IntroductionBackground

The lifetime prevalence of eating disorders (EDs) globally is 8.4% for women and 2.2% for men []. Health care services in several countries are struggling to meet the demand for ED treatment [,], particularly after the impact of the COVID-19 pandemic, which saw an increase in ED symptoms and referrals [,]. Health care services are struggling with long waitlists [], causing specialist ED services to prioritize the treatment of severe cases. This has led to longer durations of untreated EDs [], which are associated with a worsening of symptoms, such as dangerously low BMI (<13 kg/m2) and life-threatening emergencies []. This situation leaves individuals with mild to moderate EDs with limited treatment options and at risk of disease progression.

Similarly, people with EDs may have difficulty accessing treatment due to logistical barriers such as lack of transport [], financial barriers to private psychotherapy [], and language barriers []. For those who struggle to access traditional treatment, digital interventions can offer a more accessible means of managing ED symptoms and distress [].

Individuals with EDs may also encounter nonstructural barriers to seeking treatment. A substantial issue is the fear of stigma surrounding mental health difficulties and EDs. Individuals with EDs are often believed to be responsible for their mental health difficulty as EDs are commonly perceived to be a “lifestyle choice” as opposed to a mental health issue [-]. These attitudes may also be held by health professionals [], deterring individuals from seeking treatment due to feelings of shame or embarrassment about their ED. This, in turn, contributes to further barriers to receiving professional help [].

Digital interventions can address structural and logistical barriers to ED treatment by offering greater flexibility and accessibility. They can be accessed via the internet on PCs or mobile devices, often at a low cost, and do not involve lengthy waitlists. Digital interventions can also address nonstructural barriers, such as fear of stigma, by providing a more discrete form of treatment without fear of judgment as opposed to face-to-face approaches.

Digital interventions may be delivered through smartphone apps or websites, and typically adhere to cognitive behavioral models []. These models often incorporate psychoeducation, cognitive restructuring, and guided or unguided self-help, and frequently include self-monitoring of symptoms, binges, and compensatory behaviors. Such approaches align with the National Institute for Health and Care Excellence guidelines for ED treatment, which recommends ED-focused cognitive behavioral therapy (CBT) for the treatment of anorexia nervosa, binge ED, and bulimia nervosa [].

An example of a digital intervention is the Recovery Record mobile phone app [,], designed for use either as a self-management tool or tool for clinicians to monitor patient’s thoughts and behaviors between treatment sessions. The app records data on user’s meals, behaviors, feeling, and thoughts and uses gamification of app functions to incentivize logging, rewarding users for every entry. In addition, the app provides meal reminders, positive affirmations, and personalized goals and coping strategies.

An additional example is the web-based cognitive behavioral intervention everyBody Plus []. This intervention is intended as a guided self-help intervention for adult women and offers 8 weekly psychoeducation modules covering ED-related topics such as balanced eating, binge eating and purging, improving body image, and dealing with emotions. Similar to the Recovery Record app, everyBody Plus includes functions for logging thoughts and behaviors. However, it specifically focuses on monitoring ED symptoms, such as frequency of binge eating and compensatory behaviors. This intervention represents a more structured approach to digital interventions for EDs, with each session requiring approximately 1 hour to complete and incorporating homework tasks and a group forum.

Digital Interventions for EDs

Evidence from systematic reviews and meta-analyses support the effectiveness of digital interventions for EDs [-]. One systematic review and meta-analysis, which included 23 studies, revealed that use of a digital intervention decreased ED behaviors, ED attitudes, and depression after treatment with medium effect sizes that were sustained at follow-up []. Another review investigated the efficacy of internet-based programs for EDs and found reductions in shape and weight concern, drive for thinness, and bulimic symptoms with small to medium effect sizes [].

However, existing research indicates digital interventions for EDs can have high dropout rates. For instance, a study exploring web-based CBT for patients with EDs reported a dropout rate of 37.6% []. A systematic review of internet-based treatments for EDs found that dropout rates range from 5.3% to 76.8%, with an average rate of 26.3% []. These dropout rates are consistent with digital interventions for other psychological disorders, where dropout rates range from 2 to 83%, with an average of 31% [].

When comparing dropout rates between digital and face-to-face interventions, digital interventions generally see higher dropout rates [,,]. Specifically, internet-based CBT had a dropout rate of 25% to 30%, whereas face-to-face CBT saw lower dropout rates of 24% to 25% []. The higher dropout in digital interventions may be due to engagement challenges [], while dropout in face-to-face therapy is more likely attributed to therapeutic relationships [].

Further research is needed to understand why digital interventions are acceptable [] and effective [,] yet have poor treatment adherence, as this limits their potential effectiveness. Examining user experiences and user preferences of digital ED interventions may provide some insights into the factors affecting adherence and engagement.

