Design

Intervention effectiveness will be tested with a mixed-method approach by a randomised controlled trial in which change in outcomes of patients with EDs such as self-efficacy, will be compared across “Maze Out” + TAU vs. TAU alone groups. Patients who fulfil the inclusion criteria will be randomised to the two conditions. Randomisation will be conducted using the built-in randomisation module in REDCap (Research Electronic Data Capture) [56] from the Odense Patient Data Explorative Network (OPEN). To ensure adequate allocation concealment, the random allocation sequence will be generated before patient enrolment begins, by a member in the research group (RB) who is independent and not otherwise involved in the study. The researcher in charge of obtaining written informed consent will initiate the randomisation procedure when the patient has agreed to participation and completed the baseline measures. Patients will be informed of the results of randomization immediately after the procedure has been conducted. No stratification will take place. Preliminary comparisons based on stratification will be explored in order to inform future research using Maze Out.

A subgroup of patients randomised to Maze Out + TAU will be selected for qualitative interviews, taking into consideration both patients who have played the 15 intervention weeks and patients who have played a few times or not at all.

Participants

Participants will be recruited through their contact person or clinician from a broad set of treatment institutions in Denmark, offering treatment and care to patients diagnosed with ED. The treatment institutions include psychiatric centres within the mental health care services, psychiatric clinics, municipal ED teams, mental health care institutions, and endocrinology services specialised in EDs. All regions in Denmark will be invited to participate. Recruitment from a wide range of treatment institutions and locations is intended to reflect everyday practice and nuances in clinical approach to EDs in Denmark.

To be eligible to participate, patients must: (1) agree to participate in the study and sign written informed consent; (2) be aged 18 or above; (3) speak, read and understand Danish; (4) have a registered ED diagnosis at the treatment site, according to ICD-10 and, made by at list a clinical interview and biometrics such as BMI (i.e., not only ED symptoms) and (5) receive support or treatment for ED. Exclusion criteria: There is no exclusion criteria. The ability to speak, read and understand Danish is necessary as Maze Out is currently only available in Danish.

Treatment interventionsED treatment as usual (TAU)

TAU offered at the participating units is diverse and mainly consisting of psychological treatment combined with nutritional counselling and support for everyday functioning. Psychological treatment involves elements of Cognitive Behavioural Therapy (CBT) [57], Mentalization-based treatment (MBT) [10] and physiotherapy. These methods are utilised predominantly at mental health care services and psychiatric clinics, while the nutritional counselling is the major component in endocrinology services, and support for everyday functioning is utilised in municipal ED teams. Some patients receive treatment in a group setting, while others receive individual treatment or both. Psychological treatment is delivered by trained therapists, such as psychiatric nurses, social workers and clinical psychologists. Nutritional advice is delivered by dieticians, nurses and physicians. Patients receiving TAU only will not have access to Maze Out for 15 weeks after recruiting. After 15 weeks, and after they have completed their measures, these patients will receive a link that allows them to download and play Maze Out to make participation in the study ethically fair. An online self-rapport questionnaire will be used at baseline to gain understanding of what TAU consists of (Additional file 3: Appendix 3).

Maze Out + ED treatment as usual (TAU)

The group allocated to this condition will receive TAU and access to play Maze Out. Patients will be asked to play at least once a week or as long as they wish for 15 weeks in total. Maze Out will be provided through a link whereby they can download the game on their own Android or Apple devices such as smartphones and tablets.

Procedure

Patients will be informed about the study by the healthcare staff they are usually in contact with (e.g., therapist or contact person) and asked if they are willing to meet with a trained research assistant who will provide further study information. If the patient agrees, the research assistant will provide written and oral information about the study. Patients will then have at least one week to decide if they want to participate in the study, before being asked to provide written informed consent. Participants will be able to leave the study at any time without reason and without any consequences for current or future treatment. After obtaining informed consent, baseline measurements will be obtained.

Data management

Outcomes will be assessed using information from patient case notes and self-report questionnaire (Additional file 4: Appendix 4) provided via a secure email system (e-Boks). Data from self-report questionnaires will be returned via secure routes to a secure data base (REDCap) [56]. Data collected from patients’ health records will be entered directly to REDCap by a trained research assistant. Back-end data from Maze Out will be collected pseudonymized and transferred to RedCap by a trained research assistant.

