal claims, let alone the precise mechanisms of causation, and further research is needed to support recent calls for psychosocial interventions to target loneliness in psychosis (Lim et al., 2018).

Perceived discrimination has been demonstrated as a risk factor for psychosis (Karlsen, Nazroo, McKenzie, Bhui, & Weich, 2005; Veling et al., 2007), and there have also been suggestions that perceived discrimination could be a major contributory mechanism in the well-established observation of higher psychosis rates in ethnic minority populations (Janssen et al., 2003). In a clinical high risk group, Stowkowy et al. (2016) found that perceived discrimination was an even stronger predictor of later conversion to psychosis than childhood trauma and bullying. In cognitive models, it is suggested that experiences of discrimination and/or perceived discrimination have an influence on the formation of beliefs and schemas, resulting in a paranoid attribution style (Janssen et al., 2003). In evolutionary psychology accounts, discrimination and marginalisation experiences are likely to activate social rank mechanisms (Birchwood et al., 2007), which is addressed in the following section.

The consistent evidence for higher incidence of psychosis among minority ethnic populations has not only increased the focus on discrimination experiences, but also on migration more generally. In a review of the literature, Morgan, Charalambides, Hutchinson, and Murray (2010) identified a number of social variables, in addition to perceived discrimination, that may contribute to the relationship between migration and psychosis. These include childhood factors, such as separation from a parent at a young age, as well as adult social disadvantage factors, such as being unemployed, living alone, being single, poorly educated, and having a limited social network. Other potential factors include the greater likelihood of migrants to live in urban areas (see above), as well as having exposure to the inevitable stressors of transitioning between countries, such as “unfamiliar cultural practices and beliefs, different climate and environment, challenging interactions with government institutions, and for some a new language” (p 658, Morgan et al. (2010)).

24

2.2.2. Social context of anomalous experiences

Research on the effects of social factors on anomalous experiences is based on the assumption that these psychotic-like anomalous experiences exist on a continuum throughout the population (van Os et al., 2009), and that while some people with these experiences may transition into developing a psychotic disorder, the majority will continue with their lives with no detrimental effect to wellbeing. It is therefore important to establish which factors are involved in transitioning to psychosis, and which factors may be protective. One research strategy has been to compare clinical and non-clinical groups with anomalous experiences (Brett et al., 2007; Heriot-Maitland, Knight, & Peters, 2012; Peters et al., 2017). In these two groups, who do not differ in prevalence of childhood trauma (Peters et al., 2016), a number of cognitive factors have been identified that likely increase the risk of developing psychosis and a need for care, for example, where experiences are appraised as more threatening and less controllable (Peters et al., 2017).

In one of the earliest studies comparing clinical and non-clinical groups, Jackson and Fulford (1997) suggested that social feedback is likely to be an important factor in determining how an anomalous experience becomes evaluated and embedded in a person’s life. They describe the case of ‘Sara’ who firstly described her experiences as terrifying before she received validation from a priest. The authors suggest that the kinds of social responses people receive from others have an important bearing on the consequences of that experience (possibly more so from an authority figure like a priest or even a doctor) (Jackson & Fulford, 1997). In a study using interpretive phenomenological analysis (IPA), the inter-personal context of anomalous experience was identified as being a key factor differentiating clinical from non-clinical groups (Heriot-Maitland et al., 2012). We found that more clinical than non- clinical participants had received invalidating social responses about their experiences. A quote from ‘Daniel’, a non-clinical participant, illustrated the subjective importance to him of a validating social experience: “I needed affirmation, that’s what I needed… to help me contexualise it and make sense of it… I suppose I did need kind of affirmation from other people that it was all ok” (p 46).

Other studies have emphasised the helpfulness of socially validating contexts. For instance, Brett, Heriot-Maitland, McGuire, and Peters (2014) identified perceived social support/understanding as a predictor of lower distress among people with anomalous experiences, and Powers III, Kelley, and Corlett (2016) found that ‘psychics’ who hear voices report more positive social reactions from their peers than do clinically diagnosed participants who hear voices. Among people with a diagnosed psychotic disorder, social support is also found to be a protective factor in the course of psychosis; for

25 example, higher levels of social support were associated with lower psychotic symptoms and fewer hospitalisations over the three years after a first episode psychosis (Norman et al., 2005).

An important finding of Brett et al. (2007), and later replicated (Brett et al., 2014), is the emergence of spiritual appraisals as a protective factor. In a study of the socio-demographic characteristics of groups with anomalous experiences, Peters et al. (2016) reported that a greater proportion of non- clinical participants (over 90%), compared to control and clinical participants, described themselves as spiritual. From a cognitive perspective, the interpretation is that spiritual appraisals may themselves be beneficial (i.e. less distressing or less threatening ways of making sense of an anomalous experience); however, from a social perspective, the interpretation differs; that spirituality and spiritual appraisal may be indicative of people’s access to accepting, validating social groups, with the protective benefits coming more from the social experience of feeling safe, connected and supported. Most likely, perhaps, it is a combination of the two, but it would be helpful to disentangle these social and cognitive mechanisms. This also applies to the interpretation, and potential re-interpretation of some of the historical research in this field. Traditionally, cognitive models of psychosis have focused on how adverse social experiences contribute to maladaptive cognitive appraisals of experiences; for example, Garety et al. (2001) state that “it seems likely that social marginalization, difficult or traumatic experiences or unsupportive family environments contribute to the development of negative schemas” (p 192). As seen in the above section, and will be seen in the next sections, there are other ways of conceptualising and interpreting these pathways, e.g. in terms of dissociation, attachment, and social rank processes.

2.2.3. Social rank mechanisms

According to social rank theory, there are information processing systems that evolved specifically for monitoring dominant-subordinate social roles and social threat (Gilbert & Allan, 1998). These systems are primed to detect information about the whereabouts and the intent of others, for example, whether their intent is friendly or hostile. When a person has experienced significant hostility or threat from a powerful other, the social rank systems will be attuned towards threat and can be easily activated (Heriot-Maitland et al., 2019). Social rank theory has been applied to understanding voices (Birchwood, Meaden, & Trower, 2014), paranoia (Seo & Choi, 2018), and bi-polar disorder (Gilbert, McEwan, Hay, Irons, & Cheung, 2007). A slightly different, but highly relevant model, has also been applied to understanding the more global concept of ‘schizophrenia’, namely the social defeat hypothesis (Selten, van der Ven, Rutten, & Cantor-Graae, 2013).

26

The social defeat hypothesis of schizophrenia (Selten et al., 2013) is based on the observation that social defeat may be a common mechanism underlying many of the major schizophrenia risk factors. The authors argue that social defeat processes can explain associations between schizophrenia with urban living, migration, low IQ, childhood trauma, and use of illicit drugs. Crucially, Selten et al. (2013) also link social defeat mechanisms with the well-established neuro-chemical substrates of psychosis, namely dopamine over-activity in the mesolimbic pathway. (For a narrative review of the biological arguments linking social defeat to psychosis see Selten, Booij, Buwalda, and Meyer-Lindenberg (2017).) In a large scale study of the 6646 participants in the Netherlands Mental Health Survey (NEMISIS-2), social defeat was found to be a mediator in the relationship between childhood trauma and psychosis (van Nierop et al., 2014). This provided strong support for the social defeat hypothesis, but with cross-sectional data only, the authors were unable to conclude that social defeat causes psychosis. In a later study, Seo and Choi (2018) also tested a mediation model of social defeat, but this time, in the relationship between childhood trauma and paranoia. Their sample was 199 Korean psychiatric patients, and they used structural equation modelling to demonstrate a pathway from childhood trauma to paranoia, through social defeat.

Social defeat and social rank are slightly different, but related, concepts. Social ranking is an evolutionary psychology concept, referring to the way humans have evolved to co-ordinate social (group) living, whereby low rank members are subordinate to high rank members, and that group cohesion (and self-preservation) is achieved by ‘hard-wired’ patterns of mental organisation, perception, feeling, behaviour, etc, called social ‘mentalities’ (Gilbert, 2005). So, for low rank members this involves perceiving oneself as inferior, thinking that others look down on them, and behaving submissively. Social defeat, on the other hand is the (current) experience of one being put down by a dominant other. Therefore, perceptions of social defeat may be highly associated with perceptions of low social rank (Gilbert, Allan, Brough, Melley, & Miles, 2002). Also, social defeat experiences, particularly those which are repeated or prolonged, are likely to activate and attune social rank patterns. Other experiences that might activate social rank ‘mentalities’ include entrapment (i.e. being unable to escape an uncontrollable situation), being criticised, bullied, shamed, or disempowered, as well as some internal experiences (in the self-to-self relationship) such as self-criticism and self-stigma.

The observations and evidence in support of the social defeat hypothesis of schizophrenia are therefore relevant in the application of social rank theory in psychosis. What social rank theory adds, however, is an understanding of the evolutionary context and function of these mechanisms. It also provides a framework for understanding the contribution of the self-to-self relationship of people with

27 psychosis, and, for voice-hearers, the relationship people have with their voices. For example, as suggested by Heriot-Maitland et al. (2019), social rank theory might help us understand why voice- hearing can involve “an internal ‘playing-out’ of both the hostile-dominant and the (reciprocal) threatened-subordinate social roles” (p 152). In other words, both the high rank and the low rank poles of the social rank system are activated in the relationship between voice and voice-hearer.

In the past, social rank processes have most commonly been connected to depression (Gilbert, 2016; Gilbert & Allan, 1998) and therefore applications of social rank theory in psychosis have often focused on affective pathways, with social rank being implicated in negative symptoms and depression outcomes in psychosis. For example, in a large sample of 2350 online participants, Jaya, Ascone, and Lincoln (2017) analysed the effects of three mediators (social rank, negative schemas, and loneliness) on the relationship between social adversity and psychosis (positive and negative symptoms). Low social rank, as measured by the Social Comparison Scale (SCS) (Allan & Gilbert, 1995), was found to be a significant mediator in the pathway to both negative symptoms and depression, but not in the pathway to positive symptoms. In their study, only negative schema mediated the relationship between social adversity and positive symptoms, and their results suggest that social rank may be more relevant to affective and depressive aspects of psychosis. However, other studies by Birchwood’s group have highlighted the role of social rank mechanisms in voice-hearing (a positive symptom) (Birchwood et al., 2004; Birchwood, Meaden, Trower, Gilbert, & Plaistow, 2000; Birchwood et al., 2007; Gilbert et al., 2001). In a review of this work, Birchwood et al. (2014) not only provide evidence that voice-hearing reflects a perception of low social rank, but also propose the applicability to other symptoms, including delusions and negative symptoms, as well as to broader aspects of the psychosis experience:

“Recent research has given strong support to the application of social rank theory to psychosis and shows that apparently disparate aspects of the psychotic experience—from voices to family relationships to diagnosis and hospitalisation—are all facets of the same process. This process involves a catastrophic loss of status in social rank terms, resulting in involuntary subordination, humiliation, and loss of self-esteem, and entrapment by powerful others. This can lead to anxiety, depression, and relapse in an ever-worsening decline in social status.” (p 144, Birchwood et al. (2014))

Another important social rank variable is shame (Keen, George, Scragg, & Peters, 2017; McCarthy- Jones, 2017; Michail & Birchwood, 2013; Turner, Bernard, Birchwood, Jackson, & Jones, 2013; Upthegrove, Ross, Brunet, McCollum, & Jones, 2014; Wood & Irons, 2016). Shame, and a host of related concepts, such as internalised stigma (Pyle et al., 2015a; Wood, Byrne, Varese, & Morrison, 2016) and self-stigma (Corrigan, Larson, & Ruesch, 2009; A. C. Watson, Corrigan, Larson, & Sells, 2007)

28 all reflect social rank patterns, in either self-to-other or self-to-self relationships, and can all result from experiences of being victimised or stigmatised. The experience of being seen as negative or inferior in the eyes of others (external shame) or perceiving oneself as inferior (internal shame) can be understood in terms of ‘social rank threats’, in that they are threats to one’s social self, and similar to social defeat experiences, can activate and attune social rank patterns.

A recent systemic review of shame and psychosis (Carden, Saini, Seddon, Watkins, & Taylor, 2018) identified 20 eligible papers with studies of clinical (n=8), non-clinical (n=8) and mixed clinical-non- clinical samples (n=4). In summary, they found evidence of a moderate-to-strong relationship in studies of shame and psychosis (clinical) and psychotic-like experiences (non-clinical). In relation to symptom-specific associations, they found that more studies showed positive associations between shame and paranoia than voices, and, importantly, one study also showed that shame had an amplifying effect on the relationship between stressful live events and paranoia (Johnson et al., 2014). A limitation was that all 20 studies included in this review were cross-sectional, and so future longitudinal research is needed in order to identity the directionality of these relationships.

In summary, there is a considerable literature implicating social rank mechanisms in psychosis. Although different terminology has been adopted by different researchers (e.g. social defeat, shame, internalised stigma, etc.), there are clear similarities at their core in terms of the activation of dominant-subordinate social rank mechanisms. Taking the broader literature together, therefore, social rank threats and mechanisms have been linked to many different aspects of psychosis: risk factors (social adversity risks and interpersonal trauma); depression in psychosis, and positive symptoms (paranoia and voice-hearing). Shame has also been linked to the expressed emotion pathway (Cherry, Taylor, Brown, Rigby, & Sellwood, 2017; Wasserman, de Mamani, & Suro, 2012), and as mentioned above, the biology of social defeat has been connected to the biological (e.g. dopaminergic) profile of psychosis. Finally, as Birchwood et al. (2014) highlight, social rank processes are important in understanding the wider social context and consequences of psychosis. However, as much of the research is cross-sectional, further studies are needed to test these processes with longitudinal designs.

2.2.4. Attachment mechanisms

Like social rank theory, attachment theory is an evolutionary model that explains the organisation of social interactions. Where social rank explains how social groups organise themselves hierarchically,

29 attachment explains how relationships are organised for caring and nurturing, ultimately in the interests of survival. According to attachment theory, human infants have a biological (‘built in’) drive to seek closeness to a protective caregiver, and to feel safe / secure within this affiliative bond (Bowlby, 1973). Securely attached infants use this relationship with the caregiver as both a safe haven (to calm distress) and secure base (from which to explore). However, in some inter-personal developmental contexts, these attachment relationships may not be possible, and so instead people develop what are called ‘insecure’ attachment patterns and styles. These patterns have been delineated in subgroups, e.g. anxious, avoidant, and disorganised attachment style, and each pattern has an important bearing on the growing child’s social functioning and emotion regulation.

A review of the literature on attachment and psychosis by Berry, Barrowclough, and Wearden (2007) highlighted the high rates of insecure attachment among people with psychosis, and suggested a number of ways in which attachment ideas could contribute to psychological understandings and treatments of psychosis. A later systematic review by Gumley, Taylor, Schwannauer, and MacBeth (2014) identified good evidence to support the relationship between avoidant attachment and both positive and negative symptoms of psychosis. They also found modest evidence to support a relationship between anxious attachment and positive symptoms. In their review, they also identified two studies that linked attachment insecurity to trauma, and particularly interpersonal traumatic events (Berry, Barrowclough, & Wearden, 2009; Picken, Berry, Tarrier, & Barrowclough, 2010). Recently, Lavin, Bucci, Varese, and Berry (2019) reviewed the literature specially looking at the relationship between insecure attachment and paranoia in psychosis. They found significant associations in 11 of 12 studies, and a pattern within these results whereby anxious attachment style was more strongly associated with paranoia than avoidant attachment style.

The theoretical literature on attachment and psychosis has mostly centred around disorganised attachment, which is an insecure pattern that often results from childhood trauma, and stems from a conflict between the attachment system and the threat-protection system (Barker, Gumley, Schwannauer, & Lawrie, 2015). In their theoretical account, Liotti and Gumley (2008) highlight the roles of attachment disorganisation, trauma and dissociation in psychosis, suggesting that attachment disorganisation has multiple influences on the pathway to psychosis, specifically through i) hindering affect regulation, ii) facilitating dissociative responses to trauma, and iii) creating difficulties with mentalising. The dissociation and mentalising pathways also have indirect contributions to affect regulation. For a review of the evidence and mechanisms of mentalising deficits on the psychosis continuum, see Debbane et al. (2016).

30

In summary, the attachment literature points towards disorganised attachment and dissociation as being implicated in the eliciting of psychotic experiences (e.g. voices), towards insecure attachment being implicated in the appraisal of experiences (Berry & Bucci, 2015), and towards insecure anxious attachment specifically in more paranoid thinking/appraisal styles (Lavin et al., 2019). Again, the further testing of these relationships in longitudinal studies is needed to evidence causality.

2.3. Where the two paths meet: Interaction of trauma and social pathways

The review of pathways to psychosis in the previous section has identified a number of potential mechanisms, some traumatic/dissociative, and others social/inter-personal. It also identified areas where their empirical and theoretical strands overlap; for example, the case of inter-personal trauma, which is linked to multiple different pathways, both dissociative and social. It may be that each mechanism could be activated by different (independent) determinants, or, in the case of inter- personal trauma, by the same determinant. There is also a chance that one or more mechanisms could interact, and potentially become determinants of each other, which is the topic of this section. Already noted has been one important model of integrated mechanisms, namely CAV (Berry et al., 2017), which describes the dissociation and attachment interactions in voice-hearing. In another integrated model, although not in the psychosis field, Sloman and Taylor (2016) outline an evolutionary account of childhood maltreatment in terms of the interaction between attachment and social rank systems. The focus will now turn to the area of interaction between dissociation and social rank.

2.3.1. The traumatic/dissociative properties of shame

As social animals, some of the most important threats to humans are those that operate in the social realm; for example, threats of social devaluation, rejection, and isolation. As Gilbert (2000) points out, while the threat for most animals is aggression, for humans it is “more commonly related to loss of acceptance and approval” (p175). In the psychosis literature, previous studies of shame (a socially- conscious emotion) have typically focused on the relationship between shame and emotional symptoms in the context of psychosis (e.g. anxiety (Michail & Birchwood, 2013) and depression (Keen et al., 2017; Upthegrove et al., 2014; Wood & Irons, 2016)). Shame has also been researched as a consequence of psychosis (A. C. Watson et al., 2007), as well as potentially having a role in post- psychotic trauma (Turner et al., 2013). Less attention, however, has been paid to shame as a potential causal mechanism in dissociative and psychotic experiences themselves. This may partly be due to shame not traditionally being regarded in ‘threat’ terms, with e.g. traumatic-like, perhaps dissociative,

31 properties of its own. However, what social rank theory offers is an understanding of shame within its evolutionary context, and with links to survival strategies. There are strong theoretical grounds to examine shame through the lens of threat and evolved threat-protection mechanisms. From this angle, social experiences, such as stigma, shame, and other threats to social self (Tracy & Robins, 2004) can be investigated for links to traumatic, dissociative processes, and subsequently as potential causal mechanisms in psychosis.

There is evidence for a relationship between shame and dissociation; for example Talbot, Talbot, and Tu (2004) found that levels of shame were positively associated with levels of dissociation, and that the strength of this relationship was greater among participants with childhood sexual abuse. Dorahy et al. (2017) tested a causal relationship between shame and dissociation. They gave students (n=78) both shame-inducing and neutral scripts, and while scripts were being listened to and repeated, participants were assessed on measures of shame (e.g. internal, external, eye gaze) and dissociation. They found that in the presence of shame activation, dissociative states increased. These findings give evidence for threat- or traumatic-like properties of shame (Dorahy et al., 2017), which are therefore of relevance to psychosis research. In another series of studies, Matos and colleagues provided evidence for traumatic characteristics of shame memories (Matos & Pinto-Gouveia, 2010; Matos, Pinto-Gouveia, & Duarte, 2012; Matos, Pinto-Gouveia, & Gilbert, 2013), and also found that the more traumatic a shame memory, and the more central it is to one’s identity, the more it predicts paranoia in the general population (Matos, Pinto-Gouveia, & Gilbert, 2013). Again, these studies suggest a more causal role for shame in dissociative phenomena, as well as in psychotic-like phenomena.

Other studies have investigated interactions between dissociation and threat activation. Although these studies have not looked specifically at shame, they may still have relevance in terms of understanding the consequences when dissociative traits and current threat interact. For example, Holmes, Brewin, and Hennessy (2004) researched the interaction of dissociation with a current threat stimuli (of watching an aversive film) in producing intrusions among non-clinical participants. They measured the intrusions experienced by participants following exposure to the film under experimental conditions that manipulated the participants' conceptual processing of threat stimuli, relative to perceptual processing. They found that both decreased conceptual processing of threat (an experimentally simulated trait dissociation), and spontaneous increases in state dissociation, led to increased intrusions of threat stimuli after the film. In another study, Marks, Steel, and Peters (2012) showed that healthy individuals with psychotic-like experiences had lower level trait conceptual integration, and more intrusions, following an aversive film than controls. Although these studies did

32 not specifically look at shame, they do pose interesting research questions for dissociation-shame interactions. For example, if, as suggested, shame is understood to operate as a threat stimulus (social threat), might we also expect dissociation-shame interactions to lead to increased intrusions?

2.3.2. Summary of the pathways and their areas of interaction

In pathways to psychosis, there is evidence for mechanisms of i) dissociation, ii) attachment disruption, and iii) social rank/shame. Inter-personal trauma has relevance as a determinant for all three mechanisms, as it has both the traumatic/dissociative aspects, as well as the social/inter- personal aspects. There are a range of other potential determinants for each mechanism, for example, those that potentially lead to i) dissociation (e.g. non-personal trauma, drug-use, existential crisis/impasse); ii) attachment disruption (e.g. parental mental health, separation, institutional living, neglect); and iii) social rank/shame (e.g. discrimination, bullying, criticism, expressed emotion, stigmatisation). There are also potential ways in which these mechanisms could interact, essentially becoming determinants of each other. This could be, for example, through the influence of either attachment (Berry et al., 2017; Liotti, 2006) or shame (Dorahy et al., 2017), or both, on dissociation.

Figure 2.1 provides a diagrammatic representation of the main areas reviewed in this chapter, with a particular aim of illustrating the areas of interaction; i.e. the social influences (attachment or social rank) on different parts of the trauma pathway to psychosis. The diagram is not intended to show the whole picture, but to illustrative where this thesis sits in relation to the literature, with the thesis research questions highlighted in bold. These research questions are elaborated in the following section.

33

Figure 2.1. Social influences on the trauma pathway to psychosis

Trauma / Attachment Social rank X adversity 4 patterns patterns

5 Attachment influence on dissociation 2 6 Dissociative properties of shame

Dissociation Thesis research question: 1 3 Does shame maintain dissociative processes and produce more PLEs? (Study 2) PLEs

Social context of anomalous experiences7 Social-rank influences in relating to PLEs8

Non-clinical Thesis research question: Clinical psychotic / no need Are social-rank and attachment patterns

for care associated with outcomes of PLEs? (Study 3) disorder

Expressed emotion influencing outcomes9

Stigma and shame influencing outcomes10

Clinical outcomes

1 2 Trauma relationship with psychosis (Bentall et al., 2014); Trauma relationship with dissociation (C. A. Ross & Keyes, 2004; Varese, Barkus, et al., 2012); 3Dissociation relationship with PLEs (Moskowitz et al., 2005; Pilton et al., 2015); 4Trauma relationships with attachment (Pearce et al., 2017; Pilton et al., 2016) and social rank/shame (Seo & Choi, 2018; Talbot et al., 2004; van Nierop et al., 2014) in psychosis; 5(Berry et al., 2017; Liotti, 2006); 6(Dorahy et al., 2017; Matos & Pinto-Gouveia, 2010); 7(Brett et al., 2014; Heriot-Maitland et al., 2012); 8(Birchwood et al., 2004; Gilbert et al., 2001); 9(Bebbington & Kuipers, 1994; Onwumere et al., 2009); 10(Keen et al., 2017; Turner et al., 2013; Upthegrove et al., 2014)

2.3.3. Aims and objectives of Section A studies

All three studies in Section A are designed to investigate and advance our understanding of psychological mechanisms in psychosis, but in different ways. Study 1 develops a questionnaire to tap PLEs in the general population, but specifically with the items, format, and psychometric properties that are needed for addressing the research questions and designs in Studies 2 and 3.

Study 2 tests the interaction of dissociation and social rank threat. As mentioned in Chapter 1, dissociation itself is not problematic, however, it has been consistently implicated in pathways to psychosis. This study is therefore interested in investigating how dissociation can become problematic

34 through its interaction with on-going experiences of threat; in this case, the social rank threat that is experienced through shame. Under ongoing conditions of social (shame-based) threat, it might be harder to reintegrate the dissociated conceptual and perceptual systems, therefore potentially driving and maintaining the unhelpful dissociative processes in psychosis. Specifically, Study 2 is designed as a longitudinal online questionnaire study, aiming to test the moderating effects of shame on the longitudinal relationship between dissociative traits and PLEs in a healthy population.

Study 3 similarly is designed to test relationships between social rank variables and PLEs by recruiting from a novel ‘non-clinical in crisis’ population with PLEs. This sample is potentially of interest from a psychological and social viewpoint as these individuals are seeking support for their PLEs through spiritual organisations, rather than through the mainstream mental health system. This means that their social experience (e.g. with stigma and shame) are potentially very different to their clinical counterparts who are also experiencing ‘crisis’ with their unusual experiences. As this is a novel population, the aims of Study 3 are to characterise the sample and look for patterns of relations between social, psychological, and PLE-related variables that could inform the development of future research hypotheses and questions with this group.

35

Chapter 3. Study 1) – Detecting anomalous experiences in the community: the Transpersonal Experiences Questionnaire (TEQ)

3.1. Abstract

3.1.1. Background There is growing recognition of the value of researching anomalous experiences in the general population to aid our understanding of the psychosis continuum. There are key differences in aims, foci and epistemologies of existing measures, with varying utility for specific research designs. This study addresses gaps in the literature by developing a measure of anomalous experiences with utility for longitudinal (time-sensitive) research, and with particular reliability for people towards the upper (high scoring) end of the continuum.

3.1.2. Methods An online sample were recruited from the general population to provide questionnaire data for two study parts: A) item selection and B) psychometric evaluation. For Part A, both classical test theory and item response theory methods were used to select which items to be included from an initial pool of 57, generated from non-clinical individuals with persistent anomalous experiences. For Part B, psychometric properties of the resulting measure were evaluated using exploratory and confirmatory factor analysis, and tests of reliability and validity.

3.1.3. Results Scores were provided by 532 participants, from which a 19-item scale, the Transpersonal Experiences Questionnaire (TEQ), was developed. The TEQ was found to be a unidimensional scale, with satisfactory internal consistency (0.85), good test-retest reliability, and convergent validity.

3.1.4. Conclusions The TEQ can be used as a unidimensional scale to detect anomalous experiences in the general population, with particular reliability for people with higher incidence of these experiences.

36

3.2. Introduction

Research into understanding the nature of psychotic phenomena is no longer limited to clinical populations, but is increasingly carried out in the general population, where psychotic-like experiences exist on a continuum (van Os et al., 2009). This has led to demand for scales to detect these experiences in community samples.

In the field of psychosis research, the term psychotic-like experiences (PLEs) has been widely used to differentiate these experiences in the general population from the psychotic symptoms of clinical populations. However, some researchers have preferred anomalous experiences (Bell, Halligan, & Ellis, 2006; Brett et al., 2007) as it still captures an out-of-the-ordinary quality of experience, but without pathological links and connotations. Another term, transpersonal experience (i.e. literally meaning experiences that go beyond the personal identity or self), has not been used within the psychosis research field, but is familiar terminology to those involved with the psychological study of spiritual phenomena (e.g. the field of Transpersonal Psychology; Friedman (2002)) as well as to the experiencer participants themselves, many of whom attach spiritual/mystical/psychic meaning to their experience. In researching these experiences, not only has the approach to terminology been varied, but so has the research measurement. Some measures take clinical (psychotic) symptoms as their starting point, while others are based on schizotypal personality traits. Symptom-anchored measures have generally either focussed on detecting a ‘proneness’ or ‘prodrome’ to psychotic disorders in the general population, or on demonstrating the continuum of PLEs in the general population, and therefore the non-pathological nature of these experiences. Schizotypy-anchored measures are focused on assessing psychosis-like personality traits distributed across the general population.

In a recent review, Grant, Green, and Mason (2018) highlight the key distinctions between the major schizotypy models, and call for future researchers to clarify the framework in which their research is being conducted; distinguishing between: i) a ‘quasi-dimensional’ framework, where schizotypal indicators represent ‘toned down’ (less explicitly expressed) manifestations of psychotic symptoms; ii) an Eysenckian ‘fully dimensional’ framework, where they represent a dimension of personality, distributed throughout the population, with psychotic illness at the extreme end; and iii) Claridge’s ‘fully dimensional’ framework (Claridge, 1994), where these indicators represent a normal personality dimension, but that this is separate to the dimension that marks a transition to psychotic illness. The latter of these has received considerable attention in the literature through the work of van Os and

37 colleagues (Kaymaz & van Os, 2010; Linscott & van Os, 2013), which dovetails with Claridge’s framework.

Item development for scales that are both ‘quasi’ or ‘fully’ dimensional is similar; i.e. items are sourced from clinical phenomena, among populations with psychosis-related diagnoses. Quasi-dimensional scales, like the Magical Ideation Scale (Eckblad & Chapman, 1983), clearly identify psychiatric symptoms as their reference point for item selection. Meanwhile, the fully dimensional scales, such as the Oxford-Liverpool Inventory of Feelings and Experiences (O-LIFE) (Mason, Claridge, & Jackson, 1995), comprises items that are “deliberately focused on trait, rather than symptom features, avoiding as far as possible stronger clinically worded items” (p205; Mason and Claridge (2006)). While the epistemological stance and wording may indeed be more deliberately tailored for non-clinical traits, the items themselves are still adapted from the symptomatology of clinical populations. The current study is aligned to Claridge (1994), in that it conceives a continuum of normal human experience. However, if conceiving these phenomena as distinct from pathology, then measurement should also be de-coupled from pathology; for example, using scales and items derived from outside the clinically diagnosed populations. This study has adopted an approach that uses the experience of non-clinical populations as the reference point, rather than the experience of clinical populations. Peters and colleagues have focused on recruiting people from spiritual / mystical / psychic organisations (Brett et al., 2014; Brett et al., 2007; Heriot-Maitland et al., 2012; Peters et al., 2016). These are individuals in the non-clinical or non ‘need-for-care’ population (i.e. those without psychosis-related diagnoses or services) with persistent anomalous experiences that they describe in spiritual or transpersonal terms. In this study, the approach was to elicit the experiences to be assessed, and the terminology of items, from this non-clinical population, and to use a title in keeping with the language of this population; namely, the Transpersonal Experience Questionnaire (TEQ).

The intention for the TEQ was to capture a broad range of experiences, across ideational and perceptual domains, and contain both positive and negative experiences (i.e. not just those linked to mental distress typically encountered in clinical services). This differentiates the TEQ from the recently published Questionnaire for Psychotic Experiences (QPE) (Rossell et al., 2019), which although capturing a broad range of experience, very much anchors its range within clinical symptomatology. This also differentiates the item development approach from many other existing measures of anomalous experience, which have often focused on specific domains, e.g. ideational (Peters et al Delusion Inventory (PDI) (Peters, Joseph, & Garety, 1999)), perceptual (Cardiff Anomalous Perceptions Scale (CAPS) (Bell et al., 2006)), or sensory phenomena (Multi-Modality Unusual Sensory Experiences

38

Questionnaire (MUSEQ) (Mitchell et al., 2017)). So not only would TEQ items be generated from non- clinical samples, they would also capture a wide variety of experience, which although out-of-the- ordinary, is not necessarily associated with distress or “need for care” (van Os et al., 2009).

For the current study, the aim was to develop the TEQ as a scale that is reliable for people with high incidence of these experiences, in order to serve future research in identifying and investigating these groups in the general population. This is also something that differentiates it from those ‘fully- dimensional’ scales that have items designed to measure a schizotypal trait that is normally distributed in the general population. While it was important to retain a wide breadth of experience, this study aimed to develop a measure with a limited number of items to facilitate use for future research. The selection preference was towards items with good reliability among high scorers; i.e. those with a variety of experiences over a given period. Item Response Theory (IRT) (Lord, 1980) allows the analysis of each item’s reliability as an indicator of a general continuum as well as each item’s reliability for different places along that continuum, i.e. where there are low, medium or high expressions of that continuum. Hence, for the current purposes, IRT could be used to identify items that are most useful for research into people with higher incidence of anomalous experience in the general population.

Another important consideration was for the measure to have research utility for longitudinal hypothesis-testing. Longitudinal studies can make causal claims about the relationships between psychological variables, and so measures are required to capture a snapshot in time, and to be sensitive to change over time. Many previous measures were not designed for this purpose and typically assessed general ‘traits’, rather than ‘states’. For example, the MUSEQ (Mitchell et al., 2017) asks participants to rate whether “there have been times when…” and the O-LIFE (Mason et al., 1995) asks about the presence of experiences in general, e.g. “do you sometimes feel…” and “on occasions, have you seen…”. This study aimed to develop items that could anchor experiences in time: “in the last 7 days” thus making the measure useful for longitudinal studies.

Therefore, the aim of this study was to establish a psychometrically robust but brief, state, measure called the Transpersonal Experience Questionnaire (TEQ) that could be used as a measure suitable for longitudinal research.

39

3.3. Methods

3.3.1. Sample and design

An online sample were recruited from the general population to participate in two parts of the study: Part A: item selection from the initial pool; and Part B: psychometric evaluation of the TEQ.

Exclusion criteria were being under 18 years, and reporting contact with mental health services for a ‘psychosis-related diagnosis’.

For Part A, a sample size of 285 was estimated to be adequate for item selection purposes (i.e., minimum of five participants per item for factor analysis (Stevens, 1996)). For item selection, the psychometric properties of all items were evaluated using classical test theory methods (endorsement, test-retest reliability, and internal consistency), item response theory methods (information/precision of each item), and content validity evaluations (experts’ input).

Part B evaluated the factor structure and psychometric properties of the TEQ measure. To evaluate test-retest reliability, it was estimated that a sub-sample of at least 50 would be required to repeat the measure seven days later. To evaluate convergent validity, a comparator (fully dimensional) measure of anomalous experiences was administered (O-LIFE-Unusual Experiences (UnEx) subscale (Mason et al., 1995)). The analysis of factor structure involved splitting participants into two sub- samples: one for exploratory factor analysis (EFA), and one for confirmatory factor analysis (CFA). The ‘EFA sample’ was the same as the ‘item selection sample’ (above) and a second, similarly sized, ‘CFA sample’ only completed the items selected for the TEQ.

3.3.2. Measures

The Transpersonal Experiences Questionnaire (TEQ): As part of an unpublished PhD study (Brett, 2005), Brett and colleagues generated a pool of questionnaire items from a non-clinical population with anomalous experiences, using the same extensive piloting, interviews, and adjusting procedures as in the Anomalous Experiences (AANEX) Interview (Brett et al., 2007). The aim was to use these questionnaire items (a total pool of 57 items) as the basis for the TEQ. Brett’s initial pool of 57 items were developed in the form of self-rated measurements of anomalous experiences in the general population. Item scores ranged from 1 (‘never experienced this’) to 5 (definitely experienced this

40

(frequently)’). For the current study, three adaptations were made: i) the items were converted to online format; ii) “In the past 7 days...” was added to anchor to a time frame and capture ‘state’; iii) response options were reduced to binary scores (Yes/No).

The Oxford-Liverpool Inventory of Feelings & Experiences (Mason et al., 1995) is a 159-item self-rated scale measuring four aspects of schizotypal personality in the general population. The current study used only the Unusual Experiences factor, a 30-item subscale pertaining to unusual perceptual experiences, hallucinations and magical thinking.

3.3.3. Procedure

The study was approved by the King’s College London PNM Research Ethics Subcommittee (ref: PNM/14/15-26) (Appendices 4.1 and 4.2). Participants were recruited via adverts placed on websites and email lists, including King’s College London (www.kcl.ac.uk) and www.experimatch.com. Participation was remunerated through a prize draw (1st, 2nd, & 3rd prizes), with an additional prize draw for those completing 7-day retests. Invitations for retest were sent to the first participants entering the study until at least 50 were completed, at which point invitations ceased.

3.3.4. Analysis

Multiple analysis methodologies were used. They are briefly described in the first paragraph of each results section for ease of presentation, and further information is provided in Appendix 3.1. The statistical software, Mplus 7 (Muthén & Muthén, 1998-2017), was used for Item Response Theory (IRT), exploratory factor analysis (EFA), and confirmatory factor analysis (CFA). For the EPA in Mplus, the WLSMV estimator was used to test for the dimensionality of the latent structure, and the rotation method used was promax. The rest of the analysis was conducted in SPSS 24 (IBM, 2016). Due to the online questionnaire tool providing complete datasets, there was no requirement for methods of handling missing data in any of the analyses.

3.4. Results

41

3.4.1. Characteristics of the samples

544 participants were recruited in total, of which 12 were excluded for a ‘psychosis-related diagnosis’, making a final sample of 532. Part A item selection included 283 participants. For Part B, the same 283 comprised the ‘EFA sample’ and a further 249 comprised the ‘CFA sample’. The majority were women (79%), white (white British 45% and white other 30%), and in the age range 18-29 (70%). Sixty-two percent had never visited services for a mental health problem. Table 3.1 shows the demographic details for each sample.

Table 3.1. Participant numbers for each part of the study, and demographics Part A Part B Item selection from initial Psychometric evaluation of TEQ pool of 57 items (N=532) (N=283) Exploratory Confirmatory factor analysis factor analysis (N=283) (N=249) Gender Male 63 (22.3%) 63 (22.3%) 44 (17.7%) Female 220 (77.7%) 220 (77.7%) 205 (82.3%) Age group 18-29 188 (66.5%) 188 (66.5%) 186 (74.7%) 30-39 44 (15.5%) 44 (15.5%) 30 (12.0%) 40-49 27 (9.5%) 27 (9.5%) 15 (6.0%) 50-59 14 (4.9%) 14 (4.9%) 8 (3.2%) 60+ 10 (3.5%) 10 (3.5%) 10 (4.0%) Ethnicity White British 130 (45.9%) 130 (45.9%) 111 (44.6%) White other 85 (30.0%) 85 (30.0%) 74 (29.7%) Mixed 15 (5.3%) 15 (5.3%) 18 (7.1%) Asian 43 (15.2%) 43 (15.2%) 31 (12.4%) Black 4 (1.4%) 4 (1.4%) 4 (1.6%) Other 6 (2.1%) 6 (2.1%) 11 (4.4%) Language English 216 (76.3%) 216 (76.3%) 193 (77.5%) Other 67 (23.7%) 67 (23.7%) 56 (22.5%) Education No degree 83 (29.3%) 83 (29.3%) 82 (32.9%) Degree 120 (42.4%) 120 (42.4%) 83 (33.3%) Higher degree 80 (28.3%) 80 (28.3%) 84 (33.7%) Mental Health services No 176 (62.2%) 176 (62.2%) 154 (61.8%) Yes 107 (37.8%) 107 (37.8%) 95 (38.2%)

42

3.4.2. Part A: item selection

Two analysis methods, a) classical test theory and b) item response theory, were employed in parallel with content validity checks, using independent evaluations by two researchers (author CHM and MSc student KH), and further content validity checks by a supervisor (EP). On occasions where an item was problematic on test criteria but had particular content value (e.g. in retaining the breadth of experiences from the initial pool of items (Brett, 2005)), the item was retained for subsequent stages of analysis. Where there was conflict between two items, content value was prioritised in decision to retain.

3.4.2.1. Classical test theory Classical test theory was used to remove items on the basis of endorsement, stability, and internal consistency criteria. Appendix 3.1 summarises the criteria used, and Appendix 3.2 shows how each item fared against these criteria. In total, 25 items were omitted at this stage. Six did not fulfil criteria but were retained because of their content validity. 32 proceeded to the next stage.

3.4.2.2. Item response theory In Stage 2, an item response theory (IRT) model (2-parameter logistic model (Lord, 1980)), was implemented to explore the precision (the information, in IRT terminology) of each item separately. As IRT models require unidimensionality, exploratory factor analysis (EFA) (Wirth & Edwards, 2007) was used on the 32 items to identify the number of IRT models required (one per factor). Although a one-factor model showed adequate fit (rel χ2=1.6, RMSEA=0.044, CFI=0.912, TLI=0.906), the 2-factor model provided a closer fit (rel χ2=1.3, RMSEA=0.033, CFI=0.955, TLI=0.949).

Figures 3.1 and 3.2 show IRT results for the two factors in the form of item information curves (IFC). Omission decisions were made from visual analysis of the curves, ensuring no replication of items (i.e. two overlapping item curves) and ensuring that reliable (highly informative) items were present in all levels of the factor (i.e. different positions along the x-axis). In line with the specific aims of this study, there was a particular interest in items that were reliable for high scorers (i.e. further right along the x-axis). Eighteen items loaded onto Factor 1 (Figure 3.1), of which seven were omitted (2, 13, 19, 29, 46, 49, 57). Fourteen items loaded onto Factor 2 (Figure 3.2), of which six were omitted (3, 5, 9, 11, 17, 18). In total, 13 items were omitted at this stage. The remaining 19 items comprised what will hereafter be referred to as the TEQ scale (Appendix 3.3).

43

Figure 3.1. IRT item information curves for Factor 1

Figure 3.2. IRT item information curves for Factor 2

44

3.4.3. Part B: Psychometric evaluation of TEQ

Participants were split into two samples to examine the factor structure: i) the EFA sample (n=283) and ii) the ‘CFA sample’ (n=249). The samples were then combined (n=532) to examine iii) reliability and validity of TEQ and iv) difficulty and discrimination of individual TEQ items. Finally, v) demographic characteristics of the combined sample (n=532), in relation to TEQ scores, are reported.

3.4.3.1. Exploratory factor analysis of TEQ In the sample correlation matrix (please see Appendix 3.1 for an explanation of why the correlation matrix is used in EFA), there were four eigenvalues above 1 (8.545, 1.966, 1.263, and 1.019). However, both a 1-factor model (rel χ2=1.4, RMSEA=0.040, CFI=0.935, TLI=0.927) and a 2-factor model (rel χ2=1.2, RMSEA=0.027, CFI=0.0973, TLI=0.966) provided close fit to the data, indicating that no further factors should be extracted. The scree plot (Figure 3.3) suggested the extraction of only one factor, but the confirmatory factor analysis was needed to decide which model provided the best fit.

Figure 3.3. Scree plot for exploratory factor analysis

3.4.3.2. Confirmatory factor analysis of TEQ Confirmatory factor analysis suggested a close fit for both the 1-factor (rel χ2=1.41, RMSEA=0.041, CFI=0.955, TLI=0.950) and the 2-factor (rel χ2=1.27, RMSEA=0.033, CFI=0.972, TLI=0.968) models.

45

Therefore, to identify whether or not the structure was unidimensional, the bi-factor model approach was used, as recommended by Reise et al (2007). The fit of the bi-factor model was close (rel χ2=1.1, RMSEA=0.021, CFI=0.990, TLI=0.987). However, in the presence of a general factor, the item loadings on the initial two factors were substantially reduced, becoming non-significant in most cases (Table 3.2). These results favour the use of TEQ as a unidimensional instrument, which is in line with the greater coherence in content for one, compared to two, factor structures. Therefore, the TEQ was considered unidimensional for the remaining analyses.

Table 3.2. Loadings for the one, two, and bi-factor two factor models 1-factor 2-factor Bi-factor 2-factor

(Μ1) (Μ2) (Μ3) F1 F1 F2 General F1 F2 Item λ p λ p λ p λ P λ p λ P TEQ01 1.11 <.01 1.22 <.01 1.06 <.01 1.26 <.01

TEQ02 1.02 <.01 1.1 <.01 1.32 .01 -.45 .32

TEQ03 .93 <.01 1.01 <.01 .86 <.01 .51 .01

TEQ04 .66 <.01 .70 <.01 .71 <.01 .05 .83

TEQ05 .88 <.01 .95 <.01 1.03 <.01 .69 .02

TEQ06 .94 <.01 1.02 <.01 .90 <.01 .40 .08

TEQ07 1.26 <.01 1.41 <.01 1.39 <.01 .17 .46

TEQ08 1.24 <.01 1.40 <.01 1.18 <.01 .74 <.01

TEQ09 1.23 <.01 1.36 <.01 1.61 .01 2.25 .03

TEQ10 1.51 <.01 1.73 <.01 1.85 <.01 .75 .11

TEQ11 1.01 <.01 1.10 <.01 .89 <.01 .90 <.01

TEQ12 .91 <.01 .98 <.01 1.03 <.01 .66 .03

TEQ13 1.47 <.01 1.75 <.01 1.50 <.01 .38 .09

TEQ14 1.09 <.01 1.20 <.01 1.19 <.01 .13 .58

TEQ15 1.18 .01 1.31 .03 1.56 .08 -.46 .34

TEQ16 .77 <.01 .83 <.01 .88 <.01 -.10 .69

TEQ17 1.05 <.01 1.16 <.01 1.01 <.01 .42 .03

TEQ18 .85 <.01 .91 <.01 1.10 <.01 -.46 .18

TEQ19 .81 <.01 .87 <.01 .85 <.01 .28 .23

3.4.3.3. Psychometric properties (reliability and validity) of TEQ The item endorsement varied from 4% (item 15) to 29% (item 5, see Table 3.3). Cronbach’s alpha was satisfactory (0.85) and no item was found to reduce the internal consistency of the scale (Table 3.2). The item-total correlations ranged from 0.35 to 0.52, further confirming internal consistency. The items retained in the final scale had at least 83% of agreement between the two time points (specifically 83 to 97% agreement between time points, Table 3.3), showing good test-retest reliability.

46

The total TEQ score was highly correlated with total O-LIFE-UnEx scores (Spearman’s rho=0.74, p<0.001), indicating convergent validity.

3.4.3.4. Psychometric properties (difficulty and discrimination) of individual TEQ items Table 3.3 presents the difficulty and discrimination parameters. Their corresponding Item Characteristic Curves (ICCs) and Item Information Curves (IFCs) are shown in Figure 3.4. The most difficult (to endorse) question was item 2, the least difficult was item 5, and the most discriminative across different levels of the continuum was item 15 (see Appendix 3.1 for all items). The 19 items of TEQ have increased precision when it comes to individuals with total scores at 1 to 3 standard deviations above the mean (as shown visually in Figure 3.4), meaning that the TEQ as a whole is able to reliably identify people at the higher end of the anomalous experience continuum.

Table 3.3. EFA/CFA loadings, IRT parameters, and reliability indices for TEQ items Reliability IRT Endorsement Loadings* Internal Stability parameters* consistency (test-retest) Item Short description N (%) EFA CFA A B AID ITC k % TEQ01 in contact 75 (14%) 0.65 1.11 2.14 -0.59 0.84 0.43 0.52 89.8 TEQ02 seeing 30 (6%) 0.59 1.02 +2.32 0.99 0.84 0.35 0.13 86.4 TEQ03 others read 73 (14%) 0.71 0.93 1.64 0.93 0.84 0.44 0.54 94.9 TEQ04 smellingthoughts 99 (19%) 0.54 0.66 1.56 0.71 0.84 0.36 0.41 81.4 TEQ05 thoughts whirl 155 (29%) 0.52 0.88 -0.93 0.74 0.84 0.40 0.10 96.6 TEQ06 'mission' revealed 54 (10%) 0.68 0.94 1.85 1.00 0.84 0.42 0.48 86.4 TEQ07 body sensations 84 (16%) 0.61 1.26 1.44 1.01 0.83 0.48 0.48 96.6 TEQ08 messages or hints 118 (22%) 0.76 1.24 1.05 1.14 0.83 0.52 0.35 89.8 TEQ09 picking up thoughts 87 (16%) 0.70 1.23 1.44 0.97 0.83 0.46 0.66 91.5 TEQ10 monitored 73 (14%) 0.67 1.51 1.45 1.22 0.83 0.51 0.40 91.5 TEQ11 others’ emotions 55 (10%) 0.72 1.01 1.81 1.02 0.84 0.43 0.62 91.5 TEQ12 isolated 104 (20%) 0.71 0.91 1.27 0.95 0.83 0.47 0.37 94.9 TEQ13 caused event 37 (7%) 0.67 1.47 1.89 1.35 0.84 0.46 0.78 96.6 TEQ14 time disorientation 103 (19%) 0.65 1.09 1.23 1.03 0.83 0.48 0.64 89.8 TEQ15 bodily movements 19 (4%) 0.84 1.18 2.29 +1.37 0.84 0.40 0.55 84.7 TEQ16 influenced by 41 (8%) 0.70 0.77 2.11 0.96 0.84 0.38 0.35 86.4 TEQ17 lossothers of identity 77 (14%) 0.78 1.05 1.45 1.13 0.83 0.50 0.47 88.1 TEQ18 hearing 47 (9%) 0.65 0.85 2.05 0.92 0.84 0.39 0.40 83.1 TEQ19 events in reference 79 (15%) 0.56 0.81 1.70 0.80 0.84 0.39 0.57 89.8 Abbreviations: N: number of positive responses. EFA: exploratory factor analysis (sample 1), CFA: Confirmatory factor analysis (sample 2), IRT: Item response theory model (complete sample). a difficulty parameter, b discrimination parameter, AID alpha if item deleted (complete sample), ITC item total correlation (complete sample), k Cohen’s kappa, % agreement (test-retest sample). *all loadings were significant (p<0.05). +, - in column a denote the most and least difficult (to endorse) item of the factor, and in column b denote the most and least discriminative item

47

Figure 3.4. Item characteristic and item information curves for TEQ items (n=532)

3.4.3.5. Demographic characteristics Non-parametric tests were used for demographics analyses due to skewed TEQ data. There were no significant differences between genders (Mann-Whitney U=23,260, p=0.706), ethnicity groups (Mann- Whitney U=25,971.5, p=0.774) or language (Mann-Whitney U=23,924, p=0.399). Age was not correlated with total TEQ (Spearman’s rho=0.04, p=0.311), nor were there significant differences found across age groups (Kruskal-Wallis ANOVA χ2=2.648, df=4, p=0.618). Education, however, was related to the scores. The higher the education, the lower the TEQ scores (Spearman’s rho=-0.26, p<0.001). Specifically, those with a higher degree scored lower than those with an undergraduate degree, who, in turn, scored lower than those without a degree (Kruskal-Wallis ANOVA: χ2=33.751, df=1, p<0.001). Those who had visited Mental Health services (for non-psychosis related difficulties) had higher scores than those who had not (Mann-Whitney U=26,094.5, p<0.001).

3.5. Discussion

3.5.1. Summary of results

The final 19-item TEQ incorporated a broad range of experiences, from ‘seeing things’, through ‘messages or hints’, to ‘time disorientation’. The final items are representative of the range of experiences emerging from Brett’s (2005) initial interviews, and showed good psychometric properties. The TEQ demonstrated good internal consistency, and factor analyses confirmed it is a unidimensional scale. TEQ also has good reliability and validity as a measure of anomalous experiences

48 in the general population. The IRT analysis demonstrated that the 19 items have particularly good precision for individuals scoring highly on the TEQ total score making it suitable for detecting and researching anomalous experiences in people at the higher end of the continuum.

3.5.2. Psychometric profile and potential uses

The unidimensionality of the scale was in contrast to the AANEX-Inventory (Brett et al (2015)), which showed a five-component structure of anomalous experiences (meaning/reference; paranormal/hallucinatory; cognitive/attention; dissociative/perceptual; first-rank symptoms). This difference may be partly due to different aims and analysis methods (Brett et al used Principal Components Analysis) but may also have been influenced by different samples. In Brett et al’s study, almost two thirds had been recruited from clinical services, which might explain the emergence of additional components of anomalous experience that are more uniquely clinical, particularly in the cognitive/attention domain.

As a scale with measurement precision at the upper end of the continuum, the TEQ potentially has research utility for studies comparing clinical vs non-clinical groups. For example, it could be used as a screening tool to match clinical and non-clinical groups for anomalous experiences. Such comparison studies are useful in examining associated psychological, social, and emotional factors (Peters et al., 2016), and with additional time-anchoring, the TEQ could be used in longitudinal research designs investigating causal relationships.

3.5.3. Individual item profiles

There were interesting findings in terms of endorsement rates for the TEQ items that mapped most closely onto ‘positive symptoms’; i.e. hallucinations (visual- item 2; auditory- item 18; olfactory- item 4) and delusions (paranoid- item 10; personal reference- item 19). Of the hallucination-type items, the highest endorsement was olfactory (15.5%), then auditory (8.1%) and visual (5.3%). With the delusion-type experiences, 12.7% had experienced ‘feeling watched or monitored’ in the past 7 days, and 13.4% had experienced ‘personal reference’. Considering this was specifically a non-psychosis population, these figures are relatively high. Studies of paranoid thoughts in the general population have reported similarly high prevalence, with 19% having the thought ‘I might be being observed or followed’ at least weekly (Freeman et al., 2005). However, these reports were not anchored to a specific 1-week period, unlike the current study. Epidemiological research has reported considerably

49 lower prevalence, with only 5.2% life-time prevalence of hallucinatory experiences and 1.3% for delusional experiences (McGrath et al., 2015). The higher prevalence of both types of experience found in this study, in a shorter timeframe (7 days), may reflect the sensitivity of the TEQ to detecting anomalous experiences, which will be useful for longitudinal research. However, an alternative explanation may be that these results were due to the self-selecting nature of the online cohort, which may have attracted respondents who had high rates of anomalous experiences.

Interestingly, the above symptom-related items are by no means the most discriminatory (i.e. able to discriminate between people with different levels of anomalous experience). Item 10 (‘feeling watched and monitored’) is one of the more discriminatory items in the scale, although still not as discriminatory as items 15 (‘bodily movements being controlled’) and 13 (‘events caused with your mind’). Item 18 (‘hearing voices’) is one of the least discriminatory items (15th of 19), which is noteworthy since hearing voices is found to be the most prevalent symptom among schizophrenia patients (WHO, 1973), occurring in around 59% with this diagnosis (Waters et al., 2014). This reinforces the ‘fully dimensional’ viewpoint (Claridge, 1994) that the construct being detected is not necessarily linked to pathology. If it were, one might expect that the major identifiers of this pathology would also come out as important identifiers and discriminators of its ‘severity’ in the population.

3.5.4. Demographic profiles

The demographic results showed no significant relationships between TEQ scores and gender, age, ethnicity, or language, but there was a significant relationship with education. Compared to a recent meta-analysis of likelihood of psychotic experience for different demographics (Linscott & van Os, 2013), these results are at odds regarding gender and age (the meta-analysis found greater likelihood among men and younger participants), but consistent regarding ethnicity and education. Peters et al (2016) also found a difference in education between clinical and non-clinical groups with anomalous experience, so it is interesting to see this difference reflected within the non-clinical population. The discrepancy with age and gender findings, compared to previous research, may be related to the sample recruited, the majority of whom were young (70% below the age of 30) and female (79%).

The other significant demographic difference was contact with mental health services for non- psychosis-related difficulties. The rate of diagnosis for a mood related condition is slightly above the general population average (in this sample 25% had a diagnosed mood disorder vs 17% in the general population (McManus, Bebbington, Jenkins, & Brugha, 2016)). Again, this could be due to the age and

50 gender bias in this sample, since the prevalence of mental health problems is elevated among women (21%), particularly young women (26% for 16-24 year olds) (McManus et al., 2016). The over- representation of people with anxiety or depression is likely to have some bearing on the prevalence of anomalous experiences, given these experiences are found to be more common among people with mood disorders (Varghese et al., 2011).

3.5.5. Limitations

One limitation is that the TEQ was developed with binary (yes/no) scoring. This means that a high total TEQ indicates that a larger number of items are endorsed. So the ‘upper end’ of the continuum refers to a greater variety of anomalous experiences and says nothing about quality of experience (e.g. intensity, frequency, duration etc). For instance, someone with different ‘low level’ or mild anomalous experiences would score higher on the TEQ than someone with a fewer number of ‘high level’ anomalous experiences. Other binary measures (Peters et al., 1999) have included secondary dimensions (e.g. distress level 1-5), so this may be a consideration for future TEQ developments.

Another limitation is online recruitment, which can make it harder to monitor and ensure data quality. This study attempted to control quality by keeping the questionnaires short and by giving participants the expected completion time before starting; however, there was no system of monitoring time actually spent. Another limitation of online questionnaires is the potential for sampling selection bias, in that respondents are only those who i) happen to have internet access, ii) happen to visit websites where research studies are advertised, and iii) decide to participate. Finally, the narrow demographic range (young, female) limits the generalisability of the results, and combined with broader limitations to the external validity of online methods, these TEQ results are only applicable to the current sample.

3.5.6. Conclusions

This study has developed and evaluated a reliable and valid tool for measuring state anomalous experiences in the general population. The TEQ is a unidimensional scale and has particular utility for longitudinal research among those who have a high prevalence of anomalous experiences, which is its most unique and useful contribution to the literature. Limitations are outlined, and future research should determine whether the TEQ shows similar promising psychometric properties in specific populations, for example clinical and non-clinical populations with persistent anomalous experiences.

51

Chapter 4. Study 2) – Social influences on the relationship between dissociation and psychotic-like experiences

4.1. Abstract

4.1.1. Background Shame is experienced as a threat to social self, and so activates threat-protective processes and responses. There is evidence that shame has trauma-like characteristics, which supports a move towards understanding these ‘threats to social self’ within the same conceptual framework as traumatic and dissociative processes. Evidence for causal links between trauma/dissociation and psychosis thus warrant the investigation of shame as a potential causal mechanism in psychosis. In this study, it was hypothesised that there would be an interaction effect between dissociation and shame in predicting psychotic-like experiences six months later. It was also hypothesised that social safeness would moderate this interaction, by buffering against these interacting threat processes.

4.1.2. Methods A longitudinal, online questionnaire design was used, with two time points (six months apart). Participants from the general population provided questionnaire data on the following scales: Dissociative Experiences Scale (DES); Other as Shamer Scale (OAS); Social Safeness and Pleasure Scale (SSPS); and Transpersonal Experiences Questionnaire (TEQ). A moderation model of the interaction DES x OAS, and a moderated moderation model of the interaction DES x OAS x SSPS, on TEQ6m (6 months later) scores, were tested using multiple regressions with bootstrap procedures.

4.1.3. Results 314 participants provided scores at both timepoints. Although there was no direct effect of DES on TEQ6m, as predicted the moderation effect of DES x OAS on TEQ6m was significant. SSPS was found to significantly moderate this DES x OAS interaction effect. Visual analyses of the interaction plots for ‘high’, ‘average’ and ‘low’ OAS scorers revealed complex patterns in the directions of these effects: For high OAS scorers, higher DES scores predicted higher TEQ6m scores, but for low OAS scorers, higher DES scores predicted lower TEQ6m scores. The interaction effects were negligible for low SSPS scorers, unexpectedly becoming more pronounced in the context of higher SSPS scores.

4.1.4. Conclusions The results demonstrate evidence for an interaction between dissociation and shame on its impact on psychotic-like experiences, which manifests more for those experiencing higher social safeness. This suggests a potential role of social mechanisms in both the aetiology and treatment of psychosis, which warrants their further testing in clinical populations. Evidence for the impact of shame on psychotic- like experiences, particularly among those with high dissociation, may also be relevant for psychosis interventions, in that they could be developed to target shame-reduction in psychosis. However, as the impact of the dissociation-shame interaction is more pronounced with higher levels of social safeness, interventions could seek to establish social safeness as the context in which to target shame.

52

4.2. Introduction

Evidence for dissociative and social pathways in psychosis and for the traumatic/dissociative properties of shame (Dorahy et al., 2017) supports a theoretical framework in which shame, a ‘threat to social self’, could influence dissociative processes as a potential causal mechanism in psychosis. The current study investigates the interplay of these key threat-related processes in psychosis: dissociation (a threat-response) and shame (a threat to social self). Specifically, it tests a dissociation-shame interaction model, whereby dissociation, a primary threat-based process that is related to psychotic experiences, becomes magnified or maintained through its interaction with shame, a (secondary) socially based threat. The suggestion, in line with social rank theory (Gilbert, 1992), is that negative social ranking experiences, which threaten human survival needs for acceptance and connection (threats to social self), will activate threat-responses, drive dissociative processes, and maintain vulnerability to psychotic-like experiences. One of the limitations of previous studies on dissociation and psychosis is that many have used cross-sectional designs, which cannot determine direction of causality. This study aimed to improve on this by using a longitudinal design. In order to create a large sample for greater precision, the study recruited from the general population, which is a commonly used recruitment strategy for studying mechanisms in psychosis, due to the evidence for a continuum of psychosis throughout the population (Peters et al., 2016; van Os et al., 2009).

Social safeness (pro-social, affiliative experiences) may also play a protective role in buffering or regulating these interacting-threat processes. Social support is found to have an important role in protecting against stress generally (Hostinar & Gunnar, 2015) and against psychosis symptoms more specifically (Norman et al., 2005). Building social safeness experience is one of the key aims of Compassion Focused Therapy (CFT) (Gilbert, 2009b, 2014), which is still in the early days of being applied and researched in interventions for psychosis (Braehler et al., 2013; Heriot-Maitland & Russell, 2018; Mayhew & Gilbert, 2008). According to the CFT model, and the neuroscience model of emotions on which it is based (Depue & Morrone-Strupinsky, 2005), experiences of social safeness and social threat are linked to different emotion systems, with distinct functions and physiologies (in the same way that para-sympathetic and sympathetic nervous systems are functionally and physiologically distinct). Hence, in this study, these two social mechanisms (threats to social self and social safeness) were explored as distinct mechanisms, each with potentially different influences on dissociative and psychotic processes. As distinct mechanisms, it is entirely plausible that both social threat and safeness systems could be activated in parallel; and hence, why a three-way interaction (dissociation x shame x social safeness) is important to test in a moderated moderation model.

53

Specifically, the following hypotheses were tested: 1) Dissociative traits will predict psychotic-like experiences (PLEs) six months later (Figure 4.1A). 2) Shame will moderate the longitudinal relationship between dissociative traits and PLEs (Figure 4.1B). 3) Social safeness will moderate the magnitude of the moderating effect of shame on the longitudinal relationship between dissociative traits and PLEs (Figure 4.1C).

4.3. Methods

4.3.1. Design and participants

This study used a longitudinal, online questionnaire design with two time points (six months apart) to test three models (Figure 4.1). The study sample was recruited from the general population, with the only exclusion criterion being that participants should not be under 18 years old. This web-based design has been employed previously in the general population to research psychotic-like experiences and predictive factors (Freeman et al., 2005; Oliver, McLachlan, Jose, & Peters, 2012)

Figure 4.1. Models tested A. Direct effect model. Dissociative traits (DT) predict psychotic-like experiences (PLEs) six months later, controlling for PLEs at baseline

DT PLE6m

PLE

B. Simple moderation model. Shame moderates the longitudinal relationship between dissociative traits (DT) and psychotic-like experiences (PLEs)

Shame

6m DT PLE

PLE 54

C. Moderated moderation model. Social safeness (SS) moderates the magnitude of the moderating effect of shame on the longitudinal relationship between dissociative traits (DT) and psychotic-like experiences (PLEs)

Shame

SS

DT PLE6m

PLE

4.3.2. Measures

Demographics Age, gender, ethnicity, education level, first language, and contact with MH services were collected.

