The goal of this Mixed study, combining a randomized controlled trial and an Interpretative Phenomenological Analysis (IPA-qualitative study) is to evaluate the effects of the EDEN College courses on quality of life, compared with usual group interventions provided in mental health services in adult people who hear voices (or present Auditory Verbal Hallucinations (AVH). Volonteers must be between 18 and 65 years old and hear perturbating voices for 6 months or more; they must receive care from a psychiatrist or a psychiatric team.

The main questions it aims to answer are: Does EDEN College improve quality of life in a better way than usual group interventions (RCT part)? What processes are linking changes in quality of life and changes in the relationship between voice-hearers and their voices (IPA part)?.

EDEN college is a ten days representing about 50 hours course. Usual group interventions (named CAU for "Care As Usual") will combined an individualized program mixing Social skills training, cognitive remediation, psychoeducation, mediation therapy and social therapy and representing between 40 and 60 hours of training.

Our primary hypothesis is that EDEN College Courses will provide a significant better improvement of Quality of life than usual groups provided in post intervention. the voice hearers Quality of life will be assessed through the Mental Well Being (main criteria is Warwick Edimburg Mental Well Being Scale- WEMWBS), the Self Estime and the internalized Stigma of Mental Illness processes (secondary criterias).

Our secondary hypothesis are:

• EDEN College course will decrease voice hearing / AVH phenomena measured with PSYRATS-AH and AHRS /HCS scales in post intervention
• EDEN College course will improve the quality of relationship between the voice hearer and his voices, measured by BAVQ-R and VAAS 9 scales
• EDEN College effects on quality of life and on AVH phenomena will be stables one year after the end of interventions.
• Improvement on quality of life and improvement on quality of relationship between the voice hearers and his voices are linked (IPA part of the study)

Participants will be asked to:

• Participate to an individual meeting with an investigator be explained the goals of this study and ask all the questions they have on it before giving their consent to participate.
• Participate to 7 individual meetings (around 1 hour each) with members of research team (2 before interventions; 1 during the intervention; 2 after the intervention and 2 one year after the end of intervention) to answer researchers questions on their voices, needed to score the PSYRATS-AH and AHRS/HCS scales, and quote 5 questionnaires on mental well being, self esteem, internalized stigma of mental Illness and relationship with voices.
• Participate to EDEN College or CAU sessions (about 50 hours of training ) dispended in a 8 mounth period
• Answer a mounthly phone call (about 10 minutes) with a member of the research team regarding their participation on the study and any mental healthcare or support services received outside the study
• A subset of participants may be invited to take part in a one-hour interview after completing the groups or EDEN training programmes. The purpose of this interview is to explore participants' perspectives and experiences of the groups or training programmes, and their perceptions of any changes in their quality of life (qualitative part of the study).

EDEN-RCT study will employ a multicenter mixed-methods design using a connected phases approach, combining a randomized controlled trial for the quantitative component with an Interpretative Phenomenological Analysis (IPA) study for the qualitative component.

The study will be conducted over a period of 44 months, including 24 months of participant recruitment and 21 months of follow-up (± 30 days).

Participants recruitment:

Recruitment will be conducted among the active patient populations receiving care within the clinical departments of the eight centers participating in the study.

All healthcare professionals working in each participating center will be informed of the study initiation and its general procedures, namely the comparison of EDEN College with Care-as-usual (CAU) group interventions. These professionals may provide the study information sheet to adult voice-hearers and refer interested individuals to an appointment with one of the study investigators. A minimum period of 24 hours will be respected between the provision of the information sheet and the appointment with the investigator.

During the month preceding study initiation, posters will be displayed at the study sites and within their affiliated hospitals. These posters will invite adult voice-hearers to contact their healthcare professionals or the center investigators for further information about the study. Individuals who identify themselves as adult voice-hearers will be provided with the study information sheet and offered an appointment with one of the study investigators. A minimum period of 24 hours will be respected between the provision of the information sheet and the appointment with the investigator.

Method:

The single-blind randomized controlled trial (1:1 allocation ratio) will compare the effects of the EDEN College course with those of groups routinely provided in standard care that offer an equivalent number of intervention hours over the same time period. Conducting the study across eight centers in France will enable assessment of whether the effectiveness of the EDEN College intervention is independent of the facilitators delivering it.

Assessment of outcome measures at baseline (T1), post-intervention (T2), and 12 months after completion of the intervention (T3) will allow evaluation of both the effectiveness of the EDEN College course and the sustainability of their effects over time.

The primary analysis will be conducted according to the intention-to-treat principle using a random-effects linear mixed model to account for the repeated-measures structure of the data (pre- intervention, post-intervention assessments) and the comparison between the two study groups (EDEN and CAU). Fixed effects will include intervention group, assessment time point, and the group × time interaction. Recruitment centre will be included as a random effect to account for within-centre correlation and to allow for differences in mean outcome levels across centers. The model will be adjusted for relevant covariates identified a priori, including age, sex, and diagnosis.

The primary intervention effect will be assessed by comparing pre-intervention (T1) and post-intervention (T2 = T1 + 9 months) outcomes and will be estimated from the coefficient of the group × time interaction term.

The impact of the intervention on secondary outcomes at T2 and T3 will be evaluated using the same modelling approach. Given the number of questionnaires administered, secondary outcomes will be prioritised according to their relative importance and, where appropriate, adjustments for multiple comparisons will be applied using the Holm correction procedure.