Prior Work on User Experience and Preferences

User experience describes the way people use an interactive product [] and includes the combination of the following components: (1) content, presentation, the esthetic appeal of the product; (2) functionality, the capabilities of the product; (3) interactivity, the way users engage with and navigate through the product; (4) manipulation, the control users have in the product’s features and functions; (5) stimulation, the sensory experience from the product, such as visual and auditory feedback; (6) identification, how users perceive and connect with the product, including their sense of ownership; and (7) perceived user control and evocation, the cognitive responses generated in relation to the product, and their consequences [, ]. User experience influences initial commitment to an intervention, therefore has a key role in whether it is adopted [].

Previous studies suggest users typically find digital ED interventions easy to use and useful [,]. Evidence supporting this includes a thematic analysis on 9 participants who used an internet-based CBT intervention, where all participants agreed the intervention was easy to use []. However, other users report that certain digital interventions for EDs have poor usability due to counterintuitive interfaces and basic technical malfunctions [].

Digital interventions are valued for offering confidentiality, privacy, and anonymity, which are substantial factors contributing to their appeal []. However, users have raised data privacy concerns and questioned the accuracy and credibility of the information provided []. Negative user experiences are often attributed to a lack of personalization and ability to meet individual needs (ie, diagnosis, ED subtype, and stage of treatment) [,], and some users have felt overwhelmed by the amount of content and tasks available []. While psychoeducational content has been appreciated for maintaining user engagement with the intervention [], there are concerns that such content may be potentially triggering for some users [,]. Many users indicated that their motivation to adhere to the intervention increased when they had an opportunity to speak with a health care professional, as they provided a sense of safety and support []. However, a minority of users found the involvement of a professional to be distressing and experienced a sense of surveillance [,].

User preferences outline the features and functionalities that individuals seek in interactive products, including aspects such as digital functionality, device types, and content-delivery formats []. However, many digital interventions are designed without user input and therefore do not solve problems most relevant to users []. By identifying user priorities in advance, interventions can be designed to better meet users’ needs, potentially increasing engagement and commitment to treatment [].

One study with 722 community-based participants used a web-based survey to investigate preferences for various eHealth functions aimed at treating and preventing ED []. The study found that preferences and intentions to use the app were consistent across subgroups. Functions receiving >80% endorsement included clinical support, tailored feedback, strategies to change ED cognitions, screening scales to assess symptoms, ED psychoeducation, and just-in-time intervention prompts. Users showed preference for visual content (such as videos and graphs) over audio and text, and they valued opportunities to customize content delivery (eg, through text, images, videos, or audio recordings). The study found that users preferred fewer motivational pop-ups and reminders, in contrast to other studies, which indicated users valued notifications and reminders because they helped prompt healthier food choices [] and provided structure for meal timings [,]. This suggests users may appreciate notifications that support their recovery efforts but are less receptive to less meaningful notifications simply reminding them to use an app. In addition, food logging emerged as a controversial function as some users desired to continue meal tracking habits as a means of relieving stress and reducing fear of gaining weight, whereas others considered it harmful [].

While user experiences and user preferences address different aspects of interaction with a product, they are inextricably linked, with each influencing the other. For example, a user’s preference for a food logging function may enhance their experience if they find that it meets their functional needs and provides interactivity. Similarly, a particularly positive user experience with a digital intervention might reveal valued functions and features that users had not previously recognized.

Due to rapid advancements in technology, 32,000 new health apps were introduced from 2017 to 2021 [,]. Therefore, an updated systematic review was required to synthesize more recent qualitative research in this area. In addition, prior research on user experience and preferences has been limited by small sample sizes and a lack of sufficient qualitative data for in-depth synthesis and insight [,,-,]. By conducting a peer review of existing literature, we aimed to collate and interpret the available data for a broader and more generalizable overview of user experiences and preferences.

Our Study

This study aimed to investigate user preferences for digital ED interventions and how they are experienced. We systematically reviewed studies using self-help with minimal or no guided support. This review investigated the following questions:

What are user experiences of digital interventions for reducing ED symptoms in adults?What are user preferences for digital interventions for reducing ED symptoms in adults?

This review synthesized qualitative studies to allow for an in-depth exploration of attitudes, views, and experiences that are critical to intervention acceptability, outcome, and engagement [].

Rationale

A previous systematic review of qualitative research explored users’ experiences of computer-based and book-based guided and unguided self-help for EDs []. The review compared the experience of using conventional and computer-based self-help but did not explore user preferences for digital interventions. Researchers searched for studies using either guided or unguided self-help. While 38% (3/8) of the included studies were self-guided interventions, the remaining 50% (4/8) of the studies were fully guided self-help interventions.

Methods

The methods are reported according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) checklist [] (). The protocol was preregistered on PROSPERO on May 19, 2023 (CRD42023426932). Ethical approval was not required for this study.

Eligibility Criteria

outlines the specific inclusion and exclusion criteria. We specified a relatively broad criteria, as a preliminary search revealed limited studies on digital interventions for EDs.