Psychotic-like experiences (PLEs) The Transpersonal Experiences Questionnaire (TEQ) (Chapter 3) a 19-item self-rated scale measuring psychotic-like, anomalous, experiences. Items are rated Yes/No according to whether an individual has had the experience ‘in the past 7 days’. Total scores range 0-19. The TEQ has been validated for use in the general population and has good internal consistency (Cronbach’s alpha .85).

Dissociative traits (DT) The Revised Dissociative Experiences Scale (DES-II) (Carlson & Putnam, 1993) is a 28-item self-report scale measuring trait dissociation. Items are rated as percentages (in 10% increments from 0% = never to 100% = always) according to what percentage of time they have the experience. The DES-II captures feelings of depersonalisation, derealisation, and psychogenic amnesia. Total scores range 0-280. It is the most widely used measure of dissociation and has high internal consistency (Cronbach’s alpha .90).

55

Shame The Other as Shame Scale (OAS) (Goss, Gilbert, & Allan, 1994) is an 18-item self-report measure of external shame. Items are rated on a five-point scale according to the frequency of evaluations about how others judge the self (0 = never to 4 = almost always). Items tap a number of socially based concerns, such as ‘I feel other people look down on me’ and ‘other people see me as somehow defective as a person’, making it a suitable measure of threat to social self in this study. Total scores range 0-72, and it has high internal consistency (Cronbach’s alpha .92).

Social safeness (SS) The Social Safeness and Pleasure Scale (SSPS) (Gilbert et al., 2009) is an 11-item self-report scale measuring the extent to which people experience their social worlds as safe, warm and soothing. Items are rated on a five-point scale (1 = almost never to 5 = almost all the time). Items tap feelings of belonging, acceptance, such as ‘I feel connected to others’ and ‘I feel a sense of warmth in my relationships with people’, making it a suitable measure of social safeness in this study. Total scores range 11-55, and it has high internal consistency (Cronbach’s alpha .92).

4.3.3. Procedure

The study received ethical approval by the King’s College London PNM Research Ethics Subcommittee (ref: PNM/14/15-26) (Appendices 4.1 and 4.2). Participants were recruited via adverts placed on websites and email lists, including King’s College London (www.kcl.ac.uk) and www.experimatch.com. Upon responding to an advert, and providing informed consent, participants accessed the electronic questionnaires, hosted on the Bristol Online Surveys platform (www.onlinesurveys.ac.uk).

Participation was remunerated through a prize draw (1st £100, 2nd £50, & 3rd £25 prizes). Invitation for follow-up retest was offered to all participants, with an additional prize draw for those who completed this. Those opting-in were asked to leave their email address, so that a link to the retest questionnaire could be sent 6 months later. The two sets of responses (baseline and follow-up) were matched by participant identification codes.

4.3.4. Data analysis

In testing distribution normality, histogram and Q-Q plots of dependent variable (TEQ6m) demonstrated positive skewness, and a Shapiro-Wilk’s test confirmed non-normality (W=0.719,

56 p<0.001). Therefore, non-parametric tests were used for the correlation analyses (Spearman’s rho). For the regression analyses, in accordance with Russell and Dean (2016), the data were not log- transformed, but instead analysed using the recommended method of bootstrapping with the original (positively-skewed) TEQ6m scores. Correlation analyses were used to check the stability of DES scores over time, as required for its intended use as a ‘trait’ measure, and secondly, to check whether DES and TEQ were empirically distinct, as required for the testing of moderation models. A series of multiple linear regressions were then used to test the models. All analyses were conducted in SPSS 24 (IBM, 2016). The moderation models (B and C) were run in the PROCESS v3 macro (Hayes, 2017) for SPSS, using 5000 bootstrap samples. To test moderation, PROCESS uses a simultaneous entry method, as opposed to a hierarchical (i.e. entry in steps) method, with each of the models A, B, C being tested independently as a ‘stand-alone’ analysis. Despite using simultaneous entry, PROCESS is programmed to yield an R² change value within each moderation model, where R² change is mathematically identical to the squared semi partial correlation for the interaction term (Hayes, 2017); hence why it is not necessary to calculate an R2 change by comparing incremental improvements in fit between hierarchically different models (Hayes (2017), p 290). The directions of moderation effects were then interpreted visually using interaction plot graphs.

4.4. Results

4.4.1. Longitudinal sample selection

A sample of n=314 provided questionnaire responses at both baseline and 6-month follow-up. The number entering the study at baseline was 544, meaning that 58% comprised the final longitudinal sample. To examine selection bias in the sample, comparisons between completers (n=314) and non- completers (n=230) were tested using chi square tests (demographic data) and Mann Whitney tests (questionnaire data; data were non-normally distributed). See comparison table in Appendix 4.3. There were no significant differences found between completers and non-completers on gender, language, mental health service use, psychotic-like experiences, shame, or social safeness. However, completers were significantly older (35% over 30 years, compared to 24.8%), more highly educated (72.6% degree or higher, compared to 64.3%), more likely to be of a white ethnic background (80.6%, compared to 69.1%), and had lower dissociation scores (mean 13.78, compared to 16.74) than non- completers.

57

4.4.2. Demographics

In the final sample, most participants were women (80%), in the 18-29 age range (65%) and white (white British 52%, white other 29%), were highly educated (with 73% to degree level), and the main first language was English (79%). 57% of participants had never been in contact with services regarding mental health, and of those who did report contact, 34% had received a diagnosis (26% mood disorder, 1% psychotic disorder, 7% other) and 9% had not.

4.4.3. Correlations

The directions of all correlations were as expected (Table 4.1), with positive correlations between the measures of PLEs (TEQ), dissociative traits (DES) and shame (OAS), and negative correlations between each of these and the measure of social safeness (SSPS). Due to multiple correlations, p-values for detecting significance were adjusted to p<.01. All correlations were statistically significant at p<.01, with one exception, SSPS and TEQ6m, which had a non-significant negative correlation. The dissociative traits measure (DES) was re-administered at 6 months to check stability of scores over time. A strong positive correlation between DES scores at both timepoints (r=.78, p<.01) confirmed stability. The correlations between DES and TEQ at both time-points were only moderate (r=.57 (baseline) and r=.55 (6m follow-up), p<.01), meaning that the two variables of dissociative traits and PLEs could be regarded as empirically distinct, as is required for the testing of moderation models.

Table 4.1. Means, standard deviations, medians, interquartile ranges, and Spearman’s inter- correlations of variables at baseline and 6 months (n=314). Mean SD Med IQR TEQ DES OAS SSPS TEQ6m TEQ 2.45 2.99 1.00 4.00 DES 13.18 10.57 10.00 11.88 .57* OAS 22.72 12.61 21.50 16.00 .29* .36* SSPS 37.75 8.78 38.00 12.00 -.15* -.22* -.52* TEQ6m 2.11 3.02 1.00 3.00 .70* .47* .27* -.10 DES6m 12.08 10.33 9.29 11.52 .52* .78* .36* -.22* .55* *p<.01

4.4.4. 4.4.5. Model A: Direct effects

The results in Table 4.2A show that DES did not significantly predict TEQ6m scores at 6 months, controlling for TEQ scores at baseline (b=.021, t(311)= 1.636, p=.103), meaning that the first hypothesis (1) was not supported. However, the absence of a direct effect does not prevent

58

subsequent testing of moderation effects (Hayes, 2017), and therefore testing models B and C for moderation effects could proceed.

Table 4.2. Regression analysis results (n=314): Testing Models A, B & C, with TEQ6m as dependent variable. Coeff. (b) SE t p Model A: Direct effects Constant .129 .188 .686 .493 DES .021 .013 1.636 .103 TEQ .692 .046 15.082 **<.001

R2=.531, MSE=2.072, F(2,311)= 176.081, p<.001 Model B: Simple moderation Constant .506 .387 1.308 .192 DES -.038 .026 -1.458 .146 OAS -.009 .016 -.586 .558 DES x OAS .002 .001 2.313 *.021 TEQ .690 .046 15.049 **<.001

R2=.544, MSE=4.199, F(4,309)= 92.273, p<.001 Model C: Moderated moderation Constant .623 1.783 .349 .727 DES .003 .111 .026 .980 OAS .026 .057 .447 .655 DES x OAS -.004 .003 -1.269 .205 SSPS -.003 .041 -.073 .942 DT x SSPS -.001 .003 -.509 .611 OAS x SSPS -.001 .001 -.669 .504 DES x OAS x SSPS .0002 .0001 2.393 *.017 TEQ .710 .045 15.928 **<.001

R2=.584, MSE=3.887, F(8,305)= 53.441, p<.001 *p<.05; **p<.001

4.4.6. Model B: Simple moderation

Table 4.2B shows that OAS significantly moderated the relationship between DES and TEQ6m, controlling for TEQ at baseline (b=.002, t(309)=2.313, p=.021), with the interaction term (DES x OAS) explaining a small but significant amount of variance in TEQ6m (R2 change=.008, F(1,309)=5.350, p=.021). The direction of these effects is illustrated by the plots in Figure 4.2, which show the interaction slopes at different levels of OAS. For people with high OAS scores (the top slope), the higher the DES score, the higher the TEQ6m score. This is the expected direction (in hypothesis 2). However, the other plots show that for people with average OAS scores, there is only a marginal

59 positive relationship between the two, and for low OAS scorers, there is a negative relationship (i.e. the higher the DES score, the lower the TEQ6m score). Although there is an overall significant interaction effect, the interaction plots reveal that the direction/s of this effect are more complex than first anticipated.

Figure 4.2: Plots of the interaction effects (DES x OAS) on TEQ6m 3.5 Low OAS Av. OAS High OAS 3

2.5

6m 2 TEQ 1.5

1

0.5

0 Low DES Av. DES High DES

4.4.7. Model C: Moderated moderation

Table 4.2C shows that SSPS significantly moderated the magnitude of the moderating effect of shame on the relationship between DES and TEQ6m (b=.0002, t(305)=2.393, p=.017). The three-way interaction term (DES x OAS x SSPS) again explained a small but significant amount of variance in TEQ6m scores (R2 change=.008, F(1,305)=5.276, p=.017). Figure 4.3 displays the three-way interaction plots, showing how the above DES x OAS interaction operates at different levels of SSPS. The plots reveal that the interaction only seems to be occurring in the context of average and high SSPS scores. In the context of low SSPS scores (top graph), the different levels of OAS do not impact on how DES scores are influencing TEQ6m scores. So, again, although the three-way interaction effect is statistically significant, the plots reveal more complexity in the directions of effects, and actually showed SSPS to have a moderating effect in a direction opposite to what was expected (hypothesis 3), in that higher SSPS scorers had more pronounced DES x OAS interactions.

60

Figure 4.3: Plots of the interaction effects in the context of different levels of SSPS a) In the context of low SSPS: 3.5 Low OAS Av. OAS High OAS 3

2.5

6m 2 TEQ 1.5

1

0.5

0 Low DES Av. DES High DES b) In the context of average SSPS: 3.5 Low OAS Av. OAS High OAS 3

2.5

6m 2 TEQ 1.5

1

0.5

0 Low DES Av. DES High DES c) In the context of high SSPS: 3.5 Low OAS Av. OAS High OAS 3

2.5

6m 2 TEQ 1.5

1

0.5

0 Low DES Av. DES High DES

61

4.5. Discussion

The findings show evidence of an interaction effect between dissociation and shame, a social rank threat, in predicting PLEs six months later. This supported the moderation hypothesis (2). Importantly, this moderation / interaction effect was significant even in the absence of a direct effect (i.e. dissociation predicting PLEs at six months, which was not supported (hypothesis 1)). The interaction plots showed interesting differences in the predictive effects of dissociation on PLEs at different levels of shame: At high levels of shame (or specifically, experiencing others as shamers), higher dissociation predicted more PLEs; at average levels of shame, there was no predictive effect; and at low levels of shame, higher dissociation predicted fewer PLEs. The other main finding was that social safeness, which was negatively correlated with all other variables, had a significant moderating influence on the dissociation-shame interaction. While a significant moderated moderation effect was hypothesised (hypothesis 3), the interaction plots showed that directions of effects were the opposite of what was expected. At high levels of social safeness, the dissociation-shame interaction was more pronounced, in that shame levels were having a greater impact on the predictive effect of dissociation on PLEs. With average levels of social safeness, the influence of shame on dissociation was less marked, and in the context of low social safeness, there was no interaction between shame and dissociation in predicting PLEs.

4.5.1. No direct effect of dissociation on PLEs

The finding that dissociation did not predict PLEs directly at six months was at odds with some of the existing literature. For example, in a review and meta-analysis of 19, mostly cross-sectional studies, Pilton et al (2015) reported a significant positive relationship between dissociative experiences and voice-hearing. This discrepancy in findings might be due to a number of differences in the study design; firstly, the current study tested a greater range of PLEs than just voice-hearing (only one of the 19 TEQ items relates to voices), and also used a longitudinal design, controlling for PLEs at baseline, whereas the majority of studies reviewed by Pilton et al (2015) were cross-sectional. Looking only at the cross- sectional data in this study, there were significant correlations of moderate strength between dissociation and PLEs, so cross-sectionally there were relationships consistent with Pilton et al. (2015). This highlights the importance of longitudinal designs in differentiating between relationship effects and directional effects.

62

4.5.2. The interaction effect of dissociation-shame on PLEs

The second finding, that dissociation predicts PLEs at follow-up when in interaction with shame, shows that it is not simply a case of dissociation not being predictive of PLEs, but that further factors need to be present. At high levels of shame, high levels of dissociation lead to higher levels of PLEs. This supports the dissociation-shame interaction model. The more complex finding to interpret is why, at low levels of shame, high levels of dissociation would lead to lower levels of PLEs. To interpret this finding, it may be helpful to return to some of the basic theoretical principles of the function of dissociation. As noted in the introduction, dissociative processes are (themselves) understood to be fundamentally helpful: dissociation is regarded as an adaptive mechanism for dealing with life adversities, threats, shame, etc., as well as with more significant traumas and threats to survival. In threatening situations, dissociating can help to keep arousal down, which is helpful in preventing us from becoming overwhelmed, and in enabling us to function.

If we start from the theoretical premise that dissociation can be helpful, we can then consider circumstances in which a (helpful) dissociative process could become problematic. One such circumstance may be when dissociation gets overwhelmed by additional layers of threat, such as socially based threats to ‘social self’. In the introduction, the distinction was made between a primary threat-based process (dissociation) and a secondary, socially based, threat (shame), and perhaps it is the interacting influences of these that determines whether or not dissociation is successful in keeping threat arousal down. One interpretation of the interaction plots (Figure 2) is that in the context of low shame (bottom line), dissociation may be more successful at reducing arousal, creating less vulnerability to PLEs. This may be why we see the bottom line sloping down: the more dissociation people have, the better. However, in the context of high shame (top line), it may be that the dissociation is having to work much harder, in that it is not only having to deal with the reasons for a primary dissociation (which could be a range of factors from drug use, stress, through to historical trauma and adversity), it is also having to deal with shame. It may be that this double impact of primary and secondary processes creates overload in the threat emotion levels, which ‘normal’ (helpful) dissociation cannot manage, leading to increased mental divisions, sensory intrusions, and PLEs. This interpretation has appeal as it also fits with the first finding that dissociation, alone, is not a direct predictor of PLEs; it may only be when dissociation becomes overwhelmed (by interacting with other factors) that PLE vulnerability occurs.

63

There are other possible interpretations, for example around potential issues with how the constructs of dissociation, PLE and shame were measured. The dissociation measure, DES, which was used as a total score contains three separate factors (C. A. Ross, Joshi, & Currie, 1991). It may be that the different slope directions are to do with the different factors each having a slightly different relationship with shame and PLEs. A post-hoc analysis showed that although only two of the three dissociation factors were statistically significant (‘activities of dissociated states’ and ‘depersonalisation-derealisation’), the slopes for all three showed a similar pattern, with high shame producing an upwards slope and low shame producing a downward slope. The TEQ measure of PLEs also requires scrutiny when interpreting these results, because when reporting a high TEQ score does not necessarily mean that someone is having more PLEs in a week; it means that they are having more types of PLE in a week. More types of PLE may signal that PLE frequency has increased, but there also could be situations where someone may be having relatively few PLEs, but of two or three different types, compared to someone who is having lots of PLEs of the same type. This is an important consideration in understanding the findings; however, does not undermine the interpretation above (that dissociation becomes overwhelmed, leading to increased PLEs), as this could apply to either an increased variety of PLEs, or to an increased frequency of PLEs, or both. Further research is needed to disentangle these two dimensions of PLEs, perhaps along with other dimensions, such as intensity, emotional salience, etc. Finally, there are issues to consider around how OAS captures shame. Firstly, it only captures one aspect: external shame (perceiving the negative attitudes of others towards self), as opposed to internal shame (perceiving oneself negatively). Secondly, there might be some overlap between OAS and paranoia, which is one of the 19 types of PLE in the TEQ.

4.5.3. Social safeness influences on the interaction effect of dissociation-shame on PLEs

The third finding was that social safeness moderates the dissociation-shame interaction, suggesting that social safeness experiences, e.g. feeling connected to others and feeling accepted / understood by people, may have a role in influencing the dissociation-shame effect on PLEs. However, the interaction plots showed effect slopes that were in the opposite direction to the predictions. The initial hypothesis was based on the well-known finding that social support buffers stress and mental health problems, and particularly on the CFT approach to mental health and wellbeing (Gilbert, 2014), which aims to build people’s experiences of, and capacities for, social safeness and compassionate relating. When applied to people with psychosis, where threat-based dissociative processes are typically highly activated (Gumley et al., 2010), CFT aims to increase social safeness and compassionate mentalities to aid threat-regulation and re-integration of dissociated emotional parts. Therefore, it was expected

64 that higher levels of social safeness would have the effect of down-regulating the dissociation-shame interaction. However, what the plots in Figure 3 showed is that in contexts of high social safeness, the dissociation-shame interaction effect is more pronounced. With high social safeness, high shame was influencing dissociation to predict more PLEs and low shame was influencing dissociation to predict fewer PLEs. Hence, the interpretive task here is to consider possibilities for why people who feel less socially connected seem to be less influenced by differing shame levels, and why people who feel more socially connected seem to be more influenced by differing shame levels, in terms of the predictive effects of dissociation on PLEs.

One interpretation may be that when people are less connected, the experience of shame does not have such threatening (trauma-like / dissociative) properties. These people may be more isolated, so what they believe other people think of them is not such a concern or threat to their ‘social self’, and less likely to activate threat-protective responses. So external shame may be present, but not in such a socially threatening way. In the context of high social safeness, however, where people are more socially connected, engaged and integrated, shame is potentially more threatening; and hence more likely to have the trauma-like / dissociative properties. This could lead to the double-impact described above (i.e. both the primary and secondary threat pressures on dissociation). So, in this context of high social safeness, shame may be experienced as a more imminent threat to social self, creating more dissociation overload, and leading to increased PLEs. In other words, the key to this tentative interpretation is considering the different conditions in which shame may or may not have dissociative properties.

A consideration in interpreting these findings again requires a closer look at the measures, and in this case, particularly the items themselves. For some people, there may potentially have been an overlap between one or two items on the SSPS and the TEQ. For example, one SSPS item seems to have the potential for tapping social safeness/connectedness in more of a spiritual sense (item 4: “I feel part of something greater than myself”), which might overlap with one of the more ‘spiritual-type’ PLEs on the TEQ (e.g. item 14: “have you had an experience of a loss of your individual identity and a sense of being part of some greater whole that extends far beyond you?”). However, looking at the correlations in Table 1, it was found that SSPS and TEQ were negatively correlated (SSPS-TEQ significantly, and SSPS-TEQ6m non-significantly). These are quite low negative correlations, but the fact that they are negative at all suggests this is not likely to have confounded the results. Another point worth mentioning about items is the relationship between SSPS and OAS items. At first glance, some items on one could read as polar opposites. For example, “I feel insecure about others’ opinions of me” (OAS

65 item 4) and “I feel secure and wanted” (SSPS item 6). However, the negative correlation of -.52 is too low to suggest a polar opposition, combined with the fact that the spread of their scores are very different (skewness: SSPS -.21 and OAS .85, kurtosis: SSPS -.29 and OAS 1.18). This also supports the conceptual approach taken in this study of measuring social threat and social safeness as two distinct mechanisms, linked to different emotion systems and physiological process. This approach is line with Gilbert’s ‘three systems’ approach to emotion and emotion-regulation (Gilbert, 2009b, 2014) (also described in Chapter 6).

One final consideration in interpreting the three-way interaction is to think about the effects that social safeness experiences might have on accessing dissociated emotions and experiences, conceptualising them, and relating to them. In line with ideas from attachment theory, the social safeness experience could itself represent a ‘secure base’ (in attachment language) from which one is more able to safely access and open one’s mind to dissociated phenomena, as opposed to suppressing and avoiding them, which is what could happen when feeling less safe. This may be a further line of consideration, in that these results could reflect a greater awareness, articulation, or reporting of PLEs among people scoring high on dissociation who feel socially safe, compared to those who feel socially unsafe. Feeling socially unsafe might lead to more attempts to shut down / avoid PLEs, which doesn’t mean they are not there, just that they are not explored (or even embraced). The measurement of PLEs in this study was unrelated to the emotional tone of these experiences (e.g. whether they were positive or negative, distressing or benign), so it could be that socially safe participants have more openness and access to what dissociation-shame interactions were creating in their minds, some of which might be welcomed or valued (see Study 3 in Chapter 5, which explores different emotional outcomes with PLEs in a spiritual support-seeking population).

4.5.4. Research and clinical implications

This study has implications for fine-tuning models of mechanisms in pathways from social rank threat to psychosis (via dissociation). It suggests that future studies could be designed to test a modified, updated model, testing further the role of dissociation in psychosis with and without shame. The suggestion that dissociation, on its own, may not lead to PLEs (and may even reduce them potentially via lowered threat arousal), but may lead to PLEs when combined with ‘secondary’ (social) threats, is important for focusing attention towards social influences on psychological mechanisms. More generally, understanding socially based threats, such as shame, within a similar conceptual framework as traumatic and dissociative processes, could prove a fruitful direction for future research

66 into understanding psychosis. This is consistent with studies by Matos & Pinto-Gouveia (2010) and Dorahy et al (2017), who have previously investigated the ‘trauma-like’ qualities of shame, and it also aligns to Gilbert’s evolutionary analysis, which links social rank experiences, such as, stigma and shame to evolved threat-protection mechanisms (Gilbert, 2000, 2014).

These processes require further testing within clinical populations with psychosis, to see whether these patterns are consistent among those who have a received a psychosis-related diagnosis and are likely in many cases to be experiencing far higher levels of dissociation (Carlson & Putnam, 1993) and shame (Wood & Irons, 2016) than the current general population sample. While further research is needed within these specific populations, the already considerable evidence for a psychosis continuum throughout the population (Peters et al., 2016; van Os et al., 2009) strengthens the applicability of these results in understanding psychosis, whether clinically diagnosed or not.

These findings also have implications for the treatment of psychosis, for example, in focusing attention on social factors, not only as consequences of psychosis, but also as causal mechanisms. Interventions that specifically target shame, especially among those with high dissociation scores, could potentially have an impact on the occurrence of PLEs. Also, as the three-way interactions showed, the impact of targeting shame-reduction could be even more pronounced within the context of high social safeness, where dissociation-shame interactions were found to have the strongest effects. If these findings were replicated in psychosis samples, there may be implications for creating therapeutic experiences of social safeness, first, before targeting shame. The implications of this go beyond the therapy room, for instance, in improving and resourcing peer-support initiatives, like Hearing Voices Network groups. These groups are good opportunities for social safeness experience to be accessed and facilitated inter-personally. A final implication for clinical practice is that perhaps we should be more careful about not meddling with naturally occurring, adaptive processes. Rather than using treatments that interfere directly with processes such as dissociation, there may be a case for allowing these psychologically adaptive mechanisms to take place. Instead, the focus of interventions could be more about creating the social contexts that facilitate (and reducing those that impede) the helpfulness of dissociative processes that occur naturally.

4.5.5. Strengths, limitations, and future research

There are a number of strengths and limitations of this study. A strength is the longitudinal design which is better able to test the direction and causal relationships between variables. Online, web-

67 based designs can help with achieving better-powered longitudinal research studies; however, they are not without limitations. For example, there was a lack of quality control over the participant responses (e.g. effort, concentration, time taken on answering the questions). The study attempted some control by giving participants the expected completion time before starting; however, it was not possible to check that this was the actual time taken. Another limitation was the study sample, which was not representative of the general population; namely, the overrepresentation of students (highly educated, aged 18-29), and women. There was also evidence of some selection bias within the study, in that those completing 6 month follow-ups were older, more highly educated, more likely to be of a white ethnic background, and had lower dissociation scores than those not completing follow-ups. There were other limitations with regards to the measures used. For example, there may have been some overlap in the language between spiritually-connected-type items on both the SSPS and TEQ (e.g. SSPS item 4 ‘I feel part of something greater than myself’ could have transpersonal meaning to some). There may also have been overlap between DES and TEQ, for example, both contain an item on hearing voices. There was also the limitation of using a general population sample, rather than a clinical sample, meaning that the clinical implications discussed are speculative until further research is carried out.

Future research should seek to recruit more representative longitudinal samples from the general population, and to also adapt these hypotheses for testing within clinical populations with psychosis. Further clinical research is also indicated to test some of these findings within the context of interventions for people with psychosis that aim to reduce shame, within contexts of enhanced social safeness. This might include structured 1-to-1 therapy interventions, like CFT (Gilbert, 2014), but also extends to some of the more community-based social interventions offered by organisations such as the Hearing Voices Network and the Spiritual Crisis Network. Evaluating these socially based approaches, with both qualitative and quantitative methods could help to enhance our understanding of the role of social factors in both driving and helping psychotic phenomena. Another suggestion for future research would be to look specifically at the emotional dimensions of PLEs; for example, it would be interesting to explore the impact of these social mechanisms on PLE-related distress, as opposed to just on the occurrence and range PLEs (which could be malign or benign). Finally, future studies could include areas that were not measured in this study, but which may be highly relevant; for example, past or current traumas, adversities, and threats, or some of the other possible causes of what, in this study, have been referred to as primary dissociative processes.

68

4.5.6. Conclusions

This study tested interaction effects between dissociative processes and social factors in predicting PLEs in the general population. The results identified a moderating effect of shame on the link between dissociation and PLEs, and a further moderated moderation effect of social safeness. These results imply that a direction of future research would be to consider social experiences, particularly those linked to social ranking, as having a causal role in the aetiology of psychosis. The results also imply that closer examination is warranted on the direction of interaction effects, as more complexities and subtleties were revealed than first anticipated (for high, average, and low scorers). This calls for more nuanced future study designs to detect and highlight these patterns, with the aim of improving our understanding of the role of social factors in dissociation and threat processing in psychosis.

69

Chapter 5. Study 3) – Seeking spiritual support for psychotic like experiences: characteristics of individuals, contexts, and experience

5.1. Abstract

5.1.1. Background People in the general population who have psychotic-like experiences (PLEs) but with no need for care can inform our understanding of protective processes in psychosis. Some of these people appraise their experiences in a spiritual framework and may belong to cultural sub-groups who provide validation for their experiences. While previous studies have recruited such individuals from specialist sources such as psychics and new religious movements, this study specifically aimed to recruit a spiritual support-seeking population; i.e. a ‘non-clinical-in-crisis’ group, seeking support outside of traditional mental health services. As well as exploring the experiences and the social, psychological, and spiritual characteristics of this population, the study aimed to examine how these characteristics are related to different dimensions of PLEs.

5.1.2. Methods Thirty participants were recruited from spiritual crisis support organisations: International Spiritual Emergence Network (ISEN) and Spiritual Crisis Network (SCN). They completed standardised questionnaires and interviews to assess demographic, clinical, support-seeking, social, psychological, spiritual, and PLE characteristics. Partial correlations were used to test the relationships between social, psychological, and spiritual variables with PLE-dimension variables, namely whether PLEs are experienced as ‘distressing’, ‘safe’, ‘pleasant’, ‘disruptive’, or ‘enriching’, controlling for overall PLE scores.

5.1.3. Results The sample reported a high number of PLEs, significant trauma histories, and high levels of dissociation. They also reported high rates of previous contact with mental health services, which they had perceived as unhelpful. Higher total PLEs was related to more distress, however, once the total score of PLE endorsement was controlled for, these experiences were rated by participants as more ‘enriching’, ‘safe’ and ‘pleasant’, than ‘distressing’ and ‘disruptive’. The variables that were significantly correlated with positive PLE dimensions included spirituality, post-experience growth, and social safeness. The main variable that correlated with negative PLE dimensions was depression.

5.1.4. Conclusions The sample was found to have many characteristics similar to clinical populations with psychosis (e.g. trauma, dissociation, PLEs, and receiving support from crisis organisations). However, unlike their clinical counterparts, they reported more enhancing, and less distressing, relationships with their PLEs. The correlational analyses provide some preliminary indications of which social, psychological and spiritual variables may be related to more positive PLE outcomes, which could be the focus of future empirical studies to investigate causal effects and mechanisms.

70

5.2. Introduction

Psychotic-like experiences (PLEs) are present in the general population, not just in clinical populations who receive a mental health diagnosis and services. Identifying which factors are related to different dimensions and outcomes of PLEs can inform our understandings of psychosis and approaches to treatment. One strategy is to recruit people in the general population who report experiences of an equivalent nature to those with psychosis, but whose lives have not been negatively affected in terms of functioning and well-being (Brett et al., 2007; Heriot-Maitland et al., 2012; Peters et al., 2016). Studying this population has helped to establish which factors are protective against distress with PLEs. For instance, Brett et al. (2014) identified a number of protectors against distress, such as spiritual appraisals, perceived social support/understanding, perceived controllability of experiences, and reacting to experiences with a neutral response. A number of studies have found that, compared to clinically diagnosed groups, non-clinical participants report greater control over their PLEs, and are also less distressed by them (Baumeister, Sedgwick, Howes, & Peters, 2017; Johns & van Os, 2001). Powers III et al. (2016) looked specifically at a sample of clairaudient psychics who received auditory messages (akin to voice-hearing in psychosis), and found that the social integration of voice-hearing was different between the populations, with psychics receiving a more positive social reaction to their voices from peers, whereas clinical voice-hearers experienced more negative social reactions. These differing social acceptance and validation experiences were also a major finding in an earlier qualitative study comparing clinical and non-clinical groups with PLEs (Heriot-Maitland et al., 2012).

One of the benefits of sampling these, so called, ‘no need for care’ populations (van Os et al., 2009) is that they can provide researchers with information about protective factors in the absence of clinical procedures and interventions, e.g. psychiatric diagnosis and medication, which potentially could confound investigations of psychological and social mechanisms. However, a limitation of ‘no need for care’ population studies to date is that typically the PLEs have been present for years (average of 31 years in Peters et al. (2016)) and therefore what they are assessing is a long term adaptation to persistent PLEs. These studies cannot therefore provide information on factors that might be helpful for people whose PLEs are emerging and who are in crisis. For this reason, it is important to study samples who fall somewhere in between these populations, i.e. who are both outside of clinical services (to avoid the potential confounding of medication, diagnosis, stigma etc), and actively support-seeking and in crisis (to closely resemble the emotional context of clinical populations with a need for care). Research into ‘at risk mental state’ (ARMS) populations goes some way to address this, as these individuals have early signs of PLEs, but are not at the stage of attracting a diagnosis or

71 medication. However, the ARMS group are receiving support within the context of mental health services, and so will not provide information about the experience of support-seeking for PLEs outside of a clinical context. Even the language of ‘at risk’ is itself clinical and may therefore have psychological and social implications for how PLEs are perceived and incorporated.

This study aimed to identify a non-clinical-in-crisis population in order to investigate the characteristics and experiences of people seeking support for PLEs outside of the clinical context. Participants were identified through organisations (mainly charities) that specialise in helping people through spiritual, mystical or paranormal phenomena. One such UK organisation is the Spiritual Crisis Network (SCN), which is part of the International Spiritual Emergence Network (ISEN). People contacting these networks for support might not only be struggling with their PLEs (by virtue of accessing a “crisis” organisation) but might also be exempt from potentially confounding variables associated with either those who are already clinically diagnosed. In line with the previously researched non-clinical groups with PLEs, it was expected that this group would have adopted a spiritual framework and would differ from clinical groups by having socially validating reactions from others in relation to their PLEs.

This currently spiritual support seeking population is unstudied and so the research approach and questions were mainly exploratory, e.g. Who is this population? What kinds of experiences do they have? What support do they access? Do they find this helpful? However, it was possible to make some exploratory hypotheses about which social, psychological and spiritual factors might be related to positive outcomes, in terms of reduced distress and improved wellbeing. For example, following Brett and colleagues (Brett et al., 2014; Brett et al., 2007), it was predicted that spiritual frameworks of understanding would themselves be related to lower distress, perhaps due to helping people feel more connected to a meaning, or indeed, a greater meaning (see also Peters et al. (2017)). It was also predicted that socially safe and validating experiences would be associated with more positive dimensions of PLEs, e.g. experiencing them as more safe, pleasant, and enriching. Brett et al. (2014) also speculated that not only may protective benefits come through the support, advice, and companionship of a social group, but also through the inter-personal validation and normalisation of unusual experiences, with potential to reduce distress associated with social stigma and shame.

It is already well established that supportive social experiences regulate threat processes, at both psychological and biological levels (Hostinar et al., 2014). One possible psychological mechanism is through dissociative processes; i.e. social safeness exerts a regulatory effect by reducing threat- activated dissociation and facilitating mental integration. Previous research in non-clinical populations

72 has identified the important roles of dissociation (Cole, Newman-Taylor, & Kennedy, 2016; Humpston et al., 2016) and trauma (Arseneault et al., 2011; Moriyama et al., 2018) in psychotic-like experiences, which is consistent with the vast literature linking trauma and dissociation in clinical populations with psychosis (Pilton et al., 2015; Varese, Barkus, et al., 2012). It was predicted that high levels of trauma and dissociation would be reported in this spiritual support-seeking population and that social safeness experience (i.e. self-reported feelings of social connection and affiliation) would be associated with more positive and healthier integration of PLEs. It was predicted that positive PLE dimensions would be related to post-experience growth (Peters et al., 2016) and that PLE-related distress and disturbance would be associated with self-stigma appraisals (Pyle et al., 2015b), insecure attachment styles (Berry, Band, Corcoran, Barrowclough, & Wearden, 2007), and emotional symptoms (Hartley, Barrowclough, & Haddock, 2013). It was anticipated that this population would report a high endorsement of spiritual/religious-type PLEs.

In summary, this study was exploratory, with a focus on characterising this ‘non-clinical-in-crisis’ population. However, based on research evidence the following tentative hypotheses were made: 1) This sample will be characterised by a) higher levels of trauma and dissociation and b) higher endorsement of spiritual/religious-type PLEs1, in comparison to the general population 2) Self-reported levels of i) social safeness, ii) post-experience growth, and iii) spirituality will be positively related to safe, pleasant, and enriching dimensions of PLEs, and negatively related to distressing and disturbing dimensions of PLEs 3) Self-reported levels of i) self-stigma appraisals, ii) insecure attachment styles, and iii) emotional symptoms will be positively related to distressing and disturbing dimensions of PLEs, and negatively related to safe, pleasant, and enriching dimensions of PLEs

5.3. Methods

5.3.1. Design and participants

This was a cross-sectional questionnaire design. Participants were recruited from spiritual crisis support organisations: International Spiritual Emergence Network (ISEN), and its UK subsidiary, the Spiritual Crisis Network (SCN). The study was intentionally inclusive, and inclusion criteria were that participants: i) had current psychotic-like experiences (PLEs), identified by scoring at least 1 on the

1The definition of spiritual/religious-type PLEs in this study used, as a benchmark, the Alister Hardy Question: “Have you ever been aware of or influenced by a presence or power, whether you call it God or not, which is different from your everyday self” (Hardy, 1979).

73

TEQ (Chapter 3), which taps experiences “in the past 7 days”; ii) had recently contacted ISEN or SCN for support within the last 2 months; iii) were in crisis (by virtue of being actively/recently support- seeking). Participants were excluded if they were: i) under 18 years old; ii) unable to speak good English; and iii) unable to comprehend and consent to the research.

5.3.2. Measures

Demographic and support-seeking characteristics: age; gender; ethnicity; first language; education level; current employment/education/training; marital status; mental health diagnosis and treatment history, drug and alcohol use; and family mental health history. A support-seeking questionnaire to measure the sources, extent, and helpfulness of support-seeking was designed for this study. This included: Are there any distinct periods of time or episodes where you reached out for support and/or guidance about your experiences? From: a. family, b. friends, c. mental health services, d. other organisations/networks (if yes, list), each rated by i) extent 1 – 5 and ii) helpfulness 1 – 5.

Psychotic-like experience (PLE) characteristics

The Transpersonal Experiences Questionnaire (TEQ) (Chapter 3) is a 19-item self-report scale measuring psychotic-like, anomalous, experiences. Items are rated Yes or No for whether the individual has had the experience ‘in the past 7 days’. Total scores range 0-19. The TEQ has been validated for use in the general population and has good internal consistency (Cronbach’s alpha .85).

The TEQ-dimensions measure (Appendix 5.3) was adapted from TEQ by adding scores for associated emotion and impact on functioning for each item scored ‘yes’. The associated emotion was rated for distressing 1 (not at all) – 7 (very much so); pleasant 1 – 7; and safe 1 – 7 and the impact on functioning on disruptive 1 – 7; and enriching 1 – 7. Total TEQ-dimensions scores in each of these five domains are calculated as the mean score for all items endorsed (potential range of scores: 1-7). TEQ items that are not endorsed do not receive a dimension rating and are therefore not included in the mean.

Social, psychological and spiritual characteristics

The Social Safeness and Pleasure Scale (SSPS) (Gilbert et al., 2009) is an 11-item self-report scale measuring the extent to which people experience their social worlds as safe, warm and soothing. Items

74 are rated on a five-point scale and total scores range 11-55 (higher scores representing higher social safeness). The scale has high internal consistency (Cronbach’s alpha .92).

The Personal Beliefs about Experience Questionnaire (PBEQ) (Taylor, Pyle, Schwannauer, Hutton, & Morrison, 2015) is a 13-item self-report measure of appraisals of psychotic-like experiences, in the domains of negative appraisal of experience, external shame, and internal shame/defectiveness. Items are rated on a four-point scale (1 = strongly disagree to 4 = strongly agree). Although the measure has three scales, they have variable internal consistency so for this study we use only the total score, range 13-52 (higher scores representing more negative appraisals of psychotic-like experiences).

The Psychological Well-Being Post-Traumatic Changes Questionnaire (PWB-PTCQ) (Joseph et al., 2012) is an 18-item self-report measure for post-traumatic changes and growth, but in this study, we used the adaptation used in a previous study (Peters et al., 2016) that substitutes ‘post-trauma’ with ‘post- anomalous experiences’. Items are rated on a five-point scale reflecting the perceived change as a result of their anomalous experiences (5 = much more so now to 1 = much less so now). Total scores range 18-90 (higher scores representing higher psychological well-being). The questionnaire has good internal consistency (Cronbach’s alpha .87) (Joseph et al., 2012).

The Psychosis Attachment Measure (PAM) (Berry, Wearden, Barrowclough, & Liversidge, 2006) is a 16-item self-report scale measuring avoidant and anxious attachment. Items are rated on a four-point scale (‘not at all’ to ‘very much’) in response to questions about thoughts, feelings and ways of behaving in relationships. The measure produces two subscales: avoidant and anxious attachment. Total scores range from 0-3 (avoidant) and 0-3 (anxious), with higher scores representing higher attachment difficulties. Subscales have satisfactory Cronbach’s alphas of .75 and .82 respectively.

The Trauma History Questionnaire (THQ) (Hooper, Stockton, Krupnick, & Green, 2011) is a 24-item self-rated measure of trauma history where individuals rate whether a traumatic experience happened and if it did, the number of times it happened and their age at that time. The questionnaire covers ‘crime-related’ experiences, ‘general disaster and trauma’ questions, and questions about ‘physical and sexual experiences’.

The Revised Dissociative Experiences Scale (DES-II) (Carlson & Putnam, 1993) is a 28-item self-report scale measuring feelings of depersonalisation, derealisation, and psychogenic amnesia. Items are

75 rated as percentages (in 10% increments from 0% = never to 100% = always) according to the percentage of time they have the experience. Total scores range 0-280 (higher scores representing higher dissociation). It is the most widely used measure of dissociation and has high internal consistency (Cronbach’s alpha .90).

The Depression, Anxiety and Stress Scale (DASS-21) (Lovibond & Lovibond, 1995) is a 21-item self- report measure of emotional symptoms in three subscales: depression, anxiety and stress (seven items for each scale). Items are rated on a four-point scale (0 = never to 4 = almost always) to indicate the extent to which the statements apply over the past week. Total scores range 0-42 (depression), 0- 42 (anxiety), and 0-42 (stress), with higher scores representing higher emotional symptom levels. The subscales have good internal consistency (Cronbach’s alpha .94 for depression, .87 for anxiety, and .91 for stress).

The Duke University Religion Index (DUREL) (Koenig & Bussing, 2010) is a 5-item self-report scale measuring three dimensions of religiosity (religious attendance; private religious activities; and intrinsic religiosity). Total scores range 1-6 (attendance, 1 item), 1-6 (activity, 1 item), and 3-15 (intrinsic, 3 items). The internal consistency of these subscales ranges from acceptable to high (Cronbach’s alphas = 0.78 to 0.91). In this study, the wording ‘religious’ was replaced with ‘spiritual’ to make it more applicable to the sample.

5.3.3. Procedure

The study received ethical approval from the King’s College London PNM Research Ethics Subcommittee (ref: HR-16/17-3698) (Appendices 5.1 and 5.2). Recruitment adverts were placed on ISEN and SCN websites (www.spiritualemergencenetwork.org and www.spiritualcrisisnetwork.uk), and an opt-in link was placed on the email responses to people in crisis who had contacted these organisations. Once a participant had opted-in they were screened for eligibility using the TEQ over email or phone. Eligible participants met once with CHM via Skype to discuss their experiences and crisis. The self-report questionnaires were administered during the same Skype call, using the Bristol Online Surveys platform (www.onlinesurveys.ac.uk). Participants received £15 honorarium following this assessment.

76

5.3.4. Data analysis

Descriptive statistics and frequencies were used to characterise the sample’s demographics, support- seeking characteristics, and PLE characteristics. In testing normality of distribution, histogram and Q- Q plots of the TEQ showed that the data were positively skewed, and a Shapiro-Wilk’s test confirmed non-normality (W=0.719, p<0.001). Therefore, non-parametric tests were used for the correlation analyses (Spearman’s rho). A series of partial correlations were used to test the relationships between different PLE dimensions, between total TEQ and PLE dimensions, and between social, psychological, and spiritual variables and PLE-dimensions, controlling for the number of PLEs. All analyses were conducted in SPSS 24 (IBM, 2016).

5.4. Results

5.4.1. Demographic, clinical, and support-seeking characteristics

Thirty participants completed the study. The majority were female (80%), white British (63.3%) and were aged 30-39 (43.3%) and 40-49 (20.0%). The majority reported a family history of mental health problem (76.7%), of whom eight described their family members’ diagnoses as psychosis-related (e.g. schizophrenia), seven described mood disorders (anxiety and depression), and five described bi-polar disorder or cyclothymia. The large majority (80%) reported some involvement with mental health services, with nine having received a diagnosis of a psychotic disorder at some point in their lives, six a mood disorder, and seven an ‘other’ diagnosis, which included three bipolar disorder and two complex PTSD (see Table 5.1).

The most striking characteristic was that every single participant reported a trauma history in three main categories, supporting hypothesis 1a: crime-related events (reported by 20, 66.7%), general disaster and trauma (reported by 29, 96.7%), and physical and sexual experiences (reported by 26, 86.7%). Within the third category, sexual trauma alone was reported by 22 of 30 participants, of whom 20 described their sexual trauma experiences as multiple or repeated. The total number of different types of trauma reported ranged from 1 to 14 (M=7.3, SD=3.9).

The number of different PLEs reported in the last week ranged from 3-18 (M=10.0, SD=4.6). The sample’s dissociation scores ranged from 0-43.6 (M=16.9, SD=13.8), which is considerably higher than the average reported in the general adult population (M=5.4) and even slightly higher than the

77 average reported in the schizophrenia-diagnosed population (M=15.4) (Carlson & Putnam, 1993), again supporting hypothesis 1a.

Table 5.1. Demographic and clinical characteristics (n=30) Gender Family history of mental Male 6 (20.0%) health problem Female 24 (80.0%) No 7 (23.3%) Age group Yes 23 (76.7%) 18-29 4 (13.3%) Previous contact with mental 30-39 13 (43.3%) health services 40-49 6 (20.0%) No 6 (20.0%) 50-59 3 (10.0%) Yes. If yes, diagnosis: 24 (80.0%) 60-69 3 (10.0%) No diagnosis 2 (8.3%) 70+ 1 (3.0%) Mood disorder 6 (25.0%) Ethnicity Psychotic disorder 9 (37.5%) White British 19 (63.3%) Other 7 (29.2%) White other 5 (16.7%) Trauma – all types Mixed 4 (13.3%) No 0 (0.0%) Asian 2 (6.6%) Yes. If yes, number: 30 (100%) Black 0 (0.0%) 1-5 times 3 (10.0%) Other 0 (0.0%) 6+ times 27 (90.0%) Education Trauma – crime No degree 13 (43.3%) No 10 (33.3%) Degree 8 (26.7%) Yes. If yes, number: 20 (66.7%) Higher degree 9 (30.0%) 1-5 times 14 (70.0%) Employment / training 6+ times 6 (30.0%) No 9 (30.0%) Trauma – disaster Yes 21 (70.0%) No 1 (3.3%) Marital status Yes. If yes, number: 29 (96.7%) Married/Live partner 12 (40.0%) 1-5 times 16 (55.2%) Single 9 (30.0%) 6+ times 13 (44.8%) Divorced 6 (20.0%) Trauma – physical & sexual Other 3 (10.0%) No 4 (13.3%) First language Yes. If yes, number: 26 (86.7%) English 26 (86.7%) 1-5 times 10 (38.5%) Other 4 (13.3%) 6+ times 16 (61.5%) Drugs / alcohol – past Psychotic-like experiences No 6 (20.0%) Mean 10.0 Yes 24 (80.0%) SD 4.6 Drugs /alcohol – current Dissociative experiences No 20 (66.7%) Mean 16.9 Yes 10 (33.3%) SD 13.8

Mental health services were the least used source of support (mean ‘extent’ rating M=2.8, SD=1.6), along with family support (M=2.9, SD=1.4). The most sought support was from ‘other’ sources (M=3.5, SD=1.3) and from friends (M=3.2, SD=1.2). The perceived helpfulness of the different sources followed

78 a similar trend, with ‘other’ and ‘friends’ rated as the most helpful sources of support (M=3.7, SD=1.3 and M=3.5, SD=1.1 respectively), and mental health services were rated as the least helpful (mean helpfulness 1.8 (SD=1.0), where 1 = not at all helpful and 2 = a little bit helpful). A wide range of other support was reported, from ISEN and SCN (the two main recruitment sources), through religious communities, to online forums and Facebook groups. A full list of support sources is in Appendix 5.4.

In the comments section of the support-seeking questionnaire, there were indications of why participants had not found mental health services helpful. For example, participant 5 described their mental health service contact as “a traumatizing experience, and the medication made my living almost unbearable”, and participant 6 said that they “found the staff cold and removed”. Participant 25 said that there was “little opportunity in that service to talk about belief/experiences/perceptions”, and participant 19 said “they told me it was all in my mind (the vibration) and that it was due to a chemical imbalance”. This was in contrast to some of the comments made about the helpfulness of other sources of support. For example, participant 13, who connected with a Facebook peer group, said that “when the energies came in, I shared them with the group in joy. They celebrated with me and also offered intuitive guidance relating to the energies”. Participant 11, who contacted the SCN, said that it “helps especially when feeling isolated”, and participant 21 reflected that “sharing with likewise people was crucial for my recovery”.

5.4.2. Psychotic-like experience (PLE) characteristics

All participants in the sample reported at least three different psychotic-like experiences in the week prior to assessment, as measured by the 19-item TEQ. The highest number of different experiences reported by a participant was 18. Table 5.2 shows that the highest ratings for experience-related dimensions were reported for ‘enriching’, ‘safe’ and ‘pleasant’. A post-hoc paired t-test comparing the mean of positive dimensions (enriching, safe, pleasant: combined mean = 4.34 and Shapiro-Wilk normal distribution W=0.970, p<0.568) with negative dimensions (distressing, disruptive: combined mean = 3.06 and Shapiro-Wilk normal distribution W=0.965, p<0.460) was significant (t=4.55, p<.001). There were significant correlations between the TEQ score and the ‘distressing’ and ‘pleasant’ scores (the latter inversely), but not the ‘disruptive’, ‘safe’, and ‘enriching’ scores. As expected, there were positive correlations between the ‘distressing’ and ‘disruptive’ scores, and between the ‘pleasant’, ‘safe’ and ‘enriching’ scores, although not all these correlations reached significance.

79

Table 5.2. Means, standard deviations, and Spearman’s inter-correlations of PLE dimensions TEQ- TEQ- TEQ- TEQ- M (SD) TEQ distress Pleas Safe disrupt TEQ 10.0 (4.6) TEQ-distressing 3.0 (1.3) **.53 TEQ-pleasant 4.2 (1.3) *-.37 **-.52 TEQ-safe 4.5 (1.3) -.26 **-.53 **.72 TEQ-disruptive 3.1 (1.2) .33 **.62 *-.38 -.27 TEQ-enriching 4.6 (1.3) -.03 -.36 **.82 **.76 -.25 *p<.05; **p<.01

Figure 5.1. Characteristics of individual psychotic-like experiences (n=30) a) Endorsement of each TEQ item

b) Mean TEQ-dimension ratings

80

Figure 5.1 shows a breakdown of different types of PLE experienced by participants (Figure 5.1a), along with mean dimension ratings for each different experience (Figure 5.1b). The most commonly endorsed item was TEQ8 (83% of the sample), which refers to “experiences in which things in the world around you seemed to contain messages or hints, perhaps in a metaphorical or symbolic way”. The least endorsed item was TEQ15 (23%), which refers to experiences of “your bodily movements being controlled by someone or something outside of you”. See Appendix 5.3 for the full list of TEQ items. Hypothesis 1b was supported by the finding that the sample’s top five endorsed experiences, each with at least 70% endorsement, were religious/spiritual in nature, involving being “aware of or influenced by a presence or power, whether you call it God or not, which is different from your everyday self” (Hardy, 1979). To assess whether the most endorsed PLEs in this sample were relatively more religious/spiritual than in the general population, a table was constructed (Appendix 5.4) comparing the TEQ item endorsements and rankings of this sample with those of the general population sample from Study 1 (Chapter 3). Only one of the top five items in this sample was also in the general population top five (TEQ8: ‘message or hints’). The other top five items (TEQ 1, 6, 7 and 17) are all of a religious/spiritual nature (e.g. ‘in contact’, ‘mission revealed’, and ‘loss of identity’), whereas others in the general population top five (TEQ 5, 9, 12, and 14) are non-religious/spiritual (e.g. ‘thoughts whirl’, ‘isolated’, and ‘time disorientation’).

In terms of PLE-dimensions, the experience rated as the most ‘enriching’ was TEQ17, which asks “have you had an experience of a loss of your individual identity and a sense of being part of some greater whole that extends far beyond you?” This highly enriching experience, which was reported by 73% of the sample in the past seven days, also showed very high ‘safe’ ratings. The next most enriching items, closely behind TEQ17, were items TEQ8 and TEQ6. Item TEQ8, already mentioned to be the most endorsed item, also attracted the highest equal ‘safe’ rating. The most distressing experiences were TEQ5 and TEQ12, both endorsed by around half the sample (50% and 57% respectively). Item 5 refers to the experience of having “thoughts rushing very rapidly through your mind” and item 12 refers to feeling “cut off or isolated from things and people around you”. These two items were also rated as the most disruptive experiences.

5.4.3. Relationships between social, psychological, and spiritual variables and PLE-dimensions

Partial correlations between each of the social, psychological and spiritual variables and each of the PLE-dimensions, controlling for the total number of PLEs, are presented in Table 5.3.

81

Table 5.3. Means and standard deviations of social, psychological, spiritual characteristics, their Spearman’s correlations with TEQ, and their partial correlations with TEQ-dimension TEQ-dimensions, controlling for TEQ TEQ- TEQ- TEQ- TEQ- TEQ- M (SD) TEQ distress pleas safe disrupt enrich SSPS 40.2 (8.3) -.15 -.12 .32 .36 -.34 *.41 PBEQ 26.1 (6.0) .33 .11 *-.42 -.28 .13 *-.38 PWB-PTCQ 78.9 (10.9) -.07 *-.41 **.53 .36 *-.45 **.52 PAM-anx. 1.2 (0.7) .21 .35 -.29 *-.39 .09 -.36 PAM-avoid. 1.4 (0.7) .14 -.12 .04 .03 .18 .05 DASS-D 9.0 (8.3) .17 *.38 **-.56 **-.59 .18 **-.56 DASS-A 11.0 (9.8) .35 .25 -.30 *-.40 .13 -.33 DASS-S 18.8 (10.9) .35 .29 *-.44 **-.53 .32 *-.46 DUREL-attend. 3.4 (1.7) *-.37 .11 .28 .24 -.08 .37 DUREL-activit. 4.9 (1.3) .21 -.25 *.43 .37 -.14 *.48 DUREL-intrin. 12.8 (2.8) -.15 *-.41 **.51 *.39 **-.50 **.57 *p<.05; **p<.01

PLE ratings for ‘enriching’ were significantly positively related to social safeness, post-experience growth, spiritual activities, and intrinsic spirituality; ‘pleasant’ was positively related to post- experience growth, spiritual activities, and intrinsic spirituality; and ‘safe’ was positively related to higher intrinsic spirituality. This provides some support for hypothesis 2. No significant relationship was found between ‘pleasant’ and social safeness, or between ‘safe’ and social safeness or post- experience growth. As predicted in hypothesis 2, PLE ratings for both ‘disruptive’ and ‘distressing’ were significantly negatively related to post-experience growth and intrinsic spirituality; however, no evidence was found for their significant negative relationships with social safeness.

PLE rating for ‘distress’ was significantly positively related to depression, but not to self-stigma appraisals, insecure attachment styles, or anxiety or stress. None of the significant positive relationships with ‘disruptive’ predicted in hypothesis 3 were found. Significant negative relationships were found between both ‘enriching’ and ‘pleasant’ with self-stigma appraisals, depression, and stress, which were all in line with hypothesis 3 predictions. But the hypothesis that insecure attachment and anxiety would also be negatively related was not supported. Finally, the PLE dimension of ‘safe’ was found to be significantly negatively related with all the emotional symptoms (depression, anxiety and stress), as well as to anxious insecure attachment style; however, the prediction that ‘safe’ would also be negatively related to self-stigma appraisals was not supported.

82

5.5. Discussion

In summary, this spiritual support-seeking sample was characterised as having a high incidence of prior contact with mental health services (80%) but had generally experienced these services as unhelpful. Their historical use of services differentiates this sample from those recruited from spiritual populations in previous studies (Brett et al., 2014; Heriot-Maitland et al., 2012; Peters et al., 2016). The alternative sources of support, particularly from friends and peer-support groups, spiritual networks and organisations, were experienced as more helpful. The sample was also characterised as having considerable trauma histories, in line with the hypothesis. Many types of PLE were reported, with an average of 10 different types of experience in the past 7 days. As expected, the most commonly reported types of PLEs were of a religious/spiritual nature. Although having a greater incidence of PLEs was significantly related to distress, these PLEs were experienced as more highly ‘enriching’, ‘safe’ and ‘pleasant’, compared to ‘distressing’ and ‘disruptive’. The most ‘enriching’ experience was having “a sense of being part of some greater whole that extends far beyond you”, which maps closely onto Hardy’s (1979) original articulation of religious/spiritual experience. The most common experience was that things around people contained “messages or hints, perhaps in a metaphorical or symbolic way”. In line with the hypothesis, spiritual engagement, activities, and living were strongly related to positive PLE-dimensions, particularly the measure of ‘intrinsic spirituality’, which reflects the extent to which people adopt spirituality as a guiding principle in their lives. Depression, stress and post-experience growth also showed high correlations across multiple PLE- dimension domains. ‘Enriching’ PLEs were positively correlated with social safeness levels and negatively correlated with self-stigma appraisals.

5.5.1. Demographic, clinical, and support-seeking characteristics

The demographic make-up of the sample was largely female, middle-aged, white, and fairly high functioning (in terms of education and employment). The finding that 80% of the sample had previously accessed mental health services was surprising and interesting. In designing this study, the intention and expectation was to recruit a support-seeking population outside of mental health services. Instead, this is a population which has come into brief contact with services in the past, and who did not find them helpful. These are people who seemed to have multiple support options available, so mental health services were often just one of many. Of this 80% (24 participants) who did access services, only nine had been diagnosed with a psychosis-related condition, so the majority of these mental health service episodes were reported as non-psychosis related. Family history of

83 mental health problems was at a similarly high level (76.7%), so there was certainly an awareness of mainstream mental healthcare within this sample. However, it may be that from their own, or a family member’s, negative experiences of services that people were prompted to look for alternatives.

The finding that this sample had marked trauma histories was consistent with the evidence for traumatic life events as a causal factor in psychosis and psychotic-like experiences (Moriyama et al., 2018; Varese, Smeets, et al., 2012). The finding that 100% of the sample reported at least one traumatic event in their lives is consistent with the very high rates of trauma found in severe mental illness samples (Grubaugh, Zinzow, Paul, Egede, & Frueh, 2011), in line with the 98% reported in a large sample using the same measure (Mueser et al., 1998), and with an even higher number of different experiences than in Mueser et al’s (1998) severe mental illness sample (M=7.3, compared to M=3.5). The hypothesis that dissociation would be elevated was also supported. The mean DES-II score (M=16.9) was higher than the general population’s (M=5.4), and slightly higher than the schizophrenia sample (M=15.4) that was reported in the original DES-II benchmark normative data (Carlson & Putnam, 1993). It seems that this spiritual support-seeking sample have similar psychological characteristics to a clinical psychosis population, in terms of both psychological risk factors (trauma), and psychological mediators (dissociation), and yet they currently report high levels of enrichment, pleasure, and safeness in relation their psychotic-like experiences.

The sources of support were wide and varied (Appendix 5.4). Although no systematic analysis of these sources was performed, a visual inspection shows that many of the groups accessed were focused around specific experiential phenomena (e.g. kundalini experience, spiritual crisis/emergence/healing, energy, lucid dreaming). It may be that the specificity of these groups was important for people, for example, enabling them to connect with others with similar (specific) experiences. This is not something that has been provided by mental health services, which typically categorises people by a diagnosis, rather than by experiences. One exception is hearing voices groups, which are becoming increasingly supported within mainstream services. This may highlight one of the advantages of the ever-increasing online presence in people’s lives. In the list of support sources, there are repeated mentions of online resources (e.g. forums, Facebook, Skype, Twitter). This accessibility to worldwide networks inevitably helps the establishment of more specific experience-tailored support; something which cannot be easily achieved within a local community mental health team. However, one of the implications of these findings is that this could be a useful goal for services going forward – to support the establishment of local groups around specific experiences: hearing voices, spiritual crisis, and more. The data imply that giving choices to people is of paramount importance;

84 this includes choices of setting (online, in person, groups) as well as choices of language to describe experience (e.g. medical, psychological, spiritual).

5.5.2. Psychotic-like experience (PLE) characteristics

As the TEQ captures a timeframe of just one week, having high levels of endorsement reflects the high incidence of these experiences in people’s everyday lives. To provide some comparison, in a general population sample (n=532), a mean total TEQ score of M=2.7 (SD=3.3) was found (Chapter 3), which is substantially lower than this sample’s mean M=10 (SD=4.6). However, as only participants with TEQ scores 1 or above were included in this study sample, there would naturally be a higher mean from selection criteria alone. To date there have been no reports of the TEQ being used in clinical psychosis samples, so no comparison is possible, however this would be an interesting area for future research. The individual item endorsements had some interesting similarities to the general population, where TEQ5 (“thoughts rushing very rapidly”, 24.7%) and TEQ8 (“things in the world … contain messages or hints”, 21.9%) are the most common, and TEQ15 (“body movements being controlled by someone of something outside of you”; 3.5%) are the least common PLEs (Chapter 3). In this support-seeking population, the most common was TEQ8 (83%) and the least common was TEQ15 (23%), which reflects a similar pattern, albeit at far higher levels in the current sample. These similarities are supportive of continuum models of psychosis, which recognise the continuous distribution of PLEs throughout the population (van Os et al., 2009). However, the results also showed an important difference in the pattern compared to the general population, in that there were more spiritual/religious-type PLEs amongst the most endorsed items. This was as expected due to the recruitment sources.

The finding that the number of PLEs was related to how ‘distressing’ they were may indicate that the sheer quantity and variety of different PLEs is overwhelming, rather than the nature or content of PLEs themselves. The finding that PLEs were experienced as ‘enriching’, ‘safe’, and ‘pleasant’ clearly differentiates this group from a clinical population with equivalent experiences. This is similar to that of clinical versus non-clinical comparison studies (Brett et al., 2007; Heriot-Maitland et al., 2012). However, because this sample was support-seeking, it might be expected that PLEs would be more ‘distressing’ and ‘disruptive’ than those of the ‘no need for care’ groups that participated in the comparison studies. Interestingly, the sample’s ratings of ‘distressing’ and ‘disruptive’ were fairly low (means 3.0 and 3.1 on a scale from 1 ‘not at all’ to 7 ‘very much so’), and significantly lower in comparison to the more positive dimensions of ‘enriching’, ‘safe’ and ‘pleasant’ (means 4.6, 4.5 and 4.2 respectively). Therefore, on the one hand, this sample is ‘in crisis’ and support-seeking, while, on

85 the other hand, their PLEs are more beneficial than problematic. This poses the question as to what is causing the crisis and support-seeking behaviour. It could be that people’s PLEs are driving the support-seeking, and that they are looking for kindred spirits to validate the positive aspects (unlike previous contact with mental health services); hence why there is a mixed picture emerging of part positive, part negative, PLE dimensions. It could be that PLE frequency (and perhaps relentlessness), rather than the PLEs themselves elicits ‘crisis’ and support-seeking. The finding that higher frequency PLEs relates to distress is consistent with findings from the UNIQUE study (Peters et al., 2016). It could also be that other non-PLE-related factors are contributing to support-seeking; for example the results show that mean anxiety and stress levels are slightly higher than population norms, at the lower end of the ‘moderate’ range, while mean depression levels are within the ‘normal’ population range (Lovibond & Lovibond, 1995). Moderately raised anxiety and stress could be a factor. Another factor could be linked to the finding that many participants had reported unhelpful and invalidating social experiences from their previous contact with mental health services. Previous research has suggested that social validation / invalidation of PLEs can be important in determining the outcomes and need for care (Heriot-Maitland et al., 2012; Jackson & Fulford, 1997), and there is evidence of invalidating social experiences in participants’ comments about helpfulness of services.