All data collected in the study will be treated as strictly confidential. Data will be anonymized (person sensitive data) and encrypted (group allocation) before extraction by an external data manager and transmitted through secure pathways to a secure data base. Researchers responsible for reporting the results will not have access to the data and will receive the analysis in their final form for reporting. A Statistical Analysis Plan (SAP) will be written before any analysis on the data will be performed, where the blinding is lifted for the statistician.

Measurements

Data on the use of Maze Out will be collected from the back-end of Maze Out, which consists of: (a) information on which portals the patient has been through; (b) duration and frequency the patient has played the game; and (c) answers to questions within different missions of Maze Out. These data will be collected throughout the 15-week period patients play Maze Out.

Data on socio-demographic factors: age, gender, body mass index (BMI), ED diagnoses, prior ED treatment, duration and type of current ED treatment, psychiatric comorbidity, will be collected through patients’ medical records. In addition, data on occupation, children, social network, place and duration of treatment, ED symptoms and desired areas of personal development will be collected through a self-report questionnaire as stated below (T1), administered as an online questionnaire. Self-report measures will be collected at baseline (T1), and post-intervention 15 weeks after enrolment (T3). Patients will be asked to provide data during the intervention after 8 weeks (T2) and post intervention (T3) using questionnaires, distributed through their e-Boks, a secure digital mailbox for receiving and managing official documents and communications in Denmark [58]. Data are collected at following timepoints (Fig. 1):

Fig. 1

Overview of enrolment, interventions, and assessments. EDI-3, Eating Disorder Inventory, version 3; SASB, Structural Analysis of Social Behavior

At baseline (T1):

1.

Self-report questionnaire created by the study authors and administered as an online questionnaire (Additional file 4: Appendix 4)

2.

Socio-demographic data form patients’ case notes.

3.

TAU questionnaire created to describe TAU across multiple sites (Additional file 3: Appendix 3)

At baseline (T1), 8 weeks (T2) and 15 weeks (T3):

4.

Self-efficacy: The 6-item chronic disease self-efficacy scale is a 6-item self-report questionnaire, aimed to evaluate self-efficacy in terms of making decisions about one’s own life while suffering from chronic diseases [59]. It has been adapted by the authors to make it clinically appropriate to individuals suffering from ED (5-item SE_ED) (Additional file 5: Appendix 5).

5.

Brief INSPIRE-O: is a 5-item questionnaire to measure recovery on five dimensions of personal recovery (i.e. connectedness with others, hope and optimism for the future, positive identity, meaning in existence, and empowerment) [60].

6.

Structural Analysis of Social Behavior (SASB) Intrex Questionnaire is a 16-item questionnaire based on SASB model for self-image ratings where both positive and negative aspects of the self-image are assessed [61].

At baseline (T1) and 15 weeks (T3):

Eating Disorder Inventory, version 3 (EDI-3) [62] is a self-report questionnaire that consists of 91 items organized onto 12 primary scales, consisting of 3 eating-disorder-specific scales and 9 general psychological scales that are highly relevant to, but not specific to EDs. It also yields six composite scores, one that is eating disorder specific and five that are general integrative psychological constructs (i.e., Ineffectiveness, Interpersonal Problems, Affective Problems, Overcontrol, and General Psychological Maladjustment). We will focus on 2 of the composite scores: Ineffectiveness Composite (IC) and Interpersonal Problems Composite (IPC) [45].

Statistical analysis

Socio-demographic data gathered at baseline will be summarized through descriptive statistics. Categorical variables will be expressed as frequencies and percentages, while continuous variables will be presented with mean, range, and standard deviation.

Changes in the primary outcome self-efficacy levels over time (baseline, after 8 and 15 weeks) will be modeled using linear mixed models. Participant-specific random intercepts will be included, as well as random slopes if these improve the model fit significantly. The fixed part of the model will consist of the intervention, time, and the treatment-time interaction. To assess the normality assumptions of residuals for both fixed and random effects, normal quantile–quantile plots will be employed. If deviations from normality are observed, sensitivity analyses will be conducted using non-parametric bootstrapping with 1,000 bootstrapping samples. A significance threshold of 0.05 will be applied. Assuming the dropout mechanism is missing at random (MAR), linear mixed models deal efficiently with missing values due to dropout using the maximum likelihood estimator. Therefore, all available data will used in an intention-to-treat approach (ITT).

The secondary outcomes INSPIRE-O and SAS-B will be analyzed in the same way as the primary outcome, while the model will be adjusted to include only the two measured time-points (baseline and 15 weeks) for the EDI-3, with focus on two composite scales: Ineffectiveness Composite (IC) and Interpersonal Problems Composite (IPC) from the EDI3.