Textbox 1. Inclusion and exclusion criteria.

Inclusion criteria

Study: qualitative studies or mixed methods studies with a qualitative substudy; published or unpublished studies conducted between January 2013 and July 2024, in outpatient and naturalistic settings; any language; any countryParticipants: adults (aged ≥18 y) with eating disorder (ED) symptoms; ED symptoms, either self-reported, subclinical or meeting full diagnosis; use of digital interventions to manage ED symptomsIntervention: digitally delivered intervention, delivered via a computer, mobile phone app, smartphone app, or tablet; designed to reduce, manage, or help cope with ED symptoms; primarily a self-help or self-guided intervention; interventions where contact with a clinician, therapist, or support worker was optional may be included; for interventions that include contact with a clinician, therapist, or support worker, they may be included if they adhere to the following: (1) interventions offering contact every session can only be included if 1-way communication is used (eg, weekly personalized feedback with no back and forth conversation) and (2) interventions offering unscheduled 2-way communication (eg, occasional, or drop-in back and forth conversations)

Exclusion criteria

Study: quantitative studies, conference abstracts, prevention studies, and systematic or any other reviewsParticipants: participants below the age of 18 yearsIntervention: interventions targeting a specific subgroup with comorbid physical or mental health disorder (eg, digital interventions for EDs in people with type 2 diabetes); delivered through CD-ROMs and vodcasts; all sessions are fully delivered by a clinician, therapist, or support worker (eg, content from sessions are fully guided by a professional, there is no self-guided activity or learning); scheduled 2-way communication (eg, mandatory weekly back and forth conversation during every session); designed to supplement face-to-face treatment; food or exercise monitoring apps

Although young people may be able to reflect and add valuable insights on digital interventions for EDs, participants below the age of 18 years were excluded due to substantial differences in treatment needs and recommended approaches. For instance, recommended treatment for young people often require family members such as in family-based treatment or anorexia nervosa–focused family therapy for children and young people, while adults are typically treated with individual therapy [,]. These differences make it inappropriate to generalize findings between the 2 groups.

Our inclusion criteria did not require participants to have used a digital intervention, only that they were able to provide their perspective on how a digital ED intervention might help them with their ED. Therefore, studies asking participants about the hypothetical use of a digital intervention were included as long as the hypothetical intervention met the inclusion criteria.

While digital interventions improve accessibility by removing waitlists and reducing reliance on trained professionals, guided interventions still depend on funding and professional availability. By excluding guided interventions, we focused on those that fully leverage the key advantage of digital approaches—their ability to address health care barriers independently of professional involvement.

Our scoping search revealed studies before 2013 used outdated technologies (ie, CD-ROMs and vodcasts), so we only included studies from 2013 to 2024 to allow a review of comparable technologies. We excluded prevention studies because we focused on people with existing ED symptoms, rather than the at-risk population.

We focused on mild to moderate EDs, which are likely to exist predominantly within outpatient and naturalistic settings. We avoided inpatient settings that are likely to include more severe and enduring EDs. We did not require participants to have an ED diagnosis or meet a threshold on any ED assessments. This is because people with mild to moderate EDs may choose digital interventions as an early intervention or discrete method of treatment, so may not have been formally assessed for a diagnosis or have subthreshold symptoms.

We did not require a minimum treatment duration (eg, treatment must last at least 4 weeks), nor did we require a minimum number of sessions. These were to ensure that we included users who may have had negative experiences and terminated use prematurely.

We excluded studies targeting specific subgroups with comorbid conditions, as users may have specific needs that would lead to user experiences and preferences better attributed to their comorbid condition rather than their ED.

Search Strategy

We searched for published and unpublished literature using 6 electronic databases. We used 5 electronic databases that searched for published literature: MEDLINE (OVID), PsycINFO (OVID), CINAHL (EBSCOhost), Embase (OVID), and Web of Science (Core Collection). We used the British Library’s EThOS, a repository for UK PhD theses, to search for unpublished literature.

We built on an existing search strategy [], with adjustments according to our review questions. Changes included the removal of terms that suggested the digital intervention supplemented face-to-face treatment, outdated technology, and other terms irrelevant to our review questions. A specialist university librarian from University College London was consulted and reviewed the final search strategy.

Search terms captured 3 key concepts, EDs, digital interventions, and user experiences or preferences. Search terms included Medical Subject Headings terms and free text terms ( provides the full search strategy).

The initial search using all 6 databases was conducted between April 6 and April 13, 2023. The search was rerun on July 13, 2024, to ensure the inclusion of the most recent research. However, the rerun could not be conducted using EThOS due to its unavailability following a serious ransomware attack.

Study Selection

We imported search results from each database into EndNote (version 20; Clarivate) [] for storage and Covidence for screening []. Search results were deduplicated in Covidence. Any non-English studies were translated using Google Translate.