5.5.3. Social, psychological, and spiritual predictors of PLE-outcome

This study replicated previous findings that a spiritual framework and involvement may be related to a positive relationship with PLEs (Brett et al., 2014; Heriot-Maitland et al., 2012). This study has added that regardless of the quantity of PLEs, the trauma history and the dissociation levels (which are all high in this study), a spiritual framework can potentially offer a context in which PLEs can be experienced with life-enhancing qualities. The post-traumatic growth literature is potentially relevant for contextualising any future studies that build on suggestions arising from this study. For instance, it may be that a spiritual framework offers a template through which post-experience growth can occur. Although further research with larger samples is required to explore these inter-variable relationships in depth, it may be the case that spirituality (and to a certain extent religion as well) can facilitate a number of the post-traumatic growth dimensions, particularly the ‘purpose in life’ and ‘relationships’ dimensions, perhaps ‘personal growth’ as well (Joseph et al., 2012). In the results, the spirituality and post-traumatic growth variables were among those with the strongest relationships with the PLE-dimensions variables, which warrants future investigation of the directions, and mediating / moderating effects, of these relationships. One of the implications could be that interventions for psychosis should aim to help people engage with the transformative and growth

86 potential of their PLEs. Helping people to access some of the support groups already mentioned would be one way, but there is also plenty of scope for trialling new approaches within services, whereby people may be helped to explore the value and opportunities of their PLEs.

Depression and stress were also found to have strong relationships with PLE dimensions, which is consistent with the literature linking emotional problems and psychosis (Birchwood & Trower, 2006; Freeman & Garety, 2003), as well as the evidence that depression often precedes first episode psychosis (Owens & Johnstone, 2006). An implication is that these emotional symptoms should be investigated as possible mechanisms in problematic PLE outcomes, and directly targeted in psychosis interventions. The relationship between social safeness and PLE-dimension ratings is also worthy of further investigation. In this study, not only was social safeness significantly related to ‘enriching’ PLEs, but also the socially validating aspects of spiritual groups, forums, networks etc was reflected in many of the participant comments, which indicated that the helpfulness of these networks may operate (or partially operate) through the mechanism of social safeness. Further research is needed to distinguish between which effects are due to spiritual meanings and beliefs (more cognitive) and which are due to inter-personal validation and acceptance (more social).

5.5.4. Limitations

The sample was not large enough, and therefore not powered enough, to run regression analyses to include the significantly correlated measures. This would have allowed an investigation of which variables were the most significant predictors of PLE dimensions. The cross-sectional design also restricted the ability to look at directions of relationships. Although there is some speculation drawn from the significant relationships and their directions, it is not possible to be confident in these speculations without further testing. Running multiple correlations with a sample this small also increases the likelihood of finding false positives (Type I errors). The patterns observed in these data merely provide potential hypotheses for further investigation. Another limitation is that this sample included people with and without a psychosis-related diagnosis, which means that some of the support-seeking benefits may have resulted from previous help and support from services, even if the ratings for perceived helpfulness were low. Future research might consider recruiting with additional exclusion criteria (e.g. previous contact with services) to achieve a more confound free sample. It might also be helpful for future researchers to screen or collect data on length of time since the onset of PLEs. As the current study was exploratory, the inclusion criteria were broad because of an interest in knowing, for example, whether people did arrive at spiritual support-seeking after having accessed

87 more mainstream treatments. Future studies might try to differentiate between spiritual support- seekers with and without a history of psychosis treatment, or to compare the characteristics between the two groups. It might also be useful to assess past and current use of psychotropic medication, to see if this would have any influence on the results.

5.5.5. Conclusions

This study has successfully recruited from a novel population who are seeking support for PLEs, outside of the traditional support systems of mainstream services. The characteristics of this population were similar in many ways to that of a psychiatric psychosis population, in terms of trauma history, dissociation, frequent PLEs, and a need for care. However, they also have important differences, in particular, in how they relate to their PLEs. Psychotic-like phenomena, for this sample, are experienced as more enriching, safe, and pleasant than would be expected with their clinical counterparts. This study offers a potentially useful direction for future research, not only in terms of identifying this population and recruitment strategy, but also by identifying some potential variables, effects, and mechanisms that are worthy of future investigation, namely spirituality, post-traumatic growth, social safeness, and depression.

88

Chapter 6. Introduction to Section B: Compassion Focused Therapy (CFT) for psychosis

6.1. Chapter information and overview

This chapter is co-authored with Dr Angela Kennedy (Tees, Esk and Wear Valleys NHS Foundation Trust) and is published in an edited book: Heriot-Maitland, C. & Kennedy, A. (2019). Attachment themes in compassion-focused therapy (CFT) for psychosis. In K. Berry, S. Bucci, & A. Danquah (Eds.) Attachment and Psychosis. (ISPS Book Series). Routledge.

Both authors contributed original written material to this chapter.

This chapter outlines a rationale for Compassion-Focused Therapy (CFT) for psychosis. The authors discuss the role of the attachment system in helping us to process threat and shame through its impact on emotional regulation and integrative mental processes. They then describe the elements of a CFT intervention, particularly bringing out the importance of attachment themes, and focusing on the therapeutic process of developing compassion for dissociated parts of self.

89

6.2. Background to CFT

CFT was originally developed for people with high levels of shame and self-criticism (Gilbert, 2010b; Gilbert & Irons, 2005). Such people can find it very difficult to be self-supporting or compassionate, often because they do not have experience of compassionate caregiving from which to draw and learn. CFT helps people develop the capacity for close, affiliative, relationships and to become orientated towards self-reassurance and self-compassion. The theory underpinning CFT suggests that there are three major emotion regulation systems, each with distinct evolutionary functions: threat- protection, drive-excitement, and soothing-contentment. The threat system evolved to detect and respond to threats in the world, and the other two are ‘positive’ affect regulation systems, with one focused on achieving and doing and the other focused on contentment and social soothing. This latter system is believed to play an important role in regulating the threat system. In early attachment relationships, affiliative experiences of parental (or other caregiver) calming will stimulate, develop, and integrate the soothing system as a natural regulator of threat. Aversive or inadequate early experiences with caregivers may disrupt this process, making it difficult in adulthood to access any attachment-based resources in the self as a source of integration and regulation.

6.3. CFT and psychosis: trauma, attachment and dissociation

In psychosis, people can experience both their external and internal worlds as sources of threat. In terms of the external world, it is well-established that experiences of threatening or shaming interpersonal traumas, such as physical, sexual and emotional abuse, can play a causal role in the development of psychosis (Arseneault et al., 2011; Shevlin, Dorahy, & Adamson, 2007). Also, criticism and hostility from others is found to be a key predictor of relapse in psychosis (Butzlaff & Hooley, 1998). Unfortunately, traumatic experiences can often become intertwined with attachment relationships (referred to as attachment trauma), where caregivers themselves are either the sources of threat and shaming, or unable to provide protection or nurturing in the context of threat from others. In terms of the internal world, hostile and malevolent voices are often threatening, critical or shaming, whilst paranoia is associated with perceived threat from others. Experiences of voices and paranoia are often compounded by self-criticism and associated feelings of shame.

The empirical and theoretical overlap between trauma, attachment and psychosis has, in recent times, been informed by the dissociation literature. Dell and O'Neil (2009) define dissociation as a partial or complete disruption of the normal integration of a person’s psychological functioning. Nijenhuis and

90 van der Hart (2011) use the term ‘dissociation’ to refer to divisions in the experience of self, with relationships between aspects of self dominated by phobic avoidance of each other and what each represents. Such structural accounts of dissociation essentially describe the emergence of dissociated aspects of self when the mind’s integrative capacities are overwhelmed by attachment trauma, due to the degree of uncontained and dysregulated affect. Hence, a key mechanism underpinning this theory is the suggested breakdown of integrative processes as a result of trauma-related over-arousal of the evolutionary threat system. With reduced integrative capacity, these confusing and contradictory personal experiences can lead to splits in internal representations of self and others.

In structural dissociation terms, voices can be understood as intrusions from split-off components of self (Moskowitz & Corstens, 2008), and, in particular, may often represent some disowned part of self or experience related to a trauma within an attachment relationship (Longden, 2010). Support for this theory comes from findings that voices often share characteristics with a person’s abusers (Read, Agar, Argyle, & Aderhold, 2003), and that the kind of internal relationship a person has with their voices often reflects subordination in external attachments with dominating or controlling others (Birchwood et al., 2000).

The literature on psychosis and dissociation has focused predominantly on voice-hearing. However, dissociative processes may be equally important in understanding delusions and paranoia. Delusions or paranoia may often be maintained in spite of evidence to the contrary because they can either represent fears that are dissociated from their origins or meaning or attempts to make sense of unusual experiences (Maher, 1974). Paranoia can be understood as a functional response to chronic threat or interpersonal adversity. For example, when Campbell and Morrison (2007) used Interpretative Phenomenology Analysis (IPA) (Smith, 1996), a qualitative research method which seeks to explore and understand each person’s lived experience of phenomena, to analyse 12 interviews about experiences of paranoia, all participants linked their anxiety with negative life experiences. In addition to experiences of persecution or danger, shaming and humiliating experiences are also found to be highly associated with paranoia (Matos, Pinto-Gouveia, & Costa, 2013). Shame experiences, which can similarly be understood in terms of threat, i.e. the threat of social rejection and devaluation, can themselves be recorded in the brain as traumatic and can impact on later sense of trust and safety with others. The difference between paranoia and social anxiety is hypothesised to relate to the assumptions the person makes about the intention of the person threatening, shaming or hurting them. If the victim believes that the intention of the other is malevolent, then paranoia can result. In

91 social anxiety, however, the focus is on whether the self is likeable enough to gain social acceptance and not be rejected (Matos, Pinto-Gouveia, & Costa, 2013).

In psychosis, shame-based threat from internal and external sources can maintain dissociative processes, through the phobic avoidance of aspects of self that shame sets up. This in turn prevents rebuilding of integrative capacities. Recent evidence for the traumatic-like qualities of shame comes from Dorahy et al. (2017), who showed that both external and internal shame can activate dissociative processes. It therefore follows that shame, like other threat experiences in psychosis, will orientate, through evolutionary processes, an individual’s information processing and behaviour towards the motive of self-protection, activating automatic safety responses such as avoidance and dissociation.

In CFT terms, Gumley et al. (2010) suggest that threat system sensitisation in psychosis may be related to an underdeveloped soothing system. Soothing is the capacity to self-regulate threat-based emotions such as fear or shame and is learnt through healthy attachment experiences. In the same way that a threat system will have become vigilant and sensitised through experiences of threatening environments, the development of a soothing system will have been determined by experiences of nurturing and affection. Evidence for self-soothing capacity as a protective factor in psychosis comes from Connor and Birchwood (2013), who studied both self-critical thoughts (a form of internal threat) and self-reassuring capacities (linked to internal soothing) in a sample of voice-hearers. As well as finding that self-critical thoughts of self-hatred and inadequacy were related to appraisals of voice power, which they suggest reflects an underlying perception of low social rank in relation to others, their findings also suggest that capacity to self-reassure, which was linked to less shaming voice content, may be a protective factor for voice-hearers. These findings complement evidence for self- compassion being protective against psychotic symptoms (Eicher, Davis, & Lysaker, 2013); however, since both studies are based on correlational designs, neither can directly imply causation. Nonetheless, as we have seen, there is still a strong theoretical basis for employing interventions that aim to build up capacity in the soothing system. The building of internal resources is a key part of the stabilisation phase of treatment for dissociation (Boon, Steele, & van der Hart, 2011).

6.4. CFT for psychosis: key elements

CFT was developed for helping individuals with high levels of shame and internal self-criticism, and one of its core interventions is to help people move out of mentalities or mindsets that focus attention and cognition on the (potentially harmful) power of others (which in CFT, are referred to as ‘social

92 ranking’ mentalities), towards activating cooperative, caring and affiliative processes to self and others. The evolution of attachment in mammals is crucial in understanding our capacity for compassion as attachment and compassion are each linked to evolved motives for care-giving, and each are mutually supportive; just as affiliative and secure attachment experiences will support the types of mental states that are conducive to compassion, compassionate intentions and actions will also foster experiences of affiliation and attachment. CFT aims to engage this broader caring motivational system at each of the biological, psychological and social levels. Generally speaking, a CFT intervention aims to provide contexts, practices and insights that facilitate the development of compassion in self-other, other-self, and self-self relationships as has been detailed elsewhere (Gilbert, 2009a, 2013, 2014). This section will outline the key elements of a CFT intervention, as applied to working with psychosis. Three of the five main elements (establish safeness and connection; psycho- education; cultivate / deepen the compassionate self) will be described generically, but the remaining elements (formulation; direct compassion to others, self, dissociated parts/voices) are more individually tailored and will therefore be illustrated by a clinical case, Stuart. Stuart is a fictional character but is based on a number of different people that the authors have worked with.

6.4.1. Establish safeness and connection

The establishment of safeness involves engaging the brain’s soothing system as a natural threat regulator, as well as creating environments and relationships that will support experiences of safeness and connection. CFT attends to issues of social safeness first, with its associated physiology, because of the resilience this facilitates towards engaging more directly with the frightening aspects of psychotic experiences. Social safeness also creates the conditions from which compassionate intentions and mentalities arise, harnessing the person’s commitment to engaging with their own experience. It creates a context for living that aims to develop a self that can care for itself and keep itself safe, engage with some meaningful activity and social networks, set small goals and learn new regulation skills. The focus is on developing a mindful curiosity about the experience of psychosis, strengths and survival mechanisms. The aim is to decrease the disintegration experienced as a result of stress-related arousal (Siegel, 2010).

An important way of increasing the bodily experience of safeness is through the practice of soothing rhythm breathing, which seeks to establish a gentle pace to the breath and so activate the parasympathetic system. This can be accompanied by learning to recognise what postures and activities ground and centre the person. Imagery work in CFT also facilities experiences of safeness

93 and connection with others by using images of safe places and compassionate beings/objects. This harnesses both memories of positive external attachment experiences and also the potential ideal attachment to oneself. Research has shown that such imagery can help develop soothing capacities (Rockliff, Gilbert, McEwan, Lightman, & Glover, 2008). Scripts for compassionate image work are documented in the CFT literature (Gilbert, 2013). Such images create a safe haven and secure base for exploration and the resilience to turn towards one’s psychotic experiences to consider how to cope differently or address the underlying traumas. Through these exercises, the person with psychosis can generate courage, well-being and capacity to mentalise and be mindful; skills that are needed to reduce dissociative avoidance of aspects of self and experience.

6.4.2. Psycho-education about evolved brains, emotion systems, and multiple selves

Psycho-education has a key role in CFT in that it establishes a de-shaming, evolutionary understanding about the human brain. Clients are educated about the brain’s evolution alongside the development of social groups and about attachment as an improved reproductive strategy. They are guided through understanding the problems and benefits naturally arising from a brain’s evolutionary design, with the aim of recognising that their problematic thoughts, emotions and behavioural reactions are not their fault, but rather understandable and biologically typical reactions to difficult experiences. Psycho- education in CFT also helps to build a sense of common humanity; e.g. an understanding what we all share as human beings, and what we’re all up against in the flow of life. This can be particularly important for people with psychosis, who often experience stigma and shame in relation to their experiences, which can leave them feeling lonely and alienated from others.

A helpful way to normalise experiential divisions within the self as functional is through psycho- education on the concept of multiple selves (Gilbert, 2014). What we call the ‘self’ is really our personal experience of continuity, whilst consisting of various different mindsets, mentalities, or emotional states. For instance, we have an ‘angry self’, and ‘anxious self, and a ‘sad self’, each of which shape our attention / thinking in very different ways, and set up different responses. Pathological dissociation is where a division occurs between selves that would otherwise be integrated were it not for the overwhelming stress or trauma associated with this. Education about the functionality of such divisions, the unique function of each dissociated state, and also of voice- hearing as a common human experience, can be shared through both personal accounts (Romme, Escher, Dillon, Corstens, & Morris, 2009) and studies of anomalous experiences in non-clinical populations (Heriot-Maitland et al., 2012). Dissociative experiences have been described as a ‘logical

94 way of being’, given the extreme nature of abuse occurring within attachments (First Person Plural (2011)). Dissociation protects the person from the full force of awareness or emotion in order to enable the person to function in day to day life. However, when people get bombarded by the dissociated parts, it can be confusing and frightening.

6.4.3. Formulation not diagnosis

A formulation aims to help people understand how early interpersonal and attachment experiences have contributed, alongside other experiences, to the development of both external and internal safety strategies to manage key threats and fears. Again, the de-shaming narrative is crucial here, meaning that clients will be helped to respect the function of these strategies as completely understandable coping responses. To illustrate, consider the case of Stuart, who came to therapy with difficulties relating to critical, threatening voices. Stuart reported a history of bullying at school and experiences of feeling different from other children. He also described unresponsive and invalidating parents, and a distinct lack of places to turn to for comfort, security and nurturing. Stuart and his therapist decided not to try and formulate the presence of external voices to begin with. Instead, they identified an inner critic, which seemed to relate to Stuart in a similar way as the voices. This critic targeted his insecurities, and directed similar emotions towards him. Stuart’s inner critic, or ‘self- critic’, was formulated as an understandable internal strategy aimed at self-protection. The self-critic was linked to Stuart’s fear of social rejection, protecting him from humiliating himself in front of others by self-correction, keeping his attention focused on his own weaknesses, and ensuring that he avoids social situations. Later in the therapy, Stuart assigned a similar role to his critical voices. The dissociative voice-hearing experience was understood in the context of disorganised early attachment relationships, and an underdeveloped soothing system which prevented him being able to deal with his anxieties in ways other than phobic divisions and submission to his inner critic. Figure 6.1 includes illustrative examples of Stuart’s formulation for both the (a) development and (b) maintenance of his difficulties.

95

Figure 6.1a. CFT formulation (development) – combination of ↑ threat and ↓ safeness

Figure 6.1b. CFT formulation (maintenance)

96

6.4.4. Cultivate / deepen the compassionate self

In CFT, the client is helped to cultivate a compassionate self, which provides an inner ‘secure base’ from which to explore their fears and dissociated emotional parts. While continuing to build up the client’s capacity for safeness, soothing and affiliative emotion, the focus here is on cultivating a part of the client that has the required qualities and attributes to slowly engage with dissociated aspects of self and intrusions. Key compassionate capacities, such as wisdom, strength and caring- commitment, are built around this sense of identity, using behavioural practices, and practices with posture, voice tone, facial expression, and imagery. This often begins with the client imagining themselves in a compassionate role. The image can be oneself at one’s best (perhaps a memory of being kind to another), another archetypal being (such as a character from literature) or a perceptual representation (maybe a colour). The compassionate self can then be used to understand and moderate the bodily responses of threatened states of mind. It is a place of safeness and strength within the self to both come from, and go to. In the language of attachment, an inner secure base and safe haven. This parallels the work of a lot of therapy for dissociative disorders, where the ‘functional adult’ self is encouraged to liaise with and moderate the responses of ‘emotional’ or ‘child’ parts (Van der Hart, Nijenhuis, & Steele, 2006). There is emerging evidence of the value of such deliberate image work in creating good internal attachments (e.g. between the self and an imagined compassionate self or other) and the regulation associated with this (Lincoln, Hohenhaus, & Hartmann, 2013).

For many people with psychosis, particularly those who have experienced trauma, this process of cultivating compassionate qualities may be met with resistance. This means that in CFT, a major aspect of the therapy becomes identifying and navigating these blocks to compassion. For instance, safeness and connection are sometimes too frightening for traumatised people, because such compassion is contrary to the perceived value of threat protection strategies. As dissociation can be understood as a phobia of aspects of experience (Van der Hart et al., 2006), gently addressing this avoidance becomes the main route to recovery. For example, witnessing domestic violence could lead to a fear of experiencing anger, but could also lead to internalising the representation of an abusive ‘other’. Both the client’s own disowned rage and the imagined rage of another could emerge as an abusive voice that is not experienced as part of the self. CFT can initially help the client to focus on small changes in bodily state with the intention to be aware of such changes as dissociative triggers and desensitise the client to them. Boon et al. (2011) describe other useful ways of addressing such fear that is applicable to people who hear voices. The person may need to have a variety of safe places that meet the needs of different dissociated elements. Some angry elements may be able to tolerate safeness only once

97 they realise that their style of protection is ultimately counterproductive now that the person is no longer being abused. All this may require negotiation and experimentation. For some clients, a dissociative framework can be useful, and formulating their symptoms as linked to traumatic events and adaptive responses can more closely mirror the client’s own experience of themselves.

6.4.5. Direct compassion to others, self, dissociated parts/voices

Once the person can access a compassionate self, with compassionate qualities of wisdom, strength and caring-commitment, they can begin to use this in a deliberate way to address their issues and goals, asking ‘what do I need this aspect of myself to do in order to …?’ This uses the compassionate self both as a safe, secure base and also as a wise, internal attachment figure that can advise and regulate. The benefit of enhancing the compassionate self-qualities, and that of the whole system of mind, is that it facilitates a harmonious but firm acceptance of unwanted aspects of self. It can resource the person with the qualities needed to manage internal conflicts, and paves the way for supportive internal dialogue between parts. Identifying where to direct the compassionate self will be a collaborative process with the client. Some examples of where the person with psychosis may decide to direct their compassionate self are:

• Compassionate dialogue between voices that would aim to harmonise the relationships a person has between themselves and their voices; • Compassion to multiple experiences of themselves that may react to voices in different ways and set up internal conflicts, e.g. avoidant self, submissive self, self-critic (whether dissociated from their own sense of self, or simply different self-states); • Compassion to the part that fears he or she will be rejected, neglected, abandoned, abused, attacked, and developing wisdom and empathy about causation of such attachment based fears and how they link to adapting to previous relationships (e.g. inner critic, focused threat attention); • Compassion to the dissociated part that still lives exclusively in ‘trauma time’ and will intrude its terror, horror, images and pain into consciousness in ways that may or may not be contextualised (PTSD, somatisation, paranoia); • Compassion to other emotions (e.g. anger, sadness) that may be blocked because they have been learned or conditioned to be associated with threat, so that they can be accessed and expressed in a safe way; • Compassion to other people in order to facilitate a more rewarding social world.

98

In Stuart’s case, it did not feel safe, initially, to dialogue with his voices. What felt safer was to create internal characters representing the two main emotional parts of him that he wished to work with. Billy and David were the characters’ names; essentially parts of Stuart that organised the way he thinks, feels, and acts. Billy was Stuart’s compassionate self – wise, kind, strong, and committed to caring for self and others. David was Stuart’s self-critic – suspicious, untrusting, critical, and always erring on ‘better safe than sorry’. Stuart engaged with dialogues between David and Billy. He was able to develop empathy and compassion for David; for example, understanding where this part of him originated. He recognised that David had probably developed earlier in his life at a time when he was scared about what other people thought and how other people acted towards him. He recognised that David might have originally been his mind’s way of trying to protect him from the things he was afraid of. He thought the reason why David, the self-critic, had always been so suspicious of people is that he’d always been trying to protect him from what he feared would be the worst outcome. David remembered that people were not very accepting of Stuart at school, made him feel different, and that sometimes he was treated badly at home. Stuart could empathise with why David kept warning him to be suspicious of people. Stuart discovered that David was very quick to criticise him because he wanted to try and stop him from doing things in his life; from taking risks. Again, this is protective. The process of identifying the function of voices is done by the client; the therapist simply facilitates this process using guided discovery.

In CFT, the internal relationships between the compassionate self and other parts or voices can be facilitated using a variety of techniques common to other therapeutic approaches, such as imagery, chair work (using different chairs to represent different self-states or voices), and letter writing (composing a kind and wise letter to self that is based on the understanding of how the voices developed). In the case of voice dialoguing, a CFT therapist would use similar techniques to those described in chapter 13 in this book and also Kennedy and Dillon (submitted), but with the addition of a compassionate self as a moderator of the dialogue. For instance, in the compassionate approach to voice dialogue using chairs, one example would be to designate one chair for a critical voice, one chair for the part of self that receives this criticism, and then a third chair for the compassionate self. The compassionate self can then draw on qualities of wisdom, authority and warmth to engage with the emotions and functions of the parts in the other chairs. The compassionate self can either address voices and parts directly from their compassionate stance, or take more of a reflective overview of the conflicts between different parts, moving towards resolution and integration. Importantly, whenever the client takes the chair of compassionate self, it is helpful to stop and create space to

99 bring the soothing system online. This may be through, for example, soothing breathing, posture, imagery of compassionate qualities, as outlined previously.

Making peace with voices

Based on fifty narratives of recovery, Romme et al. (2009) suggest a number of factors relating to internal relationships with dissociated elements that are critical to well-being: having a framework to normalise the experience as a response to life events, being curious about the voices, making one’s own choices, changing the power relationship to voices, and recognising which emotions are difficult to learn better ways of dealing with them. A mindful, compassionate self can greatly facilitate these relational processes. As time went on for Stuart, with further cultivation of his compassionate self, and further compassionate relating with his self-critic, there was an important shift in the relationship between Stuart and his voices. The main shift was the voices becoming less critical and threatening and more supportive and caring. Towards the end of therapy, Stuart decided that he now felt safe, confident, and ready to start a dialogue with the voices themselves.

One month after this therapy (of 25 weekly sessions) had ended, Stuart reflected that it had been an incredibly helpful experience. He felt that his mood and self-confidence had improved, which had allowed him to start paid work, which he was enjoying. Stuart reported that he still heard two voices, and that the content of these voices was half negative and half positive. The negative voices were critical, while the positive voices tried to boost his self-esteem by saying things like ‘You’re in control’ and ‘You’re eating well’. Stuart acknowledged a link between his relationship with himself and his relationship with the voices, especially around self-care; for example, he noticed that if he didn’t look after himself (in terms of going to the gym or eating healthily), the voices would be more critical. Stuart said that he found the negative voices slightly distressing but he had learnt to manage them. He said that he didn’t feel the voices disrupted his life anymore and his concentration and focus had improved. Stuart felt that with the help of the techniques he had learnt in therapy he was more at peace with the voices. He said that now, when he felt stressed, he would use breathing exercises and this would help him feel more calm and peaceful. Stuart reported to have not experienced any distressing beliefs at all in the month following therapy.

100

6.5. Previous studies of CFT with people with experiences of psychosis

In a small case series of compassionate mind training involving three people who heard malevolent voices, Mayhew and Gilbert (2008) helped participants to visualise the compassionate part of themselves, with associated feelings of warmth, and to develop empathic understanding for their distress and self-criticism. Interestingly, they found that working with self-critical thoughts (as opposed to with critical voices directly) led to a reduction in the malevolence of voices. There is also evidence that group-based CFT is potentially beneficial for people recovering from psychosis, Braehler et al. (2013) studied 22 participants in CFT groups, which involved applying compassionate practice and skills in relation to internal and external threats, such as shame, stigma, paranoia, self-attacking, and hostile voices. Their main finding was that the CFT group participants reported significant reductions in depression associated with psychosis, compared to treatment as usual. Braehler et al. (2013) also analysed the recovery narratives of CFT participants and showed that more than half felt the therapy group had helped them in their recovery journey, whilst very few of the control group had made much progress. Compassion was increased in participants and this was associated with less fear of relapse, less depression and fewer negatives beliefs about their psychosis. To complement these preliminary studies, a series of first person account papers has recently been published by an individual with psychosis receiving CFT, who has been in services for 20 years (Ellerby, 2014b; Kennedy & Ellerby, 2016). This includes an account of engaging with voices from the compassion perspective (Ellerby, 2014a).

6.6. Summary

This chapter has highlighted how attachment traumas can lead to problems in regulating responses to threat, which cultivates the conditions for dissociation. It has outlined the rationale for CFT for psychosis by highlighting how delusions and hallucinations can be understood in terms of dissociation, and also by highlighting the role of attachment in both CFT formulations and interventions. In addition to attachment theory, CFT also draws on evolutionary psychology and neuroscience. CFT aims to help people regulate threat processing by building internal feelings of safeness and affiliation, and by providing contexts, practices and insights that facilitate the development of compassion to self, others, and dissociated parts. The focus is on helping people feel safe in relation to their experiences and their social worlds. The outline of a (non-linear) five stage approach to applying CFT to psychosis is provided. Although there is currently very little in the way of published evaluations of this relatively new approach to helping people with psychosis, the strong theoretical rationale outlined in this

101 chapter has recently attracted funding to conduct a programme of research, led by the first author. This research may not only inform the design of future therapies for people with psychosis, but may also inform service-level approaches to reducing stigma and shame by, for example, developing a culture of acceptance and empathic connection, as opposed to the still sadly not uncommon culture aimed at increasing insight into an ‘illness’ narrative about delusions and hallucinations as being the meaningless manifestations of an abnormal brain. In our view, the more that therapies and systems can promote external and internal experiences of social safeness, the greater chance of dissociative integration and recovery from psychosis.

102

Chapter 7. Developing the CFT for Psychosis intervention

7.1. Abstract

7.1.1. Background Intervention development studies are under-reported in the literature but to improve scientific transparency and rigour, it is important to report the decision-making that takes place between articulating a theory of intervention to starting to test the intervention. To improve the acceptability of our interventions, it is increasingly acknowledged that user-involvement is crucial in these development stages, and hence why it is vital to report the knowledge exchange that occurs in informing the intervention development. The aim of this study was to develop a CFTp manual, accompanied by a measure for rating therapist adherence to this manual.

7.1.2. Methods There were two phases of intervention development: Phase 1, prior to the evaluation period (pre-Dec 2014) and Phase 2, during the evaluation period (2015 – Dec 2018). Phase 1 involved a working group with an expert by experience and two experts by profession. Phase 2 involved engagement and feedback activities with four different sources: i) NHS participants receiving the therapy; ii) a co- produced film about the therapy; iii) co-produced events/workshops about the therapy; iv) a co- produced peer-supervision group. Feedback from each engagement source was used to inform the final CFTp manual.

7.1.3. Results Intervention developments resulted from these knowledge exchange activities, including re-ordering sections of the CFTp manual, adding a new level focused on “establishing safeness” first, and adding new items for rating therapist competence. Overall, the CFTp manual increased from 30 pages at the end of Phase 1, to 50 pages at the end of Phase 2. An analysis of the overlaps in feedback from different sources showed that a number of changes were suggested by more than one source, and two particular changes were suggested by all four sources: the suggestions to i) add the film as a therapy tool; and ii) add specific guidance on compassion for voices.

7.1.4. Conclusions The intervention development procedures involved an enriching knowledge exchange collaboration between science, lived experience, and practice. These collaborations led to substantial intervention developments, aimed to improve the acceptability and utility of CFTp.

103

7.1. Introduction

7.1.1. Intervention development studies

The development and evaluation of complex interventions requires a number of key steps and decision-making processes, and best practice is to design and report these steps systematically (P. Craig et al., 2008). Once an idea for a novel intervention has been conceived, with its basis in a theoretical understanding of change processes, there are decisions that must be taken before this intervention is then ready for feasibility and pilot testing. Historically, the decisions that occur in this phase, i.e. during the production of an intervention manual, before feasibility testing begins, have been missing or under-reported in the literature (Hoddinott, 2015). However, without specificity and transparency around this phase of the research, it is harder for future researchers to identify how an intervention might be further developed and improved for future trials and adaptations. It is important to specify not only what decisions were made in this phase, but also why, how, when, and by whom. The ‘by whom’ is particularly important with the increasingly recognised role of knowledge exchange in research, and in particular, the role of people with lived experience as collaborators and knowledge exchange partners (Thornicroft & Tansella, 2005).

7.1.2. User-involved development

For a psychosocial intervention to be implemented successfully in practice, it has to be acceptable to its users. This includes the recipient (end) users of the intervention, for whom the intervention is intended to help, as well as the clinician users (who deliver the intervention). Trivedi and Wykes (2002) highlight a number of ways in which service users can contribute to the development and testing of interventions, arguing that service-user partnerships can positively benefit the development process; for example, in terms of designing more ‘user-friendly’ interventions that are more relevant (and ready) for clinical practice. Although the importance of service user involvement in mental health research has been consistently recognised and re-iterated (Thornicroft & Tansella, 2005), there are still many challenges in creating genuine collaborative user partnerships and in avoiding user- involvement tokenism (Sangill, Buus, Hybholt, & Berring, 2019). Recent guidance for user-centred design of psychosocial interventions recommend that target users (i.e. representative practitioners and patients) are identified and engaged early on in the process, and that intervention developers convene with users into “collaborative redesign teams that tailor interventions to new contexts while retaining the core components responsible for their effectiveness” (pE2, Lyon and Bruns (2019)).

104

7.1.3. Expertise by experience and user-led developments in psychosis

In recent years, the psychosis field has seen the successful growth and global expansion of a service- user-led initiative for helping and supporting people with their experiences; the Hearing Voices Network (HVN). The HVN, established in 1988, promotes an approach that creates validating and supportive spaces for people with psychosis, whereby members of hearing voices groups are valued for the ‘expertise by experience’ that they each bring (Longden et al., 2018). The HVN’s success has highlighted a user-centred process that goes beyond collaborative working to improve the acceptability / usability of an intervention. The hearing voices movement philosophy explicitly aims to create user ownership, empowerment, and affiliation, and this itself may be considered as a mechanism of change. This process is particularly relevant from a Compassion Focused Therapy (CFT) perspective, which, as outlined in the previous chapter (Chapter 6), is concerned with understanding and cultivating social safeness and de-shaming mechanisms. User-involvement has potential for creating the change mechanisms purported (in the CFT model) to improve outcomes for people with psychosis. This leads to consideration of multiple roles for people with lived experience (or ‘experts by experience’) in a CFT intervention; from design, through testing and engagement, to dissemination, training, and delivery. This can significantly contribute to key CFT mechanisms of change, such as building common humanity and affiliative experiences, as well as reducing the stigmatising, shaming, and social ranking experiences associated with psychosis. In many ways, it may be easier (and more powerful) for these processes to be fostered by experts by experience than by professionals or researchers. In terms of actually delivering a CFT intervention, there are potentially a number of ways in which experts by experience could be involved, both directly (e.g. peer support workers/ groups that directly practice the CFT and CMT exercises together) and indirectly (e.g. co-producing videos, written materials, and other resources that are used ‘in session’ by CFT therapists).

7.1.4. Overall aims of intervention development

The aim was to develop an individual Compassion Focused Therapy for Psychosis (CFTp) intervention that would be acceptable to people with psychosis and would be deliverable and testable in routine clinical services. The development process aimed to create a genuine knowledge exchange collaboration between researchers, practitioners, and service-users, seeking to bring together the knowledge of science and of lived experience. An important aim, running through the development process, the manual content, and through all communications and engagements around the intervention, was to cultivate a sense of the validation, ownership and empowerment of lived

105 experience. By involving experts by experience in all development and engagement activities, the aim was to set the same affiliative and de-stigmatising tone around the intervention that would ultimately be targeted within the intervention itself.

The intervention development process was split into two phases: 1) Developing the initial CFTp manual prior to the start of evaluation (pre- Dec 2014) 2) Further developing the CFTp manual during the evaluation period (2015 - Dec 2018)

In both phases, there was knowledge exchange collaboration with both experts by profession (CFT researchers and therapists) and experts by experience (people with lived experience of psychosis). The specific aims, methods, and results of each phase (1 and 2) are now outlined in turn.

7.2. PHASE 1 - Developing the initial CFTp manual (pre- Dec 2014)

7.2.1. Phase 1 aims

The aim of Phase 1 was to develop an intervention manual that could be used to commence the evaluation of CFTp. At the start, an initial draft version of the manual was necessary to get the evaluation process underway, in order to guide the design and research ethics assessment of a case series study, the intervention for the first cases recruited, and therapist adherence and assessment of fidelity.

7.2.2. Phase 1 methods

Knowledge exchange partners and information sources

Expert by experience Dr Eleanor Longden (EL) is a research psychologist and voice-hearer. She has expertise by experience of hearing voices, trauma, receiving psychiatric diagnoses and services (both inpatient and community), and of receiving interventions (both medical and psychological) for distressing experiences in psychosis.

106

Experts by profession Prof Paul Gilbert (PG) is the founder of Compassion Focused Therapy (CFT). He is the leading international expert in the science, application, and training of CFT, and he was the main supervisor and advisor for the CFT components of this study. Dr Chris Irons (CI) is a CFT therapist, trainer, and supervisor. He was the secondary supervisor and advisor on the CFT components of this project.

Other information sources Attendance by CHM at CFT training events, including a 3-day Introductory workshop in CFT (2011), a 3-day Advanced workshop in CFT (2013), and a 4-day Mindful Compassion experiential workshop (2014). Each of these workshops had slides and handbooks, with supplementary material and reading. The key published sources of information used were: Braehler et al. (2013) (CFT for psychosis groups); Gumley et al. (2010) (CFT for psychosis theory); Mayhew and Gilbert (2008) (Compassionate Mind Training for voice-hearers); Gilbert (2010a) (CFT distinctive features).

Procedures

Therapy development working group A working group was established by the lead researcher (CHM) with the therapy development collaborators (PG, EL, and CI), which involved a series of monthly meetings, some face-to-face (Derby) and some online (Skype). These meetings were supported by email exchanges and by a shared online folder, which was used for writing and commenting on drafts of the CFTp manual.

The structure of the manual was proposed by CHM, based on key CFT elements identified from the workshops and published materials (outlined above). Some illustrative examples of therapist guidance notes were then provided by PG to help set the tone of manual. An example of this is in the section on definitions of compassion (CFTp manual, section S.3), where a warm, listening, and validating tone is being encouraged in the manual to guide a client’s discovery of the meanings and feelings around the word compassion.

“Is it okay if we begin to explore the therapy a little bit? So this research is looking at therapy called Compassion Focused Therapy. Have you heard about it? Do you have any thoughts about it?” “So this word compassion can sometimes be a bit odd. How does the word sound to you? What does it mean to you?” (client’s descriptions) “When you think about compassion that way, do you have any concerns or worries about it?”

107

(agree with these, validate etc) “it might be useful if I share with you what we mean by compassion” (CFTp manual, section S.3)

EL provided guidance on how to create safeness in the early stages of an intervention; for example, in the sections on what does a person need to feel safe and what kind of checking in with voices do we need at the start, asking the voice’s permission, etc (section S.2).

It may also be appropriate to check in with the client’s voices (if they hear voices) to try and establish safeness with them too, e.g. asking if it’s okay to talk about all this with their voices, and asking whether the voices are here now, and if so, if they want to say anything about the process, about being here in the room, about meeting the therapist, etc. (CFTp manual, section S.2)

CHM drafted the manual and received feedback and revisions from each of the collaborators, with redrafting, until a final draft was approved for ethics submission and the start of CFTp testing.

7.2.3. Phase 1 results

As a result of these Phase 1 intervention development procedures, an initial 30-page manual was produced: CFTp manual (version 1, Dec 2014). This manual was designed to be delivered in up to 26 weekly sessions, over a 6-9 month period, and was structured into six levels: Level 1. Starting therapy Level 2. Psycho-education Level 3. Assessment / formulation Level 4. Cultivate / deepen compassionate self Level 5. Compassion to others, self, dissociated parts Level 6. Ending therapy

This met the aims of Phase 1 in terms of having a manual that could be submitted to research ethics assessment and used to guide therapy (and therapist fidelity) for the first case series participants. This initial manual (version 1) was comprised mainly of generic CFT guidance, firmly grounding the intervention within Gilbert’s CFT approach, but with some additional instructions on how to extend the standard CFT techniques (such as compassionate relating to self and emotional parts) to the types

108 of difficulties experienced by people with psychosis (such as compassionate relating to, e.g. voices and paranoia). It was not until Phase 2, through more specific engagements, feedback and learning from people with lived experience of psychosis that the intervention could be developed with more specific adaptations and tailoring to this population.

7.3. PHASE 2 - Further developing the CFTp manual (2015 – 2018)

7.3.1. Phase 2 aims

The aim of Phase 2 was to produce a final version of the CFTp manual, and accompanying adherence and competence measure (CFTp-ACM), which incorporated feedback and learnings from experts by experience and experts by profession through engagement with four different activities: 1. active collaborative learning with NHS case series participants 2. engagement and feedback from a co-produced film 3. engagement and feedback from co-produced events/workshops 4. engagement with a co-produced peer-supervision group

The specific aims in terms of output, as a result of these engagement activities, were to produce a CFTp manual and CFTp-ACM that were: i) ready for use in a pilot randomised controlled trial ii) acceptable for experts by experience who would be using the intervention iii) acceptable for clinical staff and services who would be delivering the intervention

7.3.2. Phase 2 methods

Knowledge exchange partners and information sources

Experts by experience The case series study participants (Chapter 8) were recruited not just as therapy recipients, but also as active collaborators in developing the intervention. EL’s collaboration continued through Phase 2, and her partnership with CHM developed into co-producing CFT engagement events and resources for people with psychosis and their families / carers, as well as CFTp workshops for clinicians in the psychosis field. An additional collaboration was created with Rufus May (RM) and Elisabeth Svanholmer (ES), who both have lived experience of psychosis symptoms, delusions (RM) and voices

109

(ES). The collaboration with RM and ES involved co-producing engagement events and resources, as well as establishing an online international peer supervision group for people interested in using compassionate approaches to relating to voices. Finally, the wider group of experts by experience involved with these engagement activities also became knowledge exchange partners through providing feedback and ideas that had an influence on Phase 2 development.

Experts by profession The collaboration with CFT experts, PG and CI, continued in the form of monthly or two-monthly supervision throughout the Phase 2 period. The clinicians who attended CFTp workshops were given space to bring their cases for discussion, and to provide written feedback. Many of the clinicians attending workshops were psychologists and therapists, but different professional staff were also involved, including psychiatrists, nurses, and occupational therapists, which was helpful for engaging, and testing acceptability of the intervention, with the broader multi-disciplinary team. The international group of clinicians who attended the online supervision group (monthly) also brought case discussions around the challenges of applying this model with both individuals and teams.

Other information sources In Phase 2, the development of the CFTp intervention was heavily influenced by Romme and Escher’s published work on making sense of voices (Romme & Escher, 1994, 2000) as well as the Voice Dialogue approach developed by Stone and Stone (2011). What these approaches offered were some specific tools, techniques and guidelines for helping people to develop understanding and different ways of relating to different parts of them (Stone & Stone, 2011) and their voices (Romme & Escher, 1994, 2000), which is compatible with the relational work in CFT. Other published sources of information that influenced Phase 2 developments were mainly from the literature on attachment (Berry & Bucci, 2015; Berry et al., 2006; Gumley et al., 2014), dissociation (Longden, Moskowitz, Dorahy, & Perona- Garcelán, in press; Nijenhuis & van der Hart, 2011; Varese, Smeets, et al., 2012), and shame (Dorahy et al., 2017; McCarthy-Jones, 2017). The theoretical integration of these different elements within a CFT framework was written up by CHM, collaborators EL and PG, and an additional co-author Simon McCarthy-Jones (SMJ) (Appendix 7.1, also published in Heriot-Maitland et al. (2019)). In the Phase 2 period (2015-2018), a number of important developments occurred in CFT; firstly, a CFT therapist competencies scale (Horwood, Allan, Goss, & Gilbert, 2019a) was published, which was integrated into the current project for assessing competency; secondly, a substantial 10-module manual for CFT groups was drafted (Compassionate Mind Foundation, unpublised), which includes a level of detail

110 about CFT that had not previously been written. This was integrated into the current project for adding richness to the therapist guidance for the core elements of CFT.

Procedures

Active collaborative learning with NHS case series participants The case series study ran from 2015-2018, across London and Glasgow NHS, and involved seven participants. Participants were regularly asked for feedback on the intervention and were encouraged to consider what they were finding helpful / unhelpful over their 6-9 months of therapy, as well as what would be helpful / unhelpful to others receiving the intervention in future. With their permission, audio recordings of therapy sessions were made so that these could be listened to and discussed in supervision with PG and CI, to review the ongoing adaptation and tailoring of CFT ideas in this population.

Engagement and feedback from a film co-produced with EL & RM A public engagement video called ‘Compassion for Voices’ (CFV) was released on YouTube in February 2015. The video production was funded by four sources (a Medical Research Council public engagement grant, a King’s College London cultural innovation grant, a Compassionate Mind Foundation CFT development grant, and a Balanced Minds engagement grant). The short video was designed to engage people with the key elements of CFTp by following a young man, Stuart, who starts using CFTp to develop a compassionate relationship with his voices. Posted alongside the video was the link to an online feedback survey, designed to capture people’s views about the CFTp approach. The survey included specific questions about the impact of watching this film (e.g. “has it changed your understanding about voices? has it changed your attitude or approach towards people who hear voices? has it encouraged you to think about, or to do, something in a different way?”) as well as free text space to share “any additional feedback / comments / thoughts”. The aim was to use this feedback to gauge the acceptability of the CFTp approach, since CFTp has a message that would not only be new to people, but potentially controversial: the idea of developing compassion for voices, which as a concept, goes against the tide of traditional and mainstream understanding of how to relate to voices (i.e. trying to reduce or eliminate them).

Engagement and feedback from events/workshops co-produced with EL The co-produced events were developed by CHM & EL in 2015, and the first one was co-facilitated in July 2015. The community events for users and families were free to attend and organised by local

111 peer support groups. The main engagement component at these events were open Q&A sessions after a screening of the CFV film and a presentation. The workshops were paid events (funded by either individuals or their employers, e.g. NHS Trusts), although nominal rates were available for users and families to attend the mixed group workshops. Workshops were held over one or two days, and consisted of theoretical background, a lived experience account, CFTp techniques, and experiential practice. All participants were engaged with trying out techniques for themselves and giving feedback about their reactions and the challenges, pitfalls etc. In the clinician-only workshops, there was an additional component of clinical case discussion (i.e. applying the CFTp approach to participants’ own cases) as well as organised group discussion around the anticipated challenges of implementing CFTp in clinical teams. In all workshops, informal (verbal) feedback was gathered during the exercises and discussions, and in some cases, additional formal (written) feedback was collected at the end too. At one clinician workshop, written feedback was collected directly afterwards and again after 4-5 weeks had passed; i.e. once clinicians had a chance to implement the CFTp processes and techniques, and to evaluate the impact of these on their own practice. Event feedback (informal and formal) was used to inform the development of CFTp.

Engagement with a peer-supervision group co-produced with RM & ES An international (free) online monthly peer-supervision group called ‘CAVEtalk’ (Compassion Approaches to Voices and other Experiences) was established in January 2016. CAVEtalk was co- produced and co-facilitated by collaborators RM, ES, and CHM. There were regular attenders joining from different countries (including UK, Greece, Germany, USA, Denmark, and Australia), all of whom were working with people with distressing experiences in psychosis and had an interest in using compassion-focused approaches. The content of sessions was mainly discussion around clinical cases as well at the challenges of implementation in services. Engagement with the CAVEtalk group informed the development of the CFTp through the sharing of ideas and experiences of applying this approach, and compatible approaches such as Voice Dialogue (Stone & Stone, 2011), across various international settings.

7.3.3. Phase 2 results

As a result of these knowledge exchange partnerships and activities in Phase 2, a 50-page manual was produced; the revised CFTp manual (version 2, Dec 2018). This section details the developments made between versions 1 and 2, along with the engagement sources and feedback that informed these

112 changes. At the end of this section is a table summarising all the results (Table 7.1) and indicating which changes were informed by which engagement source (or multiple sources).

Active collaborative learning with NHS case series participants The collaboration with case series participants, as well as the subsequent listening to therapy session audios in supervision with PG & CI, resulted in the following intervention developments:

1. Re-order sections of CFTp manual with ‘establishing safeness’ first As a result of the engagement with case series participants, the CFTp manual sections were re- ordered, and a new therapy level called ‘establishing safeness’ was added at the beginning. In the previous version of the manual, the compassionate practices were collated together in Level 4 (steps 1 to 4), whereas in the revised version this was changed so that soothing breathing rhythm and grounding practices were moved earlier into the new ‘establishing safeness’ section at the beginning. This was in response to observation and discussion with participants indicating that some of the psycho-educational material was easier to grasp once people were feeling safer and more grounded. It was also observed that the usual therapy practice of conducting assessment first was causing anxiety for participants (in CFT language, ‘activating the threat system’). Therefore, spending time on establishing and practicing safeness first was preferable, so that participants could build soothing resources first, before accessing the threat-based memories and emotions during assessment.

2. Add CFT micro-skills items to CFTp-ACM In response to CFT supervisor feedback from audio recordings, the CFTp-Adherence and Competence Measure (CFTp-ACM) was revised to include CFT micro-skills, such as ‘pacing’ and ‘non-verbal communication’. The feedback from supervisors was that the therapist’s pacing was too quick, with the suggestion that slowing down would help the client engage with the CFT processes at a deeper, emotional level. This was tracked over the weeks of supervision, and it was felt that this should be more explicitly highlighted in the CFTp intervention fidelity monitoring procedures. As the CFT therapist competencies scale (Horwood et al., 2019a) had been newly produced, with inclusion of micro-skills, the decision was made to incorporate the whole scale within CFTp-ACM.

3. More user-friendly language (e.g. in mantras) In response to feedback from case series participants, some of the compassion language was changed to become more accessible, relevant, and user-friendly. There was feedback that some of the language, and specifically in mantras relating to compassionate qualities, was slightly out-of-touch

113 with daily life. For instance, some participants were finding it awkward to use compassion mantras about, e.g. ‘one’s heart-felt wish for others to find peace’. In the revised version, there were some simplified mantras added to avoid these kinds of difficulties; for example, the mantras: “may I be a person who is helpful to others” and “may I be helpful not harmful”.

Engagement and feedback from a film co-produced with EL & RM 140 people completed the survey, of whom 20 identified themselves as ‘a person who hears voices’, 17 as ‘a family member of a person who hears voices’ and 14 as ‘a friend of a person who hears voices’, and 89 as ‘someone who works with people who hear voices’. The engagement with these survey respondents, in particular the 20 voice-hearers, resulted in the following intervention developments:

1. Add CFV film as a therapy tool In response to feedback about the acceptability and usefulness of the CFV film (Appendix 7.2), the video itself became included in the CFTp manual as a key intervention tool. An early report posted on the King’s College London website (KCL, 2015) stated that in the first month of sharing the CFV film link, there had been 8125 YouTube views and 59 feedback survey responses, of whom 98% said that they thought this film could potentially have an impact on people’s health and welfare. By six months, there had been 18,027 views, and the video had been translated into eight languages (on request). A number of voice-hearer respondents provided feedback that alluded to the video’s helpfulness in communicating key concepts, e.g. “it really just simplifies and clarifies some basics to allow insight for both sufferers, relatives and the wider public”. Therefore, to accompany the video’s inclusion, a substantial section of the CFTp manual (section 5.1.1) was added to guide therapists in how to use the video to facilitate personal insights for clients, e.g. with suggestions of where to pause the video and what reflective questions to ask.

2. Add specific guidance on compassion for voices In response to feedback about the acceptability and usefulness of ‘compassion for voices’ as a concept (Appendix 7.2), specific guidance and illustrations about compassion for voices were added. This includes, for example, suggestions of what a ‘compassionate self’ might say to show understanding and validation to a voice (“That sounds really tough for you. I can see why you’re frustrated. That makes a lot of sense”, CFTp manual, section 5.1.1). Specific guidance was also added on how to use the technique of ‘chair work’ to facilitate compassionate relating to voices.

114

3. Add key psycho-education messages In response to feedback from voice-hearers about the importance of normalising and de-stigmatising the voice-hearing experience (Appendix 7.2), some additional psycho-education guidance was included in the manual; in particular, two key messages: i) there is nothing to ‘fix’ in the brain, and ii) there is a natural threat bias in the brain (CFTp manual section 2.4.3). There was also some information (and references) added to the manual about the continuum of out-of-the-ordinary experiences in the general population, with the aim of supporting therapists to create normalising and validating contexts for their clients.

Engagement and feedback from events/workshops co-produced with EL During the Phase 2 development period (2015-2018), there were a total of 13 co-produced events for different groups, across four different countries (UK 5, Australia 5, USA 2, Canada 1). These comprised of: i) three community events for users and families (average attendance approx. 45); ii) six workshops for clinicians only (average attendance approx. 45); and iii) four workshops for mixed groups of users, families and mental health workers (average attendance approx. 45). The estimated total number of participants for event engagement was 585. The engagement with, and feedback from, these events resulted in the following intervention developments:

1. Add ‘functional analysis of voices’ formulation As a result of the engagement with EL in communicating CFTp ideas at co-produced events, a substantial new section on ‘Functional analysis & making sense of voices / beliefs’ (section 3.4) was added to the formulation sections of the manual. Previously, the manual only included generic CFT formulation guidance (i.e. ‘3 circles’, ‘threat-focused’, and ‘social mentality’ formulations) and through the collaboration with EL in developing workshops, it became clear that more specific guidance was needed to help therapists formulate psychosis-related experiences. In generic CFT, there is a ‘functional analysis’ component, which is most commonly employed to understand the functional role of self-criticism (as a safety strategy). This was adapted and developed for voices and delusions though the collaboration between EL and CHM, particularly by drawing on EL’s expertise in Romme and Escher (2000)’s work on ‘making sense of voices’. In the CFTp workshops, EL and CHM discovered that using a ‘Wizard of Oz’ metaphor (attributed to PG) was an effective way to engage participants with the idea of ‘looking behind the curtain of our voice’ to explore its emotional function and meaning. Guidance on using this metaphor was included in the new section.

115

2. Add video demonstration footage In response to feedback from participants at the earlier workshops, some video demonstration footage was included in the later workshops. The footage shows EL and CHM role-playing a compassionate dialogue with a voice, using chair work. (The video clips were produced by the TV channel ARTE for a documentary (ARTE, 2016), with permission granted for use in CFTp training.) Subsequent feedback about the usefulness of the video demonstration resulted in this footage also being included as part of CFTp manual. The feedback from workshop participants was that the visual guidance in video format helps to build confidence in using these innovative methods in therapy.

Engagement with a peer-supervision group co-produced with RM & ES Over the development period (till end of 2018), a total of 24 CAVEtalk peer-supervision group sessions were held (90 minutes long, with average attendance of 8). The engagement with this group resulted in the following intervention developments:

1. Highlight integration with Voice Dialogue techniques A number of CAVEtalk discussions on the compatibility and integration of CFTp with other therapeutic approaches (namely Voice Dialogue, Non-Violent Communication, and Internal Family Systems (IFS)) helped to inform the revised manual guidance in section 5.1 on ‘Directing compassion to self, multiple selves, emotion parts, & voices’. Explicit recommendations were added in the manual about the application of Voice Dialogue techniques in CFTp, with reference to Stone and Stone’s (2011) original work, as well as to Corstens, Longden, and May’s (2012) adaptations for voice hearers.

2. Add a series of 15 short videos to guide different stages As a result of the CAVEtalk collaboration and engagement meetings, a series of 15 short videos were co-produced by ES, RM & CHM (filmed May 2018, published on YouTube March 2019, produced by MindWick (2019). One of the issues regularly raised by CAVEtalk participants was that although many clinicians were interested and on board with the idea of compassionate engaging with voices, they were still afraid or hesitant to actually start using the techniques with their clients. In the discussions, this was partly attributed to a fear of negative judgement by their non-therapist mental health colleagues, for whom engaging with voices may be seen as a direct contradiction to the more traditional treatments, such as antipsychotic drugs, which are used to act (quite literally) anti, in opposition to, or against psychotic experiences. Another suggestion was that there may also be some fear that engaging with voices might be too distressing for their clients or could make the voices worse. The idea for a video series came about to directly address these concerns; they were intended to

116 support people who were interested in this approach and to be a guide for people through different stages of engaging with voices. The second video, for example, entitled “Things to consider when you want to engage with voices” covers some of the concerns people might have before using this approach: “creating conditions in which it feels okay to do this kind of work” (MindWick, 2019). By making the videos freely available online, the hope would be that it would be easy to communicate and share the techniques with others, including clients, mental health colleagues, families, etc.

Table 7.1. Summary of changes made and the feedback themes (and sources) that led to changes Changes made Feedback themes Engagement sources Case series Film survey Events CAVEtalk Re-order CFTp Assessment, psycho- manual with education, and X X X ‘establishing intervention only useful in safeness’ first context of safeness Add CFT micro- Therapist pacing needs to skills items to slow to help client engage X CFTp-ACM with processes at emotional level More user- Some language in friendly language compassion practices X X (e.g. in mantras) feels out-of-touch Add CFV film as a Watching CFV film is therapy tool acceptable and helpful for X X X X voice-hearers Add specific Relating to voices guidance on compassionately is an X X X X compassion for acceptable concept to voices voice-hearers Add key psycho- Important to normalise education and de-stigmatise voice- X X messages hearing Add ‘functional Need for more specific analysis of voices’ guidance around X X X formulation formulating voices and delusions Add video Demonstration is demonstration important to help people X X footage grasp approach Highlight Other approaches (e.g. integration with Voice Dialogue) are X X X Voice Dialogue compatible and techniques complementary Add a series of 15 Many clinicians are on short videos to board with the idea, but X X guide different are still afraid to use it stages with clients

117

Summary of results

Table 7.1 provides a summary of the intervention developments in Phase 2, along with feedback themes and engagement sources that led to these changes. A number of the developments were suggested and informed by more than one source, and two of the changes where influenced by engagement with all four sources (‘add CFV film as a therapy tool’ and ‘add specific guidance on compassion for voices’). Included in the table are columns to indicate where there were multiple sources of feedback and influence on intervention developments.

7.4. Discussion

The range of knowledge exchange sources and activities produced some rich and varied feedback to inform the development of CFTp. At one end of the spectrum, there were deep and prolonged engagements from the therapy collaborators, the case series participants (26 x 1 hour sessions), and the CAVEtalk members (24 x 90 minute meetings), while at the other end of the spectrum, there were brief and anonymous survey comments from members of the general public. Some of the information gathered was consistent across all sources, particularly around the therapeutic usefulness of the CFV video and its central message. Other feedback differed between sources, and therefore informed developments in different areas; for example, feedback from the case series supervision led to changes in the therapist competence measure (CFTp-ACM), which was developed in parallel to the CFTp manual. The three major developments in Phase 2, between versions 1 and 2, were the additions of a new therapy level called ‘establishing safeness and connection’ (Level 1) and two new sections called ‘functional analysis & making sense of voices’ (section 3.4) and ‘compassionate relating in sessions’ (section 5.1.1). Other important developments in Phase 2 included the co-production of video material for use in guiding CFTp therapists, their clients, families, as well as any other people who may be involved with a client’s wellbeing during CFTp.

7.4.1. Informing acceptability

An important finding that emerged in the engagement and throughout the development process was the acceptability of the ‘compassion for voices’ message. It was not anticipated that people would be so amenable to this idea, especially because voices are often so attacking and malevolent that encouraging compassion for them is not an obvious or easy message. The feedback in the first months of sharing the CFV YouTube link (early 2015) showed an overwhelming positive response from

118 members of the general public, including voice-hearers, families, and mental health workers. This set a tone of confidence and optimism for embedding the ‘compassion for voices’ message firmly within the CFTp intervention by the time the first case series participants started (late 2015). The willingness of people to accept this concept, firstly anonymous members of the public, and then NHS case series participants, may in part be due to the effectiveness of the video itself as a communication and engagement tool. This could be linked to its engaging style and/or content, but perhaps more likely, in the case of voice-hearers, could be linked to the fact that the voice-overs were those of collaborators with lived experience of psychosis. Both RM (the voice of Stuart) and EL (the voice of the narrator) already had established profiles within the hearing voices movement and beyond. It may be that their collaborations added a certain authenticity to the message being communicated. This may also have contributed to a sense of user ownership, empowerment, and affiliation, similar to what has helped the appeal and growth of the Hearing Voices Network.

There are potential implications here in terms of highlighting the importance of user involvement and co-production in development, communication, and engagement with new interventions. This is consistent with the recommendations outlined in the introduction (Lyon & Bruns, 2019; Thornicroft & Tansella, 2005; Trivedi & Wykes, 2002). In developing CFTp, it was extremely useful to have information about the acceptability of some challenging new concepts before testing the therapy with NHS cases. It would have been even more important, of course, had the feedback been that the messages were unacceptable, in which case the intervention could have been modified accordingly in a different direction. Either way, these early process of engagement and feedback are of great value and certainly recommended for future studies of developing novel interventions.

The results also have implications around the potential role of digital technology, media, and other creative methods of engaging people with novel ideas. When the challenge is to rethink culturally engrained attitudes (in this case attitudes towards psychotic experiences as pathological, scary, undesirable etc.), then what is needed is a medium that can quickly communicate a clear, direct message, and which can be readily and widely accessed. Visuals and sounds can arouse emotion in a way that written words cannot; they can operate through a more direct sensory route, which might help to bypass the traditional framework of understanding. In addition, the results have implications for the development of video material as a therapy tool itself. CFV is essentially a story-telling film, in that it charts the therapeutic progression of a young man, Stuart, from being tormented by his voices, to developing the qualities needed to engage with them through compassionate dialogue. For people with psychosis, the therapeutic value may come through engaging with this story as a template or

119 metaphor for their own recovery journey. This potentially adds support to evidence on the therapeutic value of metaphor in voice-hearing (Demjén, Marszalek, Semino, & Varese, 2019) as well as the established and growing interest around applications of digital technology in psychosis interventions (Bradstreet, Allan, & Gumley, 2019; T. K. Craig et al., 2018; O'Hanlon et al., 2016).

7.4.2. Informing intervention development and theory of intervention

The primary aim of engagement, collaboration, and knowledge exchange activities was to develop an intervention that was acceptable, deliverable, and testable in routine clinical services. The engagements with users who receive treatments (e.g. via events and a case series) as well as with users who deliver treatments (e.g. via workshops and CAVEtalk) ensured that the intervention was developed in way that was acceptable to both. The feedback with the biggest impact on manual revisions was related to the importance of ‘establishing safeness and connection’ first, before other parts of the therapy. The feedback, generally, was that compassionate engaging with distressing experiences was a good idea, but that it was hard; and therefore, more time was needed to be spent on creating safe foundations, psychologically, biologically, and socially. The result from this feedback was the addition of a new therapy level, in a manual that contains just five therapy levels (or seven if you include starting and ending therapy). In version 1 (2014), there were just under 2 pages on establishing safeness and connection in the ‘starting therapy’ level. In version 2 (2018), there were still these 2 pages in the starting therapy level, as well as a new 3 pages for therapy level 1.

The significance of this feedback and development, however, extends beyond manual revisions; it also has key implications for the theory of the intervention itself, and in turn, also has implications for evaluation research design. The implication for theory of intervention is that safeness precedes compassion. And furthermore, safeness precedes improvements on outcomes (such as more peaceful, less distressing relationship with voices and delusions). These intervention development results therefore allow us to make predictions about the order in which things will change in CFTp. For instance, from this result, the prediction would be that improvements in compassionate relating to self, voices, and delusions would be more likely among those who have first improved on social safeness. This is a testable hypothesis in a study design with more than one assessment point in therapy; for example, mid-therapy and end-of-therapy point. There could be more assessment points; however, the challenge would be to find the balance between additional timepoints, without burdening participants or without confounding the results through multiple testing (e.g. with issues of participants remembering questions).

