Efficacy and cost-effectiveness of a digital guided self-management intervention to support transition from intensive care to community care in anorexia nervosa (TRIANGLE): pragmatic multicentre randomised controlled trial and economic evaluation

Summary Background There is uncertainty regarding how best to support patients with anorexia nervosa following inpatient or day care treatment. This study evaluated the impact of augmenting intensive treatment with a digital, guided, self-management intervention (ECHOMANTRA) for patients with anorexia nervosa and their carers. Methods In this pragmatic multicentre randomised controlled trial and economic evaluation, patients with a diagnosis of anorexia nervosa or atypical anorexia nervosa, aged 16+ and attending one of the 31 inpatient or day-patient services in the UK were randomised with one of their carers to receive ECHOMANTRA plus treatment as usual (TAU), or TAU alone. ECHOMANTRA was hosted on a digital platform and included a workbook, recovery-oriented video-clips and online facilitated groups (patients only, carers only, joint patient-carer). Participants were randomised on a 1:1 ratio using a minimisation algorithm to stratify by site (N = 31) and severity (defined by BMI <15 and ≥ 15 kg/m2 at baseline). The primary outcome was patient depression, anxiety, and stress at 12 months. Primary and secondary outcomes were compared between trial arms on an intention-to-treat basis (ITT). This trial is registered with the ISRSTN registry, ISRCTN14644379. Findings Between July 01, 2017 and July 20, 2020, 371 patient-carer dyads were enrolled and randomly assigned to ECHOMANTRA + TAU (N = 185) or TAU alone (N = 186). There were no significant differences between trial arms with regards to the primary outcome (completed by N = 143 patients in the TAU group, Mean = 61.7, SD = 29.4 and N = 109 patients in the ECHOMANTRA + TAU group, Mean = 58.3, SD = 26.9; estimated mean difference 0.48 points; 95% CI −5.36 to 6.33; p = 0.87). Differences on secondary outcomes were small and non-significant (standardised effect size estimates ≤0.25). Five patients died (2 from suicide and 3 from physical complications) over the course of the trial, and this was unrelated to their participation in the study. Interpretation ECHOMANTRA added to TAU was not superior to TAU alone in reducing patient depression, anxiety, and stress symptoms. This may be explained by limited engagement with the intervention materials and changes in usual care practices since the beginning of the trial. Funding 10.13039/501100000272National Institute for Health Research (NIHR), under its 10.13039/501100000664Health Technology Assessment Programme (HTA) Programme (Grant Reference Number 14/68/09). 10.13039/100019418NIHR Maudsley Biomedical Research Centre (BRC), 10.13039/100009362South London and Maudsley NHS Foundation Trust and 10.13039/100013376Institute of Psychiatry, Psychology and Neuroscience, and King’s College London. 10.13039/501100023232NIHR Applied Research Collaboration South London (NIHR ARC South London) at 10.13039/100010872King's College Hospital NHS Foundation Trust.


Protocol amendments
Amendment Changes From Protocol To Protocol N, Version Date Effective Substantial amendment 1 Change to intervention delivery 1:1 guidance sessions replaced with 8 online moderated and facilitated group sessions for patients and carers.
Change to measures Patient Difficulties in Emotion Regulation Scale, Carer Parents Versus Anorexia Nervosa and Accommodation and Enabling Scale for Eating Disorders questionnaires were removed to reduce participant burden.