The first researcher (LC) initially screened titles and abstracts. An independent second reviewer (PT) screened 100 randomly selected studies at the title and abstract screening stage. A third researcher (EB) randomly selected and screened 50% of all titles and abstracts. Any disagreements or uncertainties between researchers about the inclusion or exclusion of studies were resolved through discussion. If disagreement persisted, a fourth reviewer was involved in the discussion.

Full texts were retrieved and checked each against the eligibility criteria. Reasons for exclusion were noted in Covidence. The primary researcher (LC) completed full-text screening and screening of supplementary materials, followed by the independent reviewer (PT), who also reviewed all studies. A third researcher (EB) randomly selected and screened 50% of all titles and abstracts. Any disagreements or uncertainties were resolved through discussion. If disagreement persisted, a fourth reviewer was involved in the discussion.

The screening process involved repeated comparison of title and abstracts against the inclusion and exclusion criteria (). During the full-text screening, the criteria were repeatedly referred back to after reviewing the introduction, methods, and results sections. Given the variation in digital interventions, ranging from fully unguided to fully guided, researchers LC and PT had frequent meetings to discuss any unforeseen ambiguities. Because of the discussions, the inclusion criteria were refined to address any ambiguities.

Data Extraction

Data on study characteristics and findings were extracted using a predefined data extraction sheet. Extracted data included study information (authors, publication year, and country), study design, population (sample size, sex, ethnicity, age range, age mean, and SD), ED symptoms, inclusion and exclusion criteria, recruitment and setting, digital intervention (type of technology, contact with clinician, content or aims, and frequency or duration), main findings (user experiences and user preferences), and quality appraisal score.

Data extraction was completed by 2 researchers independently (LC and EB). Any disagreements or uncertainties were resolved through discussion. If disagreement persisted, a third reviewer was involved in the discussion.

Quality Appraisal

Included studies were critically appraised using the Critical Appraisal Skills Programme (CASP) systematic review checklist []. The checklist includes 10 domains: aims, methodology, research design, recruitment strategy, data collection, reflexivity, ethics, analysis, findings, and the value of the research. Individual studies were scored on each domain from 0 to 3. If <25% of the criteria was met, the domain was given a score of 0; if 25% to 49% was met, a score of 1; if 50% to 74% was met, then a score of 2; and if 75% to 100% was met, then a score of 3 was given [].

Scores from all 10 domains were totaled for each study. The maximum possible overall score was 30. The higher the total score, the higher the quality of the study. A total score of <15 indicated the study was poor quality, 15 to 22.4 indicated moderate quality, and 22.5 to 30 indicated high quality [].

Quality appraisal was conducted by 2 researchers independently (LC and EB). Any disagreements or uncertainties were resolved through discussion. If disagreement persisted, a third reviewer was involved in the discussion. To be inclusive, studies were not excluded based on quality score or quality category. However, quality category was used to weight findings during data analysis.

Qualitative Analysis

Included studies were exported into NVivo (version 14; Lumivero) [] for inductive coding. Any section in the included studies titled “results” or “findings,” including any “results” or “findings” in the abstract or discussion section, acted as our data and were coded. This approach was chosen to prioritize the original authors’ interpretation and synthesis of data, rather than reinterpreting the primary data. By focusing on reported results, we aimed to capture the key themes identified as substantial by the original authors, aligning with the focus of this review to synthesize existing literature rather than reanalyzing primary data. Primary data were not coded because they are rarely made available in electronic databases and accessing them through direct contact with authors could not reliably be guaranteed.

We conducted a thematic synthesis of the data [] using inductive line-by-line coding, in which each line of data was coded according to its meaning and content []. After each line was interpreted, codes were grouped into descriptive themes, meaning concepts that were overlapping or similar from one study to another were combined and reevaluated and similarities and differences in codes were compared across studies []. We anticipated the coding stage to involve a comprehensive interpretation of text, which would subsequently inform the development of our own themes in relation to our review questions. We used open coding as opposed to a codebook, no specific themes were expected in advance, rather this approach was to allow the results to emerge from the data itself without preconceptions. Codes and initial descriptive themes were organized using mind maps and tables.

One researcher conducted the initial coding to allow for deep familiarity with data, allowing a better ability to recognize subtle patterns and underlying meanings in the data, and greater consistency in interpretation. Initial descriptive themes were reviewed and discussed with researcher PT, then collaboratively refined with reflexivity in mind until clearly defined themes were apparent. These formed analytical metathemes and subthemes relevant to the review questions and go beyond the content of the data into the interpretation of the key messages. Both researchers developed the final analytical themes.

User experience and user preferences were distinct research questions to ensure a comprehensive exploration of each area. Although these were examined separately, their findings are integrated in the Results section to reflect their substantial overlap and mutual influence.

Patient and Public Involvement

Two female volunteers with lived experience of ED were recruited to support data analysis. They were recruited using convenience sampling by researchers asking their social circles for volunteers.