120

In addition to this hypothesis, a number of other hypothesis and predictions directly emerge from the intervention development results, including: i) therapist pacing competency will improve compassionate engagement measures for clients; ii) simpler language in mantras will improve measures of compassion for clients; iii) watching the CFV video will improve measures of stigma and shame for clients; iv) using co-produced therapy material will improve measures of stigma and shame for clients; and v) use of video demonstration will improve therapist competency.

Another important output from the intervention development collaborations was the theoretical synthesis complied and published by CHM, SMJ, EL, PG (Appendix 7.1) (Heriot-Maitland et al., 2019). In this theoretical paper, we highlight other testable hypotheses that emerge from the CFTp approach applied specifically to voices. In this account, we state that the aim of CFTp “is not to eliminate voices but to make them easier to live with”, predicting that CFTp would lead to a “reduction in the negative affective valence of voice-hearing”. As such, CFTp has similar overarching aims to other psychological interventions, such as CBTp. However, because our theoretical model of voice-hearing is rooted in the activation of evolved threat-based (dominant-subordinate) motivational systems, we also tentatively predict that through reducing the activation of these systems, e.g. through building regulatory resources, CFTp could potentially reduce or eliminate voices themselves (“even if this was not the therapeutic goal”) (Heriot-Maitland et al., 2019). Furthermore, we predict that by targeting the key CFT processes (i.e. reducing social rank patterns and increasing caring and affiliative patterns), the beneficial outcomes to voice-hearers would be evidenced by reducing submission, depression and dissociation. In summary, the specific predictions are that i) CFT will improve process measures of stigma, shame, and compassion, and ii) improvements in these processes will be accompanied by improvements in voice-related distress, voice frequency, depression and dissociation. From the theoretical paper, the expectation would be for voices and voice-related distress to change simultaneously with process changes. So, as social rank threat reduces, voices simultaneously improve. This, again, could be tested by a study design including ‘within-therapy’ assessment points.

7.4.3. Limitations

The knowledge exchange partners were many and varied, which produced a richness of knowledge to inform intervention development. However, there were some limitations that may have resulted in not capturing all the knowledge available, and which could be improved in future projects with similar aims. In the case series there was no systematic method of collecting feedback from participants as the primary aim was for them to be engaged with a therapeutic process, rather than in an intervention

121 development process, and these were potentially competing processes for them. The gathering of feedback had to be timed carefully and opportunistically, so as not to interfere with the therapeutic process. One recommendation for future studies would be to involve an independent person, not the therapist, to collect feedback in a more structured way. This would also help to avoid feedback bias. In the current study, this limitation was partly due to the financial constraints of conducting the project as part of a PhD. With regard to the CFV film, there were more systematic and independent procedures in place (i.e. an anonymous feedback survey); however, the questions were quite open and broad, and relating to the impact of the film. It was therefore hard to draw specific conclusions about intervention development, other than concluding that the central message is acceptable and helpful, and that the film itself could be used as a therapeutic tool. In future surveys of this kind it might be helpful to have more specific questions, e.g. ‘what do you anticipate would be the most helpful / most challenging aspects of applying this method to yourself / others?’ And perhaps asking more specifically ‘is there anything you would change, or want to add, if you were this applying this approach for yourself?’. There are of course benefits, however, of keeping surveys broad and not being too directive, so much will depend on the aims of the engagements.

Another limitation is that the engagement activities and collaborators were more representative of voice-hearing experiences than of other experiences in psychosis. This was mainly due to the CFV film being specifically focused on voices; and hence attracted more engagement around these specific experiences. However, the CFTp intervention is designed for people with other types of experience as well (e.g. delusions), and as it may be less well developed for these, one recommendation for future engagement activities around CFTp would be to specifically engage experts by experience from groups other than the Hearing Voices Network, for example, the National Paranoia Network, and indeed the Spiritual Crisis Network who collaborated on Study 3 of this thesis (Chapter 5).

The main limitation of the clinician workshops and CAVEtalk groups was that these were self-selecting individuals who had an interest in using CFTp approaches in their clinical work. Nevertheless, there were sceptical voices and challenging discussions in the workshops. Some of the scepticism was around, for example, whether the CFTp psycho-education and relational techniques would be too complex for some people with psychosis. But in the main, people were curious and inspired by the CFTp approach. In future projects, it may be helpful to try and engage groups of more dissenting clinicians, or clinicians who hold more bio-medical views about psychosis causation. In order for a new intervention to be truly acceptable to services, there needs to be engagement and knowledge exchange with those coming from multiple viewpoints. One of the future goals for the development

122 and dissemination of CFTp could be to specifically engage with people who believe that compassionate engaging with symptoms would be unhelpful, or even harmful.

7.4.4. Conclusions

The intervention development procedures aimed to create a genuine knowledge exchange collaboration between research, lived experience, and practice. The main outcome of these collaborations was to develop CFTp to be an acceptable, user-informed, and where possible, user- owned intervention. The procedures succeeded in these aims and produced a CFTp manual that can be used for therapy evaluation and testing. The outputs and results of the engagement activities allowed predictions not only about what processes and outcomes will change with CFTp, but also when they will change. This has implications for designing CFTp evaluation studies; for example, using a design with ‘mid-therapy’ assessment to test which measures change first (quickly) and which measures are slower to change. A more general recommendation is for the inclusion and publishing of intervention development studies, not only for transparency and rigour, but also for the improvement of our interventions.

123

Chapter 8. Study 4) – A case series study of Compassion Focused Therapy for distressing experiences in psychosis

8.1. Abstract

8.1.1. Background CFT is a psychological intervention that aims to improve mental health by building experiences of safeness and compassion, by regulating threat-related emotions, and by countering social rank threat experiences such as self-criticism and shame. Shame, dissociation, and other threat-driven processes are common among people with psychosis, which merits the application and testing of CFT in this population. This study aimed to develop and test the acceptability of a 26-session individual CFT for distressing experiences in psychosis (CFTp).

8.1.2. Methods This study used a case series design, where participants with distressing psychotic experiences were randomly allocated to one of the multiple baseline periods (2, 4 and 6 weeks), before starting CFTp with a psychologist. Brief measures of social safeness and dissociation were collected every week throughout the study, and a full battery of process and outcome measures were collected at five timepoints (baseline, pre-therapy, mid-therapy, post-therapy, and 6-8 week follow-up).

8.1.3. Results Seven out of eight participants completed the therapy, which supports the acceptability of CFTp in this population. Among the completers, clinically reliable and significant improvements in key measures of process and outcome were found in the intervention phase, compared to baseline phase. At the single case level of analysis, over half of participants showed clinically reliable improvements in outcome measures of depression (5/7), stress (5/7), general wellbeing (5/7), anxiety (4/7), and voices (3/5). One participant showed a deterioration in anxiety (1/7) and dissociation (1/7). At the group level of analysis (n=7), there were significant improvements in outcome measures of depression, stress, wellbeing, voices and delusions. The significant group-level improvements in general wellbeing, voices and delusions were sustained at 6-8 weeks follow-up, but depression and stress dropped slightly to trend level improvements.

8.1.4. Conclusions CFTp is a feasible and acceptable intervention for people with distressing experiences in psychosis. This intervention warrants further investigation with a pilot Randomised Controlled Trial. Future evaluation should involve different trained therapists to deliver the interventions, as well as independent, blinded researchers to collect and analyse the data.

124

8.2. Introduction

8.2.1. Background

Compassion-Focused Therapy (CFT) was developed for helping individuals with high levels of shame and internal self-criticism (Gilbert, 2009b), and one of its core interventions is to help people move out of social ranking ‘mentalities’ (see Chapter 2) that focus attention on the (potential harmful) power of others, towards activating cooperative, caring and affiliative processes to self and others. Consequently, CFT provides contexts, practices and insights that facilitate the development of compassion in self-other, other-self, and self-self (as well as self-voice) relationships. There is now growing evidence for the effectiveness of CFT in targeting these processes across different mental health populations, e.g. complex difficulties (Gilbert & Procter, 2006), personality disorders (Lucre & Corten, 2013), and eating disorders (Gale, Gilbert, Read, & Goss, 2014).

The literature outlined in Chapter 6 suggests that CFT is also ideally suited to address the kinds of threat-based difficulties experienced by people with psychosis, and to date, this has been investigated in two clinical studies with this population: Firstly, a small case series (n=3) of compassionate mind training (which refers to the specific practices and activities employed in CFT) for people who hear malevolent voices (Mayhew & Gilbert, 2008). In this study, participants were helped to visualise the compassionate part of themselves, with associated feelings of warmth, and to develop empathy for their distress and self-criticism. Interestingly, the researchers found that working with self-critical thoughts (as opposed to with critical voices directly) led to a reduction in the malevolence of voices. Secondly, Braehler et al. (2013) studied 22 participants receiving CFT groups, which involved applying compassionate practice and skills in relation to internal and external threats, such as shame, stigma, paranoia, self-attacking, and hostile voices. Their main finding was that the CFT group participants reported significant reductions in depression associated with psychosis, from pre to post therapy; reductions that were not seen in a treatment-as-usual control group. These were promising signs for potential effectiveness; however, as these changes were from within-group analyses, rather than between-group comparisons (all of which were non-significant), further research is needed before there is evidence for effectiveness of CFT groups for people with psychosis.

125

8.2.2. Summary and aims

This study aimed to develop and test the acceptability of a 26-session individual CFT for distressing experiences in psychosis (CFTp). As described in Chapter 7, the initial draft version of the CFTp manual was developed in collaboration with experts in CFT and experts by experience, and then continued to be developed and evolved as the study progressed, and as more was learnt (e.g. from service-user participants and collaborators) about applying the model in this population. The study also aimed to investigate the effectiveness of the CFTp intervention, and its processes of change, by using a multiple baseline, single-case series design. The aim was to have a final version of the CFTp manual that could be taken forward for use in a pilot trial, and with acceptability, process, and outcome indicators to inform whether such a trial would be justified.

8.2.3. Case series methodology

Case series are often used in the early phases of evaluating novel interventions. They can provide evidence to support and justify future studies with stronger design. They can provide initial evidence on acceptability and contribute to refinements to manuals and protocols. These are useful, informative steps to take before the more costly, and resource-intensive, procedures of pilot and full trials. Case series studies can also be used to provide initial indicators of outcomes and evaluation. They can test whether an intervention is successful at targeting the processes and mechanisms it intends to target, and whether these relate to clinically relevant outcomes. In terms of evaluation, case series studies are limited in the strength of conclusions that can be drawn, due to small samples and absence of control groups. However, there are designs that can strengthen their conclusions and implications for future evaluations, for example, by assessing ‘no intervention’ baseline phases before treatment, which can act as comparators for the intervention phase (P. J. Watson & Workman, 1981).

8.2.4. Measuring processes of change

As outlined in Chapter 6 (see also Heriot-Maitland et al. (2019)), the CFT theoretical framework predicts that for people with psychosis, threat-based dissociative and psychotic processes would be accentuated by external and internal social ranking signals (such as shame and self-criticism), and attenuated by external and internal communication signals that indicate caring, supportive, and social safeness experience (e.g. such as compassion to/from self and others). The CFTp intervention in this study therefore aimed to target social rank threat as the putative mechanism underpinning distressing

126 psychotic experiences, and the way CFTp was designed to target social rank threat processes was by helping people to build experiences and capacities for social safeness (as a threat-regulator), and by helping them develop compassionate motives towards themselves, others, and their distressing experiences.

A variety of process measures were chosen to capture various aspects of these targeted change processes. The measure chosen to tap changes in social safeness was the Social Safeness and Pleasure Scale (SSPS) (Gilbert et al., 2009). This had been previously used by community psychosis settings to capture the processes of change for a brief CFT group (Heriot-Maitland, Keen, Lawlor, Kane, & Peters, in preparation), and because it had shown good sensitivity to change over short periods of time, and was quick to administer (11 items), this was deemed a suitable session-by-session (weekly) measure for the current study. Alongside this, a measure of Heart Rate Variability (HRV) was included, which reflects the influence of parasympathetic system on the heart. HRV is becomingly increasingly used in CFT research as a physiological indictor of the key processes targeted in the compassion training and CFT (Kirby et al., 2017).

In terms of other process measures, the Self-Compassion Scale-Short Form (SCS-SF) (Raes, Pommier, Neff, & Van Gucht, 2011) was chosen to capture the process of developing compassion towards the self, and for social rank threat processes, there were four measures chosen: Social Comparison Scale (SocCS) (Allan & Gilbert, 1995), the Forms of Self-Criticising/Attacking and Self-Reassuring Scale (FSCSR) (Gilbert, Clarke, Hempel, Miles, & Irons, 2004), the Other as Shamer Scale (OAS) (Goss et al., 1994), and the Personal Beliefs about Illness Questionnaire – Revised (PBIQ-R) (Birchwood, Jackson, Brunet, Holden, & Barton, 2012). The reason for having a selection of measures was that, between them, they would be able to tap slightly different aspects of external and internal social ranking signals; for example, the SocCS and OAS would be capturing social rank patterns in self-to-other (external) relationships, and the FSCSR would be capturing social rank patterns within the self-to-self (internal) relationship. The PBIQ-R was developed to access five concepts of social ranking theory, including shame and social marginalisation.

Deciding on which process measures to include involved consideration of the trade-off between psychological richness and minimizing multiple testing and Type 1 errors. As a compromise, the composite total score for two of the measures was used (PBIQ-R and SCS-SF, rather than their alternatives of five and six subscales respectively). The three subscales of the FSCSR (inadequate, reassured, hated) were retained, however, as these different forms of self-relating had clinical

127 relevance to the CFTp approach. A further aim of doing this early evaluative work (i.e. case series) is to inform the most acceptable scales as well as those most able to capture change.

8.2.5. Measuring outcomes

If, as according to social rank theory, distressing psychotic experiences are related to the activation of social rank mentalities, then it would be expected that people with these experiences would also display a range of other social ranking characteristics; for example, vulnerability to feelings of defeat, inferiority, rejection, and shame, high sensitivity to social comparison, as well as a range of protective strategies, such as submission, depression, and dissociation (see Heriot-Maitland et al. (2019) for a review of theory and empirical support). Therefore, in terms of outcomes, the study was not only interested in outcomes for the positive symptoms themselves (voices and delusions), but also for dissociation and depression. Additionally, from a clinical service perspective, it was important to assess whether CFTp could be considered for implementation in the context of routine NHS service delivery, and as such it was also interested in tapping more generic, clinically-relevant outcomes, such as wellbeing, functioning, and risk/harm.

The measure chosen for sessional (weekly) outcome recording was a brief 3-item dissociation scale. The same three items had previously been used to capture fluctuations in dissociation in an Experiencing Sampling Method (ESM) study (Varese, Udachina, Myin-Germeys, Oorschot, & Bentall, 2011), and was therefore deemed a suitable scale for monitoring dissociation changes over short time- periods. To measure voices, delusions, dissociation and depression at the five assessment points, the measures chosen, respectively, were: the Psychotic Symptom Rating Scales (PSYRATS) (Haddock, McCarron, Tarrier, & Faragher, 1999); the Revised Dissociative Experiences Scale (DES-II) (Carlson & Putnam, 1993); and the Depression, Anxiety and Stress Scale (DASS) (Lovibond & Lovibond, 1995). To measure general clinically-relevant outcomes, the Clinical Outcomes in Routine Evaluation (CORE) (Barkham Chris Evans Frank Margison, 2009) was chosen, as this was the measure already being used routinely for all patients of the NHS trust from which recruitment commenced.

8.2.6. Research questions and hypotheses

• Is the CFTp intervention feasible and acceptable?

128

• Can the CFTp intervention make targeted changes to key social processes (i.e. increasing social safeness, compassion, and decreasing social rank threat), with the desired effect on clinical outcomes (i.e. reducing psychotic states, dissociation, and depression)?

To address this research question, the following hypotheses were made about which changes would occur, and when, at the analyses levels of both i) single-case and ii) group: 1) Significant changes in outcome and process measures will occur during the intervention, and not the baseline, phase. 2) Significant changes in process measures will precede changes in outcome measures 3) Significant changes in outcome and process measures will still be evidenced 6-8 weeks after the intervention has finished 4) Session-by-session measures will significantly improve in the intervention, compared to the baseline, phase

8.3. Methods

8.3.1. Design and participants

A non-concurrent multiple-baseline, single case design (Watson & Workman, 1981) was used. There were three phases to the study: a baseline, intervention, and follow-up phase. Participants were randomly allocated to one of the multiple baseline periods (2, 4 and 6 weeks), before commencing the intervention (26 weekly sessions over 6-9 months). They were then followed-up after finishing the intervention (6-8 weeks later). The random baseline allocation is included in Appendix 8.1.

Participants were secondary care mental health service-users with distressing psychotic experiences and a psychosis-related diagnosis, recruited from two NHS sites: one in London, UK (South London and Maudsley NHS Foundation Trust) and one in Glasgow, UK (NHS Greater Glasgow and Clyde). The inclusion criteria were as follows: i) over 18 years old; ii) psychosis-related diagnosis (F20-39); iii) distressing positive symptoms of psychosis (voices and/or delusions, with distress evidenced by an ‘intensity of distress’ score of 2+ on the PSYRATS) (Haddock et al., 1999); and iv) not currently engaged in cognitive behavioural therapy, or having completed a 12+ sessions of therapy within the last 3 years.

129

8.3.2. Measures

Outcome measures

Psychotic symptoms Psychotic symptoms were measured by the Psychotic Symptom Rating Scales (PSYRATS) (Haddock et al., 1999), which is a 17-item interview-rated scale measuring various dimensions of auditory hallucinations (11 subscales) and delusions (6 subscales). Scores for each subscale are rated 0 – 4, with total scores range 0 – 44 (hallucinations) and 0 – 24 (delusions). A total score for both hallucinations (all subscales) and delusions (all subscales) is used in the main analysis, and then a distress score using only two subscales (amount and intensity of distress) is used for both hallucinations and delusions in an additional analysis. The PSYRATS has been found to have high inter-rater reliability and was designed to be sensitive to changes following psychological interventions with psychosis.

Depression and anxiety symptoms Depression, anxiety and stress was measured by the Depression, Anxiety and Stress Scale (DASS-21) (Lovibond & Lovibond, 1995), which is a 21-item self-report measure of emotional symptoms in three subscales: depression, anxiety and stress. Items are rated on a four-point scale (0 = never to 3 = almost always) to indicate the extent to which the statements apply over the past week. Total scores range 0 – 42 (depression), 0 – 42 (anxiety) and 0 – 42 (stress). The subscales have good internal consistency (Cronbach’s alpha: .94 depression, .87 anxiety, .91 stress), which has been replicated in a smaller psychosis sample (.93 depression, .91 anxiety, .93 stress, Huppert, Smith, and Apfeldorf (2002)).

General wellbeing General clinical outcomes were measured by the Clinical Outcomes in Routine Evaluation (CORE) (Barkham Chris Evans Frank Margison, 2009), which is a 34-item self-report scale measuring domains of subjective wellbeing, symptoms, functioning, and risk/harm. Total mean scores range 0 – 4.

Dissociation symptoms Dissociation was measured by the Revised Dissociative Experiences Scale (DES-II) (Carlson & Putnam, 1993), which is a 28-item self-report scale. Items are rated as percentages (in 10% increments from 0% = never to 100% = always) according to what percentage of time they have the experience, and scores are then converted to a 0 – 10 rating. Total scores range 0 – 280. The DES-II captures feelings of depersonalisation, derealisation, and psychogenic amnesia. It is the most widely used measure of

130 dissociation, has high internal consistency in clinical and non-clinical populations (mean Cronbach’s alpha .93, Vanijzendoorn and Schuengel (1996)), and has been used extensively with psychosis samples. This study used a state-adapted version, i.e. asking specifically about ‘in the past week…’

Process measures

Social rank threat, safeness, and compassion The key targeted social processes of (internal/external) social rank threat, safeness, and compassion were measured by: i) the Social Comparison Scale (SocCS) (Allan & Gilbert, 1995), which is an 11-item self-report scale (rating 1 – 10, total scores range 11 – 110) measuring social rank and relative social standing; ii) the Forms of Self-Criticising/Attacking and Self-Reassuring Scale (FSCSR) (Gilbert et al., 2004), which is a 22-item self-report scale (rating 0 – 4) measuring self-criticism (‘inadequate self’ 9 items with total scores range 0 – 36 and ‘hated self’ 5 items with total scores range 0 – 20) and ‘self- reassurance’ (8 items with total scores range 0 – 32); iii) the Other as Shamer Scale (OAS) (Goss et al., 1994), which is an 18-item self-report measure of external shame (rating 0 – 4, total scores range 0 – 72); iv) the Self-Compassion Scale-Short Form (SCS-SF) (Raes et al., 2011), which is a 12 item self-rated scale measuring self-compassion (rating 1 – 5, total scores range 12 – 60); and v) the Personal Beliefs about Illness Questionnaire – Revised (PBIQ-R) (Birchwood et al., 2012), which is a 20-item self-report measure (rating 1 – 4, total scores range 20 – 80) of social rank variables in psychosis, with five subscales: shame; loss; entrapment; control over illness; and social marginalization/group fit.

Heart rate variability At the start of each assessment session, participants had their heart rate monitored using a single channel ECG waveform recorder (Actiwave Cardio, CamNtech Ltd.) that was connected to their chest by two ECG pads. For 5 minutes, a resting Heart Rate Variability (HRV) recording was taken as participants sat normally in a chair, and then for 2 minutes, a further HRV recording was taken as participants engaged with a soothing breathing rhythm exercise, using a guided breathing app (Breathing Zone). In the first minute of the breathing exercise, a visual cue was used to gradually slow participants’ breathing rate from 8 to 6 breaths per minute, and for the second minute, the visual cue kept their breathing at the target rate of 6 breaths per minute. The measurement used in this study was the RMSSD (Root Mean Square of Successive Differences in the inter-beat intervals) for the full 2- minute breathing exercise. RMSSD is a recommended measure for the time-domain assessment of HRV (Malik et al., 1996), and increases in the RMSSD score signal improvements in HRV, which is indictive of improved parasympathetic regulatory influence on the heart.

131

Session-by-session measures

Social safeness Session-by-session social safeness was measured by the Social Safeness and Pleasure Scale (SSPS) (Gilbert et al., 2009), which is an 11-item self-report scale measuring the extent to which people experience their social worlds as safe, warm and soothing. Items are rated on a five-point scale (1 = almost never to 5 = almost all the time, total scores range 11 – 55). The scale has high internal consistency (Cronbach’s alpha .92).

Dissociation Session-by-session dissociation was measured by a brief 3-item self-report scale (1 = not at all to 7 = very much, total scores range 3 – 21). The three items are: “since the last session I’ve found it difficult to focus on what was happening around me”; “since the last session I’ve been easily distracted”; “since the last session I’ve found myself doing things without paying attention”. These same three items were used in a previous study to measure fluctuations in dissociation (Varese et al., 2011), and had been validated against the Revised Dissociative Experiences Scale (DES-II) (above) by the study authors.

8.3.3. Procedure

The study received ethical approval from the London Dulwich Research Ethics Committee (REC reference 15/LO/0198) (Appendices 8.2 and 8.3), and Research & Development approval from the two NHS sites in London (Appendix 8.4) and Glasgow (Appendix 8.5). Participants were identified, and invited to participate, by their clinical teams. Those interested were then invited to meet with the researcher to discuss the study and to provide written consent.

The study involved five data collection points: T1 (start baseline); T2 (end baseline); T3 (mid therapy); T4 (end therapy); T5 (follow-up). Each assessment lasted about one hour. Brief sessional measures were also administered every week that participants were in the study (during baseline and intervention, but not follow-up phase). During the intervention phase, these were administered at the start of sessions. Most of the measures were self-report questionnaires, answered electronically by participants on a laptop. The only two measures that required researcher involvement were PSYRATS and Heart Rate Variability measure. PSYRATS was scored by the lead researcher (CHM) and Heart Rate Variability was measured with assistance from research nurses from both the NIHR/Wellcome Trust King’s Clinical Research Facility at King’s College Hospital and the Glasgow Clinical Research Facility.

132

Participants received an honorarium and travel expenses for each of the five assessment points. All participants were asked for their consent to record the intervention sessions, by audio, video, or both. This was partly for CFT clinical supervision purposes, but also for purposes of fidelity checking.

8.3.4. Intervention description

The intervention was provided by a clinical psychologist and developer of the CFTp intervention (author CHM), with adherence to a 50-page CFT for Psychosis (CFTp) manual that was developed in collaboration with CFT experts and experts by experience, as detailed in Chapter 7. In line with recommended guidelines for how to describe interventions for trials (Hoffmann et al., 2014), Table 8.1 shows the Template for Intervention Description and Replication (TIDieR) checklist. A weblink to access the CFTp manual is included in Appendix 8.6.

Table 8.1. Template for intervention description and replication (TIDieR) checklist 1. Brief name Compassion Focused Therapy for Psychosis (CFTp) 2. Why To target changes to key social processes (i.e. increasing social safeness, compassion, and decreasing social rank threat), with an effect on clinical outcomes (i.e. reducing psychotic symptoms, dissociation, and depression) 3. What (materials) A CFTp manual 4. What (procedure) The CFTp manual is divided into the following levels: Starting therapy 1. Establishing safeness and connection 2. Learning about evolved (tricky) brains, emotional systems, & multiple selves 3. Understanding how my emotions and mind have become shaped 4. Building the compassionate self 5. Directing compassion to self, others, emotional parts, and voices Ending therapy Full descriptions and guidance for each level is provided in the manual 5. Who provided Clinical Psychologist with 10+ years’ experience working with people with psychosis, with monthly supervision from a CFT expert. As the therapist was the developer of the intervention, there was no specific training in CFT for Psychosis; however, the therapist had trained in (generic) CFT, at both the introductory and advanced levels. 6. How Face-to-face sessions, individually 7. Where NHS therapy/consultation room. For five participants, the room was in an NHS site in London, and for two, the room was in an NHS site in Glasgow. 8. When and how 26 x 1-hour weekly sessions, over 6-9 months; much 9. Tailoring The CFTp manual includes the following instructions for tailoring: “The ordering of 1-5 levels is to guide the therapist through the therapy content. In reality this is a process-driven therapy, so therapists will be

133

following the client, using clinical judgment, supervision, and collaborative discussion to transition to/from each level. This may involve re-ordering levels 1-5, blending one with another, or skipping one out completely. The phrases below are also to guide, rather than to quote. The therapist will be talking to clients in their own language, at their own pace, using Socratic questioning, and attending to process.” (CFTp manual, Appendix 8.6, p1) 10. Modification Modifications were made to the manual over the intervention period (2015-2018), as more was learnt about applying CFT with this population. Full details of the modifications made between the 30-page CFTp manual v1 (Dec 2014) and 50-page CFTp manual v2 (Dec 2018) can be found in Chapter 7. 11. How well (planned) A CFTp Adherence and Competence Measure (CFTp-ACM) was developed to assess fidelity to the manual and therapist competence in CFT. In the CFTp-ACM, there are 21 manual adherence items and 23 therapist competence items. All are rated on a five-point scale from 0 (“absent or inappropriate”) to 4 (“skilful enactment”). The 21 adherence items relate to the key elements in the manual, which had been developed specifically for the current (psychosis) population, and the 23 competence items were taken from an existing scale that had been developed for generic CFT research, namely the Compassion Focused Therapy Therapist Rating Scale (CFT-TCRS) (Horwood, Allan, Goss, & Gilbert, 2019b). The 23-item CFT-TCRS is comprised of 14 unique competencies (e.g. “compassionate mind training” and “multiple selves”) and 9 microskills (e.g. “pacing” and “mentalisation”). The CFTp- ACM is included in Appendix 8.7. 12. How well (actual) Five audios were sent to an external CFTp expert (Dr Tammy Lennox) for fidelity checking against the CFTp-ACM. To ensure that fidelity checking procedures covered a suitable breadth of different therapy levels, and a suitable breadth of participants in the sample, audios were selected so that each of the five different therapy levels were represented by a different participant. The adherence ratings averaged 3.86 (on the scale 0 “absent or inappropriate” to 4 “skilful enactment”), and the therapist competence ratings averaged 3.67 on the same scale. For the two subscales within the therapist competence scale, the ratings averaged 3.80 for CFT unique competencies, and 3.64 for CFT Microskills. Overall, the therapist was rated as “Excellent” as a CFT therapist for 3 of the sessions, “Very good” for 1 session and “Good” for 1 session. See Appendix 8.8 for a full table of items and ratings. These ratings confirmed that the therapist in this study was adherent to the CFTp manual and competent in providing the CFTp intervention.

8.3.5. Data analysis

The first hypothesis (1), that significant changes in process and outcome measures would occur in the intervention, and not the baseline, phase, was analysed at both the single-case and the group levels. At the single-case level, individual scores were compared against a Reliable Change Index

134

(RCI) (Jacobson & Truax, 1991) to determine which changes across phases were greater than would be expected from the standard variability of measures. In line with the internal consistency method for clinical populations (Martinovich, Saunders, & Howard, 1996), the RCI analysis required reliability scores (internal consistency) and standard deviations of each measure, which were sourced from existing normative data (see Appendix 8.9 for normative scores and sources). The RCI is computed by dividing the change scores (in either baseline, T1→T2, or intervention, T2→T4, phases) by the standard error of change between the scores. If the RCI is greater than 1.96, the change is considered to reliable (Jacobson & Truax, 1991). If the RCI is lower than 1.96, the change is not reliable, as it could have occurred due to measurement error. At the group level, a Wilcoxon Signed Rank Test was used to compare changes in group mean scores in the baseline phase (T1→T2) with those in the intervention phase (T2→T4).

Hypothesis 2, that changes in process measures would precede changes in outcome measures, was analysed using the same procedures as above, but with the additional use of the mid-therapy timepoint (T3) to enable comparisons between a ‘first half’ intervention phase (T2→T3) with the intervention phase (T2→T4), to see which measures were quicker to change.

Hypothesis 3, that changes in outcome and process measures would still be evidenced after the intervention had finished, was analysed by repeating the above analysis procedures with the follow- up timepoint (T5) replacing end of therapy (T4) in all calculations. So, essentially creating a new ‘intervention + follow-up’ phase (T2→T5) as the comparator to baseline phase (T1→T2).

Hypothesis 4, that session-by-session measures would significantly improve in the intervention, compared to baseline, phase, was also analysed at both the single-case and group levels. At the single-case level, the Tau-u statistic was used to analyse whether there was a significant difference between the baseline and intervention phases (or specifically, a significant degree of non-overlap between the two phases) (Parker, Vannest, Davis, & Sauber, 2011). At the group level, the same Tau- u scores for each participant were combined to produce a Tau-u Omnibus score. All Tau-u calculations were run using an online tool: http://www.singlecaseresearch.org/calculators/tau-u. For session-by-session measures, there was no equivalent of hypothesis 2, i.e. comparing ‘first half’ and ‘intervention’ phases, because the two measures, SSPS and dissociation, were expected to be more concurrent in their change patterns, rather than one preceding the other. So the Tau only analysed baseline (T1→T2) and intervention (T2→T4) trends.

135

8.4. Results

8.4.1. Feasibility and acceptability

The CONSORT diagram in Figure 8.1 shows the flow of participant recruitment and attrition in this study. Of the 21 who were identified and referred to the study by clinical teams, nine were recruited, eight started the intervention, and seven completed.

Figure 8.1. Flow diagram of participant recruitment

Referred to study N=21 Excluded (N=12) Not eligible (6, of whom 3 had no F20-39 diagnosis, 2 had no distressing positive symptoms, and 1 was seeing a therapist) Declined (3) Male therapist not appropriate (1) Uncontactable (2) Recruited and started baseline N=9

Excluded (N=1) Unable to attend around work hours (1)

Started intervention

N=8

Excluded (N=1) Due to intoxication, switched to providing clinical support, rather than CFTp (1)

Completed intervention N=7

136

The finding that seven of the eight starters completed the full course of therapy (i.e. engaged with all 26 sessions) is evidence for the overall acceptability of the CFTp intervention. The only starter who did not complete therapy finished at the mid-therapy point (after 13 sessions) due to a relapse in heavy alcohol use from around session 7 or 8 onwards. Clinical support continued for this individual, but because of high levels of intoxication it wasn’t possible to continue CFTp, and therefore the data could not be used. Alcohol abuse had been a chronic pre-existing condition for this participant but had not been an immediate problem at the start of recruitment.

Of the seven who completed therapy, and comprised the final study sample, one had distressing voices only, two had distressing delusions only, and four had both distressing voices and distressing delusions. Table 8.2 shows clinical and demographic information of these seven participants.

Table 8.2. Clinical and demographic profiles of participants (n=7) Positive symptoms Participant Age Sex Site Ethnicity Diagnosis Voices Delusions (distress) (distress) P1 53 Male London White F21 Yes Yes “Steve” British (3 marked) (3 marked) P2 64 Male London White F29 Yes Yes “Greg” British (4 extreme) (4 extreme) P3 37 Male London White F29 Yes Yes “Thomas” Irish (4 extreme) (4 extreme) P4 48 Male London Black F32 No Yes “Tosin” British (3 marked) P5 58 Female London Black F32 Yes No “Charmaine” British (4 extreme) P6 36 Female Glasgow White F20 Yes Yes “Amanda” British (2 moderate) (2 moderate) P7 57 Male Glasgow White F20 No Yes “Gareth” British (3 marked)

In terms of feasibility and acceptability of research procedures, all seven participants provided full outcome and process measures across all five assessment points. This included providing a full complement of HRV readings, although some readings had to be excluded due to issues with incorrect software set-up (4 readings) and poor ECG connection (4 readings), which were later discovered when processing the data. The missing HRV data were due to technical issues, rather than issues of participant acceptability or willingness to provide HRV data.

Six of the seven participants consented to audio recording, but none opted for video recording. The one person who opted out of any form of recording (P1) explained that this was to do with paranoia-

137

related concerns, i.e. as a precaution to prevent information being obtained by others with harmful intent. The session-by-session measures were acceptable for six of the seven participants for the full therapy. One participant (P1) did start providing weekly measures initially, but then asked to stop after 10 sessions, saying that this was due to finding the measures irritating and confusing. It was the same participant (P1) who opted out of both audio recording and sessional measures.

8.4.2. Single-case level analysis

Table 8.3 shows a summary of the reliable change directions in process and outcome measures, across four phases of the study. A full table of RCI results is included in Appendix 8.10. The phases are broken down into baseline phase (T1→T2), first half phase (T2→T3), intervention phase (T2→T4) and follow up phase (T2→T5).

Table 8.3. Directions of reliable changes in process and outcome measures, showing only those that were significant against a Reliable Change Index (RCI). (Full RCI scores table in Appendix 8.10.) Arrows in bold signal the direction of improvement for that measure

Summary of all 7 cases T1→T2 T2→T3 T2→T4 T2→T5 T1→T2 T2→T3 T2→T4 T2→T5

Process measures (RCI) Outcome measures (RCI) SocC ↑ 3 ↑ 4 ↑ 4 PSYRATS-V (n=7) ↓ 2 (n=5) ↓ 3 ↓ 4 FSCSR-Inad PSYRATS-D (n=7) ↓ 1 ↓ 2 ↓ 4 ↓ 5 (n=6) ↓ 1 ↓ 2 ↓ 1 FSCSR-Reas ↑ 1 ↑ 3 ↑ 3 ↑ 2 DASS-Dep ↑ 1 (n=7) ↓ 1 ↓ 2 ↓ 1 (n=7) ↓ 1 ↓ 5 ↓ 5 ↓ 5 FSCSR-Hate ↑ 1 DASS-Anx ↑ 1 ↑ 1 ↑ 1 ↑ 2 (n=7) ↓ 5 ↓ 5 ↓ 4 (n=7) ↓ 2 ↓ 4 ↓ 4 ↓ 3 OAS DASS-Str ↑ 2 ↑ 1 ↑ 1 (n=7) ↓ 2 ↓ 3 ↓ 3 ↓ 5 (n=7) ↓ 1 ↓ 5 ↓ 5 ↓ 4 SCS-SF ↑ 1 ↑ 1 ↑ 4 ↑ 3 CORE ↑ 1 (n=7) ↓ 1 (n=7) ↓ 4 ↓ 5 ↓ 5 PBIQ-R DES-II ↑ 1 (n=7) ↓ 6 ↓ 6 ↓ 7 (n=7) ↓ 3 ↓ 2 ↓ 4 RMSSD (ms) ↑ 1 ↑ 2 ↑ 2 (n=variable) ↓ 1 ↓ 1 ↓ 1 (n=4) (n=6) (n=5) (n=5)

138

Breakdown by single case P1 T1→T2 T2→T3 T2→T4 T2→T5 T1→T2 T2→T3 T2→T4 T2→T5

Process measures (RCI) Outcome measures (RCI) SocC1 - - - - PSYRATS-V - - - - FSCSR-Inad - - - - PSYRATS-D - - ↓ - FSCSR-Reas - ↑ ↑ - DASS-Dep - ↓ - ↓ FSCSR-Hate - ↓ ↓ - DASS-Anx ↓ ↑ ↑ ↑ OAS - - - ↓ DASS-Str ↑ ↓ ↓ ↓ SCS-SF - - ↑ - CORE - - - - PBIQ-R - ↓ ↓ ↓ DES-II - - ↑ - RMSSD (ms) - ↑ md2 - P2 T1→T2 T2→T3 T2→T4 T2→T5 T1→T2 T2→T3 T2→T4 T2→T5

Process measures (RCI) Outcome measures (RCI) SocC - ↑ ↑ ↑ PSYRATS-V - - ↓ ↓ FSCSR-Inad - - - ↓ PSYRATS-D - ↓ ↓ ↓ FSCSR-Reas - ↑ ↑ ↑ DASS-Dep - - ↓ ↓ FSCSR-Hate - ↓ ↓ ↓ DASS-Anx - ↓ ↓ ↓ OAS ↓ ↓ ↓ ↓ DASS-Str ↑ ↓ ↓ ↓ SCS-SF - - ↑ ↑ CORE - ↓ ↓ ↓ PBIQ-R - ↑ ↓ ↓ DES-II - ↓ ↓ ↓ RMSSD (ms) ↑ md md ↓ P3 T1→T2 T2→T3 T2→T4 T2→T5 T1→T2 T2→T3 T2→T4 T2→T5

Process measures (RCI) Outcome measures (RCI) SocC - - - - PSYRATS-V - - - ↓ FSCSR-Inad - - ↓ - PSYRATS-D - - - - FSCSR-Reas - - - - DASS-Dep - ↓ ↓ ↓ FSCSR-Hate - ↓ - - DASS-Anx ↓ ↓ - - OAS - - - - DASS-Str - - ↓ - SCS-SF - ↑ ↑ ↑ CORE - ↓ ↓ ↓ PBIQ-R - ↓ ↓ ↓ DES-II - - - ↓ RMSSD (ms) ↓ - ↑ - P4 T1→T2 T2→T3 T2→T4 T2→T5 T1→T2 T2→T3 T2→T4 T2→T5

Process measures (RCI) Outcome measures (RCI) SocC ↓ ↑ ↑ ↑ PSYRATS-V nv3 nv nv nv FSCSR-Inad - - ↓ ↓ PSYRATS-D - - - - FSCSR-Reas ↓ ↑ ↑ ↑ DASS-Dep - ↓ ↓ ↓ FSCSR-Hate - - ↓ ↓ DASS-Anx - ↓ ↓ ↓ OAS ↓ - - - DASS-Str - ↓ ↓ ↓ SCS-SF - - - ↑ CORE ↑ ↓ ↓ ↓ PBIQ-R - ↓ ↓ ↓ DES-II - - - - RMSSD (ms) md ↑ - md P5 T1→T2 T2→T3 T2→T4 T2→T5 T1→T2 T2→T3 T2→T4 T2→T5

Process measures (RCI) Outcome measures (RCI) SocC ↓ ↑ ↑ ↑ PSYRATS-V - - ↓ ↓

139

FSCSR-Inad - ↓ ↓ ↓ PSYRATS-D nd4 nd nd nd FSCSR-Reas ↑ ↓ - ↓ DASS-Dep - ↓ ↓ ↓ FSCSR-Hate - ↓ ↓ ↓ DASS-Anx ↑ ↓ ↓ ↓ OAS - ↓ ↓ ↓ DASS-Str - ↓ ↓ ↓ SCS-SF - - ↑ - CORE - ↓ ↓ ↓ PBIQ-R - ↓ ↓ ↓ DES-II - ↓ ↓ ↓ RMSSD (ms) md - - - P6 T1→T2 T2→T3 T2→T4 T2→T5 T1→T2 T2→T3 T2→T4 T2→T5

Process measures (RCI) Outcome measures (RCI) SocC - - ↑ ↑ PSYRATS-V - - ↓ ↓ FSCSR-Inad - ↓ ↓ ↓ PSYRATS-D - - - - FSCSR-Reas - - - - DASS-Dep ↓ ↑ - - FSCSR-Hate - - - - DASS-Anx - - - - OAS - - ↓ ↓ DASS-Str ↓ ↑ - - SCS-SF - - - - CORE - - ↓ ↓ PBIQ-R - ↓ ↓ ↓ DES-II - ↓ - ↓ RMSSD (ms) - - - - P7 T1→T2 T2→T3 T2→T4 T2→T5 T1→T2 T2→T3 T2→T4 T2→T5

Process measures (RCI) Outcome measures (RCI) SocC - - - - PSYRATS-V nv nv nv nv FSCSR-Inad ↓ - - ↓ PSYRATS-D - - - - FSCSR-Reas - ↓ - - DASS-Dep - ↓ ↓ - FSCSR-Hate ↑ ↓ ↓ ↓ DASS-Anx - - ↓ ↑ OAS - ↓ - ↓ DASS-Str - ↓ - ↑ SCS-SF ↑ ↓ - - CORE - - - - PBIQ-R - ↓ - ↓ DES-II - - - - RMSSD (ms) md ↓ ↑ md 1SocC = Social Comparison Scale; FSCSR = Forms of Self-Criticising/Attacking and Self-Reassuring Scale (Inad = Inadequate- self, Reas = Self-reassurance, Hate = Hated-self); OAS = Other as Shamer Scale; SCS-SF = Self-Compassion Scale Short-Form; PBIQ-R = Personal Beliefs about Illness Questionnaire-Revised; RMSSD (ms) = Root Mean Square of Successive Differences (milliseconds); PSYRATS = Psychotic Symptoms Ratings Scales (V = Voices, D = Delusions); DASS = Depression Anxiety and Stress Scales (Dep = Depression, Anx = Anxiety, Str = Stress); CORE = Clinical Outcomes in Routine Evaluation; DES-II = Revised Dissociative Experiences Scale. 2md = missing data; 3nv = no voices; 4nd = no delusions

Examining the reliable (significant) changes in measures across each of the different study phases in Table 8.3, the first three hypotheses (1, 2, 3) were tested as follows:

1) Significant changes in outcome and process measures will occur during the intervention, and not the baseline, phase.

In the intervention phase (T2→T4), there were cases of reliable improvement in both process and outcome measures. With the process measures, there were six reliable improvers on PBIQ-R, five on FSCSR-Hate, four on SocC, FSCSR-Inad, and SCS-SF, three on FSCSR-Reas and OAS, and two on RMSSD. These numbers of reliable change cases were all higher than in their equivalent baseline phases. In

140 the baseline phase, there were some reliable changes in process measures, but for only one or two (maximum) cases per measure, so most cases were not registering baseline changes. With the outcome measures, there were five reliable improvers on DASS-Dep, DASS-Str, and CORE, four on DASS-Anx, three on PSYRATS-V (of n=5), and two on DES-II and PSYRATS-D (of n=6). There was one case (P1) where two measures changed in the opposite direction (P1 had increases in both DASS-Anx and DES-II scores). This is in contrast to the baseline phase, where there were no changes for any participant in three of the seven outcome measures (PSYRATS-V, PSYTRATS-D, DES-II) and only one or two changes in DASS-Dep (one decrease P6), DASS-Anx (one increase P5 and two decreases P1 and P3), DASS-Str (two increases P1 and P2 and one decrease P6), and CORE (one increase P4). The higher incidence of reliable improvement cases in the intervention phase, compared to baseline, provided support for hypothesis 1 at the single-case level of analysis.

2) Significant changes in process measures will precede changes in outcome measures

This hypothesis would predict result patterns whereby more cases would register process changes in first half (earlier), in comparison to outcome changes, which would not register until later. However, the results show that participants who recorded reliable changes did so in both process and outcome measures in the first half of therapy. In Table 8.3, the only measures that seemed to show differences in how many cases were changing in ‘first half’, compared to ‘intervention’, are FSCSR-Inad (from two to four cases), SCS-SF (from one to four cases), and PSYRATS-V (from zero to three cases). Only one of these was an outcome measure (PSYRATS-V), so aside from PSYRATS-V, which changed late for three cases (P2, P5, and P6), there was not much evidence from the RCI results to support the hypothesis that process changes would precede outcome changes. Most of the other outcome measures were fairly stable in their numbers of first half changers, compared to intervention changers. Also, as already mentioned, two of the process measures were late-changers for some: FSCSR-Inad for two cases (P3 and P4), and SCS-SF for three cases (P1, P2, and P5). Overall, hypothesis 2 was not supported at the single-case level of analysis.

3) Significant changes in outcome and process measures will still be evidenced after the intervention has finished

Findings from the follow-up phase (T2→T5) show that most reliable improvements recorded by the end of therapy continued to be reliable improvements after therapy ended. In some measures, there was loss of one reliable improver at follow-up (FSCSR-Reas; FSCSR-Hate; SCS-SF; PSYRATS-D; DASS-

141

Anx; DASS-Str), but in other measures, there was more improvers post-therapy; for example, OAS was a reliable improvement for three cases at end of therapy (P2, P5 and P6), but by follow-up, was a reliable improvement for five (P1, P2, P5, P6 and P7). FSCSR-Inad had also reliable improvement for five cases (up from four) at follow-up, and for PBIQ-R, all seven cases had recorded reliable improvements by follow-up. An exception was RMSSD, which had registered two cases of improvement in the intervention (P3 and P7) (of n=5), but neither of these remained at follow-up, and one case (P2, who had missing data for the intervention phase), showed a RMSSD deterioration. In terms of outcomes at follow-up, four (of n=5) cases were found to have improvement in PSYRATS-V (P2, P3, P5 and P6), and one had improved on PSYRATS-D (P2). Five had DASS-Dep improvements, four had DASS-Str improvements, and three had improved on DASS-Anx. The only outcomes that changed in the opposite direction (a deterioration) at follow-up were DASS-Anx for two cases (P1 and P7) and DASS-Str for one (P7). Finally, five cases had improved scores on CORE, and four had improved DES-II scores at follow-up. Overall, these results support hypothesis 3 at the single-case level of analysis, with the exception of RMSSD, which had no evidence of improvements remaining post-therapy in the single-case analyses.

As a post-hoc, the visual inspection of PSYRATS distress scores (for both voices and delusions) was of particular interest since distress associated with experiences was the explicit target of the CFTp intervention. PSYRATS contains two distress-related scores (amount and intensity of distress) so a composite item of both (a 10-point scale) was used to represent visually the changes in distress associated with voices and delusions over time for each case (Figure 8.2). These graphs clearly show overall trends of distress reduction for all cases, across all five assessment points; however, the patterns of reduction are slightly different. For distressing voices, there was a pattern of distress staying quite stable (high) for the first three timepoints (T1, T2 and T3), and then dropping for the final two timepoints (T4 and T5). This drop is also greater for four cases (P2, P3, P5, P6) than for one (P1). For distressing delusions, there is a similar late reduction in distress for four cases (P1, P3, P6 and P6), but in one case (P2) there is a quick reduction by mid-therapy (T3), and for another case (P4), a slow reduction that does not occur until follow-up (T5). These results looking specifically at distress (visually), offer some support for all three above hypotheses (1, 2, 3); however, no statistical analyses were performed on these distress data.

142

Figure 8.2. Changes in distressing voices and distressing delusions for each case

PSYRATS Voices - distress 10 9 8 7 6 5 4 3 2 1 0 P1 P2 P3 P4 P5 P6 P7

T1 (base) T2 (start) T3 (mid) T4 (end) T5 (FU)

PSYRATS Delusions - distress 10 9 8 7 6 5 4 3 2 1 0 P1 P2 P3 P4 P5 P6 P7

T1 (base) T2 (start) T3 (mid) T4 (end) T5 (FU)

The final single-case level analysis was regarding the testing of hypothesis 4 as follows:

4) Session-by-session measures will significantly improve in the intervention, compared to the baseline, phase

143

This hypothesis was tested, firstly, with visual inspection of the plots of session-by-session measures (Figure 8.3), and secondly, with the Tau-u analysis of significant degree of non-overlap between baseline and intervention phases (Tables 8.4 and 8.5). One participant (P1) opted out of the weekly measures so his data were excluded from the Tau-u analysis.

Figure 8.3. Plots of session-by-session measures P1 “Steve” (2-week baseline)

SSPS Dissociation 60 25

50 20 40 15 30 10 20 10 5 0 0 T1T2 T3 T4 T5 T1T2 T3 T4 T5

P2 “Greg” (6-week baseline)

SSPS Dissociation 60 25

50 20 40 15 30 10 20 10 5 0 0 T1 T2 T3 T4 T5 T1 T2 T3 T4 T5

P3 “Thomas” (4-week baseline)

SSPS Dissociation 60 25

50 20 40 15 30 10 20 10 5 0 0 T1 T2 T3 T4 T5 T1 T2 T3 T4 T5

P4 “Tosin” (2-week baseline)

144

SSPS Dissociation 60 25

50 20 40 15 30 10 20 10 5 0 0 T1T2 T3 T4 T5 T1T2 T3 T4 T5

P5 “Charmaine” (2-week baseline)

SSPS Dissociation 60 25

50 20 40 15 30 10 20 10 5 0 0 T1T2 T3 T4 T5 T1T2 T3 T4 T5

P6 “Amanda” (4-week baseline)

SSPS Dissociation 60 25

50 20 40 15 30 10 20 10 5 0 0 T1 T2 T3 T4 T5 T1 T2 T3 T4 T5

P7 “Gareth” (6-week baseline)

SSPS Dissociation 60 25

50 20 40 15 30 10 20 10 5 0 0 T1 T2 T3 T4 T5 T1 T2 T3 T4 T5

145

Figure 8.3 shows that all participants ended the study with a higher SSPS scores, and lower dissociation scores, than when they started, which was the expected trend. The significance of these trends depended not only on the magnitude of change, but also the relative changes within the baseline phases before therapy started. The biggest SSPS changes were for cases P2 and P5; however, interestingly, P5’s considerable improvement over therapy T2→T4 (from SSPS score 19 to 46) was not a significant result in the Tau analysis (Tau = .37, p = .30) because there had also been improvement in the short 2-week baseline phase T1→T2 (from SSPS score 16 to 20), meaning that there was not a significant degree of non-overlap between the two phases. On the other hand, what visually looks like a very flat line for P4’s Dissociation trend (from score 20 to 18) was statistically significant in the Tau analysis (Tau = -.96, p < .01) due to the even flatter baseline period (with Dissociation score remaining constant at 21, which is the maximum score).

A visual inspection of the SSPS and Dissociation plots together shows a clear mirroring of each other’s trend in three or four cases. For P2 and P5 this mirroring is very clear, with Dissociation decreasing simultaneously as the SSPS increases. P3 and P7 also have mirroring trends, but because the changes were of a smaller magnitude, this is not so clear in the plots. P3 seems to have hit a ceiling in SSPS fairly quickly (up to score 49 by session 4, on a scale with total range 11-55), and a floor in Dissociation around the same time (down to score 3 in session 4, with total range 3-21). P7’s scores look fairly flat in both SSPS and Dissociation, but interestingly, the Tau analysis reveals that P7’s SSPS intervention scores are in a significant non-overlap with the baseline scores, in a negative direction (Tau = -.63, p = .01). This is because the P7’s baseline phase shows a steady increase in SSPS scores, which then stop increasing in therapy.

The cases who showed significant increases in SSPS, compared to baseline, outweighed the cases who did not, or showed a significant negative trend. This is the result of the Tau-u Omnibus test, in which all six participant scores were combined (bottom row of Table 8.4). For SSPS, the Tau omnibus score showed a significant upward trend (Tau = .48, p < .001). For Dissociation, the Tau omnibus score showed a non-significant downward trend (Tau = -.17, p = .17). Overall, hypothesis 4 was therefore supported for the SSPS measure, but not for the Dissociation measure.

146

Table 8.4. Summary of Tau analysis for sessional process measure (SSPS) Participant Tau SDTau p-value 85% conf. interval

Baseline P2 “Greg” .24 .32 .45 -.22, .70 (baseline trend) P3 “Thomas” .00 .49 1.00 -.71, .71 P4 “Tosin” .67 .64 .30 -.25, 1.00 P5 “Charmaine” .33 .64 .60 -.59, 1.00 P6 “Amanda” .40 .41 .33 -.19, .99 P7 “Gareth” .14 .32 .65 -.31, .60 Intervention P2 “Greg” .95 .25 <.001 .59, 1.00 (phase change) P3 “Thomas” .98 .32 <.001 .52, 1.00 P4 “Tosin” 1.00 .36 <.001 .48, 1.00 P5 “Charmaine” .37 .36 .30 -.14, .89 P6 “Amanda” .45 .28 .11 .04, .86 P7 “Gareth” -.63 .25 .01 -.99, -.27 Combined (phase change) P2+P3+P4+P5+P6+P7 .48 <.001 .30, .66 Significant results in bold

Table 8.5. Summary of Tau analysis for sessional outcome measure (Dissociation)

Participant Tau SDTau p-value 85% conf. interval

Baseline P2 “Greg” .38 .32 .23 -.08, .84 (baseline trend) P3 “Thomas” .00 .49 1.00 -.71, .71 P4 “Tosin” .00 .64 1.00 -.92, .92 P5 “Charmaine” .00 .64 1.00 -.92, .92 P6 “Amanda” .00 .41 1.00 -.59, .59 P7 “Gareth” -.05 .32 .88 -.50, .41 Intervention P2 “Greg” -.35 .25 .16 -.71, .01 (phase change) P3 “Thomas” -.06 .32 .84 -.52, .39 P4 “Tosin” -.96 .36 <.001 -1.00, -.44 P5 “Charmaine” -.81 .36 .02 -1.00, -.30 P6 “Amanda” .66 .28 .02 .26, 1.00 P7 “Gareth” .18 .25 .47 -.18, .54 Combined (phase change) P2+P3+P4+P5+P6+P7 -0.17 .17 -.35, .01 Significant results in bold

8.4.3. Group-level analysis

The first three hypotheses (1, 2, 3) were also tested at the group-level of analyses, meaning that scores from all seven participants were combined to test whole group changes across the four study phases:

147 baseline (T1→T2), first half (T2→T3), intervention (T2→T4), and follow up (T2→T5). Table 8.6 shows the mean scores for the group (n=7, unless otherwise specified) at each timepoint.

Table 8.6. Means and standard deviations for outcome and process measures at five assessment points Start baseline Start therapy Mid therapy End Therapy Follow up (T1) (T2) (T3) (T4) (T5) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Outcome measures PSYRATS-V1 22.71 (16.03) 22.14 (15.33) 18.86 (13.64) 11.00 (10.91) 10.43 (10.03) PSYRATS-D 14.71 (6.99) 14.29 (6.78) 12.00 (8.85) 7.86 (5.61) 7.86 (5.76) DASS-Dep 28.00 (4.76) 27.71 (5.22) 18.57 (6.60) 14.29 (6.87) 12.86 (10.45) DASS-Anx 21.14 (7.10) 20.57 (9.22) 15.43 (8.85) 13.43 (8.77) 13.43 (11.98) DASS-Str 27.43 (11.53) 28.29 (10.67) 19.14 (10.67) 15.43 (9.36) 15.14 (11.65) CORE 2.24 (0.51) 2.26 (0.47) 1.56 (0.69) 1.19 (0.62) 1.06 (0.71) DES-II 23.27 (18.61) 24.80 (19.11) 16.12 (13.07) 15.71 (17.13) 13.37 (15.21) Process measures SocC 35.29 (9.05) 29.43 (16.87) 47.14 (15.48) 60.57 (16.69) 52.57 (16.13) FSCSR-Inad 29.71 (4.79) 27.86 (4.95) 21.86 (3.89) 19.29 (5.65) 15.00 (10.38) FSCSR-Reas 13.57 (7.89) 14.00 (8.94) 15.29 (4.03) 18.71 (5.79) 16.00 (9.31) FSCSR-Hate 10.00 (5.42) 12.29 (3.25) 4.86 (2.12) 4.00 (2.38) 4.43 (5.44) OAS 48.71 (15.99) 44.43 (15.74) 31.57 (13.79) 24.71 (10.64) 20.14 (13.18) SCS-SF 28.29 (6.80) 30.14 (6.44) 32.29 (7.59) 37.29 (7.34) 38.57 (9.91) PBIQ-R 68.43 (9.95) 68.71 (6.80) 50.86 (9.49) 43.00 (12.25) 45.71 (12.63) RMSSD (ms) 0.05 (0.02) 0.04 (0.03) 0.04 (0.03) 0.06 (0.07) 0.04 (0.03) 1All measures and timepoints are n=7, except for PSYRATS-V (n=5 for all timepoints), PSYRATS-D (n=6 for all timepoints), and RMSSD (n=4 for T1, n=7 for T2, n=6 for T3, n=5 for T4, n=5 for T5). For RMSSD, all n=7 provided data at all timepoints, however some were excluded due to incorrect software setup (4 occasions) and poor ECG connection (4 occasions).

Table 8.7 shows the Wilcoxon Signed Rank Test results, which compared changes in group mean scores across the four different study phases, enabling testing of hypotheses 1, 2 & 3 as follows:

1) Significant changes in outcome and process measures will occur during the intervention, and not the baseline, phase.

The results showed significant improvements in six process measures (SocC, FSCSR-Inad, FSCRSR-Hate, OAS, SCS-SF, and PBIQ-R) in the intervention phase, none of which changed significantly in the baseline phase. Only two process measures did not change significantly (FSCSR-Reas and RMSSD), although both had improvement that were at trend level (p = .072 and p = .080 respectively). There were also significant improvements on most outcome measures (PSYRATS-V, PSYRATS-D, DASS-Dep, DASS-Str, and CORE), with the exception of DASS-Anx and DES-II, which were not significant in the intervention

148 phase. Again, the significant changes occurred in the intervention phase, and not the baseline phase, which shows no significant changes in any measure. Overall, these results provided support for hypothesis 1 at the group level of analysis.

2) Significant changes in process measures will precede changes in outcome measures

The majority of process measures that had significant changes in the intervention (T2→T4) showed their significant changes were already occurring within the first half (T2→T3). The only exception was the SCS-SF, which only registered significant change by the end of therapy. Among the outcome measures, three changed in the first half (PSYRATS-V, DASS-Str and CORE), and the remaining two changers (PSYRATS-D and DASS-Dep) were slower to improve. This result suggests differential patterns in when outcome measures were changing in relation to process measures. Some outcome measures changed more concurrently with changes in process measures (e.g. PSYRATS-V), but others changed after the bulk of process measures had changed (PSYRATS-D and DASS-Dep). To complicate matters, one of the process measures was late to change anyway (SCS-SF), so these results do not support hypothesis 2 with any great certainly, although there is some evidence of a pattern of processes preceding outcomes in relation to the PSYRATS-D and DASS-Dep outcomes.

3) Significant changes in outcome and process measures will still be evidenced after the intervention has finished

The follow-up results in Table 8.7 show that five of the process measures (FSCSR-Inad, FSCSR-Hate, OAS, SCS-SF, and PBIQ-R) and three of the outcome measures (PSYRATS-V, PSYRATS-D, and CORE) remained as significant improvements post-therapy. The one process measure (SocC) and two outcome measures (DASS-Dep and DASS-Str) that had lost their significant change by the post-therapy point, had only done so very marginally (p = .091, p = .063 and p = .061 respectively). Interestingly, the DES-II, which had not registered a significant improvement during the intervention phase (z = -1.352, p = .176) continued to improve post-therapy, at trend level (z = -1.859, p = .063). Overall, hypothesis 3 was supported for most measures, in that improvements were still evidenced 6-8 weeks after therapy.

149

Table 8.7. Wilcoxon Signed Rank Tests for group-level changes across phases Baseline phase 1st half phase Intervention phase Follow-up (T1→T2) (T2→T3) (T2→T4) (T2→T5) Z sig Z sig Z sig Z sig Outcome measures PSYRATS-V -.378 .705 -2.032 .042 -2.023 .043 -2.023 .043 PSYRATS-D -1.000 .317 -1.084 .279 -2.207 .027 -2.232 .026 DASS-Dep -.172 .863 -1.947 .051 -2.124 .034 -1.859 .063 DASS-Anx -.255 .799 -1.472 .141 -1.614 .106 -1.153 .249 DASS-Str -.412 .680 -2.032 .042 -2.371 .018 -1.876 .061 CORE .000 1.000 -2.028 .043 -2.366 .018 -2.366 .018 DES-II -1.018 .309 -1.185 .236 -1.352 .176 -1.859 .063 Process measures SocC -1.214 .225 -2.201 .028 -2.366 .018 -1.690 .091 FSCSR-Inad -.734 .463 -2.201 .028 -2.207 .027 -2.197 .028 FSCSR-Reas .000 1.000 -.314 .753 -2.371 .072 -.426 .670 FSCSR-Hate -1.625 .104 -2.371 .018 -2.371 .018 -2.201 .028 OAS -.847 .397 -2.028 .043 -2.371 .018 -2.366 .018 SCS-SF -.511 .610 -.762 .446 -2.028 .043 -2.117 .034 PBIQ-R .000 1.000 -2.371 .018 -2.366 .018 -2.371 .018 RMSSD (ms) -.365 .715 -.734 .463 -1.753 .080 -.674 .500 Significant results in bold

8.5. Discussion

8.5.1. Summary of results

The CFTp intervention was found to be feasible and acceptable to participants with distressing experiences (voices and/or delusions) in NHS community psychosis services. Nine out of 21 referrals (43%) were eligible and willing to participate in the study, although one person did not start the intervention due to work commitments. Only one of eight participants did not complete the intervention, and the seven who did complete, engaged with a full course of 26 sessions. The attrition rate of 1 in 8 (12.5%) is slightly less than 16% reported in a meta-analysis of CBT for psychosis (Lincoln, Suttner, & Nestoriuc, 2008). These seven participants were not only engaged in the therapy, but also as active collaborators in the therapy development process. All seven provided a complete set of process and outcome measures at all five timepoints, although eight (of 35) HRV recordings were later excluded due to technical issues. Only one participant opted out of providing weekly measures and audio recordings of sessions, and all participants opted out of video recording.

150

A series of both single-case-level and group-level analyses produced results that support the further testing of CFTp as an intervention option for people with psychosis. This study has found that CFTp can improve key measures of process and outcome, and that most improvements are clinically reliable and significant, compared to a baseline (no intervention) phase. The results suggest that, overall, CFTp was able to target successfully the intended social mechanisms (e.g. significantly reducing shame and self-criticism, and increasing self-compassion), with the desired impact on outcomes (e.g. significantly reducing voices, delusions, depression, and general wellbeing). The outcomes that showed reliable improvement for the most cases were depression (5/7), stress (5/7), general wellbeing (5/7), anxiety (4/7), and voices (3/5), and when the whole group were combined, the outcomes that showed significant improvement were depression, stress, general wellbeing, voices and delusions. It cannot be concluded from the results whether changes in social processes preceded these outcome changes, or whether these occurred concurrently. However, there was some evidence that self-compassion (a process) and delusions and depression (outcomes) were slower to change than other measures. The results also provided evidence that most of the improvements brought about by CFTp were still apparent 6-8 weeks after the intervention had finished. The single-case results showed that not all participants benefited to the same extent. Some had measures that did not change significantly, and on a small handful of occasions, that changed significantly in the opposite direction. However, overall, it can be concluded that CFTp shows sufficient feasibility, impact and promise to warrant further testing in a pilot trial.