The qualitative component will adhere to the methodological standards of Interpretative Phenomenological Analysis (Smith et al., 2021). Particular attention will be paid to participants' lived experiences and the meanings they attribute to them. These data will contribute to a deeper understanding of the relationship between changes in quality of life and changes in the relationship between voice-hearers and the voices they hear.

Interview participants will be selected from each of the eight participating centers. The sample will include individuals with relatively low levels of mental well-being and individuals with relatively high levels of mental well-being at T2, compared with the overall sample within each center. Participant selection will be carried out by the coordinating investigator. Selected participants will take part in an in-depth semi-structured interview conducted by the coordinating investigator. The analysis will be based on transcripts of these interviews, as well as reports from the review interviews conducted at T1, T1 + 4 months, and T2.

Thematic analyses will be performed by three researchers independent of the intervention groups and representing three disciplinary backgrounds: an advanced practice nurse, a sociologist, and a peer support worker with lived experience of hearing voices.

The themes identified by this multidisciplinary research team will be triangulated with a lexicometric analysis conducted using NVivo software.

Measures Implemented to Minimise Bias

• Use of a control group receiving an equivalent number of intervention hours and a comparable intervention format to the EDEN College group, in order to minimise confounding bias.
• Randomisation will be performed using randomly varying block sizes of 4 and 6 to ensure adequate balance between study arms while limiting the predictability of treatment allocation.
• Randomisation will be conducted through the REDCap platform. The investigator conducting the inclusion visit will be informed of the allocation. This investigator will schedule subsequent assessment visits with an associated researcher responsible for conducting the T1, T2, and T3 evaluations, while maintaining that researcher blinded to treatment allocation.
• Selection of a self-administered questionnaire, completed in the absence of group facilitators, as the primary outcome measure in order to minimise assessment bias.
• Use of self-report questionnaires or outcome assessments conducted under single-blind conditions for secondary outcomes to reduce assessment bias. The investigator conducting the T1, T2, and T3 assessments will remain blinded to treatment allocation through the randomisation procedure.
• Intention-to-treat analysis will be performed to minimise attrition bias.
• Qualitative interviews will be conducted by the coordinating investigator, who is not involved in delivering the interventions, in order to reduce follow-up and interviewer bias.
• Qualitative interview transcripts and related documents will be independently coded by three researchers and analysed using qualitative data analysis software (NVivo®) to reduce subjective interpretation bias.
• Triangulation of coded qualitative data by researchers from different disciplinary backgrounds and knowledge traditions (sociology, nursing science, and experiential expertise) will be used to minimise subjectivity bias.
• Participation in Accept Voices© interventions and Hearing Voices Groups delivered by healthcare professionals within the participating centres will not be permitted throughout the study period, due to their conceptual proximity to certain components of the EDEN College and the resulting risk of confounding.
• Different facilitators will be assigned to the CAU groups and the EDEN College to ensure comparable levels of facilitator engagement and to minimise contamination between intervention methods.
• The five self-report questionnaires will be completed in the presence of a healthcare professional (without assistance except to clarify items that are not understood). In addition, questionnaire administration will be divided across two appointments at each assessment time point to reduce participant burden and minimise attrition bias.
• Information regarding all psychiatric consultations and therapeutic encounters received in parallel with study participation will be collected monthly from both clinical records and participant reports throughout the inclusion and follow-up period in order to minimise confounding bias.
• Information on antipsychotic, mood stabiliser, antidepressant, and anxiolytic medications will be collected at T1, T2, and T3 to account for potential confounding effects of pharmacological treatment.

Data Collected (data will be recorded directly in the electronic Case Report Forms):

Demographic data: age and gender. Psychiatric diagnostic status: psychiatric diagnosis according to DSM-5 criteria or absence of a psychiatric diagnosis.

Type and dosage of prescribed psychotropic medications at T1, T2, and T3, (antipsychotics, mood stabilizers, antidepressants, and/or anxiolytics).

Name of the treating psychiatrist or the Advanced Practice Nurse (APN) responsible for the participant's care.

Presence of a legal protective measure concerning property or financial affairs, and the nature of such measure.

Name of the legal representative, where applicable. Responses and scores for the eight psychometric instruments administered at T1, T2, and T3: WEMWBS, ISMI, RSE, PSYRATS-AH, ARHS, HCS, BAVQ-R, and VAAS-9.

Reports of participant review interviews conducted at T1, T1 + 4 months, and T2 (transcripts or digital copies).

Full transcripts of the in-depth qualitative interviews conducted with the 18 to 24 participants selected for the qualitative component.

Monthly summary records of psychiatric care services received outside the study during the participant's 21-month study participation period.

Inclusion Criteria:

• Have received written and oral information about the study and have provided written informed consent prior to participation.
• Be aged between 18 and 65 years.
• Be affiliated with, or entitled to, the French national health insurance system.
• Be under the care of a psychiatrist or a psychiatric care team.
• Have experienced auditory verbal hallucinations (AVH) for at least 6 months.
• Have a PSYRATS-AH total score ≥25 at inclusion day.
• Have sufficient proficiency in spoken and written French to understand the study procedures and complete the self-report questionnaires.
• Participants subject to french legal protective measures (tutelle or curatelle) may be included in the study.

Exclusion Criteria:

• Auditory verbal hallucinations occurring exclusively in the context of substance use.
• Auditory verbal hallucinations attributable to an organic neurological or medical condition.
• Severe cognitive impairment preventing participation in half-day group sessions.
• Behavioural disturbances incompatible with participation in half-day group sessions
• Refusal to participate in the study

The clinician-rated outcome measures (PSYRATS-AH, AHRS and HCS) will be administered by assessors who are blinded to treatment allocation.