A 60-minute online video call was conducted with each individual to discuss descriptive themes and final analytical themes. The video call consisted of a brief explanation on the focus of the review and volunteers’ role and importance of their involvement. The volunteer was presented with a textbox containing the review’s results, the textbox outlined the metathemes and subthemes identified. Each metatheme and subtheme was explained and findings from across the studies were summarized. The researcher then asked questions about the volunteers’ thoughts, whether they resonated with the findings, any disagreements or contrasting opinions they had to the research, and any further thoughts they had that the results did not mention.

Themes were presented to volunteers for feedback, such as the “value of support” theme, which highlighted varied communication functions in digital interventions (eg, direct chat with professionals and group forums). The researcher explained the theme further, for instance, while many users valued these features, some felt pressured or sensed a power imbalance. Volunteers were asked how the themes resonated with their experience of EDs and whether any quotes were misinterpreted. Their lived experiences informed their feedback, helping to improve understanding of the data, reevaluate and reshape themes where needed, and improve the validity of the thematic synthesis. They were compensated with a £20 (US $25) Love2Shop e-gift card for their time.

Reflexivity

To ensure high-quality qualitative research, consideration of researchers’ views and biases on design and analysis were included []. LC is a Chinese British, female postgraduate student, without lived experience of an ED. PT is a White British, female PhD student without lived experience of ED. Both researchers are digital helpline volunteers for an ED charity, so understand the current “treatment gap” in ED service provision in the United Kingdom []. LC and PT discussed their respective backgrounds and potential biases that might influence their methodological choices and interpretation of data using the Social GGRRAAACCEEESSS model [].

During the coding process, reflexivity involved continuous critical reflection on factors that may influence the interpretation of data. Researchers actively considered any of their personal experiences of eating problems, awareness of stereotypes, and the impact of media narratives on EDs. They acknowledged that their understanding of EDs was shaped by Western academic institutions and Western theoretical approaches to mental health and eating behaviors and held this in mind to consider various cultural views on mental health and treatment. Initial codes were discussed among researchers, with open questioning of why certain codes were applied. This reflective dialogue aimed to promote openness and curiosity, allowing for consideration of diverse perspectives and a more balanced, nuanced interpretation of the data.

ResultsStudy Selection

presents a PRISMA flowchart, outlining the screening process that identified the final 8 included studies from 3695 search results. Studies excluded during full-text screening are shown in along with the reasons for exclusion.

‎Figure 1. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart for study selection.

Of the 8 included studies, 2 (25%) [,] asked participants about their hypothetical preferences for digital interventions rather than preference based on actual use. One study did not require participants to have ED symptoms for inclusion, only self-reported body image issues []. However, all participants had a formal ED diagnosis except one. Given body image disturbances are both a risk factor and a symptom of an ED [], this study was included.

Study Characteristics

Included studies were published between 2013 and 2022, with sample sizes ranging from 7 to 24 participants and a sample age range of 18 to 67 years. Of the 8 studies, 5 (63%) provided details on the mean sample age, with an average mean age of 35.5 years. Overall, 75% (6/8) of the studies included all-female samples, with only 4 (4%) male participants across all studies. Of the 8 studies, 3 (38%) described the sample’s ethnicities, all of which were predominantly Caucasian or White (n=29, 59%), with remaining ethnicities including Asian (n=6, 12%), African American (n=5, 10%), Pacific Islander (n=3, 6%), Mixed (n=2, 4%), and Eastern European (n=1, 2%).

All included studies were conducted in high-income countries [-]: United States (3/8, 37%), United Kingdom (3/8, 37%), and Australia (2/8, 25%). Studies were conducted in community settings (5/8, 63%), or both community and specialist settings (3/8, 37%). Participants were recruited through universities (2/8, 25%), universities and the community (3/8, 37%), or solely through the community (3/8, 37%).

outlines study characteristics and intervention details for each study. Digital interventions used include smartphone apps (4/8, 50%), web-based interventions (2/8, 25%), or no specific digital intervention (2/8, 25%). ED symptoms were based on self-report (5/8, 63%), a screening assessment (2/8, 25%), or either self-reported symptoms or assessment (1/8, 13%).