8.5.2. Acceptability of CFTp

The acceptability of this novel intervention was the most important finding of this study. The kinds of techniques employed by CFTp are not standard techniques that participants would have been likely to encounter before in mental health services, or even heard about. Indeed, many of the CFTp techniques may have been in direct contradiction to how psychosis service-users had been accustomed to relating to distressing experiences. For example, the technique of compassionate engaging with voices is in a stark contrast to some ways of relating to voices that are traditionally advanced by services. Historically, the primary aim of psychosis treatment has been to reduce or eliminate psychotic experiences (e.g. with antipsychotic drugs), viewing these as the undesirable symptoms of a disordered brain. The CFTp approach encourages an alternative way of relating to these experiences, moving away from avoidance and suppression, to more compassionate listening and empathy, where voices might instead be considered as having possible emotional functions (e.g. protective), and possibly holding useful information about their emotional concerns. As participant P2

151 pointed out, this concept of compassion applied to distressing voices may have initially been hard for people to accept:

“I know it’s hard to accept that to love and care for these evil voices is the correct way to go. But think about it, by giving this caring love to them you are really giving yourself this love. The voices are you, they are not external they are part of you. Be compassionate to yourself.”

P2, quoted in Heriot-Maitland and Russell (2018)

CFTp also requires participation in experiential practices, e.g. with breathing rhythm, compassionate imagery, and body posture. Again, these are concepts that might be new to people, and therefore make the acceptability findings of this study even more important. The fact that seven participants were engaging with such exercises during sessions for many months, amid distressing voices and delusions, is highly promising for the future development and research of CFTp in this population. The one participant who did not complete therapy did not report any issues with the therapy content or procedure itself; the attrition in this case was due to relapse in a pre-existing alcohol condition. It was not clear whether or not the therapy had some role in triggering an alcohol relapse, e.g. by eliciting distressing content, or whether there would have been a relapse without therapy. Nevertheless, overall, this study demonstrated good acceptability and feasibility of applying CFT psycho-education, processes, and interventions for people with psychosis.

8.5.3. Single case variability in results

At the single-case level, it was possible to distinguish which participants were most helped by the intervention, and which were not, or less so. Five participants (P2, P3, P4, P5, and P6) showed multiple improvements across multiple outcome measures, which accounted for the significant changes detected at group-level. Two participants (P1 and P7), however, had only one or two outcomes that improved, along with one or two that deteriorated. For example, both P1 and P7 reported an increase in anxiety over the whole study period from starting therapy to follow-up (T2→T5), even though, like the rest of the group, a number of their process measures had improved over that same period, and their scores on voice-related and delusion-related distress had reduced.

There were notable differences in anxiety and psychotic symptom results between the single-case and group analysis. Although anxiety improved reliably for four of the seven cases (single-case analysis), it did not reach significance at group level. This is likely due to the fact that anxiety also deteriorated reliably for one case (P1), which affected the group mean. Regarding psychotic symptoms, the single-

152 case analysis found more cases with reliable improvement in voices (three of five) than delusions (two of six), whereas in the group level analysis, the significant improvement for delusions was of a higher magnitude than for voices. This can be made sense of through closer inspection of the Reliable Change Indices (RCIs) (Appendix 8.10), where it can be seen that for the remaining four cases for whom delusions did not improve reliably, the RCIs were only marginally short of the reliable change threshold of -1.96 (their RCIs were -1.63, -1.90, -1.63, and -1.90).

The results from session-by-session measures showed that, overall, CFTp produced an improvement in experiences of social safeness over therapy; however, again, the single case plots and analyses show that this effect was being caused by three cases with significant improvements relative to baseline (P2, P3, P4), one with a large improvement that was not significant relative to baseline (P5), and one with a steady, but small, improvement, not significant relative to baseline (P6). One case (P7) reported social safeness levels that actually improved less in the therapy than in baseline, registering a negative trend. So while it can be concluded that, overall, CFTp was producing the intended changes for most, there were still one or two exceptions. This of course is normal for mental health treatments, where it is unlikely that a treatment can help everyone, and especially so for treatments with a complex population at the more ‘severe and enduring’ end of mental health.

The discussion will now consider possible reasons for less successful responses to CFTp on the outcome and sessional measures used. Gareth (P7) had a range of diagnoses, namely Paranoid Schizophrenia, Tourette’s Syndrome, Autism, and Obsessive-Compulsive Disorder. He had been in mental health services for 35 years, had spent most of his adult life in supported accommodation, and was accompanied to each session by a member of his full-time residential support team. As the therapy sessions progressed, one aim was for Gareth to spend more time alone with the therapist, which he was able to do, so long as the support worker was sitting outside. However, this did cause anxiety for him, and a lot of the work was about managing that separation anxiety. By the follow-up timepoint, Gareth had shown a reliable reduction in self-criticism (both inadequate and hated types), as well as in shame and personal belief about illness. These were positive shifts in his social ranking patterns but did not seem to translate into positive shifts in outcomes. Although his depression score had reduced during therapy, but not at follow-up, his anxiety score had increased at follow-up. Within the therapy period, Gareth’s anxiety was on a trend (not significant) of reducing from 8 (start) to 6 (mid) to 5 (end); however, it then shot up to 14 at follow-up. This may have been partly to do with a separation anxiety (upon separation with the therapist), similar to his previous experiencing of separation with his support staff. The anxiety was unlikely to be related to social rank threats (which

153 had improved) or to distressing delusions (his distress scores had also improved, as displayed in Figure 8.2). It may be that this was a ‘new’ anxiety actually caused by the therapy ending. Given Gareth’s attachment anxieties, and accompanying cognitive difficulties, the completion of 26 sessions of therapy was itself a major achievement. The clinical team had reported that multiple interventions had been tried over Gareth’s many decades of mental health service involvement, and that this CFTp intervention had been a marked difference in terms of his engagement. Gareth’s Tourette’s tics were barely noticeable by the end of therapy, but this was not measured as a study outcome.

The other participant who did show some positive shifts in social rank patterns, but not in outcome measures, was P1, Steve. This man had experienced multiple repeated traumas in his life from a very early age, including childhood sexual abuse, and then had multiple repeated exposure to violence, sexual abuse, and torture. The clinical impression working with Steve was that 26 sessions of therapy was not nearly long enough. It took 26 sessions just for Steve to feel comfortable and safe sitting in the therapy room, e.g. without having to check for hidden cameras and wires. Steve did not agree to providing session-by-session measures, but the clinical impression was that he did eventually start showing signs of safeness experience towards the end of therapy. Steve did provide a single SSPS score at the follow-up timepoint (score 24), which showed that his social safeness rating had doubled since the start (score 12). Although this is a clinical judgement, the therapist’s impression is that Steve could have benefited from CFTp if the intervention had been extended for longer (e.g. a further 20-26 sessions). This is the only participant for whom this conclusion was reached. For Gareth (P7), it was felt that an extension would not have made much difference, and for the other five cases (P2, P3, P4, P5, P6), it was felt that 26 sessions was adequate to produce desired outcomes. The overall recommendation, therefore, is that CFTp courses should be offered as 26 sessions standard, but with an option of extending for cases with specific clinical justification.

8.5.4. Limitations

One limitation of this study is that the CFTp therapist was not only the developer of the intervention, but also the researcher, which has potential for evaluation bias. Having non-blinded clinicians collecting their own data is a serious limitation, but this was all that was practically available, as this study was conducted in the context of a PhD Fellowship, without funding and resources for additional research staff. However, the research procedures were designed to minimise bias; for example, self- report measures were used as far as possible, and all measures were adapted into electronic format so that participants could work through the questionnaires without researcher involvement. The

154 researcher was on hand to answer queries, but was not actively involved, and was not able to see any of the answers as they were being entered on sequences of webpages on the laptop screen. The only measure that involved a degree of (subjective) interpretation from the clinician/researcher was the PSYRATS; however, the PSYRATS items have enough guidance in the questions to elicit a more objective, quantitative answer from participants anyway, and it was the participant’s own assessment of their delusions and voices that was the main guide for scoring.

Another limitation is with regard to the RCI analyses, where for a number of measures, the normative data (Appendix 8.9) was obtained from non-psychosis populations. Normative data would ideally be well-fitted to the study sample, however, a number of measures had either not previously been used at all with psychosis samples, or only in very small samples without reliability statistics. PSYRATS, DES- II, and PBIQ-R were the only measures widely used in psychosis research, and normative data from psychosis samples was obtained for PSYRATS and PBIQ-R. In the case of DES-II, however, the current study used an adapted ‘state’ measure (‘in the past week’), whereas the existing psychosis research has typically used DES-II as a trait measure. In all cases where normative data could not be validly derived from the published literature, the preference was to calculate normative data from the large baseline sample (n=544) in Study 2 of the current thesis. The possible limitations of not having well- fitting normative data for some measures must be considered in interpreting the single case (RCI) results, and an implication for future research is to establish normative data for these CFT-relevant measures in psychosis populations.

The loss of some Heart Rate Variability data (8 of 35 readings, or 23%) was a further limitation to this study, and also potentially has implications for how these data may be collected differently in future studies. There is a variety of hardware options for collecting heart rate readings, ranging from electrode-based (ECG) devices, to ear-clips, to chest straps and wrist straps. The initial reason for choosing an ECG measurement option was for accuracy, even though ECG procedure was more complex than alternatives, and required expert assistance from clinically trained nurses. However, in the last few years, since this study was designed, technology has advanced such that some of the more accessible, ‘wearable’, devices have not only become cheaper and more widely available, but have also been found to have improved accuracy, comparable to that of classic ECG (Georgiou et al., 2018). There have also been technological advancements in the software for converting heart rate readings into RMSSD measurements. This conversion can now be done more automatically, with cheaper and user-friendly smartphone apps. In the current study, data were lost due to poor ECG connections (four readings) and software set-up issues (four readings), which would both potentially be eliminated or

155 reduced with wristbands and app software. Future CFTp research might consider ‘wearable’ devices as preferred alternatives to ECG, without sacrificing accuracy.

Another limitation in this study is that only five audio recordings were sent for fidelity checking against the CFTp-ACM. This was due to time constraints and limited availability of experts, among a very small pool of people with expertise in CFT for Psychosis. Given these practical restraints, the selection of audios was chosen to maximise the breadth of coverage for fidelity checking (i.e. to ensure that one audio was rated from each of the five therapy levels, with a different participant to represent each of these levels). This is a limitation more with regards to the assessment of manual adherence, rather than therapist competence, because, as there was only one therapist in the study, five audios were considered adequate for the assessment of one therapist’s competence. An improvement for future studies would be to rate a greater number of audios for manual adherence.

8.5.5. Conclusions and implications for future research

CFTp is a feasible and acceptable intervention for people with distressing experiences in psychosis. There were enough indictors of effectiveness to warrant further investigation in the form of a pilot Randomised Controlled Trial, with the aim of establishing the design parameters for a full Randomised Controlled Trial. In the next stage of evaluation, it is important for other therapists (not the developer) to provide the interventions, and for independent, blinded researchers to collect and analyse the measures. Most of the process and outcome measures used in this study were found to be feasible and were sensitive to change in response to the intervention over time. For a pilot trial, researchers may need to reduce the number of measures used to reduce the chance of Type 1 errors, and therefore to improve the validity of analyses and evaluation conclusions. One suggestion would be to focus on the PSYRATS distress scores, rather than total score, as a primary outcome measure. The visual inspection of PSYRATS distress (Figure 8.2) showed clear improvement trends in all seven cases, and this outcome would be more closely aligned to the specific target of CFTp, which is primarily concerned with shifting the threat-based emotional relationship with psychotic symptoms. This study also supports the inclusion of depression as an important outcome measure in CFTp, as this showed good improvements. This is consistent with the findings and recommendations of Braehler et al. (2013) in their pilot evaluation of group-based CFTp. There are no specific recommendations about which measures to exclude, as all are relevant to CFTp, and each measure showed group-level trends in the expected directions, which may have registered as statistically significant changes with a larger

156 sample size. However, one suggestion for future studies would be to select measures that have developed by different research groups. In this study, four measures (SSPS, SocC, FSCSR, and OAS), were developed by Gilbert’s research group, and future testing could benefit from assessment tools derived from a greater variety of sources, which would add to the external validity of the research.

To conclude, this study shows promise for CFTp as a treatment option for people experiencing distressing voices and/or delusions in the context of a psychosis-related diagnosis, and warrants progression to the more robust methodology of an RCT.

157

Chapter 9. Discussion

9.1. Thesis summary

9.1.1. Summary of aims and objectives

This thesis aimed to advance the research field in understanding pathways to psychosis through three studies: through 1) questionnaire development with utility for longitudinal research in the general population, particularly among those with high incidence of PLEs; 2) testing interactions between dissociative and social mechanisms in pathways to psychosis longitudinally; and 3) characterising a novel ‘non-clinical-in-crisis’ population, considered to be of particular interest for future research into dissociation-social interactions due to different social experiences around PLE ‘crisis’.

The second overarching aim of the thesis was to advance psychological interventions for people with psychosis by developing a novel intervention based on social evolutionary theories (social rank and attachment theories) that is acceptable and helpful for people with distressing psychotic experiences. To improve the acceptability of the intervention, the therapy manual was developed through engagement and collaboration with experts by experience and clinicians. A case series study with NHS patients with distressing experiences was carried out to provide initial acceptability and initial indications of outcome.

9.1.2. Summary of main findings

Study 1 analysed cross-sectional data from 532 online participants to select items for, and psychometrically evaluate, a new scale called the Transpersonal Experiences Questionnaire (TEQ). This 19-item scale was found to be unidimensional, with satisfactory internal consistency (0.85), and with reliability for high scorers, i.e. people at the upper-end of the PLE continuum.

Study 2 analysed longitudinal data from 314 participants, finding evidence for an interaction effect of dissociation and shame on PLEs 6 months later, despite there being no significant direct effect of dissociation on PLEs. A three-way interaction analysis found that social safeness moderated this dissociation-shame interaction, but contrary to the hypothesised direction of this effect, higher social safeness was associated with more pronounced dissociation-shame interaction effects on PLEs.

158

Study 3 characterised a sample of 30 participants from a spiritual support-seeking population, finding that participants had high incidence of PLEs, significant trauma histories, high levels of dissociation, and significant contact with mental health services in the past. Correlational analyses found that higher incidence of PLEs was related to more distress, and that PLEs were rated as more ‘enriching’, ‘safe’ and ‘pleasant’, than ‘distressing’ and ‘disruptive’. The more positive PLE dimensions were correlated with spirituality, post-experience growth, and social safeness, while the negative PLE dimensions were correlated with depression.

Study 4 tested the acceptability and usefulness of CFT for Psychosis (CFTp). Seven out of eight participants completed the full 26 sessions, which demonstrated acceptability of this intervention. Single-case level and group level analyses indicated that it was possible to target key identified processes with the intervention, with desired outcome changes, such as improvements in depression, stress, wellbeing, voices and delusions. There were different patterns of change across individual cases, but overall there was support for the further testing of this intervention in a pilot RCT.

9.2. Strengths and limitations

9.2.1. Strengths

The breadth of this thesis reflects the breadth of research activities needed to advance the treatment of mental health conditions. To help people in mental distress, we need to understand the factors and mechanisms that contribute to their mental distress (ideally, across each of the bio-, psycho-, and social levels). In order to investigate and identify these mechanisms, we need appropriate research designs, with suitable research tools and relevant recruited samples. Once identified, these mechanisms can be translated into therapy targets, to inform therapy development in a way that is both acceptable (to its users) and amenable to research testing.

This thesis has encompassed the full range of these different elements: from designing studies and measurements, through identifying samples and mechanisms, to developing and testing therapies. Although there are limitations within each of the four studies, in general terms, the aims and objectives were successfully met, with robust designs, methodological rigour, and with implications to advance the field of psychological approaches to psychosis.

159

As one of its outputs, this project has produced a new measure, the Transpersonal Experiences Questionnaire (TEQ), that is valid and reliable, and has measurement precision for people in the general population at the upper end of the psychosis continuum. This tool could be useful for researchers using longitudinal research designed to establish causal relationships between social and psychological processes in psychosis. It could also be used in comparison studies of clinical vs non- clinical populations with PLEs (i.e. with and without a ‘need for care’, Peters et al. (2016)).

This thesis also identified interesting interaction effects between dissociation and shame that have important implications for the fine-tuning of psychological models. Specifically, the longitudinal research design in Study 2 has shown that dissociation (alone) is not enough to explain the pathway to PLEs, but that an interacting dissociation-shame model may be needed. This not only has implications for the design of future studies but also has implications for the re-interpretation of previous research linking dissociation and psychosis.

Even more intriguing interaction patterns were produced in the three-way interactions between dissociation, shame, and social safeness. It was expected that high social safeness would reduce the influence of shame on dissociation in producing PLEs (by a direct social buffering/regulating effect, e.g. Hostinar et al. (2014)), but the results demonstrated that experiencing high social safeness heightened the influence of shame on dissociation in producing PLEs. One explanation might be that for someone who feels highly socially engaged, connected and attuned but at the same time is experiencing other people as shamers, it is likely that this social shaming may be experienced as more threatening for this person than for someone else who is socially disengaged / disinterested. Another possibility is that this pattern of scoring could be tapping into disorganised attachment. For instance, drawing from models of attachment and dissociation (see Chapter 2), it may be that participants with unresolved trauma could display the kind of scoring shown in the three-way interaction plots (Figure 4.3C), namely high social safeness, high dissociation, high shame, and high PLEs. People with disorganised attachment often lack a coherent narrative about their social world, for example, may experience their key relationships as sources of both safeness and threat at the same time (Liotti, 2004). One of the strengths of this thesis is that it advances the field by drawing attention to the importance of interacting dissociative and social processes in pathways to psychosis and poses some interesting questions about these interactions for future investigation.

Another strength of this thesis is its thorough and transparent reporting of the intervention development process (Chapter 7). Historically, this stage of the process (i.e. the stage between having

160 a theory of intervention and translating it into an actual intervention manual) has either been left out, not reported, or at best, under-reported (Hoddinott, 2015). The reporting and publishing of this stage of the research will hopefully become more commonplace going forward with the continuing drive towards research transparency.

The level of knowledge exchange and engagement with people with lived experience, and with the broader general public is a level of engagement that is not typical for such an early stage in the development of a new psychological approach. However, this early stage is arguably the most important point to be engaging with different people’s views and feedback; i.e. before the more costly, larger-scale evaluations commence. Most engagement activities were funded externally, which may be one factor in why this level of engagement is not very common, especially within a PhD project. For example, to make the Compassion for Voices video (Cultural Institute at King's, 2015), funding had to be secured from four independent sources. Although challenging in terms of financial and time- commitment, the engagement activities greatly enhanced the intervention development, and therefore contributed significantly to the second main aim of the thesis; i.e. to develop an acceptable and useful intervention for people with psychosis.

The ethos of sharing ‘ownership’ of the intervention with people with lived experience (described in Chapter 7) is a strength. This approach was taken in alignment to the central CFT process of fostering affiliative experience and was also modelled on the ethos of the Hearing Voices Network. While it would not have been possible to give ownership of CFTp to quite the same extent as in the Hearing Voices Movement (as it was developed within the context of an academic degree by someone without lived experience), there were certainly considerable efforts made to make this as close as possible, for instance, multiple examples of co-production (videos, therapy materials, events, training workshops), as well as co-authored publications with a case series participant (Heriot-Maitland & Russell, 2018) and a collaborator with lived experience (Heriot-Maitland et al., 2019). Although this is subjective, and not a research finding, it was felt that the co-production elements of CFTp contributed to the actual targeted mechanisms of CFTp, namely the fostering of affiliative, co-operative and compassionate social mentalities.

9.2.2. Limitations

This section concentrates on limitations for meeting the overarching aims of the thesis.

161

Online data collection (Studies 1, 2 and 3): Online data collection allows the recruitment of large numbers of participants (Studies 1 and 2) and from groups of participants that are hard to recruit (Study 3, a novel population). The achievement of Study 2 to recruit 314 participants for a longitudinal study (i.e. providing data twice) would have been extremely challenging without the help of online tools, and impossible within the timeframe of a PhD. But potentially online data collection may compromise data quality. It was not possible to monitor the time, effort or concentration of participants. Were they thoughtfully choosing their answers or selecting any answer at random? These are the kinds of issues that can at least be mitigated (although not eliminated) in a face-to-face interview. However, face-to-face also potentially introduces other biases, such as pleasing the researcher, which would not apply as much to data provided online and anonymously. The steps taken to improve the data quality were giving an estimated completion time in the recruitment advert and keeping the battery of measures to a minimal length, with simple short questions. The only measure that required a fair amount of reading was the Revised Dissociative Experiences Scale (DES-II) (Carlson & Putnam, 1993), but to answer each DES question, participants had to open a ‘drop-down’ box to select their answer, meaning they couldn’t race through selecting answers at random (as you could with a ‘radio buttons’ layout). Future studies could improve how online data quality is monitored, for example by recording start and stop times, and excluding responses that were completed in an unrealistically quick time. Other options for checking data quality include asking the same question twice, or the same question with slightly different wording, or the exact reverse/opposite of the question. The other online method used was Skype (for Study 3 interviews), which did not seem to compromise the quality of data collected.

Potential non-representation of the participant population: Some samples were not representative of the population for whom the thesis was ultimately aimed to help; i.e. people with distressing psychotic experiences. The psychosis continuum model (van Os et al., 2009) justifies the widely adopted strategy of investigating psychotic phenomena in the general population. However, the psychosis continuum is not without its critics (David, 2010), and it is important that any conclusions reached in this thesis are not automatically assumed to apply to people struggling with clinically diagnosed psychotic conditions. The sample in Study 1 specifically excluded people with psychotic disorder; in Study 2, there was only a small number (1%); in Study 3 the proportion was higher (30%); and, of course, everyone (100%) in Study 4 had psychosis-related diagnoses. It is important that rigorous testing of all these findings are carried out with samples from psychosis populations before claims are made about psychosis and its mechanisms.

162

That nine people (30%) in Study 3 reported previous contact with mental health services for a psychotic disorder was a limitation of this study in meeting its aims. The aim was to identify a group who were seeking support for their PLEs outside of mental health services. The sample were indeed currently seeking support outside of mental health services (i.e. they had contacted a spiritual support organisation within the last 2 months), but what was not anticipated was that many had previously been in mental health services (and perhaps in some cases, rejected them), and were now seeking alternative avenues of support within a spiritual framework. Data were not collected on the time of the mental health contact or diagnosis, so it was not clear how recent (or even current) these service contacts were. Nonetheless, the very fact that there was service contact does still confound the social experiences of the sample around their PLEs. Even if the nine participants had now rejected or disputed their ‘psychosis’ labels, they could still potentially have experienced some of the social stigma or shame around these diagnoses at the time of treatment. And of course, the same is true for supportive social experiences, in that they could have experienced beneficial support from services (although this is unlikely judging by their ‘helpfulness’ ratings and comments about services in section 5.4.1.). Either way, a limitation of Study 3 is the inability to recruit the ‘confound free’ sample that was ideally hoped for in order to investigate social aspects of PLEs outside of services. In retrospect, however, it would have been highly unlikely to find a sample with no exposure at all to psychiatric or pathologizing narratives around PLEs. Indeed in an earlier qualitative study of clinical and non-clinical groups with PLEs (Heriot-Maitland et al., 2012), we found that every single participant, in both groups, reported some ‘awareness’ of the pathologizing views and narratives of others around their PLEs.

One suggestion for future studies into social mechanisms in this ‘non-clinical-in-crisis’ group would be to tighten the inclusion/exclusion criteria; for example, excluding people with previous contact with mental health services (generally) or with mental health services for psychosis specifically. Another suggestion for future research would be to follow up this group longitudinally, to see how social experiences evolve and change over time in relation to their PLEs, and whether this has an impact on outcomes.

Transferability of the results: The clinical case series was a small sample, which although appropriate for a feasibility case series study, is still very limited in terms of what it can tell us about the effectiveness of CFTp. The case series tells us mainly about acceptability, but even then, only to a limited degree, as only eight people experienced the therapy. The next stage would be a pilot RCT. The headline finding from this study is that no-one dropped out due to the unacceptability of CFTp.

163

The only case (of eight) who did not complete CFTp was stopped by a clinical decision of the therapist (not the participant) to switch to offering support for accessing alcohol detoxification services.

Another limitation of the case series is that it was not designed as a targeted intervention study, and therefore not in keeping with the recommendations outlined in Chapter 1 for both delusions (Freeman, 2011) and voices (Thomas et al., 2014). Indeed this case series did precisely what Thomas et al. (2014) warned against in their concerns about the influx of novel therapies with their own “brand naming” possibly moving the field towards “yet more broad therapy approaches for broad populations, as opposed to research targeted at understanding processes” (S206, Thomas et al. (2014)). However, an argument in response to this valid concern is that this CFTp study is targeting a specific process, despite recruiting a mixed sample (distressing voices and/or delusions). In the CFTp model, the same mechanisms are purported to be underpinning voices and delusions (the activation of threat-based social rank mentalities) and the same processes are targeted as therapeutic solutions for each (switching to mentalities of caring, affiliation, compassion, or in attachment language, creating internal experiences of safe haven and secure base). So, in that sense, CFTp is very much a targeted intervention, guided by a good prior understanding, and seeking to further its understanding of key processes. Hence why at this early stage of development and evaluation there are strong theoretical grounds for testing targeted CFTp processes for people with different types of distressing psychotic experiences. In the results, there are improvements found for both delusions and voices, which supports the suggestion that both are successfully targeted. However, there are also differential patterns in the results, which could have implications for future developments and testing of CFTp. This and other implications are discussed in the clinical implications section below.

Therapy adherence and competences ratings: Only five 60 minutes therapy recordings were sent for rating by an external CFTp expert, which would need to be increased in future studies to assess whether the therapist was delivering CFTp. In the current case series, this was slightly less relevant because one of the aims was to continue developing the therapy manual throughout the case series period (2015-2018), as more was learnt from the NHS participants and other knowledge exchange partnerships (Chapter 7). Indeed, the adherence and competence rating scale itself was being developed throughout this period. Hence, at this stage of evaluation, the questions around adherence measurement were more around the acceptability / usability of the rating scale (CFTp-ACM, Appendix 8.7). The five audio recordings were specifically chosen to represent each of the five different levels of CFTp, and the indications were that the external CFTp expert was able to use the CFTp-ACM to rate

164 therapist adherence and competence across each of these levels. Recommendations about the use of CFTp-ACM in future studies are made in section 9.4 below.

9.3. Implications for theory of psychosis

In general, the results support the use of interacting mechanism models in understanding psychosis. This is most closely aligned (theoretically) to the recently developed Cognitive Attachment Model of Voices (CAV) model (Berry & Bucci, 2015; Berry et al., 2017), which describes dissociation x attachment interactions on the pathway from trauma to voices. However, there are two important differences: firstly, this thesis did not take a single symptom approach; and secondly, the thesis findings point to the inclusion of an additional interacting mechanism on the pathway (social rank threat), and thus advocates a three-way interaction approach (dissociation x social rank x attachment).

9.3.1. Targeted approaches, but targeting what? A symptom, a mechanism, or a function?

The main outcome measure, the TEQ, was found through factor analysis to be a unidimensional scale of PLEs, meaning that its 19 items were considered to be various different expressions of a single PLE continuum. Therefore, far from adopting a ‘single symptom’ approach, the focus of this thesis was in considering outcome as a range of experience along a PLE continuum. In the TEQ, there is one item that specifically taps voice-hearing (item 18), one that taps paranoia (item 10), and then a range of other items that would also be very familiar to anyone who has met or worked with people with psychosis, or has lived experience themselves; for example, perceiving “messages or hints” in the world (item 8), feeling compelled towards a “mission” (item 6), and experiencing “loss of individual identity” (item 17). This focus on a range of PLE expressions immediately sets this thesis theoretically apart from some recent literature in the psychosis field. This section will outline different ways of taking a ‘targeted approach’ in research and will consider why taking a targeted approach towards mechanisms and functions, rather than towards symptoms per se, may have some benefit for future theoretical developments in the psychosis field.

In recent years, the psychosis literature has made strides towards becoming more targeted (Bentall et al., 2014); and for most psychosis researchers, taking a targeted approach has been interpreted as identifying and researching single ‘symptoms’ of psychosis. This has rapidly produced two major streams within the literature: one modelling pathways to voices, and one modelling pathways to paranoia. In Bentall et al.’s (2014) review of pathways to specific symptoms, they specify thought

165 disorder first of all, and then, within psychotic experience, they focus entirely on voices (one type of perceptual experience) and paranoia (one type of ‘delusion’). But what about the many experiences that have different perceptual aspects? What about other (non-paranoid) thinking styles, and other types of delusion?

If the goal is to take a ‘targeted approach’, another approach would be to identify and research a specific mechanism (rather than symptom) as the starting point. So, for instance, starting with a mechanism like dissociation, and then looking at the various functions and processes around dissociation and how this manifests in psychosis. Symptoms themselves are not the phenomena that define the theoretical/research focus; they are secondary to the mechanism. The ‘targeted approach’ is towards the mechanisms themselves, not the symptoms elicited as a consequence of that mechanism. In reviewing the dissociation and psychosis research fields, Dorahy and Green (2019) make the point that focusing on either symptoms or mechanisms (as starting points) has been one of the main distinguishing features of the two fields. They observe that in the psychosis field, research has focused on “specific symptoms as a state or an outcome of abnormal threat‐related cognitive processes”, whereas in the dissociation field, research has focused on dissociation “as a process in itself that reduces the impact of threat.” (p 284). They state that whereas psychosis researchers ask, “What cognitive mechanisms underlie, or lead to, psychotic symptoms?”, dissociation researchers ask, “What cognitive mechanisms are associated with the process of dissociation?” (p 284, Dorahy and Green (2019)).

By this analysis, the thesis would be more aligned to what has traditionally been the focus of the dissociation field, and one of the main theoretical implications of this thesis is that importing the dissociation research model into the psychosis research may be a fruitful way forward for future modelling of psychosis. In this thesis, the focus is on the dissociation mechanism and the (social) processes that influence and interact with this mechanism. The TEQ outcome measure captures a range of different experiential manifestations of this interaction. As such, having a higher score on the TEQ signals that there is a greater amount of interacting-mechanism activity taking place, which, in this thesis, is a more important finding than the precise ‘content’ of these expressions; e.g. whether these are PLE experiences of energy passing through the body (item 7), picking up on other people’s thoughts (item 9), feeling one’s movements are being externally controlled (item 15), or hearing a voice when no one is around (item 18).

166

As Dorahy and Green (2019) point out, one of the hallmarks of the dissociation literature is that understanding the dissociation mechanism is very much pinned to an understanding of its functional role (in reducing the impact of threat). In the psychosis literature, if we did let our research streams become too aligned to single symptoms, there might be a danger that we lose sight of the common functional themes that may sit behind different symptoms. And if we lose sight of the common functional themes, how would this impact on treatment? As Hunter (2014) remarks, “psychosis and dissociation both stem from the same underlying processes at attempting to deal with overwhelming life experiences. And this is where treatment should be focused”.

What is useful about an evolutionary approach to psychology and psychotherapy is that an evolutionary context is very good at helping us identify underlying functional themes. Evolutionary psychology suggests that our psychological mechanisms are designed to promote evolutionary goals (ultimately survival and reproduction), rather than say, our happiness or mental health (Gilbert, 2002). This brings into focus questions about function, which tie in much more closely with understanding the mechanism itself (the process), rather than the symptom (the manifestation). Evolutionary functional analysis in psychosis has highlighted some common themes of mechanisms across symptoms; for example, the theme of social rank threat in both paranoia and voices. In their evolutionary model of paranoia, Raihani and Bell (2019) suggest that paranoia may be linked to the evolution of how people detect and respond to social threats. They specifically focus on the role of ‘coalitional threats’, which refers to how people organise themselves into coalitions to manage threats from competitive rivals (e.g. lower-ranking people may form coalitions to overcome higher-ranking competitors).

In the context of the wider literature, the findings of this thesis are more theoretically positioned to contemplate questions about mechanisms, the function of mechanisms, and the interaction of different mechanisms, in producing a broad spectrum of PLE expression. In this thesis, the PLE expression is considered unidimensional, as measured by the TEQ, and therefore the focus here is not about differentiating between different versions of PLE expression, but on which mechanisms (and interactions) lead to greater or lesser amounts of overall PLE expression. This primary focus on mechanism and mechanism function is more aligned to approaches of the dissociation literature and the evolutionary psychology literature, than to the psychosis literature. This does not mean it is not ‘targeted’, it is just targeted on mechanism and function, rather than on symptom. The main implication for future theory and research is to be cautious about jumping too quickly onto the ‘single symptom’ bandwagon, and to instead consider mechanisms and functions, which may be shared

167 across different PLEs, as being the target for scientific enquiry. One possible function, as the dissociation literature tells us, is that dissociation, and by extension PLEs, may be psychological strategies with the function of managing threat.

9.3.2. A three-way interaction approach

This thesis provided evidence for both two-way (dissociation x shame) and three-way (dissociation x shame x social safeness) interactions in influencing the amount of PLEs six months later. Importantly, while the interacting mechanisms produced an effect on PLEs, dissociation mechanism (alone) did not. This supports the approach taken by Berry et al. (2017) in developing integrated, interacting mechanism models of pathways to psychosis, and also has implications for how psychosis modelling might be further fine-tuned and developed going forward. Of particular importance is the implication of these findings in modelling social influences in the pathways to producing PLEs themselves, rather than what has traditionally been done, which is to model social influences on processes occurring after PLEs; e.g., the social influences on the appraisals of PLEs (Garety et al., 2001), the social consequences of psychosis in terms of stigma (Corrigan et al., 2009), shame (A. C. Watson et al., 2007), and internalised stigma (Pyle et al., 2015a), and the social influences on relapse in psychosis in terms of expressed emotion (Bebbington & Kuipers, 1994).

Figure 9.1 is a conceptual diagram that shows the various points along the psychosis pathway where social mechanisms may play a role, and which may therefore be of interest to investigators. These points are marked with grey circles (A to D). Circles C and D are the areas where social influences have typically been researched in the past, i.e. in influencing whether PLEs lead to clinical vs non-clinical status (C), and in influencing the outcomes of PLEs, in terms of emotion and functioning (D). These two areas have also been the focus of author CHM’s own research prior to this thesis, published in Heriot-Maitland et al. (2012) (C) and in Brett et al. (2014) (D). Circle D was also the focus of Study 3 (Chapter 5) in this thesis, which found that different social mechanisms were associated with different PLE-related dimensions (‘distressing’, ‘pleasant’, ‘safe’, ‘disrupting’, ‘enriching’) in a group of people with high levels of dissociation and significant trauma histories. The relatively new areas, which this thesis draws attention to, are circles A and B. These are the areas where social experiences are hypothesised to have influences on activating (A) and exacerbating (B) dissociative processes in the pathway to producing PLEs. Evidence for the activation of dissociation by social processes exists in relation to both social rank/shame determinants (Dorahy et al., 2017) and attachment/caring determinants (Dutra, Bureau, Holmes, Lyubchik, & Lyons-Ruth, 2009). New evidence for the

168 interaction of dissociation and social processes in producing PLEs comes from Study 2 of this thesis (Chapter 4).

Figure 9.1. A model of dissociation and social mentality (social rank x attachment) interactions in the pathway to psychotic-like experiences (PLEs)

Inter-personal threat

Other1 Other 2 Other3

Dissociative Social rank / Attachment / traits shame patterns caring patterns

Social influences on A dissociation activation 1

Dissociation

activation

Social influences on B dissociation → PLEs

PLEs

Social influences on C PLEs → C v NC status

Clinical Non-clinical

Social influences on

D PLE outcomes

PLE outcomes Ongoing shaping of social mentalities through

(emotional / functioning) new bio-psycho-social contexts & experiences

1Other determinants of dissociation, e.g. trauma, drug-use, existential crisis/impasse; 2Other determinants of social rank, e.g. bullying, criticism, stigmatising; 3Other determinants of in/secure attachment, e.g. availability / attunement of care-giving, n urturing, safeness

Figure 9.1 also models how the social influences are mediated by a combination of both social rank and attachment patterns, which, in the case of voices, is a proposed addition to Berry et al.’s (2017) CAV model. The suggestion from this thesis is that a) social rank patterns will have an additional influence to that of attachment patterns on dissociation, and that b) this three-way interaction

169

(dissociation x social rank x attachment) will not just be implicated in the pathway to voices, but in the pathway to PLEs more generally. The model in Figure 9.1 also illustrates how social rank/shame patterns and attachment/caring patterns, both social mentalities, will be continually shaped through ongoing social experiences and interactions with the environment. The potential for shaping these social mentalities is of course the cornerstone rationale for CFT interventions, such as that tested in Study 4, which specifically aim to de-activate and ‘tone down’ patterns of social rank/shame, and to activate and ‘tone up’ patterns of secure attachment and caring motive.

9.4. Implications for psychosis interventions

This thesis has not only advanced the theoretical rationale for applying CFT for psychosis through its specific targeting of social mentalities, but it has also made significant steps towards the development, testing and evaluation of CFTp. There are implications for the next steps in evaluating CFTp specifically, and for the development and evaluation of complex interventions generally.

This thesis has demonstrated that CFTp is acceptable as a 1-to-1 therapy for people with distressing psychotic experiences in NHS mental health services. It is feasible to recruit people to start this intervention, as evidenced across two different sites (London and Glasgow), for people to provide assessment measurements, and importantly, to complete the full 26-sessions of therapy. It is also feasible to use audio recording of sessions, and to use the CFTp-ACM as a scale for rating audio recordings. All these acceptability / feasibility findings point towards the implication of taking this forward for further evaluation testing. The recommendation is to run a pilot Randomised Controlled Trial, with the aim of assessing feasibility and design parameters for a full trial. In the pilot RCT, some suggestions for specific feasibility questions would be: Is it feasible to randomise participants into a CFTp group and a control group? (and what type of control group?); What is the drop-out rate in CFTp and control groups? Is it feasible for other therapists (not the developer) to deliver CFTp, by following the CFTp manual with adherence and competence? What is the training and supervision requirement for other clinicians to deliver CFTp with adherence and competence? What sample size would be needed for a subsequent full RCT? What are the best measures to assess primary and secondary outcomes? In general, the recommendation would be for future researchers to proceed with evaluating CFTp by following the specific Medical Research Council (MRC) guidance on evaluating complex interventions (P. Craig et al., 2008).

170

Future studies should continue to engage and co-produce with people with lived experience of psychosis. Co-production contributed to the incorporation of key therapeutic elements that might add to the beneficial effects of the intervention. In terms of limitations, future studies should be advised to conduct more rigorous assessment of therapist adherence. Now that the CFTp manual and the CFTp-ACM are both finalised, the recommendation would be for future studies to send a greater number of audio recordings for external fidelity checking, and for these to be selected at random. Manual adherence and therapist competence will be crucially important at the next stage (pilot RCT) as the intervention starts to be delivered by different therapists.

A pilot RCT should follow the CONSORT guidance, particularly for raters to be unaware of group allocation. The choices of measures is described in Chapter 8. The arguments made so far in this chapter are in favour of targeting specific mechanisms in samples of people with a range of psychotic experience, so this would mean recruiting participants with different positive symptoms (hallucinations and/or delusions). As has been outlined in this thesis, there is strong theoretical grounds for why, for example, both voice-related and paranoia-related outcomes should benefit from targeting the same social mentality mechanisms described. Nonetheless, there is still an important question for future research, which is whether different versions of CFTp could be specifically adapted and tailored for people with different psychotic experiences. There is some evidence from the case series, for instance, that with CFTp, voices improve quicker than delusions. Therefore, one implication could be that interventions for voice-hearers could potentially be shorter than for people with delusions, even if the focus remains the same. There is also a pattern in the process measure results (Table 8.7) that social rank/shame mechanisms (as measured by, e.g. Social Comparison Scale and Other as Shamer Scale) change quicker than the attachment/caring mechanisms (as measured by Self Compassion Scale). Could this mean, for example, that social rank mechanisms are more closely tied to voices? And if so, could future researchers use this information in future developments and iterations of CFTp? These are questions for the future. However, for the immediate next step, the combination of strong theoretical grounds and initial data from Study 4 suggest that a pilot RCT should test the CFTp manual with people with voices and/or delusions.

Beyond the specific implications for CFTp evaluation, there are more general implications arising from this thesis for the development and evaluation of psychosis interventions. Much of this is centred around the processes described in Chapter 7, which is one of the more unique strengths of this thesis. There is wide recognition of the importance of service user involvement in mental health research (Thornicroft & Tansella, 2005; Trivedi & Wykes, 2002) and more recently, specific guidance on user-

171 involvement in the actual design of interventions (Lyon & Bruns, 2019). As there is not yet much literature, nor guidance on how to do this, Chapter 7 might serve as an initial example. Future developers can also learn from the limitations identified in Chapter 7 about how to improve the processes of knowledge exchange.

In terms of other general implications, this thesis has demonstrated an integrated bio-psycho-social approach that may be useful for future studies. As mentioned in Chapter 1, CFT is recognised as a bio- psycho-social approach, and hence one focus of the CFT research has been seeking to understand the biological processes, and to evaluate biomarkers of change (Kirby et al., 2017). This thesis specifically investigated the Heart Rate Variability (HRV) biomarker to index changes in CFT processes. This could help guide future inter-disciplinary research collaborations on treatment evaluations in psychosis. Another possibility for the future would be to develop this bio-psycho-social approach further into designing evaluation studies that incorporated other biological markers, such as fMRI, endocrinological, and epigenetic data. This corresponds to some developments in the CBT evaluation research, where Kumari et al. (2011) have used functional brain imaging to demonstrate the neurological effect of CBT on threat processing in psychosis.

There are further implications of this thesis beyond individual therapies for psychosis. For example, the finding that social rank threat influences dissociative and psychotic processes has wider implications for the development of family and group therapies that reduce external social threats. At the level of services, these findings also have implications for promoting a culture of compassionate care and initiatives to tackle stigma (both in services and the wider society). The results of this thesis add scientific backing to these service-level strategies by highlighting the direct role of service ‘cultures’ in the underlying mechanisms of psychotic symptoms. Finally, there are also implications for the non-statutory sector, particularly organisations working to de-stigmatise psychosis and encourage social acceptance and inclusion. Demonstrating the importance of social rank threat in the course of psychotic symptoms, enhances the scientific case for organisations such as the Hearing Voices Network and Intervoice, who work tirelessly to promote socially validating experiences for people with psychosis.

9.5. Concluding remarks

This thesis has reported on a broad and varied PhD project that aimed to advance our understandings and treatments of psychosis. The chapters of this thesis have demonstrated successful advances in

172 different areas, specifically in measurement tools, theory, and intervention. The results and outputs have implications for the future design of theory testing studies, for the further evaluation of CFTp, and for the broader understanding and shaping of social contexts for people with psychosis. There is a clear message of validating and engaging with the voice of lived experience that runs throughout all aspects of this thesis; from identifying social mechanisms in models of psychosis, through intervention development partnerships, to targeting pro-social, affiliative processes in therapy. This thesis puts the social and the inter-personal firmly at the heart of our scientific efforts to help this population.

173

References

Allan, S., & Gilbert, P. (1995). A Social-Comparison Scale - Psychometric Properties and Relationship to Psychopathology. Personality and Individual Differences, 19(3), 293-299. doi:Doi 10.1016/0191-8869(95)00086-L Arseneault, L., Cannon, M., Fisher, H. L., Polanczyk, G., Moffitt, T. E., & Caspi, A. (2011). Childhood trauma and children's emerging psychotic symptoms: A genetically sensitive longitudinal cohort study. Am J Psychiatry, 168(1), 65-72. doi:10.1176/appi.ajp.2010.10040567 ARTE. (2016). Hearing Voices [online video]. Retrieved from https://www.youtube.com/watch?v=TPk4wjlvUw4 Barker, V., Gumley, A., Schwannauer, M., & Lawrie, S. M. (2015). An integrated biopsychosocial model of childhood maltreatment and psychosis. Br J Psychiatry, 206(3), 177-180. doi:10.1192/bjp.bp.113.143578 Barkham Chris Evans Frank Margison, M. (2009). The rationale for developing and outcome batteriesfor routine use in service settings and psychotherapy outcome research implementing core. Journal of Mental Health, 7(1), 35-47. doi:10.1080/09638239818328 Baumeister, D., Sedgwick, O., Howes, O., & Peters, E. (2017). Auditory verbal hallucinations and continuum models of psychosis: A systematic review of the healthy voice-hearer literature. Clin Psychol Rev, 51, 125-141. doi:10.1016/j.cpr.2016.10.010 Bebbington, P., & Kuipers, L. (1994). The clinical utility of expressed emotion in schizophrenia. Acta Psychiatr Scand Suppl, 382, 46-53. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/8091997 Beck, A. T. (1979). Cognitive therapy and the emotional disorders: Penguin. Bell, V., Halligan, P. W., & Ellis, H. D. (2006). The Cardiff Anomalous Perceptions Scale (CAPS): a new validated measure of anomalous perceptual experience. Schizophr Bull, 32(2), 366-377. doi:10.1093/schbul/sbj014 Bentall, R. P., de Sousa, P., Varese, F., Wickham, S., Sitko, K., Haarmans, M., & Read, J. (2014). From adversity to psychosis: pathways and mechanisms from specific adversities to specific symptoms. Soc Psychiatry Psychiatr Epidemiol, 49(7), 1011-1022. doi:10.1007/s00127-014- 0914-0 Bentall, R. P., & Morrison, A. P. (2002). More harm than good: the case against using antipsychotic drugs to prevent severe mental illness. Journal of Mental Health, 11(4), 351-356. Bentall, R. P., Wickham, S., Shevlin, M., & Varese, F. (2012). Do specific early-life adversities lead to specific symptoms of psychosis? A study from the 2007 the Adult Psychiatric Morbidity Survey. Schizophr Bull, 38(4), 734-740. Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3406525/pdf/sbs049.pdf Berry, K., Band, R., Corcoran, R., Barrowclough, C., & Wearden, A. (2007). Attachment styles, earlier interpersonal relationships and schizotypy in a non-clinical sample. Psychol Psychother, 80(Pt 4), 563-576. doi:10.1348/147608307X188368 Berry, K., Barrowclough, C., & Wearden, A. (2007). A review of the role of adult attachment style in psychosis: unexplored issues and questions for further research. Clin Psychol Rev, 27(4), 458- 475. doi:10.1016/j.cpr.2006.09.006 Berry, K., Barrowclough, C., & Wearden, A. (2009). Adult attachment, perceived earlier experiences of care giving and trauma in people with psychosis. Journal of Mental Health, 18(4), 280-287. doi:10.1080/09638230701879185 Berry, K., & Bucci, S. (2015). What does attachment theory tell us about working with distressing voices? Psychosis, 8(1), 60-71. doi:10.1080/17522439.2015.1070370 Berry, K., Varese, F., & Bucci, S. (2017). Cognitive Attachment Model of Voices: Evidence Base and Future Implications. Front Psychiatry, 8, 111. doi:10.3389/fpsyt.2017.00111

174

Berry, K., Wearden, A., Barrowclough, C., & Liversidge, T. (2006). Attachment styles, interpersonal relationships and psychotic phenomena in a non-clinical student sample. Personality and Individual Differences, 41(4), 707-718. doi:10.1016/j.paid.2006.03.009 Bighelli, I., Salanti, G., Huhn, M., Schneider-Thoma, J., Krause, M., Reitmeir, C., . . . Leucht, S. (2018). Psychological interventions to reduce positive symptoms in schizophrenia: systematic review and network meta-analysis. World Psychiatry, 17(3), 316-329. doi:10.1002/wps.20577 Birchwood, M., 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. Psychol Med, 34(8), 1571-1580. doi:10.1017/s0033291704002636 Birchwood, M., Jackson, C., Brunet, K., Holden, J., & Barton, K. (2012). Personal beliefs about illness questionnaire-revised (PBIQ-R): reliability and validation in a first episode sample. Br J Clin Psychol, 51(4), 448-458. doi:10.1111/j.2044-8260.2012.02040.x Birchwood, M., Meaden, A., & Trower, P. (2014). Shame, humiliation, and entrapment in psychosis: a social rank theory approach to cognitive intervention with voices and delusions. In A casebook of cognitive therapy for psychosis (pp. 124-147): Routledge. 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. Psychol Med, 30(2), 337-344. doi:10.1017/s0033291799001828 Birchwood, M., Mohan, L., Meaden, A., Tarrier, N., Lewis, S., Wykes, T., . . . Michail, M. (2018). The COMMAND trial of cognitive therapy for harmful compliance with command hallucinations (CTCH): a qualitative study of acceptability and tolerability in the UK. BMJ Open, 8(6), e021657. doi:10.1136/bmjopen-2018-021657 Birchwood, M., & Trower, P. (2006). The future of cognitive-behavioural therapy for psychosis: not a quasi-neuroleptic. Br J Psychiatry, 188(2), 107-108. doi:10.1192/bjp.bp.105.014985 Birchwood, M., Trower, P., Brunet, K., Gilbert, P., Iqbal, Z., & Jackson, C. (2007). Social anxiety and the shame of psychosis: a study in first episode psychosis. Behav Res Ther, 45(5), 1025-1037. doi:10.1016/j.brat.2006.07.011 Boon, S., Steele, K., & van der Hart, O. (2011). Coping with trauma-related dissociation: Skills training for patients and therapists. New York, NY: W W Norton & Co; US. Bowlby, J. (1973). Attachment and loss: Separation: Anxiety and anger (Vol. 2). In: new york: Basic Books. Bradstreet, S., Allan, S., & Gumley, A. (2019). Adverse event monitoring in mHealth for psychosis interventions provides an important opportunity for learning. In: Taylor & Francis. Braehler, C., Gumley, A., Harper, J., Wallace, S., Norrie, J., & Gilbert, P. (2013). Exploring change processes in compassion focused therapy in psychosis: results of a feasibility randomized controlled trial. Br J Clin Psychol, 52(2), 199-214. doi:10.1111/bjc.12009 Brett, C. M. (2005). Anomalous experiences and cognitive processing in the development of psychosis. (Unpublished PhD Thesis). King's College London, London. Brett, C. M., Heriot-Maitland, C., McGuire, P., & Peters, E. (2014). Predictors of distress associated with psychotic-like anomalous experiences in clinical and non-clinical populations. Br J Clin Psychol, 53(2), 213-227. doi:10.1111/bjc.12036 Brett, C. M., Peters, E. P., Johns, L. C., Tabraham, P., Valmaggia, L. R., & McGuire, P. (2007). Appraisals of Anomalous Experiences Interview (AANEX): a multidimensional measure of psychological responses to anomalies associated with psychosis. Br J Psychiatry Suppl, 51, s23-30. doi:10.1192/bjp.191.51.s23 Brett, C. M., Peters, E. R., & McGuire, P. K. (2015). Which psychotic experiences are associated with a need for clinical care? Eur Psychiatry, 30(5), 648-654. doi:10.1016/j.eurpsy.2014.12.005 Brewin, C. R., Gregory, J. D., Lipton, M., & Burgess, N. (2010). Intrusive images in psychological disorders: characteristics, neural mechanisms, and treatment implications. Psychol Rev, 117(1), 210-232. doi:10.1037/a0018113

175

Brown, G. W., Birley, J. L., & Wing, J. K. (1972). Influence of family life on the course of schizophrenic disorders: a replication. Br J Psychiatry, 121(562), 241-258. doi:10.1192/bjp.121.3.241 Butzlaff, R. L., & Hooley, J. M. (1998). Expressed emotion and psychiatric relapse: a meta-analysis. Arch Gen Psychiatry, 55(6), 547-552. doi:10.1001/archpsyc.55.6.547 Campbell, M. L. C., & Morrison, A. P. (2007). The subjective experience of paranoia: Comparing the experiences of patients with psychosis and individuals with no psychiatric history. Clinical Psychology & Psychotherapy, 14(1), 63-77. doi:10.1002/cpp.510 Cantor-Graae, E., & Selten, J. P. (2005). Schizophrenia and migration: a meta-analysis and review. Am J Psychiatry, 162(1), 12-24. doi:10.1176/appi.ajp.162.1.12 Carden, L. J., Saini, P., Seddon, C., Watkins, M., & Taylor, P. J. (2018). Shame and the psychosis continuum: A systematic review of the literature. Psychol Psychother. doi:10.1111/papt.12204 Carlson, E. B., & Putnam, F. W. (1993). An update on the Dissociative Experiences Scale. Mar 1993. Dissociation: Progress in the Dissociative Disorders, .6(1), pp. Cherry, M. G., Taylor, P. J., Brown, S. L., Rigby, J. W., & Sellwood, W. (2017). Guilt, shame and expressed emotion in carers of people with long-term mental health difficulties: A systematic review. Psychiatry Res, 249, 139-151. doi:10.1016/j.psychres.2016.12.056 Claridge, G. (1994). Single indicator of risk for schizophrenia: probable fact or likely myth? Schizophr Bull, 20(1), 151-168. doi:10.1093/schbul/20.1.151 Cole, C. L., Newman-Taylor, K., & Kennedy, F. (2016). Dissociation mediates the relationship between childhood maltreatment and subclinical psychosis. J Trauma Dissociation, 17(5), 577-592. doi:10.1080/15299732.2016.1172537 Compassionate Mind Foundation. (unpublised). CFT Groups Manual (draft). The Compassionate Mind Foundation. Connor, C., & Birchwood, M. (2013). Through the looking glass: self-reassuring meta-cognitive capacity and its relationship with the thematic content of voices. Front Hum Neurosci, 7, 213. doi:10.3389/fnhum.2013.00213 Corrigan, P. W., Larson, J. E., & Ruesch, N. (2009). Self‐stigma and the “why try” effect: impact on life goals and evidence‐based practices. World Psychiatry, 8(2), 75-81. Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2694098/pdf/wpa020075.pdf Corstens, D., Longden, E., & May, R. (2012). Talking with voices: Exploring what is expressed by the voices people hear. Psychosis-Psychological Social and Integrative Approaches, 4(2), 95-104. doi:10.1080/17522439.2011.571705 Corstens, D., Longden, E., McCarthy-Jones, S., Waddingham, R., & Thomas, N. (2014). Emerging perspectives from the hearing voices movement: implications for research and practice. Schizophr Bull, 40 Suppl 4(Suppl_4), S285-294. doi:10.1093/schbul/sbu007 Craig, P., Dieppe, P., Macintyre, S., Michie, S., Nazareth, I., Petticrew, M., & Medical Research Council, G. (2008). Developing and evaluating complex interventions: the new Medical Research Council guidance. Bmj, 337, a1655. doi:10.1136/bmj.a1655 Craig, T. K. (2019). AVATAR therapy: a promising new approach for persistent distressing voices. World Psychiatry, 18(1), 98-99. doi:10.1002/wps.20589 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. Lancet Psychiatry, 5(1), 31-40. doi:10.1016/S2215- 0366(17)30427-3 Croft, J., Heron, J., Teufel, C., Cannon, M., Wolke, D., Thompson, A., . . . Zammit, S. (2019). Association of trauma type, age of exposure, and frequency in childhood and adolescence with psychotic experiences in early adulthood. JAMA Psychiatry, 76(1), 79-80. Cultural Institute at King's. (2015). Compassion for Voices: A tale of courage and hope [online video]. Retrieved from https://www.youtube.com/watch?v=VRqI4lxuXAw David, A. S. (2010). Why we need more debate on whether psychotic symptoms lie on a continuum with normality. Psychol Med, 40(12), 1935-1942. doi:10.1017/S0033291710000188

176

Dean, K., & Murray, R. M. (2005). Environmental risk factors for psychosis. Dialogues Clin Neurosci, 7(1), 69-80. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/16060597 Debbane, M., Salaminios, G., Luyten, P., Badoud, D., Armando, M., Solida Tozzi, A., . . . Brent, B. K. (2016). Attachment, Neurobiology, and Mentalizing along the Psychosis Continuum. Front Hum Neurosci, 10, 406. doi:10.3389/fnhum.2016.00406 Dell, P. F., & O'Neil, J. A. (2009). Dissociation and the dissociative disorders: DSM-V and beyond. New York: Routledge. Demjén, Z., Marszalek, A., Semino, E., & Varese, F. (2019). Metaphor framing and distress in lived- experience accounts of voice-hearing. Psychosis, 11(1), 16-27. doi:10.1080/17522439.2018.1563626 Depue, R. A., & Morrone-Strupinsky, J. V. (2005). A neurobehavioral model of affiliative bonding: implications for conceptualizing a human trait of affiliation. Behav Brain Sci, 28(3), 313-350; discussion 350-395. doi:10.1017/S0140525X05000063 Dorahy, M. J., & Green, M. J. (2008). Cognitive perspectives on dissociation and psychosis: Differences in the processing of threat? In A. Moskowitz, I. Schäfer, & M. J. Dorahy (Eds.), Psychosis, trauma, and dissociation : emerging perspectives on severe psychopathology (pp. p.191-208). Chichester, West Sussex, England ; Hoboken, NJ: Wiley. Dorahy, M. J., & Green, M. J. (2019). Cognitive Perspectives on Dissociation and Psychosis. In A. Moskowitz, M. Dorahy, & I. Schafer (Eds.), Psychosis, Trauma and Dissociation: Evolving Perspectives on Severe Psychopathology (pp. 283-304): John Wiley & Sons Ltd. Dorahy, M. J., McKendry, H., Scott, A., Yogeeswaran, K., Martens, A., & Hanna, D. (2017). Reactive dissociative experiences in response to acute increases in shame feelings. Behav Res Ther, 89, 75-85. doi:10.1016/j.brat.2016.11.007 Dutra, L., Bureau, J. F., Holmes, B., Lyubchik, A., & Lyons-Ruth, K. (2009). Quality of early care and childhood trauma: a prospective study of developmental pathways to dissociation. J Nerv Ment Dis, 197(6), 383-390. doi:10.1097/NMD.0b013e3181a653b7 Eckblad, M., & Chapman, L. J. (1983). Magical Ideation as an Indicator of Schizotypy. Journal of consulting and clinical psychology, 51(2), 215-225. doi:Doi 10.1037/0022-006x.51.2.215 Ehlers, A., & Clark, D. M. (2000). A cognitive model of posttraumatic stress disorder. Behav Res Ther, 38(4), 319-345. doi:10.1016/s0005-7967(99)00123-0 Eicher, A. C., Davis, L. W., & Lysaker, P. H. (2013). Self-compassion: a novel link with symptoms in schizophrenia? J Nerv Ment Dis, 201(5), 389-393. doi:10.1097/NMD.0b013e31828e10fa Ellerby, M. (2014a). Resisting Voices Through Finding Our Own Compassionate Voice. Schizophr Bull, 43(2), 230-232. doi:10.1093/schbul/sbu035 Ellerby, M. (2014b). Schizophrenia, losers and compassion-focused therapy. Psychosis-Psychological Social and Integrative Approaches, 6(4), 359-362. doi:10.1080/17522439.2013.876091 Foster, C., Startup, H., Potts, L., & Freeman, D. (2010). A randomised controlled trial of a worry intervention for individuals with persistent persecutory delusions. Journal of Behavior Therapy and Experimental Psychiatry, 41(1), 45-51. doi:DOI 10.1016/j.jbtep.2009.09.001 Fowler, D., Freeman, D., Steel, C., Hardy, A., Smith, B., Hackman, C., . . . Bebbington, P. (2006). The catastrophic interaction hypothesis: how do stress, trauma, emotion and information processing abnormalities lead to psychosis? In W. Larkin & A. P. Morrison (Eds.), Trauma and Psychosis: New Directions for Theory and Therapy (pp. 101-124). Hove: Routledge. Freeman, D. (2011). Improving cognitive treatments for delusions. Schizophr Res, 132(2-3), 135-139. doi:10.1016/j.schres.2011.08.012 Freeman, D., & Garety, P. (2014). Advances in understanding and treating persecutory delusions: a review. Soc Psychiatry Psychiatr Epidemiol, 49(8), 1179-1189. doi:10.1007/s00127-014-0928- 7 Freeman, D., & Garety, P. A. (2003). Connecting neurosis and psychosis: the direct influence of emotion on delusions and hallucinations. Behav Res Ther, 41(8), 923-947. doi:10.1016/s0005- 7967(02)00104-3

177

Freeman, D., Garety, P. A., Bebbington, P. E., Smith, B., Rollinson, R., Fowler, D., . . . Dunn, G. (2005). Psychological investigation of the structure of paranoia in a non-clinical population. Br J Psychiatry, 186(5), 427-435. doi:10.1192/bjp.186.5.427 Friedman, H. (2002). Transpersonal psychology as a scientific field. International Journal of Transpersonal Studies, 21(1), 19. Gale, C., Gilbert, P., Read, N., & Goss, K. (2014). An evaluation of the impact of introducing compassion focused therapy to a standard treatment programme for people with eating disorders. Clin Psychol Psychother, 21(1), 1-12. doi:10.1002/cpp.1806 Garety, P. A., Kuipers, E., Fowler, D., Freeman, D., & Bebbington, P. E. (2001). A cognitive model of the positive symptoms of psychosis. Psychol Med, 31(2), 189-195. doi:10.1017/s0033291701003312 Georgiou, K., Larentzakis, A. V., Khamis, N. N., Alsuhaibani, G. I., Alaska, Y. A., & Giallafos, E. J. (2018). Can Wearable Devices Accurately Measure Heart Rate Variability? A Systematic Review. Folia Med (Plovdiv), 60(1), 7-20. doi:10.2478/folmed-2018-0012 Gilbert, P. (1992). Depression: The evolution of powerlessness. 1992. New York, NY: Guilford Press; US. Gilbert, P. (2000). The relationship of shame, social anxiety and depression: The role of the evaluation of social rank. Clinical Psychology & Psychotherapy, 7(3), 174-189. doi:Doi 10.1002/1099- 0879(200007)7:3<174::Aid-Cpp236>3.0.Co;2-U Gilbert, P. (2002). Evolutionary Approaches to Psychopathology and Cognitive Therapy. Journal of Cognitive Psychotherapy, 16(3), 263-294. doi:10.1891/jcop.16.3.263.52515 Gilbert, P. (2005). Social Mentalities: A Biopsychosocial and Evolutionary Approach to Social Relationships. Gilbert, P. (2009a). The compassionate mind: A new approach to facing the challenges of life. London: Constable Robinson. Gilbert, P. (2009b). Introducing compassion-focused therapy. Advances in psychiatric treatment, 15(3), 199-208. Retrieved from http://apt.rcpsych.org/content/15/3/199.full.pdf Gilbert, P. (2010a). Compassion focused therapy: Distinctive features: Routledge. Gilbert, P. (2010b). An introduction to compassion focused therapy in cognitive behavioural therapy. International Journal of Cognitive Therapy, 32(2), 97-112. Gilbert, P. (2013). Mindful Compassion: Using the Power of Mindfulness and Compassion to Transform our Lives: Hachette UK. Gilbert, P. (2014). The origins and nature of compassion focused therapy. Br J Clin Psychol, 53(1), 6- 41. doi:10.1111/bjc.12043 Gilbert, P. (2016). Depression: The evolution of powerlessness: Routledge. Gilbert, P., & Allan, S. (1998). The role of defeat and entrapment (arrested flight) in depression: an exploration of an evolutionary view. Psychol Med, 28(3), 585-598. doi:10.1017/s0033291798006710 Gilbert, P., Allan, S., Brough, S., Melley, S., & Miles, J. N. (2002). Relationship of anhedonia and anxiety to social rank, defeat and entrapment. J Affect Disord, 71(1-3), 141-151. doi:10.1016/s0165- 0327(01)00392-5 Gilbert, P., Birchwood, M., Gilbert, J., Trower, P., Hay, J., Murray, B., . . . Miles, J. N. (2001). An exploration of evolved mental mechanisms for dominant and subordinate behaviour in relation to auditory hallucinations in schizophrenia and critical thoughts in depression. Psychol Med, 31(6), 1117-1127. doi:10.1017/s0033291701004093 Gilbert, P., Clarke, M., Hempel, S., Miles, J. N., & Irons, C. (2004). Criticizing and reassuring oneself: An exploration of forms, styles and reasons in female students. Br J Clin Psychol, 43(Pt 1), 31-50. doi:10.1348/014466504772812959 Gilbert, P., & Irons, C. (2005). Focused therapies and compassionate mind training for shame and self- attacking. Compassion: Conceptualisations, research and use in psychotherapy, 263-325.