Change to inclusion criteria
-Patients should either be admitted for inpatient care or should be attending day care for a minimum of 4 days/week at the time of consenting (previously unspecified how long patients should be attending day care for) -Patients previously only included if they suffered from anorexia nervosa, criteria changed to include subclinical/atypical anorexia nervosa (BMI > 18.5) -Clarification that patients must be able to access an electronic device and the internet to log onto the study platform -Informed consent should be signed within 2 months from admission (changed from 1 month) Change to protocol -Serious Adverse Events defined to specify whether they would be considered "related", "unrelated", "expected" or "unexpected" to clarify which kind of events will be recorded only and which events will need reporting to the REC.-Procedure for dealing with protocol violations added to protocol: protocol violations are recorded by the research assistants with justifications.These are discussed during regular meetings with the statistician and based on those discussions the study team consider whether an update to the protocol will be necessary.Study checklist changes -Changes in inclusion criteria: Informed consent will be signed at any time whilst the patient is admitted into hospital and up to 4 weeks after discharge.
-Following feedback from clinicians, age also extended to 16 years old so the team can recruit from adolescent units -Change in exclusion criteria: The patient will need to be admitted into hospital or attending day care for a minimum of 3 days/week when they are consented in the study.
-Change in recruitment into trial: Study team will be involved in all the recruitment stages.
Addition of electronic consent for carers -Due to consenting process of carers delaying randomization, carers can now provide consent electronically via the study website.
Changes in measures obtained from the clinical team and from patients -The clinical teams will no longer be asked to provide monthly updates from patients.Basic clinical information (e.g., BMI, admission/discharge date) will be collected from the clinical teams only at baseline.
-A brief monthly update questionnaire will be completed by patients instead.
Carer/family involvement questions Addition of 9 questions related to involvement of family members and carers to be asked to the participating sites.This brief questionnaire will be completed at twice (i.e., at present and 18 months).
Research perception questions Addition of 6 questions related to research perceptions.These questions will be asked to all our participating sites once.

Study flyer changes
Changes in study flyer for patients to make it clearer.
Study flyer for carers Addition of study flyer for carers.

Patients-carers joint sessions change
The content of the joint sessions will be delivered via the study website through an online group format instead of using skype.
Advertisement of the study on social media The study will be advertised on social media (e.g., Facebook, twitter).
Recruitment of participants in the community Participants will also be recruited in the community if they meet the inclusion criteria (i.e., admitted to hospital or attending day care for a minimum of 3 days/week at the start of the study).
Non-substantial amendment 7 Addition of new site New site has agreed to take part in the study as a participating site.Collection of patient BMI from clinical teams -A member of the clinical team at the hospital will be asked to provide patient's BMI measurement at baseline and every 3 months up to patient's discharge.
Please note that: a) BMI is already collected as part of routine care b) This data will be collected for research purposes c) There is ethics approval to request this data from the medical notes.d) Clinical teams were asked to provide this data monthly from the beginning of the study.As part of Amendment No. 6 and to reduce burden on clinical teams, this was changed to be provided at baseline only.However, the team has found that it is difficult to collect this data from participants (e.g.,

Typo age inclusion criteria
We would like to amend the age specified in the inclusion criteria.The approved inclusion age is 16.This was approved in amendment No.6.
There is a mistake in the study protocol.The inclusion criteria age should read "Patients aged 16 years or over" instead of "patients aged 17 years or over".

Addition of gift vouchers to the reimbursement of participants
We would like to offer to participants the option of gift voucher reimbursement as an alternative to the current cheque reimbursement.Additional qualitative data collection -To obtain participant feedback on their participation in the TRIANGLE trial and to gather further information regarding patients' views on transition from intensive treatment, we will: -Invite a selected subgroup of patients and carers (up to n=22 patients and n=22 carers, selected based on pre-specified criteria) who were randomised to the ECHOMANTRA + TAU intervention group to participate in an interview about their experience of the TRIANGLE trial and intervention.The maximum number of participants is based on the high likelihood of reaching data saturation on the topic investigated, among the study population, with less than 22 participants.-Invite a selected subgroup of patients and carers (up to n=22 patients and n=22 carers, selected based on pre-specified criteria) who were randomised to the TAU only group to participate in an interview about their experience of transitioning from the intensive treatment (inpatient or day care) they were receiving when initially recruited to the trial back to the community.The maximum number of participants is based on the high likelihood of reaching data saturation on the topic investigated, among the study population, with less than 22 participants.-Participants will be recruited who are between the 6 months and 12 months timepoints in their study participation so that they have had enough time to offer a well-formed opinion of what it meant to them to have participated in the study and to have transitioned from intensive treatment.To explore a range of opinions participants will be recruited purposively across study sites according to recruitment site location, gender, age, patient's eating disorder severity (>7 years of illness or below 3 years of illness) and carer's relationship to the patient.Sites will not be involved in the selection of participants