Table 1. Table1. Study characteristics and digital intervention details.Study; countryStudy designSample detailsDigital interventionInclusion and exclusion criteriaCASPa quality categorybJafari [], 2021; United StatesMixed methods study
The University of North Carolina at Chapel Hill students recruited via email advertisement
Community setting
N=24 (100%) female
18-30 years (mean and SD not specified)
White (n=12), African American or Black (n=3), Asian or Asian American (n=4), Hawaiian or Pacific Islander (n=3), mixed race or other (n=2), Hispanic and Latino (n=1)
No contact with clinician
Aimed to help users end restrictive dieting and resolve mindless and emotional eating using self-guided mindfulness exercises
Frequency not stated
Inclusion criteria: female, aged 18-30 years, not currently dieting, own and use a smartphone, reported ≥2 binges within the past 28 days, willing to use an app for 4 weeks
HighYim et al [], 2020; United KingdomQualitative substudy
Sample recruited through 15 NHSc Foundation Trusts, national and regional EDd charities, King’s College London email circulars, social media, and word of mouth
Community and specialist outpatient settings
N=12 (100%) female
21-60 years (mean 42.2, SD 13.7)
Ethnicity not specified
everyBody Plus
Web-based intervention
Weekly one-way communication
Feedback from a psychologist and access to an online discussion board
Aimed to support positive body image, support balanced eating, teach alternative behavioral strategies, and enhance self-confidence
Included readings, recording symptoms, binges and compensatory behaviors, behavioral exercises, and a journal
8 weekly sessions
Inclusion criteria: BNe, BEDf, or OSFEDg symptoms seeking outpatient treatment
Exclusion criteria: BMI <18.5 kg/m2, in need for inpatient ED treatment due to severity, psychiatric comorbidity needing treatment in its own right; or on antidepressant medication
HighJuarascio et al [], 2015; United StatesSample recruited from the Philadelphia community, or referred by clinicians, or a behavioral weight loss program
Community and specialist setting
N=11 (91%) female
25-67 years (mean 42.81, SD 13.85)
Caucasian (n=7), African American (n=2) unknown (n=2)
Intervention had no name
Smartphone mobile app (a prototype)
No contact with clinician
To reduce binge eating symptoms using food, emotion, and binge logging, automated data entry, CBTh-based self-help learning modules, coping strategies, and data visualization
Frequency not stated
Inclusion criteria: met criteria for BED, subthreshold BED, or had a history of binge eating
ModerateMcClay et al [], 2013; United KingdomQualitative substudy
Community-based recruitment (advertisements in mental health websites, newsletters, and magazines. Posters in public places and flyers put into university orientation packages)
Community setting
N=9 (100%) female
28-50 years (mean 33.9, SD not stated)
Ethnicity not specified
Overcoming Bulimia Online
Web-based intervention
Weekly contact with a support worker (Support was deemed optional as support time varied highly and relied on participant engagement with support)
Aimed to change users’ thoughts, feelings, and behavior with regard to food. Modules included psychoeducation, problem-solving, and planning. Each module included homework. Participants recorded mood, binging, and vomiting
8 sessions
Inclusion criteria []: aged ≥16 years, resident in the United Kingdom, BMI ≥18.5, met BN or EDNOSi criteria, established using the Eating Disorders Examination []
Exclusion criteria: currently participating in any other treatment-based ED research, self-reported drug or alcohol addiction, take street drugs daily or weekly, and no regular access to broadband with speakers or headphones
HighLinardon et al [], 2022; AustraliaMixed methods study
Advertisement distributed on the author’s open-access psychoeducational website for EDs []
Community setting
N=14 (79%) female
19-47 years (mean 32.71, SD 8.26)
Caucasian (n=10) Asian (n=2), Eastern European (n=1), Hispanic (n=1)
Break Binge Eating App
Smartphone mobile
No contact with clinician
Included psychoeducation, self-monitoring, learning healthier coping strategies. and body image
Frequency not stated
Inclusion criteria: self-reported at ≥1 objective binge eating episode in past 28 days
ModerateYim et al [], 2021; United KingdomQualitative study
Recruited sample with posters, clinicians’ referrals, and university circular emails at South London and Maudsley ED Outpatient clinic and King’s College London
Community and outpatient setting
N=15 (100%) female
Age not specified
No specific digital intervention referred to
No contact with clinician
Participants interviewed on ED needs and barriers and facilitators for online self-help programs
4×30–60-minute sessions
Inclusion criteria []: female participants aged ≥18 years. Self-identified as experiencing BN, BED, or a related subthreshold binge and purge ED
Exclusion criteria []: BMI <18.5 kg/m2, patients in need for inpatient eating disorder treatment due to the severity of the disorder, patients with substantial psychiatric comorbidity needing treatment in its own right, patients on antidepressant medication who have not been on a stable dose for at least 4 weeks, and patients not eligible for treatment coverage by health insurances or NHS due to low symptoms severity
ModerateJarman et al [], 2022; AustraliaQualitative
Recruited from a larger project (Web-Based Interventions to Reduce Eating Disorders [WIRED]) via posts on the project’s social media accountsCommunity setting
N=7 (100%) female
21-33 years (mean 25.83, SD=5.34)
Ethnicity not specified
No specific digital intervention used
Not applicable
No contact with clinician
Participants interviewed on their needs and preferences for app-based eating disorder interventions
Participants discussed user experiences and preferences hypothetically
Frequency not stated
Inclusion criteria: self-identified lived experience of ED or substantial body image issues
HighNitsch et al [], 2016; United StatesEngagement and iterative usability study
Web-based and print advertisements in the San Francisco Bay Area and on the campus of a large private university
Community setting
N=9 (100%) female
18-25 years(mean and SD not specified)
Ethnicity not specified
Student Bodies-Eating Disorders
Web-based program and mobile app
Unscheduled contact with a personal coach, plus an online, asynchronous guided discussion group []
Aimed to reduce disordered eating behaviors, improve body image, and support the development of coping skills
40 ×10-minute daily sessions.
Inclusion criteria: a positive response to ≥2 out of 5 on the SCOFFj questionnairek []
Exclusion: lack of English language fluency, hearing impairments, and involvement in any depression or anxiety intervention research study
Moderate

aCASP: Critical Appraisal Skills Programme.