178

Gilbert, P., McEwan, K., Hay, J., Irons, C., & Cheung, M. (2007). Social rank and attachment in people with a bipolar disorder. Clinical Psychology & Psychotherapy, 14(1), 48-53. doi:10.1002/cpp.508 Gilbert, P., McEwan, K., Mitra, R., Richter, A., Franks, L., Mills, A., . . . Gale, C. (2009). An exploration of different types of positive affect in students and in patients with bipolar disorder. Clinical Neuropsychiatry, 6(4), 135-143. Gilbert, P., & Procter, S. (2006). Compassionate mind training for people with high shame and self- criticism: Overview and pilot study of a group therapy approach. Clinical Psychology & Psychotherapy, 13(6), 353-379. doi:10.1002/cpp.507 Goss, K., Gilbert, P., & Allan, S. (1994). An exploration of shame measures—I: The other as shamer scale. Personality and Individual Differences, 17(5), 713-717. Retrieved from http://www.sciencedirect.com/science/article/pii/019188699490149X Grant, P., Green, M. J., & Mason, O. J. (2018). Models of Schizotypy: The Importance of Conceptual Clarity. Schizophr Bull, 44(suppl_2), S556-S563. doi:10.1093/schbul/sby012 Grubaugh, A. L., Zinzow, H. M., Paul, L., Egede, L. E., & Frueh, B. C. (2011). Trauma exposure and posttraumatic stress disorder in adults with severe mental illness: a critical review. Clin Psychol Rev, 31(6), 883-899. doi:10.1016/j.cpr.2011.04.003 Gumley, A., Braehler, C., Laithwaite, H., MacBeth, A., & Gilbert, P. (2010). A Compassion Focused Model of Recovery after Psychosis. International Journal of Cognitive Therapy, 3(2), 186-201. doi:DOI 10.1521/ijct.2010.3.2.186 Gumley, A., Taylor, H., Schwannauer, M., & MacBeth, A. (2014). A systematic review of attachment and psychosis: measurement, construct validity and outcomes. Acta Psychiatr Scand, 129(4), 257-274. doi:10.1111/acps.12172 Haddock, G., McCarron, J., Tarrier, N., & Faragher, E. B. (1999). Scales to measure dimensions of hallucinations and delusions: the psychotic symptom rating scales (PSYRATS). Psychol Med, 29(4), 879-889. doi:10.1017/s0033291799008661 Hardy, A. (1979). Spiritual Nature of Man: Study of Contemporary Religious Experience. Oxford: Oxford University Press. Hartley, S., Barrowclough, C., & Haddock, G. (2013). Anxiety and depression in psychosis: a systematic review of associations with positive psychotic symptoms. Acta Psychiatr Scand, 128(5), 327- 346. doi:10.1111/acps.12080 Hayes, A. F. (2017). Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach, Second Edition (2nd ed.). New York, NY: The Guildford Press. Hayward, M., Jones, A. M., Bogen-Johnston, L., Thomas, N., & Strauss, C. (2017). Relating Therapy for distressing auditory hallucinations: A pilot randomized controlled trial. Schizophr Res, 183, 137-142. doi:10.1016/j.schres.2016.11.019 Heriot-Maitland, C. (2012). Multi-Level Models of Information Processing, and Their Application to Psychosis. Journal of Experimental Psychopathology, 3(4), 552-571. doi:10.5127/jep.017711 Heriot-Maitland, C., Keen, N., Lawlor, C., Kane, F., & Peters, E. (in preparation). A Brief Compassion- Focused Therapy Group for Outpatients with Psychosis. Heriot-Maitland, C., & Kennedy, A. (2019). Attachment themes in compassion-focused therapy (CFT) for psychosis. In K. Berry, S. Bucci, & A. Danquah (Eds.), Attachment and Psychosis: Routledge. Heriot-Maitland, C., Knight, M., & Peters, E. (2012). A qualitative comparison of psychotic-like phenomena in clinical and non-clinical populations. Br J Clin Psychol, 51(1), 37-53. doi:10.1111/j.2044-8260.2011.02011.x Heriot-Maitland, C., McCarthy-Jones, S., Longden, E., & Gilbert, P. (2019). Compassion Focused Approaches to Working With Distressing Voices. Front Psychol, 10, 152. doi:10.3389/fpsyg.2019.00152 Heriot-Maitland, C., & Russell, G. (2018). Compassion-Focused Therapy for Relating to Voices. In C. Cupitt (Ed.), CBT for Psychosis: Process-orientated Therapies and the Third Wave. London: Routledge.

179

Hoddinott, P. (2015). A new era for intervention development studies. Pilot Feasibility Stud, 1, 36. doi:10.1186/s40814-015-0032-0 Hoffman, R. E. (2007). A social deafferentation hypothesis for induction of active schizophrenia. Schizophr Bull, 33(5), 1066-1070. doi:10.1093/schbul/sbm079 Hoffmann, T. C., Glasziou, P. P., Boutron, I., Milne, R., Perera, R., Moher, D., . . . Michie, S. (2014). Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide. Bmj, 348, g1687. doi:10.1136/bmj.g1687 Holmes, E. A., Brewin, C. R., & Hennessy, R. G. (2004). Trauma films, information processing, and intrusive memory development. J Exp Psychol Gen, 133(1), 3-22. doi:10.1037/0096- 3445.133.1.3 Hooper, L. M., Stockton, P., Krupnick, J. L., & Green, B. L. (2011). Development, Use, and Psychometric Properties of the Trauma History Questionnaire. Journal of Loss & Trauma, 16(3), 258-283. doi:10.1080/15325024.2011.572035 Horwood, V., Allan, S., Goss, K., & Gilbert, P. (2019a). The development of the Compassion Focused Therapy Therapist Competence Rating Scale. Psychol Psychother, 0(0). doi:10.1111/papt.12230 Horwood, V., Allan, S., Goss, K., & Gilbert, P. (2019b). The development of the Compassion Focused Therapy Therapist Competence Rating Scale. Psychol Psychother. doi:10.1111/papt.12230 Hostinar, C. E., & Gunnar, M. R. (2015). Social support can buffer against stress and shape brain activity. AJOB neuroscience, 6(3), 34-42. Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4607089/pdf/nihms-675235.pdf Hostinar, C. E., Sullivan, R. M., & Gunnar, M. R. (2014). Psychobiological mechanisms underlying the social buffering of the hypothalamic-pituitary-adrenocortical axis: a review of animal models and human studies across development. Psychol Bull, 140(1), 256-282. doi:10.1037/a0032671 Humpston, C. S., Walsh, E., Oakley, D. A., Mehta, M. A., Bell, V., & Deeley, Q. (2016). The relationship between different types of dissociation and psychosis-like experiences in a non-clinical sample. Conscious Cogn, 41, 83-92. doi:10.1016/j.concog.2016.02.009 Hunter, N. (2014). Trauma, Psychosis, and Dissociation [Blog post]. Retrieved from https://www.madinamerica.com/2014/08/trauma-psychosis-dissociation/ Huppert, J. D., Smith, T. E., & Apfeldorf, W. J. (2002). Use of Self-Report Measures of Anxiety and Depression in Outpatients with Schizophrenia: Reliability and Validity. Journal of Psychopathology and Behavioral Assessment, 24(4), 275-283. doi:10.1023/a:1020783101092 IBM. (2016). IBM SPSS Statistics for Windows, Version 24.0. Armonk (NY): IBM Corp. Jackson, M., & Fulford, K. W. M. (1997). Spiritual experience and psychopathology. Philosophy, Psychiatry, & Psychology, 4(1), 41-65. Jacobson, N. S., & Truax, P. (1991). Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. J Consult Clin Psychol, 59(1), 12-19. doi:10.1037//0022- 006x.59.1.12 Janssen, I., Hanssen, M., Bak, M., Bijl, R. V., de Graaf, R., Vollebergh, W., . . . van Os, J. (2003). Discrimination and delusional ideation. Br J Psychiatry, 182(1), 71-76. doi:10.1192/bjp.182.1.71 Jaya, E. S., Ascone, L., & Lincoln, T. M. (2017). Social Adversity and Psychosis: The Mediating Role of Cognitive Vulnerability. Schizophr Bull, 43(3), 557-565. doi:10.1093/schbul/sbw104 Johns, L. C., & van Os, J. (2001). The continuity of psychotic experiences in the general population. Clin Psychol Rev, 21(8), 1125-1141. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/11702510 Johnson, J., Jones, C., Lin, A., Wood, S., Heinze, K., & Jackson, C. (2014). Shame amplifies the association between stressful life events and paranoia amongst young adults using mental health services: Implications for understanding risk and psychological resilience. Psychiatry Res, 220(1-2), 217-225. doi:10.1016/j.psychres.2014.07.022

180

Joseph, S., Maltby, J., Wood, A. M., Stockton, H., Hunt, N., & Regel, S. (2012). The Psychological Well- Being-Post-Traumatic Changes Questionnaire (PWB-PTCQ): Reliability and Validity. Psychological Trauma-Theory Research Practice and Policy, 4(4), 420-428. doi:10.1037/a0024740 Karlsen, S., Nazroo, J. Y., McKenzie, K., Bhui, K., & Weich, S. (2005). Racism, psychosis and common mental disorder among ethnic minority groups in England. Psychol Med, 35(12), 1795-1803. doi:10.1017/S0033291705005830 Kaymaz, N., & van Os, J. (2010). Extended psychosis phenotype--yes: single continuum--unlikely. Psychol Med, 40(12), 1963-1966. doi:10.1017/S0033291710000358 KCL. (2015). Compassion for Voices: 1st month report (dated 27 March 2015). Retrieved from https://www.kcl.ac.uk/archive/cultural/- culturalinstitute/showcase/current/kei/compassionforvoices/compassion-for-voices- report1.pdf Keen, N., George, D., Scragg, P., & Peters, E. (2017). The role of shame in people with a diagnosis of schizophrenia. Br J Clin Psychol, 56(2), 115-129. doi:10.1111/bjc.12125 Kennedy, A., & Dillon, J. (submitted). Engaging voices in a compassionate dialogue. Psychology and Psychotherapy, Research theory and Practice. Kennedy, A., & Ellerby, M. (2016). A Compassion-Focused Approach to Working With Someone Diagnosed With Schizophrenia. J Clin Psychol, 72(2), 123-131. doi:10.1002/jclp.22251 Kilcommons, A. M., & Morrison, A. P. (2005). Relationships between trauma and psychosis: an exploration of cognitive and dissociative factors. Acta Psychiatr Scand, 112(5), 351-359. doi:10.1111/j.1600-0447.2005.00623.x Kingdon, D. G., & Turkington, D. (1994). Cognitive-behavioral therapy of schizophrenia: Guilford Press. Kirby, J. N. (2017). Compassion interventions: The programmes, the evidence, and implications for research and practice. Psychol Psychother, 90(3), 432-455. doi:10.1111/papt.12104 Kirby, J. N., Doty, J. R., Petrocchi, N., & Gilbert, P. (2017). The Current and Future Role of Heart Rate Variability for Assessing and Training Compassion. Front Public Health, 5, 40. doi:10.3389/fpubh.2017.00040 Kirkbride, J. B., Boydell, J., Ploubidis, G., Morgan, C., Dazzan, P., McKenzie, K., . . . Jones, P. B. (2008). Testing the association between the incidence of schizophrenia and social capital in an urban area. Psychol Med, 38(8), 1083-1094. Koenig, H. G., & Bussing, A. (2010). The Duke University Religion Index (DUREL): A Five-Item Measure for Use in Epidemological Studies. Religions, 1(1), 78-85. doi:10.3390/rel1010078 Kumari, V., Fannon, D., Peters, E. R., Ffytche, D. H., Sumich, A. L., Premkumar, P., . . . Kuipers, E. (2011). Neural changes following cognitive behaviour therapy for psychosis: a longitudinal study. Brain, 134(Pt 8), 2396-2407. doi:10.1093/brain/awr154 Laithwaite, H., O'Hanlon, M., Collins, P., Doyle, P., Abraham, L., Porter, S., & Gumley, A. (2009). Recovery After Psychosis (RAP): a compassion focused programme for individuals residing in high security settings. Behav Cogn Psychother, 37(5), 511-526. doi:10.1017/S1352465809990233 Laroi, 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. Clin Psychol Rev, 67, 1-10. doi:10.1016/j.cpr.2018.11.001 Lavin, R., Bucci, S., Varese, F., & Berry, K. (2019). The relationship between insecure attachment and paranoia in psychosis: A systematic literature review. Br J Clin Psychol, 0(0). doi:10.1111/bjc.12231 Leaviss, J., & Uttley, L. (2015). Psychotherapeutic benefits of compassion-focused therapy: an early systematic review. Psychol Med, 45(5), 927-945. doi:10.1017/S0033291714002141 Liddle, P. F. (1987). The symptoms of chronic schizophrenia. A re-examination of the positive-negative dichotomy. Br J Psychiatry, 151(2), 145-151. doi:10.1192/bjp.151.2.145

181

Lim, M. H., Gleeson, J. F. M., Alvarez-Jimenez, M., & Penn, D. L. (2018). Loneliness in psychosis: a systematic review. Soc Psychiatry Psychiatr Epidemiol, 53(3), 221-238. doi:10.1007/s00127- 018-1482-5 Lincoln, T. M., Hohenhaus, F., & Hartmann, M. (2013). Can paranoid thoughts be reduced by targeting negative emotions and self-esteem? An experimental investigation of a brief compassion- focused intervention. Cognitive Therapy and Research, 37(2), 390-402. Lincoln, T. M., & Peters, E. (2019). A systematic review and discussion of symptom specific cognitive behavioural approaches to delusions and hallucinations. Schizophr Res, 203, 66-79. doi:10.1016/j.schres.2017.12.014 Lincoln, T. M., Suttner, C., & Nestoriuc, Y. (2008). Wirksamkeit kognitiver Interventionen für Schizophrenie. Psychologische Rundschau, 59(4), 217-232. doi:10.1026/0033-3042.59.4.217 Linscott, R. J., & van Os, J. (2013). An updated and conservative systematic review and meta-analysis of epidemiological evidence on psychotic experiences in children and adults: on the pathway from proneness to persistence to dimensional expression across mental disorders. Psychol Med, 43(6), 1133-1149. doi:10.1017/S0033291712001626 Liotti, G. (2004). Trauma, dissociation, and disorganized attachment: Three strands of a single braid. Psychotherapy, 41(4), 472-486. doi:10.1037/0033-3204.41.4.472 Liotti, G. (2006). A model of dissociation based on attachment theory and research. J Trauma Dissociation, 7(4), 55-73. doi:10.1300/J229v07n04_04 Liotti, G., & Gumley, A. (2008). An Attachment Perspective on Schizophrenia: The Role of Disorganized Attachment, Dissociation and Mentalization. In Psychosis, Trauma and Dissociation (pp. 117- 133). Longden, E. (2010). Making sense of voices: A personal story of recovery. Psychosis-Psychological Social and Integrative Approaches, 2(3), 255-259. doi:10.1080/17522439.2010.512667 Longden, E., Moskowitz, A., Dorahy, M., & Perona-Garcelán, S. (in press). The role of dissociation in auditory verbal hallucinations. In I. S. A. Moskowitz, & M. J. Dorahy (Ed.), Psychosis, trauma and dissociation: Emerging perspectives on severe psychopathology (2nd Ed.). Chichester: Wiley-Blackwell. Longden, E., Read, J., & Dillon, J. (2018). Assessing the Impact and Effectiveness of Hearing Voices Network Self-Help Groups. Community Ment Health J, 54(2), 184-188. doi:10.1007/s10597- 017-0148-1 Longden, E., Sampson, M., & Read, J. (2016). Childhood adversity and psychosis: generalised or specific effects? Epidemiol Psychiatr Sci, 25(4), 349-359. doi:10.1017/S204579601500044X Lord, F. M. (1980). Applications of item response theory to practical testing problems. Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Lovibond, S. H., & Lovibond, P. F. (1995). Manual for the Depression Anxiety Stress Scales (2nd ed.). Sydney, Australia. Lucre, K. M., & Corten, N. (2013). An exploration of group compassion-focused therapy for personality disorder. Psychol Psychother, 86(4), 387-400. doi:10.1111/j.2044-8341.2012.02068.x Lyon, A. R., & Bruns, E. J. (2019). User-Centered Redesign of Evidence-Based Psychosocial Interventions to Enhance Implementation-Hospitable Soil or Better Seeds? JAMA Psychiatry, 76(1), 3-4. doi:10.1001/jamapsychiatry.2018.3060 Maher, B. A. (1974). Delusional thinking and perceptual disorder. J Individ Psychol, 30(1), 98-113. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/4857199 Malik, M., Bigger, J. T., Camm, A. J., Kleiger, R. E., Malliani, A., Moss, A. J., & Schwartz, P. J. (1996). Heart rate variability: Standards of measurement, physiological interpretation, and clinical use. European heart journal, 17(3), 354-381. Marks, E. M., Steel, C., & Peters, E. R. (2012). Intrusions in trauma and psychosis: information processing and phenomenology. Psychol Med, 42(11), 2313-2323. doi:10.1017/S0033291712000505

182

Martinovich, Z., Saunders, S., & Howard, K. (1996). Some comments on “Assessing clinical significance”. Psychotherapy Research, 6(2), 124-132. Mascaro, J. S., Rilling, J. K., Tenzin Negi, L., & Raison, C. L. (2013). Compassion meditation enhances empathic accuracy and related neural activity. Soc Cogn Affect Neurosci, 8(1), 48-55. doi:10.1093/scan/nss095 Mason, O., & Claridge, G. (2006). The Oxford-Liverpool Inventory of Feelings and Experiences (O-LIFE): further description and extended norms. Schizophr Res, 82(2-3), 203-211. doi:10.1016/j.schres.2005.12.845 Mason, O., Claridge, G., & Jackson, M. (1995). New Scales for the Assessment of Schizotypy. Personality and Individual Differences, 18(1), 7-13. doi:Doi 10.1016/0191-8869(94)00132-C Matos, M., & Pinto-Gouveia, J. (2010). Shame as a traumatic memory. Clin Psychol Psychother, 17(4), 299-312. doi:10.1002/cpp.659 Matos, M., Pinto-Gouveia, J., & Costa, V. (2013). Understanding the importance of attachment in shame traumatic memory relation to depression: the impact of emotion regulation processes. Clin Psychol Psychother, 20(2), 149-165. doi:10.1002/cpp.786 Matos, M., Pinto-Gouveia, J., & Duarte, C. (2012). Above and beyond emotional valence: the unique contribution of central and traumatic shame memories to psychopathology vulnerability. Memory, 20(5), 461-477. doi:10.1080/09658211.2012.680962 Matos, M., Pinto-Gouveia, J., & Gilbert, P. (2013). The effect of shame and shame memories on paranoid ideation and social anxiety. Clin Psychol Psychother, 20(4), 334-349. doi:10.1002/cpp.1766 Mayhew, S. L., & Gilbert, P. (2008). Compassionate mind training with people who hear malevolent voices: a case series report. Clin Psychol Psychother, 15(2), 113-138. doi:10.1002/cpp.566 McCarthy-Jones, S. (2017). Is Shame Hallucinogenic? Front Psychol, 8, 1310. doi:10.3389/fpsyg.2017.01310 McGrath, J. J., Saha, S., Al-Hamzawi, A., Alonso, J., Bromet, E. J., Bruffaerts, R., . . . Kessler, R. C. (2015). Psychotic Experiences in the General Population: A Cross-National Analysis Based on 31,261 Respondents From 18 Countries. JAMA Psychiatry, 72(7), 697-705. doi:10.1001/jamapsychiatry.2015.0575 McManus, S., Bebbington, P., Jenkins, R., & Brugha, T. S. (2016). Mental Health and Wellbeing in England: Adult Psychiatric Morbidity Survey 2014: a Survey Carried Out for NHS Digital by NatCen Social Research and the Department of Health Sciences, University of Leicester (1783868252). Retrieved from Michail, M., & Birchwood, M. (2013). Social anxiety disorder and shame cognitions in psychosis. Psychol Med, 43(1), 133-142. doi:10.1017/S0033291712001146 Michalska da Rocha, B., Rhodes, S., Vasilopoulou, E., & Hutton, P. (2017). Loneliness in psychosis: a meta-analytical review. Schizophr Bull, 44(1), 114-125. MindWick. (2019). Engaging with Voices. Retrieved from https://youtu.be/sdzGLNoWU8c Mitchell, C. A. A., Maybery, M. T., Russell-Smith, S. N., Collerton, D., Gignac, G. E., & Waters, F. (2017). The Structure and Measurement of Unusual Sensory Experiences in Different Modalities: The Multi-Modality Unusual Sensory Experiences Questionnaire (MUSEQ). Front Psychol, 8, 1363. doi:10.3389/fpsyg.2017.01363 Morgan, C., Charalambides, M., Hutchinson, G., & Murray, R. M. (2010). Migration, ethnicity, and psychosis: toward a sociodevelopmental model. Schizophr Bull, 36(4), 655-664. doi:10.1093/schbul/sbq051 Moritz, S., Schilling, L., Wingenfeld, K., Kother, U., Wittekind, C., Terfehr, K., & Spitzer, C. (2011). Psychotic-like cognitive biases in borderline personality disorder. Journal of Behavior Therapy and Experimental Psychiatry, 42(3), 349-354. doi:DOI 10.1016/j.jbtep.2011.02.003 Moriyama, T. S., Drukker, M., Gadelha, A., Pan, P. M., Salum, G. A., Manfro, G. G., . . . Bressan, R. A. (2018). The association between psychotic experiences and traumatic life events: the role of the intention to harm. Psychol Med, 48(13), 2235-2246. doi:10.1017/S0033291717003762

183

Morrison, A. P. (2001). The Interpretation of Intrusions in Psychosis: An Integrative Cognitive Approach to Hallucinations and Delusions. Behavioural and Cognitive Psychotherapy, 29(3), 257-276. doi:10.1017/s1352465801003010 Morrison, A. P., Frame, L., & Larkin, W. (2003). Relationships between trauma and psychosis: a review and integration. Br J Clin Psychol, 42(Pt 4), 331-353. doi:10.1348/014466503322528892 Moskowitz, A., Barker-Collo, S., & Ellson, L. (2005). Replication of dissociation-psychosis link in New Zealand students and inmates. J Nerv Ment Dis, 193(11), 722-727. doi:10.1097/01.nmd.0000185895.47704.62 Moskowitz, A., & Corstens, D. (2008). Auditory Hallucinations: Psychotic Symptom or Dissociative Experience? Journal of Psychological Trauma, 6(2-3), 35-63. doi:10.1300/J513v06n02_04 Mueser, K. T., Goodman, L. B., Trumbetta, S. L., Rosenberg, S. D., Osher f, C., Vidaver, R., . . . Foy, D. W. (1998). Trauma and posttraumatic stress disorder in severe mental illness. J Consult Clin Psychol, 66(3), 493-499. doi:10.1037//0022-006x.66.3.493 Murphy, J., Shevlin, M., Houston, J., & Adamson, G. (2012). Sexual abuse, paranoia, and psychosis: a population-based mediation analysis. Traumatology, 18(1), 37-44. Muthén, L. K., & Muthén, B. O. (1998-2017). Mplus User's Guide. Eighth Edition. Los Angeles, CA: Muthén & Muthén. Newbury, J., Arseneault, L., Caspi, A., Moffitt, T. E., Odgers, C. L., & Fisher, H. L. (2017). Cumulative effects of neighborhood social adversity and personal crime victimization on adolescent psychotic experiences. Schizophr Bull, 44(2), 348-358. NICE. (2002). Schizophrenia: core interventions in the treatment and management of schizophrenia in primary and secondary care. National Institute of Clinical Excellence NICE. (2009). Schizophrenia: core interventions in the treatment and management of schizophrenia in primary and secondary care (update). National Institute of Clinical Excellence Nijenhuis, E. R., & van der Hart, O. (2011). Dissociation in trauma: a new definition and comparison with previous formulations. J Trauma Dissociation, 12(4), 416-445. doi:10.1080/15299732.2011.570592 Norman, R. M., Malla, A. K., Manchanda, R., Harricharan, R., Takhar, J., & Northcott, S. (2005). Social support and three-year symptom and admission outcomes for first episode psychosis. Schizophr Res, 80(2-3), 227-234. doi:10.1016/j.schres.2005.05.006 O'Hanlon, P., Aref-Adib, G., Fonseca, A., Lloyd-Evans, B., Osborn, D., & Johnson, S. (2016). Tomorrow's world: current developments in the therapeutic use of technology for psychosis. BJPsych advances, 22(5), 301-310. Oliver, J. E., McLachlan, K., Jose, P. E., & Peters, E. (2012). Predicting changes in delusional ideation: the role of mindfulness and negative schemas. Psychol Psychother, 85(3), 243-259. doi:10.1111/j.2044-8341.2011.02025.x Onwumere, J., Kuipers, E., Bebbington, P., Dunn, G., Freeman, D., Fowler, D., & Garety, P. (2009). Patient perceptions of caregiver criticism in psychosis: links with patient and caregiver functioning. J Nerv Ment Dis, 197(2), 85-91. doi:10.1097/NMD.0b013e3181960e57 Owens, D. G. C., & Johnstone, E. C. (2006). Precursors and prodromata of schizophrenia: findings from the Edinburgh High Risk Study and their literature context. Psychol Med, 36(11), 1501-1514. Retrieved from ://WOS:000242507600001 Pace, T. W., Negi, L. T., Adame, D. D., Cole, S. P., Sivilli, T. I., Brown, T. D., . . . Raison, C. L. (2009). Effect of compassion meditation on neuroendocrine, innate immune and behavioral responses to psychosocial stress. Psychoneuroendocrinology, 34(1), 87-98. doi:10.1016/j.psyneuen.2008.08.011 Parker, R. I., Vannest, K. J., Davis, J. L., & Sauber, S. B. (2011). Combining nonoverlap and trend for single-case research: Tau-U. Behav Ther, 42(2), 284-299. doi:10.1016/j.beth.2010.08.006 Pearce, J., Simpson, J., Berry, K., Bucci, S., Moskowitz, A., & Varese, F. (2017). Attachment and dissociation as mediators of the link between childhood trauma and psychotic experiences. Clin Psychol Psychother, 24(6), 1304-1312. doi:10.1002/cpp.2100

184

Perry, B. D., Pollard, R. A., Blakley, T. L., Baker, W. L., & Vigilante, D. (1995). Childhood trauma, the neurobiology of adaptation, and “use-dependent” development of the brain: How “states” become “traits”. Infant Mental Health Journal, 16(4), 271-291. doi:10.1002/1097- 0355(199524)16:4<271::Aid-imhj2280160404>3.0.Co;2-b Peters, E., Joseph, S., & Garety, P. (1999). Measurement of delusional ideation in the normal population: introducing the PDI (Peters et al. Delusions Inventory). Schizophr Bull, 25(3), 553- 576. doi:10.1093/oxfordjournals.schbul.a033401 Peters, E., Ward, T., Jackson, M., Morgan, C., Charalambides, M., McGuire, P., . . . Garety, P. A. (2016). Clinical, socio-demographic and psychological characteristics in individuals with persistent psychotic experiences with and without a "need for care". World Psychiatry, 15(1), 41-52. doi:10.1002/wps.20301 Peters, E., Ward, T., Jackson, M., Woodruff, P., Morgan, C., McGuire, P., & Garety, P. A. (2017). Clinical relevance of appraisals of persistent psychotic experiences in people with and without a need for care: an experimental study. Lancet Psychiatry, 4(12), 927-936. Retrieved from ://WOS:000416228300029 Picken, A. L., Berry, K., Tarrier, N., & Barrowclough, C. (2010). Traumatic events, posttraumatic stress disorder, attachment style, and working alliance in a sample of people with psychosis. J Nerv Ment Dis, 198(10), 775-778. doi:10.1097/NMD.0b013e3181f4b163 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. Pilton, M., Varese, F., Berry, K., & Bucci, S. (2015). The relationship between dissociation and voices: A systematic literature review and meta-analysis. Clin Psychol Rev, 40, 138-155. doi:10.1016/j.cpr.2015.06.004 Plural, F. P. (2011). A Logical Way of Being: The Reality of Dissociative Identity Disorder and Other Complex Dissociative Conditions (DVD). . Retrieved from http://firstpersonplural.org.uk Powers III, A. R., Kelley, M. S., & Corlett, P. R. (2016). Varieties of voice-hearing: psychics and the psychosis continuum. Schizophr Bull, 43(1), 84-98. Pyle, M., Stewart, S. L., French, P., Byrne, R., Patterson, P., Gumley, A., . . . Morrison, A. P. (2015a). Internalized stigma, emotional dysfunction and unusual experiences in young people at risk of psychosis. Early Interv Psychiatry, 9(2), 133-140. doi:10.1111/eip.12098 Pyle, M., Stewart, S. L. K., French, P., Byrne, R., Patterson, P., Gumley, A., . . . Morrison, A. P. (2015b). Internalized stigma, emotional dysfunction and unusual experiences in young people at risk of psychosis. Early Interv Psychiatry, 9(2), 133-140. doi:10.1111/eip.12098 Raes, F., Pommier, E., Neff, K. D., & Van Gucht, D. (2011). Construction and factorial validation of a short form of the Self-Compassion Scale. Clin Psychol Psychother, 18(3), 250-255. doi:10.1002/cpp.702 Raihani, N. J., & Bell, V. (2019). An evolutionary perspective on paranoia. Nat Hum Behav, 3(2), 114- 121. doi:10.1038/s41562-018-0495-0 Read, J., Agar, K., Argyle, N., & Aderhold, V. (2003). Sexual and physical abuse during childhood and adulthood as predictors of hallucinations, delusions and thought disorder. Psychol Psychother, 76(Pt 1), 1-22. doi:10.1348/14760830260569210 Read, J., van Os, J., Morrison, A. P., & Ross, C. A. (2005). Childhood trauma, psychosis and schizophrenia: a literature review with theoretical and clinical implications. Acta Psychiatr Scand, 112(5), 330-350. doi:10.1111/j.1600-0447.2005.00634.x Reeder, C., Pile, V., Crawford, P., Cella, M., Rose, D., Wykes, T., . . . Callard, F. (2016). The Feasibility and Acceptability to Service Users of CIRCuiTS, a Computerized Cognitive Remediation Therapy Programme for Schizophrenia. Behav Cogn Psychother, 44(3), 288-305. doi:10.1017/S1352465815000168

185

Reise, S. P., Morizot, J., & Hays, R. D. (2007). The role of the bifactor model in resolving dimensionality issues in health outcomes measures. Qual Life Res, 16 Suppl 1, 19-31. doi:10.1007/s11136- 007-9183-7 Rockliff, H., Gilbert, P., McEwan, K., Lightman, S., & Glover, D. (2008). A pilot exploration of heart rate variability and salivary cortisol responses to compassion-focused imagery. Clinical Neuropsychiatry: Journal of Treatment Evaluation. Romme, M., & Escher, S. (1994). Hearing voices. Bmj, 309(6955), 670. doi:10.1136/bmj.309.6955.670a Romme, M., & Escher, S. (2000). Making sense of voices: a guide for mental health professionals working with voice-hearers.(Includes interview supplement): Mind Publications. Romme, M., Escher, S., Dillon, J., Corstens, D., & Morris, M. (2009). Living with voices: 50 stories of recovery. Living with voices: 50 stories of recovery ii, 350 pp Ross-on-Wye, England: PCCS Books; England. Ross, C. A., Joshi, S., & Currie, R. (1991). Dissociative experiences in the general population: a factor analysis. Hosp Community Psychiatry, 42(3), 297-301. doi:10.1176/ps.42.3.297 Ross, C. A., & Keyes, B. (2004). Dissociation and Schizophrenia. Journal of Trauma & Dissociation, 5(3), 69-83. doi:10.1300/J229v05n03_05 Ross, K., Freeman, D., Dunn, G., & Garety, P. (2011). A Randomized Experimental Investigation of Reasoning Training for People With Delusions. Schizophr Bull, 37(2), 324-333. doi:DOI 10.1093/schbul/sbn165 Rossell, S. L., Schutte, M. J. L., Toh, W. L., Thomas, N., Strauss, C., Linszen, M. M. J., . . . Sommer, I. E. C. (2019). The Questionnaire for Psychotic Experiences: An Examination of the Validity and Reliability. Schizophr Bull, 45(45 Suppl 1), S78-S87. doi:10.1093/schbul/sby148 Russell, C. J., & Dean, M. A. (2016). To Log or Not to Log: Bootstrap as an Alternative to the Parametric Estimation of Moderation Effects in the Presence of Skewed Dependent Variables. Organizational Research Methods, 3(2), 166-185. doi:10.1177/109442810032002 Sangill, C., Buus, N., Hybholt, L., & Berring, L. L. (2019). Service user's actual involvement in mental health research practices: A scoping review. Int J Ment Health Nurs, 28(4), 798-815. doi:10.1111/inm.12594 Savulich, G., Shergill, S., & Yiend, J. (2012). Biased Cognition in Psychosis. Journal of Experimental Psychopathology, 3(4), 514-536. doi:10.5127/jep.016711 Seikkula, J., Aaltonen, J., Alakare, B., Haarakangas, K., Keranen, J., & Lehtinen, K. (2006). Five-year experience of first-episode nonaffective psychosis in open-dialogue approach: Treatment principles, follow-up outcomes, and two case studies. Psychotherapy Research, 16(2), 214- 228. doi:10.1080/10503300500268490 Selten, J. P., Booij, J., Buwalda, B., & Meyer-Lindenberg, A. (2017). Biological Mechanisms Whereby Social Exclusion May Contribute to the Etiology of Psychosis: A Narrative Review. Schizophr Bull, 43(2), 287-292. doi:10.1093/schbul/sbw180 Selten, J. P., van der Ven, E., Rutten, B. P., & Cantor-Graae, E. (2013). The social defeat hypothesis of schizophrenia: an update. Schizophr Bull, 39(6), 1180-1186. doi:10.1093/schbul/sbt134 Seo, J., & Choi, J. Y. (2018). Social defeat as a mediator of the relationship between childhood trauma and paranoid ideation. Psychiatry Res, 260, 48-52. doi:10.1016/j.psychres.2017.11.028 Shevlin, M., Dorahy, M., & Adamson, G. (2007). Childhood traumas and hallucinations: an analysis of the National Comorbidity Survey. J Psychiatr Res, 41(3-4), 222-228. doi:10.1016/j.jpsychires.2006.03.004 Shevlin, M., Houston, J. E., Dorahy, M. J., & Adamson, G. (2008). Cumulative traumas and psychosis: an analysis of the national comorbidity survey and the British Psychiatric Morbidity Survey. Schizophr Bull, 34(1), 193-199. doi:10.1093/schbul/sbm069 Shevlin, M., McAnee, G., Bentall, R. P., & Murphy, J. (2015). Specificity of association between adversities and the occurrence and co-occurrence paranoia and hallucinations: Evaluating the stability of childhood risk in an adverse adult environment. Psychosis-Psychological Social and Integrative Approaches, 7(3), 206-216. doi:10.1080/17522439.2014.980308

186

Siegel, D. J. (2010). Mindsight: The new science of personal transformation: Random House LLC. Sitko, K., Bentall, R. P., Shevlin, M., O'Sullivan, N., & Sellwood, W. (2014). Associations between specific psychotic symptoms and specific childhood adversities are mediated by attachment styles: an analysis of the National Comorbidity Survey. Psychiatry Res, 217(3), 202-209. doi:10.1016/j.psychres.2014.03.019 Sloman, L., & Taylor, P. (2016). Impact of Child Maltreatment on Attachment and Social Rank Systems: Introducing an Integrated Theory. Trauma Violence Abuse, 17(2), 172-185. doi:10.1177/1524838015584354 Smith, J. A. (1996). Beyond the divide between cognition and discourse: Using interpretative phenomenological analysis in health psychology. Psychology & Health, 11(2), 261-271. doi:Doi 10.1080/08870449608400256 Steel, C., Fowler, D., & Holmes, E. A. (2005). Trauma-related intrusions and psychosis: An information processing account. Behavioural and Cognitive Psychotherapy, 33(2), 139-152. doi:10.1017/S1352465804001924 Stevens, J. (1996). Applied multivariate statistics for the social sciences (3rd ed.). Mahwah, N.J.: Lawrence Erlbaum Associates. Stockmann, T., Wood, L., Enache, G., Withers, F., Gavaghan, L., & Razzaque, R. (2019). Peer-supported Open Dialogue: a thematic analysis of trainee perspectives on the approach and training. J Ment Health, 28(3), 312-318. doi:10.1080/09638237.2017.1340609 Stone, H., & Stone, S. (2011). Embracing ourselves: The voice dialogue manual: New World Library. Stowkowy, J., Liu, L., Cadenhead, K. S., Cannon, T. D., Cornblatt, B. A., McGlashan, T. H., . . . Addington, J. (2016). Early traumatic experiences, perceived discrimination and conversion to psychosis in those at clinical high risk for psychosis. Soc Psychiatry Psychiatr Epidemiol, 51(4), 497-503. doi:10.1007/s00127-016-1182-y Tai, S., & Turkington, D. (2009). The evolution of cognitive behavior therapy for schizophrenia: current practice and recent developments. Schizophr Bull, 35(5), 865-873. doi:10.1093/schbul/sbp080 Talbot, J. A., Talbot, N. L., & Tu, X. (2004). Shame‐proneness as a diathesis for dissociation in women with histories of childhood sexual abuse. Journal of Traumatic Stress: Official Publication of The International Society for Traumatic Stress Studies, 17(5), 445-448. Retrieved from https://onlinelibrary.wiley.com/doi/pdf/10.1023/B%3AJOTS.0000048959.29766.ae Tarrier, N., & Wykes, T. (2004). Is there evidence that cognitive behaviour therapy is an effective treatment for schizophrenia? A cautious or cautionary tale? Behav Res Ther, 42(12), 1377- 1401. Retrieved from https://www.sciencedirect.com/science/article/abs/pii/S0005796704001767?via%3Dihub Taylor, P. J., Pyle, M., Schwannauer, M., Hutton, P., & Morrison, A. (2015). Confirming the structure of negative beliefs about psychosis and bipolar disorder: A confirmatory factor analysis study of the Personal Beliefs about Experience Questionnaire and Personal Beliefs about Illness Questionnaire. Br J Clin Psychol, 54(4), 361-377. doi:10.1111/bjc.12079 Thomas, N., Hayward, M., Peters, E., van der Gaag, M., Bentall, R. P., Jenner, J., . . . McCarthy-Jones, S. (2014). Psychological therapies for auditory hallucinations (voices): current status and key directions for future research. Schizophr Bull, 40 Suppl 4(Suppl_4), S202-212. doi:10.1093/schbul/sbu037 Thornicroft, G., & Tansella, M. (2005). Growing recognition of the importance of service user involvement in mental health service planning and evaluation. Epidemiol Psichiatr Soc, 14(1), 1-3. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/15792287 Tracy, J. L., & Robins, R. W. (2004). TARGET ARTICLE: "Putting the Self Into Self-Conscious Emotions: A Theoretical Model". Psychological Inquiry, 15(2), 103-125. doi:10.1207/s15327965pli1502_01 Trivedi, P., & Wykes, T. (2002). From passive subjects to equal partners: qualitative review of user involvement in research. Br J Psychiatry, 181(6), 468-472. doi:10.1192/bjp.181.6.468 Turner, M. H., Bernard, M., Birchwood, M., Jackson, C., & Jones, C. (2013). The contribution of shame to post-psychotic trauma. Br J Clin Psychol, 52(2), 162-182. doi:10.1111/bjc.12007

187

Underwood, R., Kumari, V., & Peters, E. (2016). Cognitive and neural models of threat appraisal in psychosis: A theoretical integration. Psychiatry Res, 239, 131-138. doi:10.1016/j.psychres.2016.03.016 Upthegrove, R., Ross, K., Brunet, K., McCollum, R., & Jones, L. (2014). Depression in first episode psychosis: the role of subordination and shame. Psychiatry Res, 217(3), 177-184. doi:10.1016/j.psychres.2014.03.023 van der Gaag, M., van Oosterhout, B., Daalman, K., Sommer, I. E., & Korrelboom, K. (2012). Initial evaluation of the effects of competitive memory training (COMET) on depression in schizophrenia-spectrum patients with persistent auditory verbal hallucinations: a randomized controlled trial. Br J Clin Psychol, 51(2), 158-171. doi:10.1111/j.2044-8260.2011.02025.x Van der Hart, O., Nijenhuis, E. R., & Steele, K. (2006). The haunted self: Structural dissociation and the treatment of chronic traumatization: WW Norton & Company. van Nierop, M., Van Os, J., Gunther, N., Van Zelst, C., De Graaf, R., Ten Have, M., . . . van Winkel, R. (2014). Does social defeat mediate the association between childhood trauma and psychosis? Evidence from the NEMESIS‐2 S tudy. Acta Psychiatr Scand, 129(6), 467-476. Retrieved from https://onlinelibrary.wiley.com/doi/pdf/10.1111/acps.12212 van Os, J., Linscott, R. J., Myin-Germeys, I., Delespaul, P., & Krabbendam, L. (2009). A systematic review and meta-analysis of the psychosis continuum: evidence for a psychosis proneness- persistence-impairment model of psychotic disorder. Psychol Med, 39(2), 179-195. doi:10.1017/S0033291708003814 Vanijzendoorn, M., & Schuengel, C. (1996). The measurement of dissociation in normal and clinical populations: Meta-analytic validation of the Dissociative Experiences Scale (DES). Clinical psychology review, 16(5), 365-382. doi:10.1016/0272-7358(96)00006-2 Varese, F., Barkus, E., & Bentall, R. P. (2012). Dissociation mediates the relationship between childhood trauma and hallucination-proneness. Psychol Med, 42(5), 1025-1036. doi:10.1017/S0033291711001826 Varese, F., Smeets, F., Drukker, M., Lieverse, R., Lataster, T., Viechtbauer, W., . . . Bentall, R. P. (2012). Childhood adversities increase the risk of psychosis: a meta-analysis of patient-control, prospective- and cross-sectional cohort studies. Schizophr Bull, 38(4), 661-671. doi:10.1093/schbul/sbs050 Varese, F., Udachina, A., Myin-Germeys, I., Oorschot, M., & Bentall, R. P. (2011). The relationship between dissociation and auditory verbal hallucinations in the flow of daily life of patients with psychosis. Psychosis-Psychological Social and Integrative Approaches, 3(1), 14-28. doi:10.1080/17522439.2010.548564 Varghese, D., Scott, J., Welham, J., Bor, W., Najman, J., O'Callaghan, M., . . . McGrath, J. (2011). Psychotic-like experiences in major depression and anxiety disorders: a population-based survey in young adults. Schizophr Bull, 37(2), 389-393. doi:10.1093/schbul/sbp083 Vassos, E., Pedersen, C. B., Murray, R. M., Collier, D. A., & Lewis, C. M. (2012). Meta-analysis of the association of urbanicity with schizophrenia. Schizophr Bull, 38(6), 1118-1123. doi:10.1093/schbul/sbs096 Veling, W., Selten, J. P., Susser, E., Laan, W., Mackenbach, J. P., & Hoek, H. W. (2007). Discrimination and the incidence of psychotic disorders among ethnic minorities in The Netherlands. Int J Epidemiol, 36(4), 761-768. doi:10.1093/ije/dym085 Waller, H., Freeman, D., Jolley, S., Dunn, G., & Garety, P. (2011). Targeting reasoning biases in delusions: A pilot study of the Maudsley Review Training Programme for individuals with persistent, high conviction delusions. Journal of Behavior Therapy and Experimental Psychiatry, 42(3), 414-421. doi:DOI 10.1016/j.jbtep.2011.03.001 Wasserman, S., de Mamani, A. W., & Suro, G. (2012). Shame and guilt/self-blame as predictors of expressed emotion in family members of patients with schizophrenia. Psychiatry Res, 196(1), 27-31. doi:10.1016/j.psychres.2011.08.009

188

Waters, F., Collerton, D., Ffytche, D. H., Jardri, R., Pins, D., Dudley, R., . . . Laroi, F. (2014). Visual hallucinations in the psychosis spectrum and comparative information from neurodegenerative disorders and eye disease. Schizophr Bull, 40 Suppl 4(Suppl_4), S233-245. doi:10.1093/schbul/sbu036 Watson, A. C., Corrigan, P., Larson, J. E., & Sells, M. (2007). Self-stigma in people with mental illness. Schizophr Bull, 33(6), 1312-1318. doi:10.1093/schbul/sbl076 Watson, P. J., & Workman, E. A. (1981). The non-concurrent multiple baseline across-individuals design: an extension of the traditional multiple baseline design. J Behav Ther Exp Psychiatry, 12(3), 257-259. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/7320215 Weng, H. Y., Fox, A. S., Shackman, A. J., Stodola, D. E., Caldwell, J. Z., Olson, M. C., . . . Davidson, R. J. (2013). Compassion training alters altruism and neural responses to suffering. Psychol Sci, 24(7), 1171-1180. doi:10.1177/0956797612469537 WHO. (1973). Report of the international pilot study of schizophrenia. Retrieved from Geneva: Wickham, S., Sitko, K., & Bentall, R. P. (2015). Insecure attachment is associated with paranoia but not hallucinations in psychotic patients: the mediating role of negative self-esteem. Psychol Med, 45(7), 1495-1507. doi:10.1017/S0033291714002633 Wirth, R. J., & Edwards, M. C. (2007). Item factor analysis: current approaches and future directions. Psychol Methods, 12(1), 58-79. doi:10.1037/1082-989X.12.1.58 Wood, L., Byrne, R., Varese, F., & Morrison, A. P. (2016). Psychosocial interventions for internalised stigma in people with a schizophrenia-spectrum diagnosis: A systematic narrative synthesis and meta-analysis. Schizophr Res, 176(2-3), 291-303. doi:10.1016/j.schres.2016.05.001 Wood, L., & Irons, C. (2016). Exploring the Associations between Social Rank and External Shame with Experiences of Psychosis. Behavioural and Cognitive Psychotherapy, 44(5), 527-538. doi:10.1017/S1352465815000570

189

APPENDIX 3.1

Appendix 3.1. Additional information about analysis methodologies Part A: item selection i) Classical test theory

‘Endorsement’ criteria used: Items with less than 10% frequency of endorsement were considered potentially problematic.

‘Stability’ criteria used: Since our data were skewed, and the Landis and Koch (1977) interpretation of Kappa’s values could be misleading (Viera & Garrett, 2005), we relied on the percentage agreement between test-retest time points as the best indicator of stability. Therefore, Items with less than 80% of agreement were considered potentially problematic.

‘Internal consistency’ criteria used: The contribution of each item to the consistency of the scale was evaluated using the item-total correlations, with values less than 0.35 been considered problematic.

ii) Item response theory

An item response theory (IRT) model (2-parameter logistic model, e.g. Lord (1980), was used. Unlike the classical test theory approach, the IRT model allows evaluations of the reliability of each item in relation to the construct/continuum being measured. So in this case, we are able to identify not only if an item is a reliable indicator of the ‘anomalous experience’ continuum, but at which level of the continuum the item is more reliable. For instance, if item is reliable for high scorers, but not for low scorers, and vice versa.

Part B: Psychometric evaluation of TEQ i) Exploratory factor analysis of TEQ

Exploratory item factor analysis for categorical data (EFA) (Wirth & Edwards, 2007) was used to identify the factor structure of the TEQ using the responses of the ‘exploratory sample’ (n=283 with complete item set). EFA is conducted using the sample correlation matrix, that is, the bivariate correlations, across all items. EFA is a method of reproducing the sample correlation matrix, allowing for stochastic errors. Kaiser’s criterion refers to the eigenvalues of the sample correlation matrix, and it is shown that the number of factors is close to the number of the eigenvalues.

ii) Confirmatory factor analysis of TEQ

Confirmatory factor analysis for categorical data (CFA) (Muthén, 1984) was used. Model fit was evaluated using the following indices: the relative chi-square (rel χ2: with preferred values close to 2; (Hoelter, 1983)), the Root Mean Square Error of Approximation (RMSEA, with preferred values less than 0.8; (Browne & Cudeck, 1993)), the Taylor-Lewis Index (TLI, with preferred values higher than 0.9; (Bentler & Bonett, 1980)), and the Comparative Fit Index (CFI, with preferred values higher than 0.9; (Bentler, 1990)).

iii) Psychometric properties (reliability and validity) of TEQ

The test-retest reliability indices of TEQ were available from Part 1 (‘stability’ analysis), because all 59 participants who did seven-day retests were among the first to enter the study.

190 APPENDIX 3.1

The TEQ’s internal consistency was evaluated using Cronbach’s (1951) alpha coefficient for the entire scale, and item-total correlations for each item separately.

To evaluate convergent validity, we used a second comparator measure that has been widely used in the literature to measure anomalous experiences (O-LIFE-UnEx, Mason et al (1995)).

iv) Demographic characteristics

Evaluations of the relationships with demographic characteristics (e.g. age, gender, education) were tested by nonparametric methods due to the TEQ total scores being skewed (namely, Mann–Whitney independent samples’ test, Kruskal-Wallis ANOVA, and Spearman’s rho correlation coefficient).

v) Psychometric properties (difficulty and discrimination) of individual TEQ items

A final IRT analysis of the 19 items was performed with the two samples combined (N=532). In addition to estimating the information (precision, reliability) of each item, the IRT model also estimates the two parameters of a) difficulty, and b) discrimination for each item. The difficulty parameter corresponds to the level (i.e. total score) required for 50% endorsement of an item. The higher the difficulty parameter, higher levels of transpersonal experience are required to endorse the item. The discrimination parameter corresponds to the change in the probability of endorsement, as the level of transpersonal experience increases. The higher the discrimination parameter, the more capable an item to discriminate between individuals with different levels of transpersonal experience. The two IRT parameters were used to gain insight into the specificity of each item.

References

Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychol Bull, 107(2), 238-246. doi:10.1037/0033-2909.107.2.238 Bentler, P. M., & Bonett, D. G. (1980). Significance Tests and Goodness of Fit in the Analysis of Covariance-Structures. Psychological bulletin, 88(3), 588-606. doi:Doi 10.1037/0033- 2909.88.3.588 Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In J. S. L. K.A. Bollen (Ed.), Testing Structural Equation Models. Newbury Park: Sage. Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16(3), 297-334. Hoelter, J. W. (1983). The Analysis of Covariance-Structures - Goodness-of-Fit Indexes. Sociological Methods & Research, 11(3), 325-344. doi:Doi 10.1177/0049124183011003003 Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33(1), 159-174. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/843571 Lord, F. M. (1980). Applications of item response theory to practical testing problems. Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Mason, O., Claridge, G., & Jackson, M. (1995). New Scales for the Assessment of Schizotypy. Personality and Individual Differences, 18(1), 7-13. doi:Doi 10.1016/0191-8869(94)00132-C Muthén, B. O. (1984). A general structural equation model with dichotomous, ordered categorical, and continuous latent variable indicators. Psychometrika, 49(1), 115-132. Viera, A. J., & Garrett, J. M. (2005). Understanding interobserver agreement: the kappa statistic. Fam Med, 37(5), 360-363. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/15883903 Wirth, R. J., & Edwards, M. C. (2007). Item factor analysis: current approaches and future directions. Psychol Methods, 12(1), 58-79. doi:10.1037/1082-989X.12.1.58

191 APPENDIX 3.2

Appendix 3.2. Reliability indices for the initial pool of 57 items (N=283) Item Internal Endorsement1 Stability1 (test-retest) Omitted TEQ item consistency1 item via: % % Item-Total Kappa CTT IRT Number Abbreviation frequencies agreement Correlation 1 15.9 0.52 89.8 0.47 TEQ1 in contact 2 27.9 0.58 84.7 0.56 yes 3 22.6 0.57 89.8 0.43 yes 4 33.9 0.38 72.9 0.31 yes 5 15.9 0.4 88.1 0.47 yes 6* 5.3 0.37 94.9 0.34 TEQ2 seeing 7 32.2 0.23 69.5 0.27 yes others read 8 14.8 0.78 96.6 0.45 TEQ3 thoughts 9 11 0.35 86.4 0.50 yes 10* 15.5 0.64 89.8 0.34 TEQ4 smelling 11 13.8 0.47 88.1 0.49 yes 12 49.5 0.5 74.6 0.51 yes 13 10.2 0.78 96.6 0.46 yes 14 4.6 0.85 98.3 0.44 yes 15 34.3 0.29 74.6 0.47 yes 16 8.5 0.74 94.9 0.42 yes 17 15.9 0.42 86.4 0.41 yes 18 33.2 0.54 81.4 0.42 yes 19 25.1 0.52 84.7 0.58 yes 20 31.8 0.42 72.9 0.38 yes 21 24.7 0.55 84.7 0.45 TEQ5 thoughts whirl 22 5.7 0.37 94.9 0.31 yes 23 9.2 0.4 91.5 0.40 yes 24 10.2 0.35 86.4 0.44 TEQ6 'mission' revealed 25 8.1 -0.07 86.4 0.39 yes 26 13.4 0.47 88.1 0.45 TEQ7 body sensations 27 7.4 0.81 96.6 0.45 yes 28 21.9 0.4 83.1 0.56 TEQ8 messages or hints 29 25.4 0.44 83.1 0.50 yes picking up 30 16.3 0.57 89.8 0.48 TEQ9 thoughts 31 12.7 0.13 86.4 0.43 TEQ10 monitored 32 5.7 0.24 91.5 0.29 yes 33 8.5 -0.02 94.9 0.46 yes 34 11.7 0.54 94.9 0.45 TEQ11 others’ emotions 35 20.1 0.41 81.4 0.53 TEQ12 isolated 36 30.4 0.38 78 0.40 yes 37* 6 0.10 96.6 0.41 TEQ13 caused event time 38 16.3 0.48 86.4 0.50 TEQ14 disorientation

192 APPENDIX 3.2

39 8.5 -0.05 89.8 0.36 yes bodily 40* 3.5 0.48 96.6 0.39 TEQ15 movements 41 16.6 0.35 86.4 0.30 yes 42 9.5 0.16 88.1 0.40 yes influenced by 43* 6 0.35 89.8 0.36 TEQ16 others 44 2.1 0.05 94.9 0.34 yes 45 8.8 0.25 91.5 0.46 yes 46 15.5 0.51 89.8 0.45 yes 47 15.9 0.66 91.5 0.54 TEQ17 loss of identity 48 6.7 -0.04 93.2 0.39 yes 49 24.7 0.34 81.4 0.47 yes 50 8.1 0.4 91.5 0.37 TEQ18 hearing events in 51 13.4 0.62 91.5 0.36 TEQ19 reference 52 24 0.19 71.2 0.41 yes 53 8.8 0.24 91.5 0.38 yes 54 6.7 0.24 91.5 0.40 yes 55 3.2 0.66 98.3 0.24 yes 56 8.8 0.88 98.3 0.46 yes 57 14.1 0.26 86.4 0.40 yes 1Criteria used for endorsement, stability, internal consistency are given in Appendix 3.1 section on classical test theory *item did not fulfil criteria but was retained due to content validity

193 APPENDIX 3.3

Appendix 3.3. Transpersonal Experiences Questionnaire (TEQ)

Please read the following questions, and select the response which you feel is the most accurate one for you in the past 7 days.

For each question please select either Yes or No.

IN THE PAST 7 DAYS…

1. have you had the experience of suddenly feeling as if you were in contact with Yes someone who is not physically present, or knowing what they were thinking or No feeling?

2. have you had the experience of seeing something that other people couldn't see, or Yes that you later found out was not there? No

3. have you had the experience of your thoughts being read or picked up by other Yes people? No

4. have you had the experience of smelling something that other people are not aware Yes of, or that is only perceptible to you? No

5. have you had the experience of thoughts rushing very rapidly through your mind, so that one idea after another comes into your head and the thoughts seem to whirl Yes around beyond your control? No

6. have you had the experience of some kind of 'mission' or duty being revealed to Yes you, and knowing that you have to fulfil this mission, or feeling compelled to do so? No 7. have you had experiences of unusual sensations in your body, not created by any obvious physical cause, for example of heat or cold, energy moving, or something Yes entering or passing through your body? No

8. have you had experiences in which things in the world around you seemed to Yes contain messages or hints, perhaps in a metaphorical or symbolic way? No 9. have you had the experience of picking up on other people's thoughts? Yes No 10. have you had the experience of feeling monitored or watched, or otherwise the subject of external attention, when there is no obvious cause for this? Yes No 11. have you had the experience of feeling emotions or thinking thoughts that were Yes actually those of other people? No

12. have you experienced being in a state in which you felt cut off or isolated from Yes things and people around you, perhaps as if there were some invisible barrier No around you that prevented a normal connection?

13. have you had the experience of observing an event happen and feeling as though Yes you had caused it with your mind? No

194 APPENDIX 3.3

14. have you had the experience of disorientation in time, so that for example, the past and the future seem distant or unavailable, and the present moment dominates, or Yes time seems to lose its meaning? No

15. have you experienced your bodily movements being controlled by someone or Yes something outside of you? No

16. have you had an experience of having your thoughts, feelings or movements Yes influenced by other people's thoughts or gestures? No

17. have you had an experience of a loss of your individual identity and a sense of being Yes part of some greater whole that extends far beyond you? No

18. have you had the experience of hearing things, like voices talking, when there hasn't Yes been anyone around? No

19. have you had the experience of feeling as though events in your environment, such as the actions or comments of other people, are in reference to you, or are directed Yes at you, even though you know that this is unlikely? No

195 APPENDIX 4.1

Dr Charles Heriot-Maitland Department of Psychology P078 B4.08 King's College London, IoPPN De Crespigny Park London SE5 8AF

23 December 2014

Dear Charles,

PNM/14/15-26 Social-rank threat, dissociation and psychotic-like experiences in an online sample

Review Outcome: Full Approval

Thank you for sending in the amendments/clarifications requested to the above project. I am pleased to inform you that these meet the requirements of the PNM RESC and therefore that full approval is now granted.

Please ensure that you follow all relevant guidance as laid out in the King's College London Guidelines on Good Practice in Academic Research (http://www.kcl.ac.uk/college/policyzone/index.php?id=247).

For your information ethical approval is granted until 22 December 2017. If you need approval beyond this point you will need to apply for an extension to approval at least two weeks prior to this explaining why the extension is needed, (please note however that a full re-application will not be necessary unless the protocol has changed). You should also note that if your approval is for one year, you will not be sent a reminder when it is due to lapse.

Ethical approval is required to cover the duration of the research study, up to the conclusion of the research. The conclusion of the research is defined as the final date or event detailed in the study description section of your approved application form (usually the end of data collection when all work with human participants will have been completed), not the completion of data analysis or publication of the results. For projects that only involve the further analysis of pre-existing data, approval must cover any period during which the researcher will be accessing or evaluating individual sensitive and/or un-anonymised records. Note that after the point at which ethical approval for your study is no longer required due to the study being complete (as per the above definitions), you will still need to ensure all research data/records management and storage procedures agreed to as part of your application are adhered to and carried out accordingly.

If you do not start the project within three months of this letter please contact the Research Ethics Office.

Should you wish to make a modification to the project or request an extension to approval you will need approval for this and should follow the guidance relating to modifying approved applications:

196 APPENDIX 4.1

http://www.kcl.ac.uk/innovation/research/support/ethics/applications/modifications.aspx

Please would you also note that we may, for the purposes of audit, contact you from time to time to ascertain the status of your research.

If you have any query about any aspect of this ethical approval, please contact your panel/committee administrator in the first instance (http://www.kcl.ac.uk/innovation/research/support/ethics/contact.aspx) We wish you every success with this work.

Yours sincerely,

James Patterson – Senior Research Ethics Officer

Cc: Emmanuelle Peters

197 APPENDIX 4.2

Social context of anomalous experiences

Q1 Welcome

Welcome to the online research study. If you decide to take part, this study will take about 25 minutes to complete. However, before you decide whether to take part it is important for you to understand why the research is being done and what it would involve for you. Please take time to read the following information carefully and discuss it with others if you wish. You can email us if there is anything that is not clear or if you would like more information. Take time to decide whether or not you wish to take part.

Please note that once you have clicked on the ‘Continue’ button at the bottom of each page you cannot return to review or amend that page.

Information

Social context of anomalous experiences

What is the purpose of the study? Recent research has shown that many people describe having anomalous (‘unusual’) experiences, such as out- of-the-ordinary perceptions, feelings, or spiritual-type experiences that are somehow different to everyday life. We are interested in studying some of the factors that influence these experiences and their outcomes, i.e. whether they are helpful or unhelpful, or whether they lead to benefits or problems in other areas of life. In particular, this study is interested in the role of social processes in anomalous experiences; for instance, whether experiences are influenced by how people see themselves in relation to others, how they think they are seen by others, as well as the kind of relationship they have with themselves.

Why have I been invited to take part? You have been invited because you are a member of general population with access to the internet. We don’t know anything about your personal experiences, nor are we making any assumptions about these. We want to recruit as wide a range as people as possible, so it doesn’t matter whether or not you think you have these kinds of experiences. We just hope you will answer the questions as accurately and truthfully as you can.

Do I have to take part? It is entirely up to you to decide whether or not to take part. If you do decide to take part, there is a short consent form to complete at the foot of this page. You are still free to withdraw at any time in the process of the study without giving a reason. If you have started the online questionnaire, but then navigate away from the webpage before reaching the end, it will be assumed that you no longer wish to take part in the study, and any data collected up to this point (from clicking a ‘Continue’ button) will not be included in the study.

What will happen to me if I take part? Taking part will involve completing this online questionnaire (approximately 25 minutes of your time). In the questionnaire are some questions about social relationships as well some questions about anomalous experiences. You will then be invited to complete a (slighter shorter) follow-up questionnaire in 6 months.