Patient measures
Demographics: Demographic features were assessed via a self-report questionnaire which included 17 items related to demographic variables (e.g., sex, ethnicity, highest level of education).
Psychological medical history: Details about patients' psychological medical history were assessed via a self-report questionnaire which included items related to clinical variables (e.g., length of eating disorder diagnosis, comorbidities, treatment under the mental health act).
Eating disorder psychopathology: Eating disorder psychopathology was assessed using the Eating Disorder Examination Questionnaire (EDE-Q; Fairburn & Beglin, 1994).The EDE-Q is a 36 item selfreport measure of eating disorder symptoms (dietary restraint, eating concerns, weight concerns, shape concerns) in the past 28 days, on a scale ranging from no days (0) to every day (6).In the main analysis, the global score was used as a marker of eating disorder symptom severity.A higher score indicates greater frequency of symptoms.
Body Mass Index (BMI): Weight and height were primarily assessed via self-report.Clinicians at participating centres were also asked to provide weight and height at admission from clinical records.
Depression, Anxiety and Stress symptoms: Levels of depression, anxiety and stress were assessed using the Depression, Anxiety and Stress Scale (DASS-21; Lovibond & Lovibond, 1995).The DASS-21 is a 21 item self-report measure of depression, anxiety and stress symptoms in the past week, on a scale ranging from 0 (did not apply to me at all) to 4 (applied to me very much, or most of the time).A higher score indicates greater frequency of symptoms.
Social Functioning: Social functioning was assessed via the Work and Social Adjustment Scale (WSAS; Mundt, Marks, Shear, & Greist, 2002).The WSAS is a five-item measure of functional impairment.In this trial the WSAS was used to assess the impact of the eating disorder on the ability to function in the areas of work, home management, social leisure activities, private leisure activities, and personal relationships.A higher score indicates greater impairment.
Social functioning: Social functioning was assessed via the informant-version of the Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997).The SDQ is a 25-item informant-rated measure of psychological attributes related to the areas of emotional symptoms, conduct problems, hyperactivity/inattentiveness, peer relationship problems, and prosocial behaviour (in the last six months).In this trial the questionnaire was administered to carers as a measure of social functioning in patients.A total score was calculated using four subscales (excluding prosocial behaviour).For the total score, a higher score indicates greater problems.
Autism Spectrum symptoms: Autistic symptoms were assessed via the Autism Spectrum Quotient (AQ-10; Allison, Auyeung, & Baron-Cohen, 2012): The AQ-10 is a 10-item self-report measure of autistic symptoms, rated on a scale from definitely agree, to definitely disagree.The total score is based on all items.A score >6 indicates high levels of autistic symptoms.
Obsessive Compulsive symptoms: Obsessive compulsive symptoms were assessed via the Obsessive Compulsive Inventory -Revised (OCI-R; Foa et al., 2002).The OCI-R is an 18-item self-report measure of obsessive-compulsive symptoms in the past month.The measure covers six dimensions including checking, washing, ordering, hoarding, obsessing, and neutralizing.Items are rated on a scale from 0 (not at all) to 4 (extremely).A score >21 indicates high levels of obsessive-compulsive symptoms.
Importance to change: Importance to change was assessed via a single-item self-report visual analogue scale, "How important is it for you to change?What score would you give yourself out of 10?" was rated on a scale from 1 (not at all important) to 10 (extremely important).
Ability to change: Ability to change was assessed via a single-item self-report visual analogue scale, "How confident are you in your ability to change?What score would you give yourself out of 10?" is rated on a scale from 1 (not at all confident) to 10 (extremely confident).
Health-related quality of life: Health-related quality of life was assessed using the EQ-5D-3L (Kind, Dolan, Gudex, & Williams, 1998).The ED-5D-3L is a five-item, self-report measure of quality of life across five dimensions including mobility, self-care, usual activities, pain/discomfort, anxiety/depression, which are rated as 1 (no problem), 2 (moderate problem) or 3 (severe problem).
Thus, a higher score indicates poorer quality of life in each domain.The scale also includes a visual analogue scale (VAS) which indicates the general health status from 0 (worst health imaginable) to 100 (best health imaginable).In the protocol paper we had indicated that the 5 level version of the EQ-5D would be used.Subsequently however, NICE recommended that the 3-level version continue to be used due to problems with the estimated UK utility weightings for the 5 level version (National Institute for Health and Care Excellence, 2019).
Number of days spent in hospital at 12-and 18-months post-randomisation.Hospital Episode Statistics data were requested from NHS Digital for England-based participants only.However, these data for the full trial period were not available.