bCASP scores indicate study quality, with a score <15 indicating poor quality, 15 to 22.4 indicating moderate quality, and 22.5 to 30 indicating high quality.

cNHS: National Health Service.

dED: eating disorder.

eBN: bulimia nervosa.

fBED: binge eating disorder.

gOSFED: other specified feeding or eating disorder.

hCBT: cognitive behavioral therapy.

iEDNOS: eating disorder not otherwise specified.

jSCOFF: Sick, Control, One Stone, Fat, Food.

kThe SCOFF questionnaire is a widely used and validated screening tool for eating disorder symptoms.

Digital interventions involved no contact with a clinician or support worker (3/8, 37%) or minimal contact with a clinician (3/8, 37%). Digital interventions with minimal contact with a clinician encapsulate those which fall within the spectrum from guided to unguided, but still meet our inclusion and criteria. The study by Yim et al [] was included due to the one-way nature of contact, the study by McClay et al [] was included as the support was deemed optional due to reported high variability in support time due to reliance on the participant engaging with support, and the study by Nitsch et al [] was included as contact was unscheduled and contact through discussion group was optional ( provides further details on contact with professionals).

Quality Appraisal

Studies were rated high quality (4/8, 50%) or moderate quality (4/8, 50%) based on the CASP systematic review checklist []. The CASP quality scores ranged from 18 to 25, with a mean of 22.1. The most common reason for low quality was a lack of reflexivity, where 50% (4/8) of the studies scored 0.

outlines CASP scores for each study and their score for each domain. Domains are labeled 1 to 10 (1=clear aims, 2=appropriate qualitative methodology, 3=appropriate research design, 4=appropriate recruitment, 5=appropriate data collection, 6=considered reflexivity, 7=considered ethical issues, 8=rigor data analysis, 9=clear findings, and 10=value of research).

Table 2. Evaluation of the quality using the Critical Appraisal Skills Programme (CASP) systematic review checklist [].StudyCASP quality assessment score for each domain
DomainCASP scorea
12345678910
Jarman et al [], 2022333330223325Juarascio et al [], 2015333310113321Linardon et al [], 2022330122112318McClay et al [], 2013333330233326Nitsch et al [], 2016333022132221Jafari [], 2021333120323323Yim et al [], 2020333123133325Yim et al [], 2021330212312118

aThe maximum overall score is 30. A total score <15 indicates the study is poor quality, 15 to 22.4 indicates moderate quality, and 22.5 to 30 indicates high quality. A score of 0 represents <25% of the criteria met, 1 being 25% to 49%, 2 being 50% to 74%, and 3 being 75% to 100%.

Qualitative AnalysisOverview

The main findings of each study are outlined in [,,,,,,,].

A total of 7 metathemes and 13 subthemes were identified, as outlined in . Quotes provided are from the studies, as opposed to the studies’ participants.

Table 3. Metathemes and subthemes.MetathemesSubthemesAppeal of digital interventionsAccessibility
Anonymity
Flexibility
Role of digital interventions in treatmentStepping stone for therapy
Alternative treatment
Value of support in treatmentSupport from a professional
Peer support
Communication at the right level—aImportance of engagement—Shaping knowledge to improve eating-disorder behaviorsIncreased self-awareness and reflection
Improvements in eating-disorder behaviors
Design of the digital interventionFunctions
Media format
User interface
Customization

aNo subthemes.

Theme 1: Appeal of Digital Interventions

Users across studies were initially drawn to using digital interventions because they were accessible [,,,,,,] and allowed for anonymity [,,,,].

Accessibility

Users were drawn to digital interventions as an accessible means of getting treatment. The intervention did not have long waitlists, unlike the traditional face-to-face interventions, and was low cost compared to other treatment options such as private psychotherapy [,,].

Users found them accessible because they already owned a device to use the intervention on and no travel was required [,]:

[Digital interventions are] accessible for everyone because technology is omnipresent.
[ 72 ]

Digital interventions were also seen as accessible for those who were non-native English speakers as they provided features such as transcripts and audio recordings []. In addition, many users expressed optimism about digital interventions, recognizing their ability to address barriers they had personally encountered or were aware of in traditional treatment options [,,,].