What are the possible risks of taking part? No risks anticipated

What are the possible benefits of taking part? If you complete this questionnaire today, you can instantly enter a prize draw involving cash prizes for three randomly selected participants (1st £100; 2nd £50; 3rd £25). You will also be given the option of entering an additional prize draw for completing the 6-month follow-up (1st £100; 2nd £50; 3rd £25), and if you are among the first 50 participants, an additional prize draw for completing a 7-day follow-up (1st £100; 2nd £50; 3rd £25). Information and Consent Form Version 1, 16th Oct 2014 King’s College Research Ethics Committee ref: PNM/14/15-26 198 APPENDIX 4.2

Will my taking part be kept confidential? All the information collected during the course of the research will be kept strictly confidential, and will be stored in accordance with the Data Protection Act 1998, secured against unauthorised access. At no point in this study will you be requested to provide your name, and the only identifiable information you will be asked for is your email address. This is purely to inform you if you are selected for one of the cash prizes, and to send you a link and password for the 6-month follow-up questionnaire. Your email address will not be seen by anyone other than the primary researcher. As soon as the questionnaires and prize draws are finished, your data will be assigned an anonymous numerical coding, and your email address will be deleted.

How is the project being funded? The project is funded through a Medical Research Council (MRC) Clinical Research Training Fellowship awarded to Dr Charlie Heriot-Maitland (ref: MR/L01677X/1).

What will happen to the results of the study? The research should be completed by mid-2017. The results will be written-up and submitted for an academic qualification and for a peer-reviewed journal. This will be openly accessible through King’s College London’s publications repository. No individual will be identifiable from the published results.

Who should I contact for further information? If you have any questions or require more information about this study, please contact me using the following contact details:

Primary Researcher Dr Charlie Heriot-Maitland, MRC Clinical Research Training Fellow King’s College London, IoPPN (PO78) De Crespigny Park, London SE5 8AF Email: [email protected] / Tel: 07710 386138

What if I have further questions, or if something goes wrong? If this study has harmed you in any way or if you wish to make a complaint about the conduct of the study you can contact King's College London using the details below for further advice and information:

Research Supervisor Dr Emmanuelle Peters, Reader in Clinical Psychology King’s College London, IoPPN (PO77) De Crespigny Park, London SE5 8AF Email: [email protected] / Tel: 020 7848 0347

Review Subcommittee The Chair, Psychiatry, Nursing & Midwifery Research Ethics Subcommittee (PNM RESC) King’s College London, Research Support Office, 5.2 Franklin Wilkins Building Stamford Street, London SE1 9NH Email: [email protected]

Thank you for reading this information and for considering taking part in this research.

Consent form

Please complete this form after you have read the information above

Title of study: Social context of anomalous experiences

King’s College Research Ethics Committee ref: PNM/14/15-26

Thank you for considering taking part in this research. If you have any questions arising from the information above, please contact the researcher before you decide whether to join in.

1. I confirm that I understand that by selecting 'Yes' in each box below I am consenting to this element of the study. I understand that selecting 'No' means that I DO NOT consent to that part of the study. I understand that by not giving consent for any one element I may be deemed ineligible for the study.

Yes No

2. I confirm that I have read and understood the information about this study. I have had the opportunity to consider the information and to contact the researcher if I have had any questions.

Information and Consent Form Version 1, 16th Oct 2014 King’s College Research Ethics Committee ref: PNM/14/15-26 199 APPENDIX 4.2

Yes No

3. I understand that my participation is voluntary and that I am free to withdraw at any time without giving any reason.

Yes No

4. I consent to the processing of my personal information for the purposes explained above. I understand that such information will be handled in accordance with the terms of the UK Data Protection Act 1998.

Yes No

5. I understand that my non-personal, anonymised information may be subject to review by responsible individuals from the College for monitoring and audit purposes.

Yes No

6. I understand that confidentiality and anonymity will be maintained and it will not be possible to identify me in any publications.

Yes No

Information and Consent Form Version 1, 16th Oct 2014 King’s College Research Ethics Committee ref: PNM/14/15-26 200 APPENDIX 4.3

Appendix 4.3. Comparison of completers versus non-completers

Completers Non-completers Comparison (n=314) (n=230) test

Gender Male 62 (19.7%) 50 (21.7%) X2(1, N=544) = .32, Female 252 (80.3%) 180 (78.3%) p=.57 Age group Below 30 204 (65.0%) 173 (75.2%) X2(1, N=544) = 6.56, 30 or over 110 (35.0%) 57 (24.8%) p=.01 Ethnicity White 253 (80.6%) 159 (69.1%) X2(5, N=544) = 9.46, Other 61 (19.4%) 71 (30.9%) p=.002 Language English 249 (79.3%) 172 (74.8%) X2(1, N=544) = 1.55, Other 65 (20.7%) 58 (25.2%) p=.21 Education No degree 86 (27.4%) 82 (35.7%) X2(2, N=544) = 4.25, Degree or higher 228 (72.6%) 148 (64.3%) p=.04 MH services No 180 (57.3%) 150 (65.2%) X2(1, N=544) = 3.47, Yes 134 (42.7%) 80 (34.8%) p=.06 DES Mean (SD) 13.18 (10.57) 16.74 (14.57) Z = -2.38, p<.05 TEQ Mean (SD) 2.45 (2.99) 3.10 (3.80) Z = -1.07, p=.28 OAS Mean (SD) 22.72 (12.61) 23.64 (13.67) Z = .873, p=.38 SSPS Mean (SD) 37.75 (8.78) 37.87 (8.97) Z = -.56, p=.58 Significant results in bold

201 APPENDIX 5.1

Charles Heriot-Maitland

6 April 2017

Dear Charles ,

Study Title: Social context of anomalous experiences in a spiritual support-seeking population

Study Reference: Review Reference

I am pleased to inform you that full approval for your project has been granted by the PNM Research Ethics Subcommittee .

For your information, ethical approval has been granted for 3 years from 6 April 2017. If you need approval beyond this point, you will need to apply for an extension at least two weeks before this. You will be required to explain the reasons for the extension. However, you will not need to submit a full re- application unless the protocol has changed.

Ethical approval is required to cover the data-collection phase of the study. This will be until the date specified in this letter. However, you do not need ethical approval to cover subsequent data analysis or publication of the results. For secondary data-analysis, ethical approval is applicable to the data that is sensitive or identifies participants.

Please ensure that you follow the guidelines for good research practice as laid out in UKRIO’s Code of Practice for research: http://www.kcl.ac.uk/innovation/research/support/conduct/cop/index.aspx

Please note you are required to adhere to all research data/records management and storage procedures agreed to as part of your application. This will be expected even after the completion of the study.

If you do not start the project within three months of this letter, please contact the Research Ethics Office.

Please note that you will be required to obtain approval to modify the study. This also encompasses extensions to periods of approval. Please refer to the URL below for further guidance about the process: http://www.kcl.ac.uk/innovation/research/support/ethics/applications/modifications.aspx

Please would you also note that we may, for the purposes of audit, contact you from time to time to ascertain the status of your research.

If you have any query about any aspect of this ethical approval, please contact the Research Ethics Office:

(http://www.kcl.ac.uk/innovation/research/support/ethics/contact.aspx)

We wish you every success with this work.

Yours sincerely,

James Patterson - Senior Research Ethics Officer

For and on behalf of

Dr Jane Petty, Chair of the PNM Research Ethics Subcommittee

Cc: Emmanuelle Peters

Page 1 of 1 202 APPENDIX 5.2

Research funded by:

Participant Information Sheet Social context of anomalous experiences in a spiritual support-seeking population

Invitation to take part in this study

We’d like to invite you to take part in our research study. Please take time to read the information carefully and discuss it with others if you wish. Ask us if there is anything that is not clear or if you would like more information.

What is the purpose of the study?

This study aims to explore relationships between social factors and anomalous experiences, like changes in perception, spiritual-type experiences, hearing voices, or extrasensory communications.

Recent research has shown that many people describe having anomalous experiences. For some people these experiences have a negative impact on their life and result in input from mental health services. For others, these experiences have a positive impact and can be life-enriching. This research aims to explore the protective social factors that, over time, might enable people to integrate their experiences in a helpful way. This study will be split into two parts: the first part is exploratory, with a focus on finding out about the experiences of people who are seeking support from a spiritual network, including social and emotional aspects; the second part will explore how social experiences may change over time (from support-seeking, to 3 months, to 6 months), and how changes in social factors might influence the relationship with, and impact of, anomalous experiences.

Who is eligible to take part?

To be eligible you will be over 18 and will have recently accessed a support network for spiritual crisis/emergence. We don’t know anything about your personal experiences, nor are we making any assumptions about these. For this study, it is not important how you describe or explain your own experiences, but the fact that you have accessed support for these experiences through a spiritual network makes you a suitable research candidate. Participants from both inside and outside the UK are eligible to take part, so long as you can speak English.

Do I have to take part?

It is entirely up to you to decide whether or not to take part. If you do decide to take part, you will be asked to sign a consent form. You are still free to withdraw at any time in the process of the study without giving a reason. After the study has finished, the final deadline for withdrawal of data will be 31st July 2018.

What will happen to me if I take part?

Taking part will involve one initial meeting with the researcher, either in person or via Skype, to discuss your experiences. This meeting will last about 1 hour, including a short interview about your experiences and some questionnaires about, e.g., your mood, your significant life events and relationships, and about how you see yourself in relation to others. At this meeting, a subsample of participants will also be invited to have a selection of the questionnaire measures repeated (online, remotely) at 3 and 6 months. You will only be eligible for this subsample if you have never received a diagnosis or treatment from mental health services in relation to your anomalous experiences. If you do participate in the follow-up study, this will involve completing an online questionnaire (lasting about 25 minutes) at 3 months and 6 months. The researcher will email you a link and login details when it is time to complete these follow up questionnaires.

What will happen if I start but then don’t want to carry on with the study?

Participant Information Sheet Version 2, 3rd Apr 2017 King’s Research Ethics Committee ref: HR-16/17-3698 203 APPENDIX 5.2

You can withdraw from the study at any time without having to justify your decision. If you decide to withdraw from the study you can tell us whether you are happy for us to use the information obtained up to that point. If you are not, any information that you have given will be destroyed and you will not be contacted by us again.

What are the possible risks of taking part?

Some of the questions are personal in nature, but you don’t have to answer anything you don’t want to, and if you do get upset by a question, the interviewer is a trained psychologist who be able to help if necessary.

What are the possible benefits of taking part?

You may find it helpful to share your anomalous experiences with someone who will not be judgemental. It may also be a positive experience to have an opportunity to contribute to our knowledge of social context of anomalous experiences, which could, in turn, be used to develop and research new social initiatives to improve wellbeing.

We are able to offer you £15 remuneration for your time. If you are eligible to participate in the online follow-up subsample, we will be able to offer you a further £10 for each assessment point, so a maximum of £35 if you complete all three assessment points (baseline, 3 months, and 6 months).

Will my taking part be kept confidential?

All the information collected during the course of the research will be kept strictly confidential, and will be stored in accordance with the Data Protection Act 1998, secured against unauthorised access. Your personal details will not be seen by anyone other than the primary researcher. Your questionnaire data will be assigned an anonymous code before they are used for supervision purposes with the research supervisors.

How is the project being funded?

The project is funded through a Medical Research Council (MRC) Clinical Research Training Fellowship awarded to Dr Charlie Heriot-Maitland (ref: MR/L01677X/1).

What will happen to the results of the study?

The research should be completed by mid-2018. The results will be written-up as part of an academic degree, and submitted for a peer-reviewed journal. This will be openly accessible through King’s College London’s publications repository. No individual will be identifiable from the published results.

Who should I contact for further information?

If you have any questions or require more information about this study, please contact me using the following contact details:

Primary Researcher Dr Charlie Heriot-Maitland, MRC Clinical Research Training Fellow King’s College London, IoPPN (PO78) De Crespigny Park, London SE5 8AF Email: [email protected] / Tel: 07710 386138

What if I have further questions, or if something goes wrong?

If this study has harmed you in any way or if you wish to make a complaint about the conduct of the study you can contact King's College London using the details below for further advice and information:

Research Supervisor Dr Emmanuelle Peters, Reader in Clinical Psychology King’s College London, IoPPN (PO77) De Crespigny Park, London SE5 8AF Email: [email protected] / Tel: 020 7848 0347

Thank you for reading this information and for considering taking part in this research.

Participant Information Sheet Version 2, 3rd Apr 2017 King’s Research Ethics Committee ref: HR-16/17-3698 204 APPENDIX 5.3

Appendix 5.3. The TEQ-dimensions measure administration and scoring

Please read the following questions, and select the response which you feel is the most accurate one for you in the past 7 days.

For each question please select either Yes or No.

For each question answered ‘Yes’, please rate associated emotion (this experience feels: distressing 1 (not at all) – 7 (very much so); pleasant 1 – 7; and safe 1 – 7) as well as impact on functioning (the impact on my day-to-day living is disruptive 1 – 7; and enriching 1 – 7)

If Yes (1-7): emot. func. IN THE PAST 7 DAYS… Y/N Dist Plea Safe Disr Enri 1-7 1-7 1-7 1-7 1-7 1. have you had the experience of suddenly feeling as if you were in contact with someone who is not physically present, or knowing what they were thinking or feeling? ______2. have you had the experience of seeing something that other people couldn't see, or that you later found out was not there? ______3. have you had the experience of your thoughts being read or picked up by other people? ______4. have you had the experience of smelling something that other people are not aware of, or that is only perceptible to you? ______5. have you had the experience of thoughts rushing very rapidly through your mind, so that one idea after another comes into your head and the thoughts seem to whirl around beyond your control? ______6. have you had the experience of some kind of 'mission' or duty being revealed to you, and knowing that you have to fulfil this mission, or feeling compelled to do so? ______7. have you had experiences of unusual sensations in your body, not created by any obvious physical cause, for example of heat or cold, energy moving, or something entering or passing through your body? ______8. have you had experiences in which things in the world around you seemed to contain messages or hints, perhaps in a metaphorical or symbolic way? ______9. have you had the experience of picking up on other people's thoughts? ______10. have you had the experience of feeling monitored or watched, or otherwise the subject of external attention, when there is no obvious cause for this? ______11. have you had the experience of feeling emotions or thinking thoughts that were actually those of other people? ______12. have you experienced being in a state in which you felt cut off or isolated from things and people around you,

205 APPENDIX 5.3

perhaps as if there were some invisible barrier around ______you that prevented a normal connection? 13. have you had the experience of observing an event happen and feeling as though you had caused it with your ______mind? 14. have you had the experience of disorientation in time, so that for example, the past and the future seem distant or unavailable, and the present moment dominates, or time ______seems to lose its meaning? 15. have you experienced your bodily movements being controlled by someone or something outside of you? ______16. have you had an experience of having your thoughts, feelings or movements influenced by other people's ______thoughts or gestures? 17. have you had an experience of a loss of your individual identity and a sense of being part of some greater whole ______that extends far beyond you? 18. have you had the experience of hearing things, like voices ______talking, when there hasn't been anyone around? 19. have you had the experience of feeling as though events in your environment, such as the actions or comments of other people, are in reference to you, or are directed at ______you, even though you know that this is unlikely?

SCORING

TEQ total scores are calculated by adding together all the ‘yes’ endorsements. TEQ scores

TEQ-dimensions scores in each domain (distressing, pleasant, safe, Tot Dist Ple Saf Disr Enr disruptive, enriching) are calculated as the mean score for all items a e i with a rating. TEQ items that are not endorsed do not receive an outcome rating and are therefore not included in the mean.

The TEQ-dimensions measure is an adapted version of the Transpersonal Experiences Questionnaire (TEQ) (Chapter 3)

206 APPENDIX 5.4

Appendix 5.4. The ‘other’ sources of support Mass National Federation Spiritual Healers (x2) Metanoia Institute Local theatre group for people diagnosed with International Spiritual Emergence Network schizophrenia (ISEN) (x 8) Church of England vicar Spiritual Crisis Network (SCN) (x 6) 12-step Shades of awakening (x 4) Buddhist Mad in America John Weir Perry's books and other spiritually Beyond Meds minded takes on psychosis Therapist Action for Happiness Spiritual celebrity (Emma Bragdon) Centre for Better Health Church (x 3) Trauma counsellor Vipassana meditation retreat, ten-day silent Books: Stanislav Grof, Joseph Campbell, John Kundalini meet up group Perry, enneagram, Jodorowsky, etc. Lucid dreaming group Jesuit Centre Retreat Turning point Crisis point ACISTE The Goddess Shrine/ Secret Facebook peer Anthroposophy group Other Facebook peer groups relating to The Prosperos School of Ontology energy A Discord Server called 'The Haven' SoulCollage® Facilitator Training Samaritans Medium Julia Wright Summer MaCool-Shamaness Books & Research on alternatives Online support group Alternative / Energy Healing areas A 'medium' Red School On-line Brain Pickings, trauma focussed therapeutic Psychosis Therapy Project (Islington Mind) organisations, spiritual teachings on Foundations of wellbeing online programme Facebook/ Twitter Spiritual direction Soteria Hearing Voices group Alcoholics Anonymous Psychiatrist Psychiatrist (integrative) Spirituality and Mental Health group Tibetan Buddhism Prem Rawat Empath, late workers, star seeds Chrism in US by skype Teach Mindfulness Training Bipolar Facebook Holistic back practitioner Blessed life coven DWP / Social Security Grof Transpersonal Facilitator Training An acupuncturist Emerging Proud Early Intervention Psychosis Art, literature, academic research and literature and poetry

207 APPENDIX 5.5

Appendix 5.5. TEQ endorsement and rankings in spiritual support sample (Study 3, Chapter 5) and general population sample (Study 1, Chapter 3) Spiritual support sample General population sample (Study 3, n=30) (Study 1, n=283) Item Abbreviation Endorse (%) Ranking Endorse (%) Ranking 1 in contact 70 5th 15.9 6th= 2 seeing 36.9 15th 5.3 18th 3 others read thoughts 50 10th= 14.8 9th 4 smelling 36.7 16th 15.5 8th 5 thoughts whirl 50 10th= 24.7 1st 6 'mission' revealed 73.3 2nd= 10.2 14th 7 body sensations 73.3 2nd= 13.4 10th= 8 messages or hints 83.3 1st 21.9 2nd 9 picking up thoughts 63.3 7th 16.3 4th= 10 monitored 53.3 9th 12.7 12th 11 others’ emotions 66.7 6th 11.7 13th 12 isolated 56.7 8th 20.1 3rd 13 caused event 26.7 17th= 6 16th= 14 time disorientation 50 10th= 16.3 4th= 15 bodily movements 23.3 19th 3.5 19th 16 influenced by others 26.7 17th= 6 16th= 17 loss of identity 73.3 2nd= 15.9 6th= 18 hearing 43.3 14th 8.1 15th 19 events in reference 46.7 13th 13.4 10th=

208 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 1

APPENDIX 7.1 HYPOTHESIS AND THEORY published: 01 February 2019 doi: 10.3389/fpsyg.2019.00152

Compassion Focused Approaches to Working With Distressing Voices

Charles Heriot-Maitland1,2*, Simon McCarthy-Jones3, Eleanor Longden4 and Paul Gilbert5*

1 Glasgow Mental Health Research Facility, Institute of Health and Wellbeing, University of Glasgow, Glasgow, United Kingdom, 2 Department of Psychology, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, United Kingdom, 3 Department of Psychiatry, Trinity College Dublin, Dublin, Ireland, 4 Psychosis Research Unit, Greater Manchester Mental Health NHS Foundation Trust, Manchester, United Kingdom, 5 Centre for Compassion Research and Training, University of Derby, Derby, United Kingdom

This paper presents an outline of voice-hearing phenomenology in the context of evolutionary mechanisms for self- and social- monitoring. Special attention is given to evolved systems for monitoring dominant-subordinate social roles and relationships. These provide information relating to the interpersonal motivation of others, such as Edited by: neutral, friendly or hostile, and thus the interpersonal threat, versus safe, social location. Roberto Cattivelli, Individuals who perceive themselves as subordinate and dominants as hostile are highly Istituto Auxologico Italiano (IRCCS), Italy vigilant to down-rank threat and use submissive displays and social spacing as basic Reviewed by: defenses. We suggest these defense mechanisms are especially attuned in some Dayna Lee-Baggley, individuals with voices, in which this fearful-subordinate – hostile-dominant relationship Dalhousie University, Canada Philip R. Corlett, is played out. Given the evolved motivational system in which voice-hearers can be Yale University, United States trapped, one therapeutic solution is to help them switch into different motivational *Correspondence: systems, particularly those linked to social caring and support, rather than hostile Charles Heriot-Maitland competition. Compassion focused therapy (CFT) seeks to produce such motivational charles.heriot-maitland@ glasgow.ac.uk shifts. Compassion focused therapy aims to help voice-hearers, (i) notice their threat- Paul Gilbert based (dominant-subordinate) motivational systems when they arise, (ii) understand their [email protected] function in the context of their lives, and (iii) shift into different motivational patterns Specialty section: that are orientated around safeness and compassion. Voice-hearers are supported This article was submitted to to engage with biopsychosocial components of compassionate mind training, which Clinical and Health Psychology, a section of the journal are briefly summarized, and to cultivate an embodied sense of a compassionate self- Frontiers in Psychology identity. They are invited to consider, and practice, how they might wish to relate to Received: 23 August 2018 themselves, their voices, and other people, from the position of their compassionate Accepted: 16 January 2019 self. This paper proposes, in line with the broader science of compassion and CFT, that Published: 01 February 2019 repeated practice of creating internal patterns of safeness and compassion can provide Citation: Heriot-Maitland C, an optimum biopsychosocial environment for affect-regulation, emotional conflict- McCarthy-Jones S, Longden E and resolution, and therapeutic change. Examples of specific therapeutic techniques, such Gilbert P (2019) Compassion Focused Approaches to Working With as chair-work and talking with voices, are described to illustrate how these might be Distressing Voices. incorporated in one-to-one sessions of CFT. Front. Psychol. 10:152. doi: 10.3389/fpsyg.2019.00152 Keywords: compassion, CFT, auditory verbal hallucinations, voice-hearing, shame, trauma

Frontiers in Psychology | www.frontiersin.org 1 February 2019 | Volume 10209 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 2

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

INTRODUCTION resources, and will offer threat displays if they are. To those to whom they are subordinate, they will monitor the intention and The experience of hearing voices that other people cannot hear, potential for stimulating aggressive intent. This leads to what variously referred to as “hearing voices,” “voice-hearing” or Chance (1967) has termed “group based attention structure,” “auditory verbal hallucinations” has been reported for millennia which can be fine-tuned to the social location of potentially (McCarthy-Jones, 2012). Extended voice-hearing experiences are helpful or hostile others. However, dominant primates can be reported by around 2.5% of the general population without a unpredictable and launch unexpected attacks on subordinates; subsequent need for psychiatric care (McGrath et al., 2015). partly as a tactic to obtain subordinate attention as well as to However, when voices are frequent, uncontrollable, and have maintain levels of stress/fear (Gilbert and McGuire, 1998). negative content, they are likely to result in distress and problems An understanding of voice-hearing can hence begin with the with functioning (Daalman et al., 2011). Diagnoses that people recognition that over many millions of years, mammals evolved distressed and impaired by voice-hearing may receive, depending attentional mechanisms that are primed for social location and on the other experiences that co-occur with the voices, include the friendliness or hostility of others with whom one might be schizophrenia, bipolar disorder, post-traumatic stress disorder, in close proximity. The roots for voice-hearing may lie in the and borderline personality disorder (Larøi, 2012). evolution of the way language has become recruited into, and Contemporary neurocognitive models of voice-hearing have utilized by, a range of self-monitoring systems that are pursuing treated the question to be answered as “why people are hearing innate motivational biopsychosocial goals. To be specific, an a voice in the absence of an external stimulus”; a query that is important part of the social brain hypothesis (Dunbar, 2016) largely driven by formal definitions of the experience in purely is the way in which human brains have evolved to represent perceptual terms (David, 2004). This has led research to focus relationships. As humans, we can represent in our minds the on the auditory system and, correspondingly, models of voice- voices of each other’s thoughts and the possible threat-thoughts hearing as an aberrance of perception. However, a more recent that others might have. We can imagine the threat-relevant conception of voice-hearing proposes it to be a hallucinated social conversations someone might have about us, which of course communication (Bell, 2013). This offers the potential for a subtly requires language, but also the representation of the speech of different explanatory path to be followed, which starts with the others. This is something that we all do, and there is a clear question of why and how we monitor for social communications. evolutionary advantage in this; for example, we can work out that if we say or do X, how another person might think about us and react. We can run mental simulations in our mind and organize Evolution and Communicative our behavior according to our predictions. If the threat system is Monitoring highly dominant, then it’s going to be those conversations which There is general agreement that many of our basic motivational will become more prominent. It may be that with voice-hearers, and emotional systems are deeply rooted in our mammalian there is a differential experience in this element of mentally heritage (Buss, 2014; Gilbert, 2016). These motivational representing the voices of others. As will be explored in the next systems serve essential reproductive and survival functions section, this may be linked to dissociative processes in response and, importantly, all require stimulus detection and behavioral to, e.g., trauma and/or shame. repertoires to successfully pursue these motives. This can be seen in relation to feeding, seeking sexual partners or avoiding Threat, Shame, and Self-Criticism harms such as predators. In addition to the basic motives brains There is a strong association between voice-hearing and evolved with, there are also various processing systems that childhood traumas (Read et al., 2003; Janssen et al., 2004; monitor stimulus configurations in terms of their “meaning.” Bentall et al., 2012; Varese et al., 2012; Kelleher et al., 2013; These seek answers to questions such as, “is this sensory Longden et al., 2016) meaning that many voice-hearers have, stimulus an indicator of food or poison?,” “is this sound or smell as children, experienced subordination to a dominant other indicative of a possible predator?,” “is this sound a distress call (Romme et al., 2009). This is likely to have led to the from offspring?,” and “is that display by conspecifics offering development of highly sensitized threat-monitoring systems, opportunities for reproduction?” In group-living animals, especially for threats from dominant powerful others who monitoring systems are specifically orientated to social location; may have malevolent intent. However, importantly, it is not that is, awareness of the physical presence (nearness or distance) just high threat activation alone that is sufficient to lead to and intentions of others. For example, in the attachment system, voices (many people experience trauma without developing a parent remains vigilant to the distress signals, and the distance voices), and therefore consideration of an individual’s threat- s/he is, from her offspring (Bowlby, 1973). regulatory resources is also crucial. Hence where there is Of potential relevance to voice-hearing is the social high threat activation (e.g., from interpersonal traumatic monitoring system of conspecifics in relation to dominant- experience) combined with low regulatory capacity in the subordinate behavior (Gilbert, 2000). Dominants and soothing system (e.g., from affiliative/attachment experiences), subordinates carefully monitor each other’s presence, there is more likelihood of such heightened monitoring of whereabouts, and behavior. For example, dominant primates threat-intentions from others. This state of affairs offers an will monitor the behavior of those subordinate to them to ensure explanation for voice-content, as the majority of voices heard they are not inappropriately accessing or attempting to access in clinical populations involve threats (Nayani and David, 1996;

Frontiers in Psychology | www.frontiersin.org 2 February 2019 | Volume 10210 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 3

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

McCarthy-Jones et al., 2014). It also offers an explanation for a terms of some abnormality in brain chemistry, it may represent common form of relating to voices, which involves feelings of the activation of evolved functions for social threat monitoring. being in relation to a more powerful figure (Chadwick et al., The basic proposition then is that humans evolved with a 2000; Hayward, 2003), as subordination to one’s voices is closely range of different motivational potentials that organize attention, linked to experiences of subordination and marginalization in thinking, behavior and physiological systems quite differently. other social relationships (Birchwood et al., 2000; Gilbert et al., When we are orientated for sexual behavior, we are in a very 2001; Birchwood et al., 2004). Importantly too, it offers a different mind state than when we are orientated to avoid threat number of potential explanations for the etiology of the voices. or for eating behavior. One of the big social motivational systems Hypervigilance for social threat has been proposed to be a key is linked to social competition, and in particular social rank in mechanism involved in at least some voice-hearing experiences the context of social hierarchies that regulate social threat. For (Dodgson and Gordon, 2009; Garwood et al., 2015). Whist this individuals who have experienced threat from powerful others, threat may be physical (Dodgson and Gordon, 2009) it may also their social threat detection systems are choreographed in the be a threat to one’s social status. Given that the latter form of brain in such a way that they can be easily activated, attentionally threat is often signaled by shame, referred to as the “affect of tuned, and can generate intrusive fears (“you are ugly and a inferiority” (Kaufman, 1989) (p. 16), this suggests a potential pervert”). Common to many voice-hearers is the permeability of role for shame in the etiology of some voice-hearing (McCarthy- the self, such that these feared aspects of the self cannot be kept Jones, 2017b). secret - others can “know about them” and “know what they are.” Shame has two components. One is linked to what is called This is the fear of detection by a dominant. external shame, which has an external focus for attention, Such considerations also point to mediating relationships thinking and coping, whereas with internal (or internalized) between shame and voice-hearing. One likely mechanism is shame, the attention is focused in on the self, with high levels dissociation. This is a common way in which people may of self-criticism. External shame involves individuals monitoring, try to deal with shame (Talbot et al., 2004; Dorahy and experiencing, or imagining that they are the focus of other Clearwater, 2012). In survivors of trauma, levels of shame are people’s criticism, rejections and desires to exclude, marginalize positively associated with levels of dissociation (Talbot et al., or even persecute them. In other words, rejecting and malevolent 2004). Dissociation additionally has a strong association with intent by others. External shame is hence the experience of voice-hearing [see Longden et al. (2012) and Pilton et al. becoming an unattractive and undesirable agent in the mind of (2015) for a review], potentially due to its eliciting of cognitive others (Gilbert, 1998). In contrast, internalized shame, is linked intrusions (Dorahy et al., 2017). With prevailing conceptual to carrying those evaluations into the sense of self. Internal shame and clinical links to psychosis, the case has been made that requires that there has to be some perception of self as actually all voice-hearing can essentially be understood as dissociative “unattractive” - not just a failure to reach a standard (Gilbert, in nature (Moskowitz and Corstens, 2008). In this respect, 1992, 2002); that is to say it is closeness to an undesired and Van der Hart et al. (2006) structural model of dissociation unattractive self rather than distance from a desired self that proposes that trauma exposure may divide the personality is at issue (Ogilvie, 1987). These observations are important into systems that are focused on daily life and functioning, because most malevolent voices attack individuals as if they are and systems that are threat-protective and fixated on survival in some way unattractive, undesirable, or an object of derision (comprising defense subsystems such as apprehend, fight, flight, or disgust – a sort of “anti-ideal” to use De Rivera and Mascolo freeze, submit). Within such a framework, voice-hearing can language – or undesired self (Ogilvie, 1987). It is now known that therefore be conceived as an experience of ‘disowned’ threat- many human competitive interactions rarely involve outright based representations of the self (or self-other relationships), aggression and much more communications and degradation intruding upon functioning-focused parts of the personality of status; that is via shaming. Men and women denigrate and or self. Hence why voices are experienced as cognitively and shame each other slightly differently, often in domains of sexual perceptually detached from autobiographical experience (Dorahy attractiveness or competency (Buss and Dedden, 2016). Legg and and Palmer, 2015). Gilbert (2006) found evidence for this in that some voice-hearers As noted by Longden et al. (2018a), a dissociative framework experienced the voices giving sexual taunts, such as “you are a for voice-hearing accounts for several aspects of the experience slut, ugly, a pervert, smelly disgusting, unattractive.” What is less that cannot be wholly understood using cognitive/perceptual clear is whether before individuals had voices, they had these models alone; including for example, why voices may often anxieties about themselves, possibly even at the unconscious speak in the second or third person, their semantic and level; i.e., whether the voice is articulating one of their internal syntactic complexity and varied personifications (e.g., different fears. In addition to insults, voices behave like hostile-dominant ages, genders and response styles) and the type of intricate individuals, often showing commands and threats. A comparison interpersonal dynamics described previously, wherein voices of the physiological consequences of being trapped by a hostile- relate to the hearer in ways that reflect broader patterns of social dominant who is regularly putting one down reveals this to be relating. not dissimilar from an internal stream of thinking that is hostile and putting oneself down. In fact, self-criticism operates through “Specifically, moving beyond the cognitive paradigm of the mind similar neurophysiological systems as being criticized by others as a computer, mental processes can be understood as a product of (Longe et al., 2010). Rather than thinking about voice-hearing in social influences through which evolutionary processes (determined

Frontiers in Psychology | www.frontiersin.org 3 February 2019 | Volume 10211 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 4

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

by culture) lead to internalized values, beliefs, and behavioral example, the fact that all mammals, including and especially patterns from caregivers and other significant figures. In turn, this humans, have monitoring systems that are designed to try internalization of cultural patterns is expressed and represented in and detect potential social threats; and that sometimes these internal dialogic processes that embody the person’s relationship systems can become highly sensitized, particularly if we’ve been with their environment. By extension, this concept invites an in traumatizing situations. We can also discuss the fact that understanding of the self, not as a unitary or centralized entity that this is not our fault (which in itself can have important de- governs the life of the person, but rather as a dynamic, complex and heterogeneous experience that is formed by biographical influences shaming potential) and that it is very easy for us to create and which allows the interplay of different self-positions that have an exaggerated sense of shame, partly because of the profound an adaptive role for the person. As such, this framework expands the importance of shame, evolutionarily, for the regulation of view of...[voice-hearing] as merely hallucinations and perceptual social behavior. Crucially, however, motivational interviewing aberrations, and instead focuses on states of consciousness reflecting techniques can be used to orientate people to the idea that they different dissociated positions of the self; positions which, in the could begin to notice these dominant-subordinate motivating majority of cases, are experienced as subjectively real and are in and monitoring systems and, when they arise, to shift to a conflict with one another” (Longden et al., 2018a). compassion system. The reason for doing this is that compassion stimulates In turn, conceptualizing distressing voices as representations a completely different set of neurophysiological systems and of one’s sense of self suggests a strong clinical rationale for finding has a profound effect on threat processing. So the focus of ways of engaging with them that can promote more peaceful, the therapy is less on preventing actual voice-hearing and positive interactions between hearer and voice, reduction of more on helping individuals switch motivational systems from dissociative divisions, and recognition of their “protective” which those voices are arising. So the essence of the therapy function by drawing attention to unresolved emotional conflicts is to set up a psychoeducation framework and then provide (Corstens et al., 2012; Moskowitz et al., 2017; Mosquera and Ross, breathing, postural and verbal tone training (described in more 2017). Jacqui Dillon, the Chair of the United Kingdom Hearing detail below). Individuals then practice engaging and imagining Voices Network, expresses the imperative in the following way: themselves at their compassionate best, rooting themselves in “Each voice is an echo of the person’s experience so an attitude compassionate courage and wisdom, and beginning to think of curiosity, understanding and compassion toward all voices about how they would wish to engage with their inner voices from is the best stance as it will encourage and support internal the position of their compassionate self. That constant practice communication and, ultimately, self-acceptance” (Dillon, 2013). of inhabiting and creating the inner patterns of compassion becomes the source of change and growth. Indeed, in a qualitative study, Waite et al. (2015) found that, in recovery from psychosis, A COMPASSION FOCUSED THERAPY self-criticism and self-compassion were linked to two different APPROACH TO VOICE-HEARING cycles of outcome: Shame based self-criticism was associated with increasing distress over psychotic experiences, whereas Social competition and hierarchical social organization is only self-compassion was associated with empowerment and growth. one of a number of social motivational systems. Another is There is also evidence that helping people to generate compassion supportive and caring behavior, and when this motivational motivation and emotion through practices such as compassionate system is orientated it organizes attention, behavior, thoughts identity or compassionate self-training may be therapeutically and physiology in very different ways. Hence one therapeutic beneficial (Braehler et al., 2013). maneuver would be therefore to switch individuals from a threat- based hierarchical competitive motivational system into a care Creating Biopsychosocial Conditions to focused one. This is not likely to be easy because these switches can involve a number of fears, blocks, and resistances; part of Facilitate Engagement With which is, of course, mistrust. If we come to locate some of Threat-Based Emotion and Processes the difficulties that people with distressing voices experience Compassion focused therapy aims to help people develop their as sensitized competitive motivational and threat-protective capacity for affiliative relating (with self, voices, and other processing systems, then one therapeutic avenue is to begin people) through the cultivation of compassion at each of the to help individuals switch out of those particular motivational bio-, psycho-, and social levels. The rationale for targeting these systems and to develop new ways of emotional regulation – ones processes and mechanisms in CFT is based on extensive research that are not simply threat based – and begin to create a sense on the physiology of emotion regulation through social and of the secure base and safe haven. Compassion focused therapy affiliative experience (reviewed by Hostinar et al., 2014), as well (CFT) attempts to do this by showing individuals how they can as evidence for how compassion exercises/practices can influence switch to different motivational orientations and learn different different physiological processes that are important for emotional emotion regulation skills. To create the insider motivational wellbeing (Pace et al., 2009; Mascaro et al., 2013; Weng et al., impetus for this work, CFT uses a lot of psychoeducation about 2013). By helping to foster these conditions within a person’s the evolved nature of the mind. Therefore, in the case of people body, mind, and social environment, CFT aims to give the voice- who are voice-hearers, it can be very useful to invite them hearer a better chance of engaging with, understanding, and to think about some of the processes discussed above. For integrating their threat-based emotions and experiences.

Frontiers in Psychology | www.frontiersin.org 4 February 2019 | Volume 10212 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 5

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

Biological Aspects: The Physiology of Social Social Aspects: Interpersonal Safeness and Affiliation Safeness The finding that dominant-subordinate social relationships At the biological level, compassion is supported by bodily are often mirrored in people’s relationships with their voices experiences of social safeness and groundedness. To (Birchwood et al., 2000; Birchwood et al., 2004) highlights a clear inform interventions, CFT has sought to understand need to extend interventions into the social sphere. Furthermore, the neurophysiological underpinnings of these processes, access to safe and validating social experiences is found to be particularly their links to activity of the vagus nerve and heart a protective factor in clinical versus non-clinical outcomes for rate variability (HRV) (Kirby et al., 2017). Indeed, HRV has those with anomalous experiences (Heriot-Maitland et al., 2012; been increasingly adopted in CFT research as an important Brett et al., 2014). In CFT, it is therefore important to consider bio-marker measurement of safeness physiology. This is reflected the interpersonal environment, and the role of relationships, in the current CFT for psychosis research underway in the interactions and social contexts in supporting the patterns and United Kingdom, which includes HRV measures at five points processes outlined above. Drawing on roots in Attachment over the 6 to 8-month course of therapy (ClinicalTrials.gov Theory (Bowlby, 1973), CFT emphasizes the links between social identifier: NCT02733575). affiliative experience and the “soothing system”; the term used in In the same way that bodily experience of safeness will support CFT to refer to the body’s natural systems of (parasympathetic) the types of mental states that are conducive to compassion slowing, calming, and settling. As Bowlby observed in infant- (e.g., mentalising, empathy, etc.), compassionate intentions and carer interactions, the soothing system is highly sensitive to actions can in turn enhance experiences of social safeness, signals of inter-personal safeness, kindness and care, and has affiliation, and connection. Hence, safeness physiology and an important role in regulating the (sympathetic) arousal of the compassion mentality are mutually supportive. threat system. These social attachment mechanisms have been It is for this reason that CFT interventions for voice- further examined and elaborated in the field of evolutionary hearers typically start with a focus on establishing social neuroscience (Porges, 2007). safeness and groundedness. These provide the foundations for Important for CFT, however, is recognizing when subsequent relational work with self, emotions, voices, and experiences of social attachment and affiliation may not activate other people. A range of practices are utilized in CFT to soothing/safeness system, but rather the threat system – as may be directly stimulate safeness physiology; e.g., soothing breathing, the case for many voice-hearers. This can be due to attachment– grounding, body posture, facial expression and verbal tone. Some threat conditioning through early interpersonal experiences, brief descriptions of these are summarized below, and for more a particularly where childhood experiences with caregivers have detailed exploration of these techniques see Kirby (2017). been aversive or inadequate. For these individuals, it will understandably be harder to access the soothing/safeness effects Grounding and body posture of caring experiences, alongside having reduced resource for This involves practicing certain body postures and movements threat regulation. In CFT, the aim would be to increase a person’s to support the desired patterns of physiology and mentality. For capacity for affiliative relating by gradual exposure to affiliative example, adopting more upright and expansive postures are more experiences, with simultaneous exposure and de-sensitizing likely to send signals of calm composure and confidence to the of the elicited threat emotions. Recruiting the physiological mind, in contrast to more inward and tighter body postures, grounding, posture and breathing techniques outlined above is which are more likely to signal anxiety, threat, and danger. an important way of supporting this challenging exposure work.

Soothing breathing practice Psychological Aspects: Compassion Mentality This involves practicing slowing down the rhythm of breathing, At the psychological level, CFT aims to help voice-hearers whilst paying mindful attention to sensations of slowing in the tap into compassionate motives and mentalities by practicing body. The practice of slow, even, smooth breathing can bring certain patterns of cognition, planning, memory, and imagery. feelings of calmness/settling and groundedness, which can be Difficulties with mentalising have been highlighted as particularly helpful for steadying and anchoring when dealing with threat- relevant among people with psychosis-related diagnoses (Liotti based emotions and experiences. and Gumley, 2008), and so cognitive work in CFT will typically focus on exercises that develop mentalising capacity. For Facial expression and voice tone example, mindfulness skills are often practiced in CFT on the This involves practicing different facial expressions and voice grounds that mindfulness can not only help people to improve tones that reflect and support compassionate intentions and their attentional skills as a tool and vehicle toward compassion, motives. There are often certain “tones” that come with threat- but also develop (mentalised) awareness of the contents and based self-monitoring (thoughts and voices). As these are patterns of their mind. Awareness of these mental states, loops driven by the threat system, i.e., serving interests of protection and patterns, particularly the threat-based self-monitoring and and survival, these tones can be loud, critical and hostile – social-monitoring described above, will enable people to develop effective for salience and attention-grabbing. Practicing self- a relationship to them, rather than be caught up within them. monitoring with friendlier, warmer, and more supportive tones In CFT, the psychoeducation about evolved “tricky brains” helps can help shift body and mind into more compassionate to ensure that this relationship is one of understanding and patterns. compassion; for example, recognizing that it’s not our fault that

Frontiers in Psychology | www.frontiersin.org 5 February 2019 | Volume 10213 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 6

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

these processes occur in our minds due to our evolutionary threat For example, in Jungian analysis, an individual is helped to protection and survival needs. This helps to reduce self-criticism communicate with various parts of their psyche, the archetypes and shame, which, as we have already seen, may be key in both (Jung, 1919), so that these can be integrated in consciousness. the etiology and the emotional consequences of voice-hearing. Similarly, Voice Dialogue, developed by Stone and Stone (2011), Memory and imagery techniques are used in CFT to support offers a framework to help people understand their many parts, the development and training of compassionate psychologies. some of which are very well known to them, and others which For example, people are invited to bring to mind remembered are more distant or disowned. Voice Dialogue encourages people and/or imagined compassionate relationships with, e.g., safe to communicate with their many selves to discover what each places, compassionate others, and themselves. CFT will always wants and needs. More recently, Greenberg has described chair seek to proceed at a pace that allows time and space to mindfully work techniques in Emotion-Focused Therapy (Greenberg and notice the bodily expressions of these memories and images. This Watson, 2006), which is similarly a way to help access and helps the therapist and client to notice and address fears, blocks relate to different emotional parts. What distinguishes CFT from and resistances as they arise (e.g., see fears of compassion scales these other “relational” therapies is that the emotional work is (Gilbert et al., 2011), and aids the process of compassion exposure conducted from the compassionate self; i.e., the self-identity or and desensitization. part that has been specifically cultivated and trained to be able to compassionately engage with distress. In attachment terms, Cultivating and Deepening the the compassionate self provides an inner secure base and safe Compassionate Self haven, from which the person can develop strength and courage In CFT, each of the bio-, the psycho- and the social- to confidently explore the emotional world of their threat system. components of compassionate mind training (outlined above) are Identifying where to direct the compassionate self is a brought together as a self-identity called the compassionate self. collaborative process with the person who hears voices. Some Essentially the compassionate self operates as a holder for all the people may wish to initiate dialogs between their compassionate various skills, practices, postures, etc that are trained; an inner self and their voices, with the aim of understanding the emotional sense of having the qualities required to bring compassion to one’s function/meaning of voices, and developing a more peaceful, self-to-self and other relationships. In the case of voice-hearers, harmonious relationship with them. For others, there may be the compassionate self is an inner sense of the qualities and skills a preference to dialog with emotional selves (e.g., angry self, that are required to bring greater understanding, care, and peace anxious self, or self-critic) that have been identified and conceived into the relationship with one’s voices. within the therapy sessions, and may therefore feel a safer route Evidence from Matos et al. (2017) has shown that the more a (initially, at least) toward accessing emotion. For those who person succeeds in embodying the compassionate mind training do directly dialog with voices, it may be important to seek in their everyday lives (i.e., via a compassionate self), the more permission from the voices first, particularly because many voices this improves their perceived feelings of safeness and their will have a role of protecting hurt or vulnerable parts (Corstens compassionate relating. As such, cultivating and deepening a et al., 2012). So unless the voice feels reassured, respected, and compassionate self can be regarded as important preparatory safe, it would make sense why it might try to (protectively) block work for compassionate engagement with voices, and indeed any process that potentially brings closeness or exposure to a with any other threat-based emotion or conflict that may be perceived vulnerability. The compassionate self can understand causing problems. The idea is that it leads the person to a point this defensive process, as well as its origins in the threat- of greater readiness to do what’s required in therapy. Whatever protection system, and can therefore meet a voice with validation the therapy goals may be (e.g., processing a trauma memory, and sensitivity. An example is given in the short YouTube film, managing dissociative states, overcoming a social fear, finding a Compassion for Voices (Cultural Institute at King’s, 2015), where job, etc.), the compassionate self becomes the place to both come the compassionate self, speaking to a critical voice, says: from and go back to when engaging with the required emotional “I want to understand you. I want to help you feel safe... I know work. In therapy, the compassionate self also takes the role of an you’re trying to help me. Thank you. Thank you for reminding me internal supportive presence and guide throughout. that I get scared. You’re right, I do.” (3:02 mins) Here the compassionate self is calmly and respectfully COMPASSION FOR VOICES, MULTIPLE validating the voice’s protective role, whilst also identifying links SELVES, AND EMOTIONAL PARTS between the voice and the emotions of the threat system behind the voice (i.e., fear of leaving the house). In CFT, this kind In this section, we focus on direct emotional work with multiple of dialog might be role-played out in the therapy room using selves as a demonstration of how compassionate self can be used different chairs, which can help the person to connect with an in CFT for people who hear voices. In CFT, this is often referred embodied sense of these emotional and motivational systems. So, to as directing the compassionate self or putting the compassionate for example, the voice-hearer might be invited to sit in one chair self to work. representing the critical voice, and another chair to represent The therapeutic process of relating to emotions through the compassionate self. When shifting between chairs, the person multiple “selves” or “parts” is not unique to CFT. There are has time and space to feel their way into each role. So if the similar constructs and processes outlined in different approaches. voice comes with feelings of frustration, anger or contempt, then

Frontiers in Psychology | www.frontiersin.org 6 February 2019 | Volume 10214 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 7

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

the individual would be invited to gradually connect with the of a hostile voice, which will necessitate considerable courage and associated postures, attitudes and mindset. Similarly with the tolerance. compassionate self, a shift into this chair will involve plenty of In addition to chair work, the process of bringing compassion time and space for the person to connect with soothing breathing, to voices and emotional parts in CFT employs other techniques groundedness, and the particular qualities required to respond such as imagery and letter writing. Compassionate imagery might compassionately to this angry voice. involve creating imagined characters to represent voices, focusing One of the important goals of compassionate dialog might on characteristics such as facial expression, size, proximity, be to understand the nature of the emotion sitting behind the voice tone, etc. These visual representations can be helpful in voice; i.e., what’s driving the voice; what feeling/experience is enhancing a voice-hearer’s understanding of different aspects and the voice protecting? What feeling is there but not yet safe intentions of their voices. They can also provide opportunities for enough to access? This information might be elicited through setting up safe imagined scenarios whereby the compassionate explorative questioning, for example asking the voice: “Why are self (or an ideal compassionate other) might encounter the you angry?” “What would happen if you stopped or couldn’t be voice to start a conversation. With visualizations, there may angry?” “What would be your concern then?” Sometimes this also be opportunities for the voice-hearer to modify some of kind of questioning may not even be necessary, because voices these characteristics over time, for example, as the image of may respond and open to the encounter (alone) of a safe and the voice receives compassion, there might a slight settling compassionate presence. So, for example, the experience of being of posture, or lightening of color, or reduction of volume. In listened to and validated (e.g., “that sounds really tough for you”; compassionate letter-writing, the voice-hearer will often write “I can see why you’re frustrated”; “that makes a lot of sense” etc.) from the perspective of their compassionate self to a voice may itself create the required safeness to start naturally eliciting or a part, bringing empathic and wise understanding to how the threat-emotional communication. it developed and acknowledging its protective role. For more Once the vulnerability is identified and linked to its existence detailed illustrations of these clinical techniques and processes, in the threat system, then this part might itself have a new chair see a recent single case study account of CFT for relating to voices allocated. For some voice-hearers, this part may be a fearful part (Heriot-Maitland and Russell, 2018). that has experienced trauma earlier in life. It may be a part that The interpersonal and behavioral work also forms an has been bullied, or experienced social shaming, discrimination, important part of CFT. This is where the individual is invited to and humiliation. This part can gradually become more a focus take the compassionate intentions and actions out in their social or recipient of compassionate intent and care. Importantly, the world of relationships and interactions, therefore developing timing and pace will proceed in communication and negotiation both the flow of compassion to others and the capacity to with the voice. One creative use of chair work is to use the receive compassion from others. As described previously, creating space and positioning of chairs in the room to represent where these wider, real-world opportunities for social safeness and the fearful part sits in relation to the voice, and in relation connection is a key relational context in which the CFT approach to the compassionate self. For example, the fearful part chair is built. In this regard, the CFT approach to voice-hearing is might start off totally hidden behind the voice chair out of sight very compatible with social network approaches in psychosis but then gradually, with the help and encouragement of the such as Open Dialogue (Seikkula et al., 2006), Peer-Supported compassionate self, come out into its own space in the room, Open Dialogue (Stockmann et al., 2017), and the Hearing Voices maybe finding its own voice, reflecting more “ownership” of this Network approach (Corstens et al., 2014; Longden et al., 2018b). emotion. The compassionate chair work can take many different forms depending on the situation. So for instance, it might be helpful to CONCLUSION have a chair for the part of self that receives the criticism from the voice. This can give the person a chance to really connect with the We have proposed a framework in which voice-hearing can dominant-subordinate roles and ranking mentality in operation. be understood as an interaction between the brain’s evolved The compassionate self might, in this case, be positioned in systems for paying attention to, decoding and responding to more of a reflective role: firstly observing and listening to the different communication signals relating to different social roles. emotional conflict being played out before responding to both These signals may be auditory but can also be visual (e.g., parts with a wise overview of the threat-based functions, and then facial expressions). We suggest that some voice-hearers can mediating these parts toward some resolution or integration. This be particularly cued into communication patterns indicating highlights the flexibility required of the compassionate self, and dominant-subordinate social relationships. When that is the hence the importance in CFT of continuing to train the range social mentality by which the individual is processing social of qualities, attributes and skills throughout therapy. In some relationships, it becomes primed to detect information about circumstances, it may be the more gentle, warm and nurturing the controlling, critical, and even attacking intent of others, in qualities of compassion that are required. This might be true, for contrast to the helpfulness. We have argued that the relationships example, when the flow of compassion is directed to an abused between voices and voice-hearers can be understood in terms of child part. Whereas, in other situations, it might be the stronger, their alignment to social-rank relationships. We have proposed more assertive and courageous qualities that are required; for that voice-hearing involves an internal “playing-out” of both the example, when first intentionally opening to the verbal attacks hostile-dominant and the (reciprocal) threatened-subordinate

Frontiers in Psychology | www.frontiersin.org 7 February 2019 | Volume 10215 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 8

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

social roles. In other words, just as some (depressed) individuals with affiliative signals and patterns, would be beneficial for can experience anger toward themselves, people who are voice- voice-hearers, and that this would be evidenced by reducing hearers are experiencing these critical condemning attacking submission, depression and dissociation. Currently there is early emotions as voices rather than as thoughts or sentiments. So evidence indicating that compassionate mind training can be while a depressed person may have the thought, “I’ve let myself beneficial to voice-hearers (Mayhew and Gilbert, 2008; Braehler down; I’m useless; there is no point in living,” the voice-hearer et al., 2013), but further evaluation is needed. will hear a voice saying, “you’ve let yourself down; you’re useless; The framework outlined in this paper can not only inform there’s no point in you living.” In other words, there is an the continuing development of psychological interventions for internal experience of an attacking, condemning part of self and those distressed by the voices they hear, but can also potentially a submissive, beaten down part of self (Gilbert et al., 2001). inform contemporary neurocognitive models of voice-hearing. A number of testable, research hypotheses emerge from this One prominent neurocognitive model of voice-hearing stresses a framework. For example, if the voice-hearing experience does role for self-monitoring deficits in their etiology (Waters et al., indeed reflect evolved social-rank processes, we would expect 2012). Our framework adds crucial context to this model, by to find evidence for voice-hearer’s minds being increasingly specifying what type of information is likely being monitored orientated toward the dominant, controlling, power of others and why. Our framework can also help inform predictive coding (and voices), and toward the vulnerable, weak, powerlessness accounts of voice-hearing. Such models conceptualize voice- of self. This could be researched through psychological studies hearing as being a perception resulting from the overweighting into, e.g., biased attentional focus and memory recall, as well of prior expectations (typically under conditions of uncertainty) as through studies into the neurological and physiological at the expense of actual sensory input (Powers et al., 2018). Our underpinnings of threat system activation. Indeed, evidence framework suggests what specific priors are likely to underpin already suggests that when individuals experience these hostile voice-hearing (presence of social threat). It also suggests that dominant voices they also experience the social world as hostile- a crucial source of information for revising these priors will dominant over them, which is sometimes linked to having come from the body. Developing a bodily sense of security been harmed by others (Birchwood et al., 2000, 2004). Based through bodily practices such as breathing and grounding, in on this framework, we would also hypothesize that voice- addition to cognitive approaches, is likely to be necessary to revise hearing experiences would be accompanied by a range of other such priors. Finally, our framework shares much in common social-ranking characteristics, such as vulnerability to feelings with the hypervigilance model of voice-hearing, which proposes of defeat, inferiority, rejection, and shame, high sensitivity to that voices are a by-product of our perceptual system which social comparison, and that voice-hearing would elicit protective has evolved to reduce false negatives in conditions of threat strategies such as submission, depression, and dissociation. In (Dodgson and Gordon, 2009). Our framework extends this model particular, we would hypothesize that the more powerful a voice, by proposing that the evolved threat in question is likely to be and the more subordinate a voice-hearer, the higher the levels of, a social threat, and offers a particular method through which e.g., depression and dissociation. Again, there is some research to reduce such perceived threats (compassion). Thus, our model evidence already suggesting that this may be the case (Gilbert can be seen to offer a useful expansion of extant neurocognitive et al., 2001). models of voice-hearing. In this paper, we have additionally proposed that these In terms of informing the development of therapeutic social-rank processes can also be moderated by helping voice- approaches for voice-hearing, our framework can be seen to share hearers activate a different evolved motivational system, linked some commonalities with other recent therapeutic developments to mammalian caring behavior. The signals generated when in the field, such as Relating Therapy (Hayward et al., 2017) and this motivational system is operated are quite different. This Avatar Therapy (Craig et al., 2018; Craig, 2019), which both work motivational system is more likely to give rise to the experience with voice-hearers to change their relationship with their voices. of a secure base and safe haven, in line with attachment As our framework involves understanding of the dominant- theory, and we have suggested that this motivational switching subordinate motivating and monitoring systems it suggests (from competitive to caring) will bring therapeutic benefit a specific remedy to the problems caused by these systems; to voice-hearers. This also leads to a number of testable compassion, which is key to aiding affiliative relating. The focus research hypotheses around the moderating effects of secure on mindfulness in Compassion Focussed Therapy is consistent attachment and affiliative experiences. So, while our framework with a trend toward utilizing mindfulness techniques to assist not only generates hypotheses around how external and internal those distressed by voice-hearing (Strauss et al., 2015). However, social-ranking signals (such as stigma, shame, self-criticism, we suggest approaches could be extended by conceptualizing the and self-stigma), would accentuate threat-based dissociative learnt skill of mindfulness as a tool and vehicle toward enabling processes for voice-hearers, it also generates hypotheses about compassion. how external and internal communication signals that indicate Earlier, we also noted the use of memory and imagery caring, supportive, and social safeness experiences (e.g., social techniques in Compassion Focussed Therapy, to support the validation/connectedness and self-compassion) would attenuate development and training of compassionate psychologies, which them. Finally, from this framework, we would hypothesize can involve bringing to mind remembered and/or imagined that interventions specifically designed to help people switch compassionate relationships with others. This bears some from social-rank patterns into caring motivational systems, resemblance to the competitive memory training (COMET)

Frontiers in Psychology | www.frontiersin.org 8 February 2019 | Volume 10216 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 9

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

approach of van der Gaag et al. (2012). COMET identifies (dominant-subordinate) motivational systems, it could also be core themes of the cognitive-emotional memory network proposed that by CFT reducing the activation of such systems, activated by voices (such as “incompetence”), creates personal voice-hearing may actually be eliminated, even if this were not examples of the positive counter theme (e.g., instances of being the therapeutic goal. The effect of CFT on voice-hearing remains competent/successful), and puts these into a scene that the to be rigorously evaluated. patient has to imagine. The patient is then taught to activate Finally, it is worth considering the specific types of voice- this memory at will, with a view to being able to hear the hearing to which this framework may be applicable. The humiliating messages of the voice and not be emotionally affected framework we have set out focusses on voice-hearers who hear by it. An RCT of the COMET approach found that although voices with a negative affective valence. However, what does it did not lessen the voices or their negative content (which this framework have to say about why 40% of people diagnosed was not the aim of the study), patients’ depression did decrease with schizophrenia-spectrum disorders hear positive, supportive (van der Gaag et al., 2012). This was fully mediated by the voices (Nayani and David, 1996)? How can it help us understand increase in self-esteem and the acceptance of voices as psychic the voices heard by people who only hear positive, supportive phenomena, and partially mediated by the attributed power voices, or people who began by hearing positive, supportive to the voices and the social ranking of oneself in relation to voices before their voices turned malevolent (McCarthy-Jones, the voices. We would be interested to know whether a focus 2017a)? Could it be that activation of the social-threat monitoring on specifically compassionate imagery could alter the negative system could lead to negatively valenced voice-hearing, but that content of voices. Indeed, whilst Van der Gaag and colleagues its opposing force, the compassion system, could also seed voice- argue that interpersonal schemata, such as social rank, may hearing in the form of benevolent supportive voices? These could mediate the relation between voice-hearing and distress, our be internal versions of the secure base of attachment theory. framework proposes these may be involved of the etiology of the Such questions remain to be elaborated. However, we suggest that voices themselves. not only can our proposed framework lead to the refinement of This raises the question about what may be hypothesized contemporary models of the causes and treatment of distressing about how Compassion Focussed Therapy (CFT) can alter the voice-hearing, but that it can also provoke new questions to help phenomenology of voice-hearing. As noted above, the aim of us reconsider the nature of voice-hearing itself. such an approach is not to eliminate voices but to make them easier to live with. It can first be hypothesized that CFT would lead to a reduction in the negative affective valence of voice- AUTHOR CONTRIBUTIONS hearing; it will make nasty voices nicer. Given the lack of evidence that cognitive behavioral therapy can reduce the negative affective All authors listed have made a substantial, direct and intellectual valance of voices (McCarthy-Jones, 2017a), and that negative contribution to the work, and approved it for publication. valence is the single largest predictor of whether a voice-hearer will have a clinical diagnosis or not (Daalman et al., 2011), if CFT could do this it would be a major clinical achievement. Indeed, a FUNDING small case series has already offered some intriguing suggestions that CFT may actually be able to reduce the negative affective CH-M was supported by a Medical Research Council Clinical valence of people’ voices (Mayhew and Gilbert, 2008). And yet, Research Training Fellowship (MR/L01677X/1) to investigate if voice-hearing is grounded in the activation of threat-based Compassion Focused Therapy for Psychosis.