Economic measures
Resource utilisation.Health service and other resource utilisation was collected using an adapted version of the Client Service Receipt Inventory (J Beecham & Knapp, 1992).The CSRI is a self-report measure which includes items related to the use of health and other services, as well as productivity losses from lost employment.Participants were asked to recall their use of services and other impacts over the previous three months.

Carer measures
Demographics: Demographic features were assessed via a self-report questionnaire which contained 15 items related to demographic variables (e.g., nature of relationship to the patient, employment status, marital status).
Caregiving skills: Caregiving skills were assessed via the Caregiving Skills Scale (CASK; Hibbs et al., 2015).The CASK is a 27-item measure of caregiving attitudes and behaviours for dealing with eating disorder symptoms.
Carer Depression, Anxiety and Stress symptoms: Carer depression, anxiety and stress symptoms were assessed via the DASS-21 (Lovibond & Lovibond, 1995), as described in the section above on patient baseline measures.
Adherence to the ECHOMANTRA intervention: Participants' adherence to the study intervention was defined as both the patient and carer attending at least four online groups.Participation in the online groups was recorded through the transcripts produced at the end of the group and saved on the study platform.To be counted as participating in an online group, participants had to have posted a message to the group at least once.Thus, participants who joined the group or read the transcript of a group were not captured.

Supplement 3.
Unit costs used in the economic evaluation ( 2022        Note.For continuous measures with skewed distributions, both the mean (SD) and median (IQR) are provided.Not related 94 (100.0)88 (98.9) 182 (99.5) outcome scales by assessment timepoint and trial arm 7 Summary of online forum sessions attended by patient and carer participants throughout the intervention period by group type 8 Mean Difference in Service Utilisation Per Participant at 12 Month Follow Up (previous 3 months) (Imputed dataset) 9 Cost-effectiveness plane -societal perspective 10 Comparison of previous 3-month costs at 12-month follow up and QALYs gained ECHOMANTRA group who did or did not complete at least 4 online group sessions (Imputed dataset) (2022 £s) 11 Patient adverse events by trial arm and overall 12 Carer adverse events by trial arm and overall Supplement 1.

**
These were associated to anxiety and upset triggered by completing the self-report questionnaires.These were associated to anxiety and upset triggered by completing the self-report questionnaires.
Question added to the consent form asking whether participants are happy to be contacted about other research projects in the future.-Patientinformation sheet (PIS) typo corrected: header "Version 3_0902" renamed to Version 4, 09/02/2017" Change to data collection Due to participant difficulty in completing the Client Service Receipt Inventory and Social Identity Map, research assistants will now provide telephone guidance to complete these assessments.Change to measures -Questions regarding perceived treatment credibility and acceptability added to monthly questionnaire for ongoing evaluation of the intervention.-Sites will now send medical records including blood test results, blood pressure, pulse, temperature, and saturation via encrypted email to the research team at baseline and when the patient is discharged.

-
Principal Investigator change at Barnet, Enfield and Haringey Mental Health Trust -from Dr Lorna Richards to Minna Raikkonen.
-Following under-recruitment from recruiting sites, TRIANGLE research assistants will now visit the participating sites on a regular basis to assist with recruiting and consenting patients.-BMIinclusioncriteria now applies at the time the patient is approached (not at consent).-Exclusioncriteria specifying that patients not admitted to inpatient/day-care for a minimum of 4 days/week applies at the time they begin the study (not at consent).Staff/research team changes -Changes to the Data Monitoring Committee (DMEC) membership Dr John Morgan has replaced Prof. Hubert Lacey as the DMEC Chair.Dr Eric Johnson Sabine has replaced Dr Robert Palmer as Psychiatrist.
of BMI as part of treatment for anorexia nervosa) and thus, would like to ask again the clinical teams to provide this data at baseline and every 3 months up to patient's discharge.
Principal Investigator change at Tees, Esk and Wear Valley's NHS Trustfrom Katie Bell to Helen McLay.
£s)Summaries of patient outcome scales by assessment timepoint and trial arm Mean Difference in Service Utilisation Per Participant at 12 Month Follow Up (previous 3 months) (Imputed dataset)*Bias corrected accelerated bootstraps ¥ each daypatient stay assumed to be one contact Comparison of previous 3-month costs at 12-month follow up and QALYs gained ECHOMANTRA group who did or did not complete at least 4 online group sessions (Imputed dataset) 2022 £s