To investigate missing data patterns and mechanisms, techniques such as descriptive statistics on missingness and analysis of missing data mechanisms (e.g., Little’s test) will be employed. Sensitivity analyses will be also conducted, investigating the complete case scenario and worst-case imputation to evaluate the robustness our imputed results.

A statistical analysis plan will be described before data analysis and posted on ClinicalTrials.gov before data analysis commences.

Sample size calculation

Since Maze Out is a novel intervention, hardly any data are available that are useful for power-calculations. Our pilot study [29] showed a pre-post mean change score of 4 based on the same self-efficacy scale as used in the current study (SD = 1.5) at 8 weeks. We hypothesize that playing the game for 15 weeks will increase mean-score self-efficacy by 25%, compared to the TAU control group. Thus, with a power of 80% and alpha = 5%, and an assumed dropout rate of 20%, we need to include a total N = 94 (i.e., N = 47 per group) to be able to find medium effect size.

Qualitative design

The qualitative exploration of Maze Out will involve individual and group interviews exploring the user experiences of using Maze Out and will seek insights into the detailed use of the SG and how it is experienced by patients with EDs as well as clinicians and significant others.

A subsample of patients from the RCT, in addition to a selection of clinicians and relatives, will be asked to participate in qualitative interviews, after playing Maze Out, with the purpose to investigate how Maze Out is experienced by its users. Purposive sampling will be used to select the participants. The purposeful sampling strategy aimed to identify participants who have played Maze Out a lot and those who played for a short time only or not at all. This strategy is chosen with the purpose to ensure that those who complete playing Maze Out and who do not complete it are both represented, and they will be included until data saturation.

Coding will be conducted using a concept map encompassing the topics of interview guide relating to the overall research questions. Codes will be grouped into semantic themes and organized according to the overall topics during the analysis process.

Nvivo 1.7.1 will be used to manage and organize qualitative data [63].

Qualitative interviews

Three interview guides will be developed prior to the interview to explore users’ experiences of Maze Out, one for each group, with the purpose of allowing patients, clinicians, and family members to describe both the challenges and opportunities associated with the use of Maze Out. The interview guides will focus on overall experience of playing the SG, impact of the SG on relationships (e.g., health care staff and significant others), whether the SG gives insight into patients’ disorders, and if so, how. It will also capture participants’ reflections on whether the SG may have an impact on shame about ED symptoms and behaviours (Additional file 6: Appendix 6).

In addition to the qualitative data, quantitative data on the use of Maze Out will be included in the study. The quantitative data will be used to help the interpretation of the interviews in terms of a deeper understanding of the different perspectives depending on how the informant used the game.

Qualitative data analysis

Thematic content analysis will be conducted using a realist methodological framework [64]. The analysis will be theoretical as it will be driven explicitly by the analysis from specific research questions, in contrast to questions evolving during the coding (i.e., an inductive approach). A theme in thematic content analysis represents both patterned responses and meanings that capture something important in relation to the research questions, but is not dependent on quantifiable measures [64]. In line with the theoretical approach, the identification of themes will be performed on a semantic level with grouping of explicit meanings and statements [65]. This implies that a unidirectional relationship is assumed between the statements of the participants and their meaning and/or motivations [66, 67]. The semantic patterns, or themes, will thus be summarized based on their surface meaning and the interpretation of these themes will be done focusing both on their explicit meaning but also on their implications and broader importance [68]. Example quotes will be provided across different themes.

Coding will be conducted using a concept map encompassing the topics of interview guide relating to the overall research questions. Codes will be grouped into semantic themes and organized according to the overall topics during the analysis process. Quantitative data on use of Maze Out from the back-end of Maze Out will be used to inform the qualitative data. For example, if there is a relation between the extend of playing and how it is experienced by the patient.

Data triangulation

Qualitative and quantitative data will be triangulated to investigate the effect of confounders, and to examine the elements that may have an influence on the quantitative outcomes and exposure to the game. We expect that by doing so we will gain a deeper understanding of the mechanisms behind the use and effect of Maze Out and increase the validity and reliability of findings. A variety of sources (e.g., semi-structured interviews, self-report questionnaires and player statistics) will be used to extract data. Triangulating these data will aid interpretation of our results; for example, qualitative interviews will provide insight into the motivations or reasons on how many times a participant played the game.