REFERENCES of a feasibility randomized controlled trial. Br. J. Clin. Psychol. 52, 199–214. doi: 10.1111/bjc.12009 Bell, V. (2013). A community of one: social cognition and auditory verbal Brett, C., Heriot-Maitland, C., McGuire, P., and Peters, E. (2014). Predictors of hallucinations. PLoS Biol. 11:e1001723. doi: 10.1371/journal.pbio.100 distress associated with psychotic-like anomalous experiences in clinical and 1723 non-clinical populations. Br. J. Clin. Psychol. 53, 213–227. doi: 10.1111/bjc. Bentall, R. P., Wickham, S., Shevlin, M., and Varese, F. (2012). Do specific early-life 12036 adversities lead to specific symptoms of psychosis? A study from the 2007 the Buss, D. M., and Dedden, L. A. (2016). Derogation of competitors. J. Soc. Pers. adult psychiatric morbidity survey. Schizophr. Bull. 38, 734–740. doi: 10.1093/ Relat. 7, 395–422. doi: 10.1177/0265407590073006 schbul/sbs049 Buss, L. W. (2014). The Evolution of Individuality. Princeton, NJ: Princeton Birchwood, M., Meaden, A., Trower, P., Gilbert, P., and Plaistow, J. (2000). The University Press. power and omnipotence of voices: subordination and entrapment by voices and Chadwick, P., Lees, S., and Birchwood, M. (2000). The revised beliefs about voices significant others. Psychol. Med. 30, 337–344. doi: 10.1017/S0033291799001828 questionnaire (BAVQ–R). Br. J. Psychiatry 177, 229–232. doi: 10.1192/bjp.177. Birchwood, M. A. X., Gilbert, P., Gilbert, J., Trower, P., Meaden, A., Hay, J., et al. 3.229 (2004). Interpersonal and role-related schema influence the relationship with Chance, M. R. (1967). Attention structure as the basis of primate rank orders. Man the dominant ‘voice’ in schizophrenia: a comparison of three models. Psychol. 2, 503–518. doi: 10.2307/2799336 Med. 34, 1571–1580. doi: 10.1017/s0033291704002636 Corstens, D., Longden, E., and May, R. (2012). Talking with voices: exploring Bowlby, J. (1973). Attachment and Loss: Separation: Anxiety and Anger (Vol. 2). what is expressed by the voices people hear. Psychosis 4, 95–104. doi: 10.1080/ New York: Basic Books. 17522439.2011.571705 Braehler, C., Gumley, A., Harper, J., Wallace, S., Norrie, J., and Gilbert, P. (2013). Corstens, D., Longden, E., McCarthy-Jones, S., Waddingham, R., and Thomas, N. Exploring change processes in compassion focused therapy in psychosis: results (2014). Emerging perspectives from the hearing voices movement: implications

Frontiers in Psychology | www.frontiersin.org 9 February 2019 | Volume 10217 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 10

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

for research and practice. Schizophr. Bull. 40(Suppl. 4), S285–S294. doi: 10.1093/ Hayward, M. (2003). Interpersonal relating and voice hearing: to what extent schbul/sbu007 does relating to the voice reflect social relating? Psychol. Psychother. 76(Pt 4), Craig, T. K., Rus-Calafell, M., Ward, T., Leff, J. P., Huckvale, M., Howarth, E., 369–383. doi: 10.1348/147608303770584737 et al. (2018). AVATAR therapy for auditory verbal hallucinations in people Hayward, M., Jones, A. M., Bogen-Johnston, L., Thomas, N., and Strauss, C. (2017). with psychosis: a single-blind, randomised controlled trial. Lancet Psychiatry Relating therapy for distressing auditory hallucinations: a pilot randomized 5, 31–40. doi: 10.1016/S2215-0366(17)30427-3 controlled trial. Schizophr. Res. 183, 137–142. doi: 10.1016/j.schres.2016. Craig, T. K. J. (2019). Avatar therapy: a promising new approach for persistent 11.019 distressing voices. World Psychiatry 18, 98–99. doi: 10.1002/wps.20589 Heriot-Maitland, C., Knight, M., and Peters, E. (2012). A qualitative comparison of Cultural Institute at King’s. (2015). Compassion for Voices: A Tale of Courage and psychotic-like phenomena in clinical and non-clinical populations. Br. J. Clin. Hope [online video] [Online]. Cultural Institute at King’s. Available at: https: Psychol. 51, 37–53. doi: 10.1111/j.2044-8260.2011.02011.x //www.youtube.com/watch?v=VRqI4lxuXAw [Accessed July 26 2018]. Heriot-Maitland, C., and Russell, G. (2018). “Compassion-focused therapy for Daalman, K., Boks, M. P., Diederen, K. M., de Weijer, A. D., Blom, J. D., Kahn, relating to voices,” in CBT for Psychosis: Process-Orientated Therapies and the R. S., et al. (2011). The same or different? A phenomenological comparison Third Wave, ed. C. Cupitt (London: Routledge). of auditory verbal hallucinations in healthy and psychotic individuals. J. Clin. Hostinar, C. E., Sullivan, R. M., and Gunnar, M. R. (2014). Psychobiological Psychiatry 72, 320–325. doi: 10.4088/JCP.09m05797yel mechanisms underlying the social buffering of the hypothalamic-pituitary- David, A. S. (2004). The cognitive neuropsychiatry of auditory verbal adrenocortical axis: a review of animal models and human studies hallucinations: an overview. Cogn. Neuropsychiatry 9, 107–123. across development. Psychol. Bull. 140, 256–282. doi: 10.1037/a003 doi: 10.1080/13546800344000183 2671 Dillon, J. (2013). “Recovery from ‘Psychosis’,” in Experiencing Psychosis: Personal Janssen, I., Krabbendam, L., Bak, M., Hanssen, M., Vollebergh, W., de and Professional Perspectives, eds J. Geekie, P. Randal, D. Lampshire, and J. Read Graaf, R., et al. (2004). Childhood abuse as a risk factor for psychotic (Hove: Routledge), 37–42. experiences. Acta Psychiatr. Scand. 109, 38–45. doi: 10.1046/j.0001-690X.2003. Dodgson, G., and Gordon, S. (2009). Avoiding false negatives: are some auditory 00217.x hallucinations an evolved design flaw? Behav. Cogn. Psychother. 37, 325–334. Jung, C. G. (1919). Instinct and the unconscious 1. Br. J. Psychol. 10, 15–23. doi: 10.1017/S1352465809005244 doi: 10.1111/j.2044-8295.1919.tb00003.x Dorahy, M. J., and Clearwater, K. (2012). Shame and guilt in men exposed to Kaufman, G. (1989). The Psychology of Shame. Theory and Treatment of Shame- childhood sexual abuse: a qualitative investigation. J. Child Sex. Abus. 21, Based Syndromes. New York: Springer. 155–175. doi: 10.1080/10538712.2012.659803 Kelleher, I., Keeley, H., Corcoran, P., Ramsay, H., Wasserman, C., Carli, V., et al. Dorahy, M. J., McKendry, H., Scott, A., Yogeeswaran, K., Martens, A., and (2013). Childhood trauma and psychosis in a prospective cohort study: cause, Hanna, D. (2017). Reactive dissociative experiences in response to acute effect, and directionality. Am. J. Psychiatry 170, 734–741. doi: 10.1176/appi.ajp. increases in shame feelings. Behav. Res. Ther. 89, 75–85. doi: 10.1016/j.brat. 2012.12091169 2016.11.007 Kirby, J. N. (2017). Compassion interventions: the programmes, the evidence, Dorahy, M. J., and Palmer, R. C. W. (2015). “Auditory hallucinations in and implications for research and practice. Psychol. Psychother. 90, 432–455. chronic trauma disorders: phenomenology and psychological mechanisms,” in doi: 10.1111/papt.12104 Comprehensive Guide to Post-Traumatic Stress Disorder, eds C. R. Martin, V. R. Kirby, J. N., Doty, J. R., Petrocchi, N., and Gilbert, P. (2017). The current and future Preedy, and V. B. Patel (Cham: Springer International Publishing). role of heart rate variability for assessing and training compassion. Front. Public Dunbar, R. (ed.). (2016). “The Social Brain Hypothesis and Human Evolution,” in Health 5:40. doi: 10.3389/fpubh.2017.00040 Oxford Research Encyclopedia of Psychology (Oxford: Oxford University Press). Larøi, F. (2012). How do auditory verbal hallucinations in patients differ from those doi: 10.1093/acrefore/9780190236557.013.44 in non-patients? Front. human neurosci. 6:25. doi: 10.3389/fnhum.2012.00025 Garwood, L., Dodgson, G., Bruce, V., and McCarthy-Jones, S. (2015). Legg, L., and Gilbert, P. (2006). A pilot study of gender of voice and gender of A preliminary investigation into the existence of a hypervigilance subtype of voice hearer in psychotic voice hearers. Psychol. Psychother. 79(Pt 4), 517–527. auditory hallucination in people with psychosis. Behav. Cogn. Psychother. 43, doi: 10.1348/147608305X70955 52–62. doi: 10.1017/S1352465813000714 Liotti, G., and Gumley, A. (2008). “An attachment perspective on schizophrenia: Gilbert, P. (1992). Depression: The Evolution of Powerlessness. New York, NY: The role of disorganized attachment, dissociation and mentalization,” Guilford Press. in Psychosis, Trauma and Dissociation: Emerging Perspectives on Severe Gilbert, P. (1998). The evolved basis and adaptive functions of cognitive Psychopathology, eds A. Moskowitz, I. Schäfer, and M. J. Dorahy (Hoboken, NJ: distortions. Br. J. Med. Psychol. 71, 447–463. doi: 10.1111/j.2044-8341.1998. Wiley-Blackwell), 117–133. tb01002.x Longden, E., Corstens, D., Escher, S., and Romme, M. (2012). Voice hearing Gilbert, P. (2000). The relationship of shame, social anxiety and depression: the in a biographical context: a model for formulating the relationship between role of the evaluation of social rank. Clin. Psychol. Psychother. 7, 174–189. voices and life history. Psychol. Soc. Integr. Approaches 4, 224–234. doi: 10.1080/ doi: 10.1002/1099-0879(200007)7:3<174::AID-CPP236>3.0.CO;2-U 17522439.2011.596566 Gilbert, P. (2002). Evolutionary approaches to psychopathology and cognitive Longden, E., Moskowitz, A., Dorahy, M., and Perona-Garcelán, S. (2018a). therapy. J. Cogn. Psychother. 16, 263–294. doi: 10.1891/jcop.16.3.263.52515 “Auditory verbal hallucinations: prevalence, phenomenology, and the Gilbert, P. (2016). Depression: The Evolution of Powerlessness. New York, NY: dissociation hypothesis,” in Psychosis, Trauma and Dissociation: Emerging Guilford Press. Perspectives on Severe Psychopathology, 2nd Edn, eds A. Moskowitz, M. J. Gilbert, P., Birchwood, M., Gilbert, J., Trower, P., Hay, J., Murray, B., et al. Dorahy and I. Schäfer (Chichester: Wiley-Blackwell). (2001). An exploration of evolved mental mechanisms for dominant and Longden, E., Read, J., and Dillon, J. (2018b). Assessing the impact and effectiveness subordinate behaviour in relation to auditory hallucinations in schizophrenia of hearing voices network self-help groups. Commun. Ment. Health J. 54, and critical thoughts in depression. Psychol. Med. 31, 1117–1127. doi: 10.1017/ 184–188. doi: 10.1007/s10597-017-0148-1 S0033291701004093 Longden, E., Sampson, M., and Read, J. (2016). Childhood adversity and psychosis: Gilbert, P., McEwan, K., Matos, M., and Rivis, A. (2011). Fears of compassion: generalised or specific effects? Epidemiol. Psychiatr. Sci. 25, 349–359. doi: 10. development of three self-report measures. Psychol. Psychother. 84, 239–255. 1017/S204579601500044X doi: 10.1348/147608310X526511 Longe, O., Maratos, F. A., Gilbert, P., Evans, G., Volker, F., Rockliff, H., et al. Gilbert, P., and McGuire, M. T. (1998). “Shame, status, and social roles: (2010). Having a word with yourself: neural correlates of self-criticism and Psychobiology and evolution,” in Series in Affective Science. Shame: self-reassurance. Neuroimage 49, 1849–1856. doi: 10.1016/j.neuroimage.2009. Interpersonal Behavior, Psychopathology, and Culture, eds P. Gilbert and 09.019 B. Andrews (New York, NY: Oxford University Press), 99–125. Mascaro, J. S., Rilling, J. K., Tenzin Negi, L., and Raison, C. L. (2013). Compassion Greenberg, L. S., and Watson, J. C. (2006). Emotion-focused Therapy for Depression. meditation enhances empathic accuracy and related neural activity. Soc. Cogn. Washington, DC: American Psychological Association. Affect. Neurosci. 8, 48–55. doi: 10.1093/scan/nss095

Frontiers in Psychology | www.frontiersin.org 10 February 2019 | Volume 10218 | Article 152 fpsyg-10-00152 January 30, 2019 Time: 17:22 # 11

APPENDIX 7.1 Heriot-Maitland et al. CFT for Voices

Matos, M., Duarte, J., Duarte, C., Gilbert, P., and Pinto-Gouveia, J. (2017). How Stockmann, T., Wood, L., Enache, G., Withers, F., Gavaghan, L., and Razzaque, R. one experiences and embodies compassionate mind training influences its (2017). Peer-supported open dialogue: a thematic analysis of trainee effectiveness. Mindfulness. 9, 1224–1235. doi: 10.1007/s12671-017-0864-1 perspectives on the approach and training. J. Ment. Health doi: 10.1080/ Mayhew, S. L., and Gilbert, P. (2008). Compassionate mind training with people 09638237.2017.1340609 [Epub ahead of print]. who hear malevolent voices: a case series report. Clin. Psychol. Psychother. 15, Stone, H., and Stone, S. (2011). Embracing Ourselves: The Voice Dialogue Manual. 113–138. doi: 10.1002/cpp.566 Novato, CA: New World Library. McCarthy-Jones, S. (2012). Hearing Voices: The Histories, Causes and Meanings of Strauss, C., Thomas, N., and Hayward, M. (2015). Can we respond mindfully Auditory Verbal Hallucinations. Cambridge: Cambridge University Press. to distressing voices? A systematic review of evidence for engagement, McCarthy-Jones, S. (2017a). Can’t You Hear Them? The Science and Significance of acceptability, effectiveness and mechanisms of change for mindfulness-based Hearing Voices. London: Jessica Kingsley Publishers. interventions for people distressed by hearing voices. Front. Psychol. 6:1154. McCarthy-Jones, S. (2017b). Is shame hallucinogenic? Front. Psychol. 8:1310. doi: doi: 10.3389/fpsyg.2015.01154 10.3389/fpsyg.2017.01310 Talbot, J. A., Talbot, N. L., and Tu, X. (2004). Shame-proneness as a McCarthy-Jones, S., Thomas, N., Strauss, C., Dodgson, G., Jones, N., Woods, A., diathesis for dissociation in women with histories of childhood sexual et al. (2014). Better than mermaids and stray dogs? Subtyping auditory verbal abuse. J. Traumatic Stress 17, 445–448. doi: 10.1023/B:JOTS.0000048959.29 hallucinations and its implications for research and practice. Schizophr. Bull. 40 766.ae (Suppl. 4), S275–S284. doi: 10.1093/schbul/sbu018 van der Gaag, M., van Oosterhout, B., Daalman, K., Sommer, I. E., and McGrath, J. J., Saha, S., Al-Hamzawi, A., Alonso, J., Bromet, E. J., Bruffaerts, R., Korrelboom, K. (2012). Initial evaluation of the effects of competitive et al. (2015). Psychotic experiences in the general population: a cross-national memory training (COMET) on depression in schizophrenia-spectrum patients analysis based on 31 261 respondents from 18 countries. JAMA Psychiatry 72, with persistent auditory verbal hallucinations: a randomized controlled 697–705. doi: 10.1001/jamapsychiatry.2015.0575 trial. Br. J. Clin. Psychol. 51, 158–171. doi: 10.1111/j.2044-8260.2011. Moskowitz, A., and Corstens, D. (2008). Auditory hallucinations: psychotic 02025.x symptom or dissociative experience? J. Psychol. Trauma 6, 35–63. doi: 10.1300/ Van der Hart, O., Nijenhuis, E. R., and Steele, K. (2006). The Haunted J513v06n02_04 Self: Structural Dissociation and the Treatment of Chronic Traumatization. Moskowitz, A., Mosquera, D., and Longden, E. (2017). Auditory verbal New York, NY: WW Norton & Company. hallucinations and the differential diagnosis of schizophrenia and dissociative Varese, F., Smeets, F., Drukker, M., Lieverse, R., Lataster, T., Viechtbauer, W., et al. disorders: historical, empirical and clinical perspectives. Eur. J. Trauma (2012). Childhood adversities increase the risk of psychosis: a meta-analysis of Dissociation 1, 37–46. doi: 10.1016/j.ejtd.2017.01.003 patient-control, prospective- and cross-sectional cohort studies. Schizophr. Bull. Mosquera, D., and Ross, C. (2017). A psychotherapy approach to treating hostile 38, 661–671. doi: 10.1093/schbul/sbs050 voices. Psychosis-Psychological Social and Integrative Approaches 9, 167–175. Waite, F., Knight, M. T., and Lee, D. (2015). Self-compassion and self-criticism doi: 10.1080/17522439.2016.1247190 in recovery in psychosis: an interpretative phenomenological analysis study. Nayani, T. H., and David, A. S. (1996). The auditory hallucination: a J. Clin. Psychol. 71, 1201–1217. doi: 10.1002/jclp.22211 phenomenological survey. Psychol. Med. 26, 177–189. Waters, F., Allen, P., Aleman, A., Fernyhough, C., Woodward, T. S., Badcock, J. C., Ogilvie, D. M. (1987). The undesired self - a neglected variable in personality- et al. (2012). Auditory hallucinations in schizophrenia and nonschizophrenia research. J. Pers. Soc. Psychol. 52, 379–385. doi: 10.1037//0022-3514.52.2.379 populations: a review and integrated model of cognitive mechanisms. Schizophr. Pace, T. W., Negi, L. T., Adame, D. D., Cole, S. P., Sivilli, T. I., Brown, T. D., et al. Bull. 38, 683–693. doi: 10.1093/schbul/sbs045 (2009). Effect of compassion meditation on neuroendocrine, innate immune Weng, H. Y., Fox, A. S., Shackman, A. J., Stodola, D. E., Caldwell, J. Z., and behavioral responses to psychosocial stress. Psychoneuroendocrinology 34, Olson, M. C., et al. (2013). Compassion training alters altruism and neural 87–98. doi: 10.1016/j.psyneuen.2008.08.011 responses to suffering. Psychol. Sci. 24, 1171–1180. doi: 10.1177/095679761246 Pilton, M., Varese, F., Berry, K., and Bucci, S. (2015). The relationship between 9537 dissociation and voices: a systematic literature review and meta-analysis. Clin. Psychol. Rev. 40, 138–155. doi: 10.1016/j.cpr.2015.06.004 Conflict of Interest Statement: CH-M and PG are practitioners, supervisors and Porges, S. W. (2007). The polyvagal perspective. Biol. Psychol. 74, 116–143. doi: trainers of CFT and have received fees for providing these services. EL is also a 10.1016/j.biopsycho.2006.06.009 trainer of CFT and has received fees for this. PG receives royalties from books he Powers, A. R. III, Bien, C., and Corlett, P. R. (2018). Aligning computational has published on CFT. psychiatry with the hearing voices movement: hearing their voices. JAMA Psychiatry 75, 640–641. doi: 10.1001/jamapsychiatry.2018.0509 The remaining author declares that the research was conducted in the absence of Read, J., Agar, K., Argyle, N., and Aderhold, V. (2003). Sexual and physical abuse any commercial or financial relationships that could be construed as a potential during childhood and adulthood as predictors of hallucinations, delusions conflict of interest. and thought disorder. Psychol. Psychother. 76(Pt 1), 1–22. doi: 10.1348/ 14760830260569210 Copyright © 2019 Heriot-Maitland, McCarthy-Jones, Longden and Gilbert. This Romme, M., Escher, S., Dillon, J., Corstens, D., and Morris, M. (2009). Living with is an open-access article distributed under the terms of the Creative Commons Voices: 50 Stories of Recovery. (Ross-on-Wye: PCCS Books), 350. Attribution License (CC BY). The use, distribution or reproduction in other forums Seikkula, J., Aaltonen, J., Alakare, B., Haarakangas, K., Keränen, J., and Lehtinen, K. is permitted, provided the original author(s) and the copyright owner(s) are credited (2006). Five-year experience of first-episode nonaffective psychosis in open- and that the original publication in this journal is cited, in accordance with accepted dialogue approach: Treatment principles, follow-up outcomes, and two case academic practice. No use, distribution or reproduction is permitted which does not studies. Psychother. Res. 16, 214–228. doi: 10.1080/10503300500268490 comply with these terms.

Frontiers in Psychology | www.frontiersin.org 11 February 2019 | Volume 10219 | Article 152 APPENDIX 7.2

Appendix 7.2. Feedback from the 20 voice-hearers who engaged with ‘Compassion for Voices’ Engagement theme Feedback from voice-hearers Feedback that helped to gauge “as a person who hears voices and struggles with them this film has the acceptability and usefulness made me understand in a positive way how to be compassionate to of the CFTp concept of my voices but it's for me a struggle and still have not overcome but the compassion for voices film in an easier understanding way to be nice to myself and my voices”

“It has encouraged me to continue practicing on accepting the voices. It has also made me understand that I'm not the only one and that I can handle it.”

“It made me rethink how I approach my voices and whether I should try and interact with them - as terrifying as that seems.”

“Yes turn to even the most abusive inner voice with self-compassion. This inner core of love will integrate even the worst of energies. Great video!!”

“To recognize and accept the voices reduces internal resistance resulting to a healthier living”

Feedback that helped to gauge “Oh my God. I can't believe the video. It is EXACTLY what goes on in my acceptability of using the film in head. This is the FIRST time I have EVER come across something that CFTp to reflect the voice- nails my experience. And the voices. Thank you SO MUCH” hearing experience “I like how this video absolutely nails many of my thoughts and feelings surrounding voices Thank you for this film!”

“The voices were eerily similar to the types I hear. Thank you for the film. It has really helped.”

Feedback that helped to gauge “The graphics in this short film are extremely sympathetic and give me acceptability of using the film as a happy warm feeling about a subject which is often dealt with in a a CFTp therapeutic tool scary or negative manner.”

“The visual depiction of the facial expressions of the voices is very powerful--more powerful than just saying they were negative, they were threatening, they were supportive, etc.”

“My therapist directed me to this and it really just simplifies and clarifies some basics to allow insight for both sufferers, relatives and the wider public”

“Usually the voices are mean and threatening and I wonder why I deserve so much pain. This film has shown me that I'm basically a good person”

“Had a psychiatrist who told me to live with it. This will help me live with it.”

220 APPENDIX 7.2

“The film was great. I watch it at least once per week.” Feedback that helped to gauge “The film can speak for those of us who find it too difficult to share our acceptability of using the film experience of hearing voices directly.” for engaging others with the. CFTp intervention (e.g. family & “Most of the people will recognize their selves and thus understand social network) others”

“May show family members how it really is”

“De stigmatise this issue”

“people may be more supportive if they understand”

“Can help people appreciate the challenges voice hearers experience, that treating them with compassion makes a difference, and it also normalizes the hearing voices experience.”

“It brings the conversation to the public and encourages people to talk about these experiences.”

Feedback that helped to inform “the film was very good for me to recognise that other people have the CFTp Manual guidance on these thoughts to and can manage them” ‘normalising’ psycho-education (CFTp manual, section 2.4.3) “This helps normalize the experience.”

“for the general public who fear people who hear voices this is a good educational film that would help the general population have a more compassionate understanding to people who hear voices”

“I wish it would but hope that the public don't take the mickey out of voice hearers”

Feedback that helped to inform “Very powerful film, scared of my voices telling me what to do. Can we the CFTp Manual guidance on all get a compassionate voice? This video shows exactly what my ‘making sense of voices’ voices are like and it doesn’t try to tell you it is all in your mind either. (CFTp manual, section 3.4) It accepts our voices, where can I learn to accept mine?”

“I am not abnormal for these feelings in me. I need to manage them better.”

“Voices don't necessarily mean sickness”

“The video does not distinguish between internal dialogue voices and the literally, audible outside voices, or the voice like mine, an intruder with a presence, messages and a personality outside of my usual self but without audible words."

221 APPENDIX 8.1

Appendix 8.1. Random allocation of multiple baselines (2/4/6 weeks)

222 APPENDIX 8.2

NRES Committee London - Dulwich Health Research Authority Skipton House 80 London Road London SE1 6LH

Telephone: 020 7972 2582 09 March 2015

Dr Charles Heriot-Maitland MRC Clinical Research Training Fellow / PhD student King's College London Institute of Psychiatry, Psychology & Neuroscience (PO78, B4.08) De Crespigny Park London SE5 8AF

Dear Dr Heriot-Maitland

Study title: A case series study of Compassion Focused Therapy for distressing experiences REC reference: 15/LO/0198 IRAS project ID: 167443

Thank you for your letter of 01 March 2015, responding to the Committee’s request for further information on the above research and submitting revised documentation.

The further information was considered in correspondence by a Sub-Committee of the REC. A list of the Sub-Committee members is attached.

We plan to publish your research summary wording for the above study on the HRA website, together with your contact details. Publication will be no earlier than three months from the date of this favourable opinion letter. The expectation is that this information will be published for all studies that receive an ethical opinion but should you wish to provide a substitute contact point, wish to make a request to defer, or require further information, please contact the REC Manager, Stephanie Hill, [email protected]. Under very limited circumstances (e.g. for student research which has received an unfavourable opinion), it may be possible to grant an exemption to the publication of the study.

Confirmation of ethical opinion

On behalf of the Committee, I am pleased to confirm a favourable ethical opinion for the above research on the basis described in the application form, protocol and supporting documentation as revised, subject to the conditions specified below.

Conditions of the favourable opinion

A Research Ethics Committee established by the Health Research Authority

223 APPENDIX 8.2

The favourable opinion is subject to the following conditions being met prior to the start of the study.

Management permission or approval must be obtained from each host organisation prior to the start of the study at the site concerned.

Management permission ("R&D approval") should be sought from all NHS organisations involved in the study in accordance with NHS research governance arrangements.

Guidance on applying for NHS permission for research is available in the Integrated Research Application System or at http://www.rdforum.nhs.uk.

Where a NHS organisation’s role in the study is limited to identifying and referring potential participants to research sites ("participant identification centre"), guidance should be sought from the R&D office on the information it requires to give permission for this activity.

For non-NHS sites, site management permission should be obtained in accordance with the procedures of the relevant host organisation.

Sponsors are not required to notify the Committee of approvals from host organisations

Registration of Clinical Trials

All clinical trials (defined as the first four categories on the IRAS filter page) must be registered on a publically accessible database. This should be before the first participant is recruited but no later than 6 weeks after recruitment of the first participant.

There is no requirement to separately notify the REC but you should do so at the earliest opportunity e.g. when submitting an amendment. We will audit the registration details as part of the annual progress reporting process.

To ensure transparency in research, we strongly recommend that all research is registered but for non-clinical trials this is not currently mandatory.

If a sponsor wishes to request a deferral for study registration within the required timeframe, they should contact [email protected]. The expectation is that all clinical trials will be registered, however, in exceptional circumstances non registration may be permissible with prior agreement from NRES. Guidance on where to register is provided on the HRA website.

It is the responsibility of the sponsor to ensure that all the conditions are complied with before the start of the study or its initiation at a particular site (as applicable).

Ethical review of research sites

NHS sites

The favourable opinion applies to all NHS sites taking part in the study, subject to management permission being obtained from the NHS/HSC R&D office prior to the start of the study (see "Conditions of the favourable opinion" below).

A Research Ethics Committee established by the Health Research Authority

224 APPENDIX 8.2

Non-NHS sites

The Committee has not yet completed any site-specific assessment (SSA) for the non-NHS research site(s) taking part in this study. The favourable opinion does not therefore apply to any non-NHS site at present. We will write to you again as soon as an SSA application(s) has been reviewed. In the meantime no study procedures should be initiated at non-NHS sites.

Approved documents

The final list of documents reviewed and approved by the Committee is as follows: Document Version Date Covering letter on headed paper [Covering letter] 2 01 March 2015 Evidence of Sponsor insurance or indemnity (non NHS Sponsors 1 14 July 2014 only) [KCL indemnity documents] GP/consultant information sheets or letters [Health professional 1 18 December 2014 information letter] Non-validated questionnaire [Therapy adherence and competence 1 18 December 2014 measure] Other [Letter about research sites] 1 15 January 2015 Participant consent form [Participant consent form] 2 24 February 2015 Participant information sheet (PIS) [Participant information sheet] 2 24 February 2015 REC Application Form [REC_Form_19012015] 19 January 2015 Referee's report or other scientific critique report [MRC reviewers 1 14 November 2013 comments on 4 studies] Research protocol or project proposal [Research protocol] 1 18 December 2014 Summary CV for Chief Investigator (CI) [CV Charles 1 18 December 2014 Heriot-Maitland] Summary CV for Chief Investigator (CI) [CV Charles 1 18 December 2014 Heriot-Maitland] Summary CV for supervisor (student research) [CV Emmanuelle 1 06 February 2014 Peters] Validated questionnaire [Outcome measure - PSYRATS Delusions] Validated questionnaire [Outcome measure - PSYRATS Hallucinations] Validated questionnaire [Outcome measure - DES-II] Validated questionnaire [Outcome measure - DASS-21] Validated questionnaire [Outcome measure - CORE] Validated questionnaire [Process measure - SRT scale] Validated questionnaire [Process measure - PBIQ-R] Validated questionnaire [Sessional measure - SSPS] Validated questionnaire [Sessional measure - Dissociation]

Statement of compliance

The Committee is constituted in accordance with the Governance Arrangements for Research Ethics Committees and complies fully with the Standard Operating Procedures for Research Ethics Committees in the UK.

A Research Ethics Committee established by the Health Research Authority

225 APPENDIX 8.2

After ethical review

Reporting requirements

The attached document “After ethical review – guidance for researchers” gives detailed guidance on reporting requirements for studies with a favourable opinion, including:

 Notifying substantial amendments  Adding new sites and investigators  Notification of serious breaches of the protocol  Progress and safety reports  Notifying the end of the study

The HRA website also provides guidance on these topics, which is updated in the light of changes in reporting requirements or procedures.

User Feedback

The Health Research Authority is continually striving to provide a high quality service to all applicants and sponsors. You are invited to give your view of the service you have received and the application procedure. If you wish to make your views known please use the feedback form available on the HRA website: http://www.hra.nhs.uk/about-the-hra/governance/quality-assurance/

HRA Training

We are pleased to welcome researchers and R&D staff at our training days – see details at http://www.hra.nhs.uk/hra-training/

15/LO/0198 Please quote this number on all correspondence

With the Committee’s best wishes for the success of this project.

Yours sincerely

Dr Michael Philpot Chair

Email:[email protected]

Enclosures: List of names and professions of members who were present at the meeting and those who submitted written comments

A Research Ethics Committee established by the Health Research Authority

226 APPENDIX 8.2

“After ethical review – guidance for researchers” [SL-AR2]

Copy to: Mr Keith Brennan Ms Jenny Liebscher, South London and Maudsley NHS Foundation Trust

A Research Ethics Committee established by the Health Research Authority

227 APPENDIX 8.2

NRES Committee London - Dulwich

Attendance at Sub-Committee of the REC meeting on 04 March 2015

Committee Members:

Name Profession Present Notes Dr Michael Philpot Consultant Psychiatrist Yes Ms Anna Ramberg Research Development Yes Manager Mr Colin Standfield Charity Worker Yes

Also in attendance:

Name Position (or reason for attending) Miss Stephanie Hill REC Co-ordinator

A Research Ethics Committee established by the Health Research Authority

228 APPENDIX 8.3

Research funded by:

Participant Information Sheet “A case series study of Compassion Focused Therapy for distressing experiences”

Invitation to take part in this study

We’d like to invite you to take part in our research study. Before you decide whether to take part it is important for you to understand why the research is being done and what it would involve for you. Please take time to read the information carefully and discuss it with others if you wish. Ask us if there is anything that is not clear or if you would like more information. Take time to decide whether or not you wish to take part.

What is the purpose of the study?

This study aims to develop and test the feasibility of a new therapy called Compassion Focused Therapy (CFT). This therapy helps people manage distressing experiences by helping people to feel safe in relation to their experiences and to develop compassion towards themselves and others. In order to keep improving the help we can provide for people with distressing experiences, we need to keep improving our therapies by testing out new approaches. A good place to start in improving current therapies is to look at some of the approaches that people are finding helpful for other mental health difficulties, such as anxiety and depression. CFT is a promising new approach that has been successfully provided for people with a range of difficulties. It is also based in the most up-to-date knowledge and science about how the mind works (both normally and under stress).

If you participate, you will have some individual sessions of CFT with Dr Charlie Heriot-Maitland, who is one of the psychologists in the service. This therapy will last up to 26 sessions (about 6 months), which is the same as standard therapy (ie Cognitive Behaviour Therapy; CBT). Not only will you be receiving this new therapy, but you will also be helping us develop and improve it as we go along; i.e. helping us learn how to help others with similar difficulties in the future.

Why have I been invited to take part?

You have been invited because you have recently been referred to see a therapist in NHS Greater Glasgow and Clyde, and you meet the study criteria because you have distressing experiences.

Do I have to take part?

It is entirely up to you to decide whether or not to take part. Your decision whether or not to take part will have no effect on the current services and treatments you are receiving, your legal rights, nor will it affect your access to the routine therapy offered by the service (CBT). If you did wish to have CBT, as well as the Compassion Focused Therapy (CFT) in this study, then this would have to start after the CFT has finished because the two therapies cannot run at the same time. You would stay on the waiting list for CBT while in the CFT study, and your place on the list would continue to move up over time as normal.

If you do decide to take part, you will be given this information sheet to keep and will be asked to sign a consent form. You are still free to withdraw at any time in the process of the study without giving a reason.

What will happen to me if I take part?

Taking part will involve up to 26 therapy sessions (1 hour weekly, over 6 months), along with 5 questionnaire sessions. The 5 questionnaire sessions will be spread out over the therapy: 1. as soon as possible; 2. just Participant Information Sheet Version 4, 6th May 2015 Research Ethics Committee ref: 15/LO/0198 229 APPENDIX 8.3

before starting therapy (2-6 weeks later); 3. in the middle of therapy; 4. at the end of therapy; and 5. 6-8 weeks after therapy. These sessions will last about 30 minutes, including a short interview about your experiences, and a few questionnaires about, e.g., how you see yourself in relation to others. We will also ask your permission to monitor your heart rate briefly for 6 minutes at the end of the questionnaire sessions, but this is optional, so if you don’t give permission, you can still take part in the study. You will be reimbursed £10 + travel expenses for each questionnaire session (so total £50 + travel). In the period before starting therapy (2-6 weeks), you will have brief contact with the therapist each week to complete a short 2-minute questionnaire (either over the phone or in person, with any travel expenses reimbursed). This short questionnaire will also be completed in the first 2 minutes of each therapy session.

As this is research, your therapist will ask your permission to audio or video record the therapy sessions. This is simply so that we can analyse and think about what’s been said, and how we responded to that. This will help us learn. The only people who will hear or watch the tapes are the therapist, the supervisor, and one external expert in CFT who can check that the therapy is adhered to. All recordings will be anonymised and confidential, so the supervisor and expert will not know your name or any other personal information. You can ask for the recordings any time you want, and you can also ask for them to be deleted at any time. All recordings will be deleted anyway at the end of the study. If you would like to have your own copies of the recordings, you are most welcome, and indeed we would be very interested in your thoughts. After all, this is really about helping to understand you and how to be more helpful for you and others.

What will happen if I start but then don’t want to carry on with the study?

You can withdraw from the study at any time without having to justify your decision. If you decide to withdraw from the study you can tell us whether you are happy for us to use the information obtained up to that point. If you are not, any information that you have given will be destroyed and you will not be contacted by us again.

What are the possible risks of taking part?

As with most therapies, you may be talking about issues that are sensitive and/or distressing for you, such as difficult experiences and relationships. However, you don’t have to talk about anything you don't want to, and actually a major component of CFT is helping people feel safe in relation to their experiences. If difficult feelings do arise, your therapist is highly trained and experienced in guiding people through emotional conversations. Your therapist will also be able to contact other professionals and services if necessary.

What are the possible benefits of taking part?

CFT is intended to have beneficial effects for people, e.g., in reducing emotional distress and improving psychological well-being, and there is growing evidence for these benefits in a range of people using mental health services. There is a chance that participants in this study will experience these therapeutic benefits. In addition, it may also be a positive experience for participants to have an opportunity to contribute to other people’s well-being by being involved with developing and researching a new therapy.

Will my taking part be kept confidential?

All the information collected during the course of the research will be kept strictly confidential, and will be stored in accordance with the Data Protection Act 1998, secured against unauthorised access. Your personal details will not be seen by anyone other than the primary researcher. Your questionnaire data and audio/video recordings will be assigned an anonymous code before they are used for supervision purposes with the research and therapy supervisors.

Your therapist will adhere to the normal NHS Trust guidelines and procedures on service-user confidentiality. With your permission, the therapist will liaise with other people involved with your care, such as GP or Care- Coordinator, and will keep brief notes about sessions for the clinical records, e.g. saying when you met etc., which are only accessible to those involved with your care. The therapist won’t write things that you don’t want them to, and you can look at these notes if you like. The only limits to confidentiality would be if, for whatever reason, things get very difficult for you, and there are issues of harm (e.g. you say something that suggests there may be a risk of harm to yourself or others), then the therapist may have to share information with others. Please note that this is likely to be a very rare occurrence.

How is the project being funded?

The project is funded through a Medical Research Council (MRC) Clinical Research Training Fellowship awarded to Dr Charlie Heriot-Maitland (ref: MR/L01677X/1).

Participant Information Sheet Version 4, 6th May 2015 Research Ethics Committee ref: 15/LO/0198 230 APPENDIX 8.3

What will happen to the results of the study?

The research should be completed by the end of 2018. The results will be written-up as part of a PhD, and submitted for a peer-reviewed journal. This will be openly accessible through King’s College London’s publications repository. No individual will be identifiable from the published results. A summary will be published in the service newsletter, and a copy of this summary will be sent to participants if they wish.

What if something goes wrong?

Any complaint about the way you have been dealt with during the study or any possible harm you might suffer will be addressed. If you have a concern about any aspect of this study, you can speak with the Primary Researcher (contact details below) in the first instance or the Research Supervisor (contact details below), who will do their best to answer your questions. If you remain unhappy and wish to complain formally, you can do this through the NHS complaints procedure (below).

Compensation for harm arising from an accidental injury and occurring as a consequence of your participation in the study will be covered by King’s College London. In the event that something does go wrong and you are harmed during the research and this is due to someone‘s negligence then you may have grounds for a legal action for compensation against King’s College London but you may have to pay your legal costs. The normal National Health Service complaints mechanisms will still be available to you (if appropriate).

Who has reviewed the study?

This research has been reviewed and funded by the Medical Research Council. People with experience of using NHS psychosis services have also been involved with advising on how to conduct this study in these services. All research in the NHS is also looked at by an independent group of people called a Research Ethics Committee, in order to protect your safety, rights, wellbeing and dignity. This study has been reviewed and approved by the NRES Committee London – Dulwich (REC ref: 15/LO/0198)

Contact details

If you have any questions relating to this research, or concerns about participation, please contact:

Primary Researcher Dr Charlie Heriot-Maitland, MRC Clinical Research Training Fellow University of Glasgow, Institute of Health and Wellbeing, Fleming Pavilion, Todd Campus, West of Scotland Science Park, Glasgow G20 0XA Email: [email protected] / Tel: 07710 386138

Research Supervisors Dr Emmanuelle Peters, Reader in Clinical Psychology King’s College London, IoPPN (PO77), De Crespigny Park, London SE5 8AF Email: [email protected] / Tel: 020 7848 0347

Prof Andrew Gumley, Professor of Psychological Therapy University of Glasgow, Institute of Health and Wellbeing, Fleming Pavilion, Todd Campus, West of Scotland Science Park, Glasgow G20 0XA Email: [email protected]

If you wish to make a complaint about the conduct of this study, you may speak to someone in your NHS clinical team initially, or if you would like to make a formal complaint, you can do this through the NHS Complaints Procedure.

Phone: 0141 201 4500 (for complaints only) Email: [email protected]

Thank you for reading this information and for considering taking part in this research.

Participant Information Sheet Version 4, 6th May 2015 Research Ethics Committee ref: 15/LO/0198 231 APPENDIX 8.4

232 APPENDIX 8.4

233 APPENDIX 8.5

Coordinator/administrator: Dr George Bakirtzis / RP Clinical Research & Development Telephone Number: 0141 232 1825 West Glasgow ACH E-Mail: [email protected] Dalnair Street Website: www.nhsggc.org.uk/r&d Glasgow G3 8SJ Scotland, UK

26/01/2018

Dr Charles Heriot-Maitland University of Glasgow Institute of Health and Wellbeing Fleming Pavilion, Todd Campus West of Scotland Science Park Glasgow G12 0XH

NHS GG&C Board Approval Dear Dr Heriot-Maitland

Study Title: A case series study of Compassion Focused Therapy for distressing experiences Principal Investigator: Dr Charles Heriot-Maitland GG&C HB site Community Mental Health Co-Sponsors King's College London; South London and Maudsley NHS Foundation Trust R&D reference: GN17MH631 REC reference: 15/LO/0198 Protocol no: Version 2 dated 06/05/2015 (including version and date)

I am pleased to confirm that Greater Glasgow & Clyde Health Board is now able to grant Approval for the above study.

Conditions of Approval 1. For Clinical Trials as defined by the Medicines for Human Use Clinical Trial Regulations, 2004 a. During the life span of the study GGHB requires the following information relating to this site i. Notification of any potential serious breaches. ii. Notification of any regulatory inspections.

It is your responsibility to ensure that all staff involved in the study at this site have the appropriate GCP training according to the GGHB GCP policy (www.nhsggc.org.uk/content/default.asp?page=s1411), evidence of such training to be filed in the site file.

2. For all studies the following information is required during their lifespan. a. Recruitment Numbers on a quarterly basis b. Any change of staff named on the original SSI form

Page 1 of 2 NHS GG&C Board 234 Approval_GN17MH631 APPENDIX 8.5

c. Any amendments – Substantial or Non Substantial d. Notification of Trial/study end including final recruitment figures e. Final Report & Copies of Publications/Abstracts

Please add this approval to your study file as this letter may be subject to audit and monitoring. Your personal information will be held on a secure national web-based NHS database. I wish you every success with this research study

Yours sincerely,

Dr George Bakirtzis Research Facilitator

Page 2 of 2 NHS GG&C Board 235 Approval_GN17MH631

C. Heriot-Maitland (lead researcher/therapist/author) E. Peters; T. Wykes; A. Gumley (research supervisors) P. Gilbert; E. Longden; C. Irons (therapy collaborators)

CFT for Psychosis – Manual Full doc: www.compassionforvoices.com/CFTpManual (click link to access full 50-page manual) Overview of therapy levels The ordering of 1-5 levels is to guide the therapist through the therapy content. In reality this is a process-driven therapy, so therapists will be following the client, using clinical judgment, supervision, and collaborative discussion to transition to/from each level. This may involve re-ordering levels 1-5, blending one with another, or skipping one out completely. The phrases below are also to guide, rather than to quote. The therapist will be talking to clients in their own language, at their own pace, using Socratic questioning, and attending to process.

Starting therapy S.1. Human connection and collaboration S.2. Feelings about being here, and hopes S.3. Definitions of compassion Establishing 1. 1 safeness and 1.1. External safeness – social safeness from the ‘outside-in’ (↑↓) 1.1.1. Connectedness (family, friends, peers, communities) connection 1.1.2. Environments (physical) 1.2. Internal safeness – social safeness from the ‘inside-out’ 1.2.1. Safeness physiology (grounding, body, breathing) 1.2.2. Safeness imagery (safe place) Learning about 2.1. Tricky brain and loops 2 evolved (tricky) 2.2. 3 circles: threat; drive; soothing (↑↓) 2.3. The power of attention brains, emotional 2.4. Normal brains operating under difficult conditions systems, & multiple 2.4.1. Multiple selves selves 2.4.2. Dissociation 2.4.3. Hearing voices and over-estimating threat Understanding how 3.1. ‘3 circles’ assessment & formulation 3 my emotions and 3.2. Threat-focused assessment & formulation (↑↓) 3.3. Social mentality assessment & formulation mind have become 3.4. Functional analysis & making sense of voices / beliefs shaped Building the 4.1. Qualities of the compassionate self 4 compassionate self 4.2. Compassionate Mind Training (CMT) (↑↓) 4.2.1. Body posture, facial expression, voice tone 4.2.2. Compassionate imagery (self and other) 4.2.3. Fears, blocks, & resistances (FBRs) 4.3. Becoming your compassionate self in daily life Directing 5.1. Directing compassion to self / multiple selves / emotional parts / voices 5 compassion to self, 5.1.1. Compassionate relating in sessions (↑↓) 5.1.2. Compassionate relating in daily life others, emotional 5.2. Developing compassion to, and from, other people parts, and voices 5.2.1. Compassionate relating in sessions 5.2.2. Compassionate relating in daily life Ending therapy E.1. Summarising session, shared as audio / written report E.2. Collating other audio / written materials for sharing E.3. Feedback and compassionate learning (for therapist)

CFT for psychosis manual Version 2, 1st Dec 2018 Research Ethics Committee ref: 15/LO/0198 236 APPENDIX 8.7 C. Heriot-Maitland (lead researcher/therapist/author) E. Peters; T. Wykes; A. Gumley (research supervisors) P. Gilbert; E. Longden; C. Irons (therapy collaborators)

Session ID: Content notes:

Selected items to be rated… CFTp manual adherence: CFT-TCRS Full doc: www.compassionforvoices.com/CFTp-ACM competence: (clink link to access full 25-page measure) Absent or Skilful inappropriate enactment CFTp manual adherence 0 1 2 3 4

For selected items only (see top for which are relevant to this session), please rate 0-4. S. Starting therapy S.1. Human connection and collaboration 0 1 2 3 4 S.2. Feelings about being here, and hopes 0 1 2 3 4 S.3. Definitions of compassion 0 1 2 3 4 1. Establishing safeness and connection 1.1. External safeness – social safeness from the ‘outside-in’ 0 1 2 3 4 Connectedness (family, friends, peers, communities) Environments (physical) 1.2. Internal safeness – social safeness from the ‘inside-out’ 0 1 2 3 4 Safeness physiology (grounding, body, breathing) Safeness imagery (safe place) 2. Learning about evolved (tricky) brains, emotional systems & multiple selvesTricky brain and loops 0 1 2 3 4 2.2. 3 circles: threat; drive; soothing 0 1 2 3 4 2.3. The power of attention 0 1 2 3 4 2.4. Normal brains operating under difficult conditions 0 1 2 3 4 Multiple selves Dissociation Hearing voices and over-estimating threat 3. Understanding how my emotions and mind have become shaped‘3 circles’ assessment & formulation 0 1 2 3 4 3.2. Threat-focused assessment & formulation 0 1 2 3 4 3.3. Social mentality assessment & formulation 0 1 2 3 4 3.4. Functional analysis & making sense of voices / beliefs 0 1 2 3 4 4. Building the compassionate selfQualities of the compassionate self 4.2. Compassionate Mind Training (CMT) 0 1 2 3 4 Body posture, facial expression, voice tone 0 1 2 3 4 Compassionate imagery (self and other) Fears, blocks, & resistances (FBRs) 4.3. Becoming your compassionate self in daily life 0 1 2 3 4 5. Directing compassion to self, others, emotional parts, and voicesDirecting compassion to self / multiple selves / emotional parts / voices 0 1 2 3 4 Compassionate relating in sessions Compassionate relating in daily life 5.2. Developing compassion to, and from, other people 0 1 2 3 4 Compassionate relating in sessions Compassionate relating in daily life E. Ending therapy E.1. Summarising session, shared as audio / written report 0 1 2 3 4 E.2. Collating other audio / written materials for sharing 0 1 2 3 4 E.3. Feedback and compassionate learning (for therapist) 0 1 2 3 4

CFTp adherence and competence measure Version 2, 1st Dec 2018 Research Ethics Committee ref: 15/LO/0198 237 APPENDIX 8.7 C. Heriot-Maitland (lead researcher/therapist/author) E. Peters; T. Wykes; A. Gumley (research supervisors) P. Gilbert; E. Longden; C. Irons (therapy collaborators) CFT therapist competence Absent or Skilful inappropriate enactment CFT Therapist Competence Rating Scale (CFT-TCRS) 0 1 2 3 4

For selected items only (see top for which are relevant to this session), please rate 0-4. > CFT Unique Competencies 1. Psychoeducation 0 1 2 3 4 2. Recognising motives and emotions 0 1 2 3 4 3. Actively working with the three-systems 0 1 2 3 4 4. Understanding the relationship between three systems 0 1 2 3 4 5. Compassionate mind training 0 1 2 3 4 6. Building motivation 0 1 2 3 4 7. Building courage 0 1 2 3 4 8. Cultivating and tolerating affiliative emotions 0 1 2 3 4 9. Cultivating and tolerating positive feelings in the drive system 0 1 2 3 4 10. Functional analysis 0 1 2 3 4 11. Fears, blocks and resistances 0 1 2 3 4 12. Unconscious emotions and processes 0 1 2 3 4 13. Formulation 0 1 2 3 4 14. Multiple selves 0 1 2 3 4 > CFT Microskills 15. Non-verbal communication to build rapport 0 1 2 3 4 16. Non-verbal communication and motivational/emotional systems 0 1 2 3 4 17. Verbal Communication 0 1 2 3 4 18. Pacing 0 1 2 3 4 19. Socratic questioning 0 1 2 3 4 20. Paraphrasing and Summaries 0 1 2 3 4 21. Agenda setting 0 1 2 3 4 22. Validation and normalisation 0 1 2 3 4 23. Mentalisation 0 1 2 3 4 How would you rate the therapist overall in this session as a CFT therapist? 0 1 2 3 4 5 6 Poor Barely adequate Mediocre Satisfactory Good Very good Excellent

Scroll down for Horwood et al’s (in press) guidance for rating the 23- item CFT Therapist Competence Rating Scale (CFT-TCRS)

Reference

Horwood, V., Allan, S., Goss, K., & Gilbert, P. (in press). The development of the compassion focused therapy therapist rating scale (CFTTCRS). Psychology and Psychotherapy: Theory, Research and Practice

CFTp adherence and competence measure Version 2, 1st Dec 2018 Research Ethics Committee ref: 15/LO/0198 238 APPENDIX 8.8

Appendix 8.8. Expert ratings for manual adherence and therapist competence Session: Level 1 (represented by “Tosin” session 4; audio ID P5S4) Adhere Item 1.1 2.4 Rating 4 4 CFT- Item 14 15 16 17 18 19 20 21 22 23 TCRS Rating 4 3 3 3 3 4 3 4 3 4

Overall: 4 “Good” Session: Level 2 (represented by “Thomas” session 6; audio ID P4S6) Adhere Item 2.2 2.4 4.1 4.2 5.1 Rating 4 3 4 4 4 CFT- Item 6 11 14 15 16 17 18 19 20 21 22 23 TCRS Rating 4 4 3 4 4 4 4 4 4 4 4 3

Overall: 6 “Excellent” Session: Level 3 (represented by “Greg” session 15; audio ID P2S15) Adhere Item 3.4 Rating 4 CFT- Item 10 13 15 16 17 18 19 20 21 22 23 TCRS Rating - - 4 3 4 4 3 1 4 4 2

Overall: 5 “Very good” Session: Level 4 (represented by “Amanda” session 17; audio ID P7S17) Adhere Item 1.2 4.2 5.1 Rating 4 4 4 CFT- Item 2 5 10 12 15 16 17 18 19 20 21 22 23 TCRS Rating 4 3 4 4 4 4 4 4 4 4 4 4 4

Overall: 6 “Excellent” Session: Level 5 (represented by “Charmaine” session 14; audio ID P6S14) Adhere Item 3.4 5.1 5.2 Rating 3 4 4 CFT- Item 6 7 15 16 17 18 19 20 21 22 23 TCRS Rating 4 4 4 - 4 3 4 4 3 4 4

Overall: 6 “Excellent” Combined Adherence Rating (mean) 3.86 CFT-TCRS Rating (mean) 3.67 (Competency items 1-14 = 3.80; Microskills items 15-23 = 3.64)

Overall: 3 x “Excellent”; 1 x “Very good”; 1 x “Good”

239 APPENDIX 8.9

Appendix 8.9. Normative data used for establishing reliable change criteria * Reliability Standard deviation SEdiff PSYRATS • Voices 0.75 7.91 5.59 • Delusions 0.70 4.75 3.68 (Kronmuller et al., 2011) (Johns et al., 2019) DASS • Depression 0.85 3.98 2.18 • Anxiety 0.81 3.38 2.08 • Stress 0.88 4.52 2.21 (Osman et al., 2012) (Osman et al., 2012) CORE 0.94 0.75 0.26 (Evans et al., 2002) (Evans et al., 2002) DES-II 0.93 12.53 4.69 (Study 2 sample, n=544) (Study 2 sample, n=544) SocC 0.89 14.96 7.02 (Study 2 sample, n=544) (Study 2 sample, n=544) FSCSR • Inadequate 0.90 8.57 3.83 • Reassured 0.88 6.55 3.21 • Hated 0.83 4.71 2.75 (Study 2 sample, n=544) (Study 2 sample, n=544) OAS 0.93 13.07 4.89 (Study 2 sample, n=544) (Study 2 sample, n=544) SCS 0.44 4.72 5.00 (Study 2 sample, n=544) (Study 2 sample, n=544) PBIQ-R 0.93 11.7 4.38 (Birchwood, on request) (Birchwood, on request) RMSSD (ms) 0.96 34.99 9.90 (Nussinovitch et al., 2011) (Nussinovitch et al., 2011) *Standard error of change (SEdiff), which was calculated by inputting the standard deviation and reliability scores into an online calculator (https://www.psyctc.org/stats/rcsc1.htm).

References

Evans, C., Connell, J., Barkham, M., Margison, F., McGrath, G., Mellor-Clark, J., & Audin, K. (2002). Towards a standardised brief outcome measure: psychometric properties and utility of the CORE-OM. Br J Psychiatry, 180(01), 51-60. doi:10.1192/bjp.180.1.51 Johns, L., Jolley, S., Garety, P., Khondoker, M., Fornells-Ambrojo, M., Onwumere, J., . . . Byrne, M. (2019). Improving Access to psychological therapies for people with severe mental illness (IAPT-SMI): Lessons from the South London and Maudsley psychosis demonstration site. Behav Res Ther, 116, 104-110. doi:10.1016/j.brat.2019.03.002 Kronmuller, K. T., von Bock, A., Grupe, S., Buche, L., Gentner, N. C., Ruckl, S., . . . Mundt, C. (2011). Psychometric evaluation of the Psychotic Symptom Rating Scales. Compr Psychiatry, 52(1), 102-108. doi:10.1016/j.comppsych.2010.04.014 Nussinovitch, U., Elishkevitz, K. P., Katz, K., Nussinovitch, M., Segev, S., Volovitz, B., & Nussinovitch, N. (2011). Reliability of Ultra-Short ECG Indices for Heart Rate Variability. Ann Noninvasive Electrocardiol, 16(2), 117-122. doi:10.1111/j.1542-474X.2011.00417.x Osman, A., Wong, J. L., Bagge, C. L., Freedenthal, S., Gutierrez, P. M., & Lozano, G. (2012). The Depression Anxiety Stress Scales-21 (DASS-21): further examination of dimensions, scale reliability, and correlates. J Clin Psychol, 68(12), 1322-1338. doi:10.1002/jclp.21908

240 APPENDIX 8.10

Appendix 8.10. Reliable Change Indices for single-case-level changes across phases – full table Baseline phase 1st half phase Intervention phase Follow-up (T1→T2) (T2→T3) (T2→T4) (T2→T5) Pt RCI Pt RCI Pt RCI Pt RCI

Outcome measures PSYRATS-V 1 .18 1 -.72 1 -1.61 1 -1.61 2 .00 2 -1.61 2 -5.01* 2 -5.01* 3 -.18 3 -.18 3 -1.79 3 -2.68* 4 No voices 4 No voices 4 No voices 4 No voices 5 -.89 5 -.89 5 -3.04* 5 -3.04* 6 .18 6 -.72 6 -2.50* 6 -2.33* 7 No voices 7 No voices 7 No voices 7 No voices

No RC = 5 No RC = 5 No RC = 2 No RC = 1 ↓ = 3 ↓ = 4 PSYRATS-D 1 .00 1 .00 1 -2.45* 1 -1.63 2 .00 2 -3.53* 2 -3.53* 2 -3.53* 3 -.82 3 1.36 3 -1.63 3 -1.90 4 .00 4 -.54 4 -1.09 4 -1.63 5 No delusions 5 No delusions 5 No delusions 5 No delusions 6 .00 6 -.27 6 -1.63 6 -1.63 7 .00 7 -1.36 7 -1.90 7 -1.90

No RC = 6 No RC = 5 No RC = 4 No RC = 5 ↓ = 1 ↓ = 2 ↓ = 1 DASS-Dep 1 1.83 1 -3.67* 1 -1.83 1 -3.67* 2 -1.83 2 -.92 2 -8.26* 2 -12.84* 3 -.92 3 -8.26* 3 -11.01* 3 -11.93* 4 .92 4 -5.50* 4 -8.26* 4 -8.26* 5 .92 5 -10.09* 5 -11.01* 5 -13.76* 6 -2.75* 6 3.67* 6 1.83 6 1.83 7 .92 7 -4.59* 7 -4.59* 7 .92 ↑ = 1 No RC = 6 No RC = 1 No RC = 2 No RC = 2 ↓ = 1 ↓ = 5 ↓ = 5 ↓ = 5 DASS-Anx 1 -7.69* 1 4.81* 1 5.77* 1 4.81* 2 .96 2 -4.81* 2 -5.77* 2 -9.62* 3 -2.88* 3 -2.88* 3 -.96 3 -1.92 4 1.92 4 -5.77* 4 -9.62* 4 -7.69* 5 2.88* 5 -6.73* 5 -9.62* 5 -15.38* 6 .96 6 .00 6 -.96 6 .00 7 1.92 7 -1.92 7 -2.88* 7 5.77* ↑ = 1 ↑ = 1 ↑ = 1 ↑ = 2 No RC = 4 No RC = 2 No RC = 2 No RC = 2 ↓ = 2 ↓ = 4 ↓ = 4 ↓ = 3 DASS-Str 1 3.62* 1 -5.43* 1 -7.24* 1 -9.05* 2 3.62* 2 -5.43* 2 -6.33* 2 -9.05* 3 .00 3 -.90 3 -2.71* 3 -1.81 4 .90 4 -6.33* 4 -10.86* 4 -9.05* 5 .00 5 -9.95* 5 -10.86* 5 -13.57*

241 APPENDIX 8.10

6 -3.62* 6 2.71* 6 -.90 6 -1.81 7 -1.81 7 -3.62* 7 -1.81 7 2.71* ↑ = 2 ↑ = 1 ↑ = 1 No RC = 4 No RC = 1 No RC = 2 No RC = 2 ↓ = 1 ↓ = 5 ↓ = 5 ↓ = 4

CORE 1 -1.70 1 -.90 1 -.68 1 -1.92 2 .79 2 -4.64* 2 -6.22* 2 -8.37* 3 -.45 3 -3.96* 3 -4.07* 3 -3.73* 4 2.38* 4 -3.62* 4 -4.41* 4 -3.96* 5 -.11 5 -5.77* 5 -7.35* 5 -9.05* 6 -1.47 6 1.13 6 -4.75* 6 -4.41* 7 1.13 7 -1.24 7 -1.36 7 -1.02 ↑ = 1 No RC = 6 No RC = 3 No RC = 2 No RC = 2 ↓ = 4 ↓ = 5 ↓ = 5 DES-II 1 -1.14 1 .84 1 2.28* 1 .61 2 .91 2 -3.05* 2 -3.12* 2 -2.59* 3 .38 3 -.61 3 -1.29 3 -1.98* 4 1.29 4 .38 4 -1.37 4 -1.22 5 .38 5 -7.61* 5 -9.06* 5 -9.67* 6 1.29 6 -3.05* 6 -1.83 6 -2.44* 7 -.84 7 .15 7 .84 7 .23 ↑ = 1 No RC = 7 No RC = 4 No RC = 4 No RC = 3 ↓ = 3 ↓ = 2 ↓ = 4

Process measures SocC 1 -1.14 1 .85 1 1.71 1 .43 2 .00 2 4.70* 2 3.99* 2 2.56* 3 1.00 3 -.14 3 1.42 3 -.14 4 -3.56* 4 3.85* 4 5.41* 4 6.27* 5 -2.42* 5 6.13* 5 10.54* 5 8.97* 6 .00 6 1.14 6 6.98* 6 5.98* 7 .28 7 1.14 7 1.00 7 -1.00 ↑ = 3 ↑ = 4 ↑ = 4 No RC = 5 No RC = 4 No RC = 3 No RC = 3 ↓ = 2 FSCSR-Inad 1 .00 1 -.78 1 -.52 1 .78 2 -.78 2 -1.57 2 -1.31 2 -5.74* 3 -1.57 3 .52 3 -2.35* 3 -1.04 4 1.04 4 -1.83 4 -3.13* 4 -2.35* 5 -.26 5 -4.18* 5 -6.01* 5 -7.05* 6 1.31 6 -2.35* 6 -2.35* 6 -4.70* 7 -3.13* 7 -.78 7 .00 7 -3.39*

No RC = 6 No RC = 5 No RC = 3 No RC = 2 ↓ = 1 ↓ = 2 ↓ = 4 ↓ = 5 FSCSR-Reas 1 .31 1 4.36* 1 2.18* 1 .62 2 1.25 2 2.18* 2 3.12* 2 5.61* 3 .31 3 .00 3 1.56 3 -.62

242 APPENDIX 8.10

4 -3.74* 4 3.12* 4 3.43* 4 2.49* 5 5.61* 5 -3.74* 5 -1.56 5 -2.49* 6 -1.87 6 -.62 6 1.56 6 .62 7 -.93 7 -2.49* 7 .00 7 -1.87 ↑ = 1 ↑ = 3 ↑ = 3 ↑ = 2 No RC = 5 No RC = 2 No RC = 4 No RC = 4 ↓ = 1 ↓ = 2 ↓ = 1 FSCSR-Hate 1 -.73 1 -4.00* 1 -2.91* 1 .00 2 .00 2 -4.36* 2 -4.73* 2 -5.45* 3 .00 3 -2.18* 3 -1.82 3 -1.45 4 .73 4 -1.09 4 -2.18* 4 -2.91* 5 1.09 5 -3.64* 5 -4.73* 5 -4.73* 6 1.45 6 -1.09 6 -1.09 6 -1.09 7 3.27* 7 -2.55* 7 -3.64* 7 -4.36* ↑ = 1 No RC = 6 No RC = 2 No RC = 2 No RC = 3 ↓ = 5 ↓ = 5 ↓ = 4 OAS 1 .61 1 .61 1 -1.84 1 -2.86* 2 -2.04* 2 -3.27* 2 -6.54* 2 -7.98* 3 -1.64 3 -.41 3 -1.23 3 -.20 4 -4.70* 4 -1.23 4 -.82 4 -1.64 5 1.02 5 -8.38* 5 -11.66* 5 -12.68* 6 1.02 6 -1.64 6 -4.91* 6 -5.73* 7 -.41 7 -4.09* 7 -1.23 7 -3.68*

No RC = 5 No RC = 4 No RC = 4 No RC = 2 ↓ = 2 ↓ = 3 ↓ = 3 ↓ = 5 SCS-SF 1 .40 1 .60 1 2.60* 1 1.20 2 1.40 2 1.60 2 2.40* 2 4.40* 3 -.40 3 2.20* 3 2.20* 3 2.60* 4 -.60 4 1.40 4 1.60 4 2.20* 5 -.40 5 .20 5 2.00* 5 1.40 6 .20 6 -.80 6 .40 6 .80 7 2.00* 7 -2.20* 7 -1.20 7 -.80 ↑ = 1 ↑ = 1 ↑ = 4 ↑ = 3 No RC = 6 No RC = 5 No RC = 3 No RC = 4 ↓ = 1 PBIQ-R 1 -.71 1 -2.48* 1 -3.01* 1 -3.54* 2 -.88 2 -1.95 2 -6.37* 2 -3.72* 3 -.18 3 -2.83* 3 -4.07* 3 -2.30* 4 .35 4 -3.19* 4 -3.36* 4 -3.19* 5 -.35 5 -7.43* 5 -9.20* 5 -9.56* 6 1.42 6 -2.12* 6 -5.66* 6 -3.54* 7 .71 7 -2.12* 7 -.18 7 -2.65*

No RC = 7 No RC = 1 No RC = 1 ↓ = 6 ↓ = 6 ↓ = 7 RMDDS 1 -.20 1 2.83* 1 1 1.11 (ms) 2 3.13* 2 2 2 -3.84* 3 -2.53* 3 1.31 3 3.23* 3 .61 4 4 2.93* 4 1.01 4

243 APPENDIX 8.10

5 5 1.21 5 1.11 5 -.51 6 -.30 6 -1.11 6 -.91 6 -1.31 7 7 -4.24* 7 13.64* 7 ↑ = 1 ↑ = 2 ↑ = 2 No RC = 2 No RC = 3 No RC = 3 No RC = 4 ↓ = 1 ↓ = 1 ↓ = 1 *p<.05

244