Flexibility

Users identified flexibility to be a substantially appealing aspect of digital interventions. They perceived digital interventions as highly flexible and practical to use. This was attributed to the normalization of phone use across various situations and ease of access to the intervention through their mobile phone. Most users thought the intervention was compatible with their lifestyle and everyday routine [,,,,,]:

A key advantage...is the flexibility and freedom for the user to engage at any time.
[ 60 ]Anonymity

Digital interventions were appealing because they offered a discrete method of help and the lack of face-to-face contact with a clinician made participants feel more comfortable [,,,,]:

Self-help approach seemed like a desirable alternative to traditional face-to-face treatment, primarily because of the private, convenient, flexible, and anonymous nature.
[ 66 ]

Anonymity was appealing because some participants expressed shame and embarrassment about their ED symptoms or subtype, so preferred to keep treatment discrete from others [,]. Some participants saw technology and the internet as protection as it allowed them to remain anonymous. This made accessing treatment less daunting because they felt less vulnerable and more autonomous over their treatment [,].

Theme 2: The Role of Digital Interventions in Treatment

Across studies users saw digital interventions as less intensive and less effective compared to face-to-face treatment [,,,].

Stepping Stone for Therapy

Many users saw digital interventions as a means of early intervention or a prior step to getting traditional treatment—psychotherapy [,,,]. Therefore, users did not see digital interventions as a sufficient treatment to be used independently, but instead as a complementary tool to be used before or alongside psychotherapy:

Most regarded self-help as a prelude to getting “proper” treatment (“one tool in the toolbox”), rather than an alternative to traditional treatments.
[ 71 ]Alternative Treatment

Many users had tried several treatment options for their ED, many of which were unsuccessful in helping them recover [,,]. Some felt they had exhausted all their options for treatment and were failed by the health care system, therefore saw digital interventions as a tool users would seek out of desperation [,]:

Participants...had tried other approaches without success...and were open to learning a novel approach.
[ 65 ]Theme 3: Value of Support

The digital interventions used in studies had varied levels of communication functions; some had direct contact with a psychologist, personal coach, or support worker [,,]; some offered a community chat forum [,], and others did not offer either [,,,,]. Nevertheless, users highly valued support from a professional [,,] and peers [,,,,].

Support From a Professional

Support from a professional helped users feel supported without being judged and kept them motivated by making them feel accountable [,]:

The support worker was a valued element...it was good to have someone there to help with their progress through the package without judgment: “...she has made a huge huge difference”
[ 66 ]

However, a minority felt pressured by the professional and disengaged as a result []. Users felt a reduced power imbalance with the professional compared to previous experiences with face-to-face therapy. However, this was at the cost of a reduced therapeutic alliance [,]. Due to this, some users still preferred face-to-face contact [].

Peer Support

In the 2 studies that provided access to a discussion group [,], users highly valued peer support. One study [] offered an intervention with neither support from peers nor a clinician, yet participants desired a discussion board to share mutual experiences and interact with others. One study, in which participants did not use an intervention but discussed their preferences hypothetically, found users valued peer support more than support from a professional []. Users thought that support from both peers and clinicians would increase motivation to use the intervention. Peer support could help users feel understood and reduce isolation and loneliness:

Being able to talk to people who are going through the same thing...is really important, because sometimes you just have questions that no one can answer even if they are a doctor or a therapist.
[ 60 ]

Similarly, a study that offered a discussion group [] found increased motivation to continue treatment, as users thought having the option to communicate with other users reduced loneliness and increased their sense of connection and belonging. However, many were concerned with the potential harm of contact with others with an ED, as some users may share harmful thoughts or encourage each other’s ED:

Most participants (3/4, 75%) also acknowledged that this may be unhelpful or even dangerous if used inappropriately.
[ 60 ]Theme 4: Communication at the Right Level

Many users suggested the intervention’s language had a key role in acceptability and motivation to continue use [,,]. If they perceived insensitive language, they felt highly demotivated to continue using it [,].

Users disliked clinical language (eg, “binge eating episode”), acronyms (eg, “BED” and “CBT”), and formal language that made the intervention feel more serious and less enjoyable [,]. Users preferred informal and caring language, as this appeared more approachable and created a more pleasant experience []. However, there was an emphasis on a careful balance between professional and informal language []:

Wording and language used was an important issue as it was found to trigger negative emotions.
[ 46 ]Theme 5: Importance of Engagement

Users highlighted facilitators that helped them initiate use and continue engagement. During initial use, users expressed the intervention’s interactivity and highly engaging content, and functions that kept them returning [,]. One user suggested a way to ensure engagement was to include frequent updates with new content and features. Users across studies explained that they were more likely to continue using if they were inputting data, specifically logging calories, moods, or binges [,,,]. If the intervention had predictable functions or familiar content, they were more likely to disengage and terminate use prematurely []:

User: “I would keep using the app if it continued to update. There has to be new features and new things about it.”
[ 46 ]Theme 6: Shaping Knowledge to Improve ED BehaviorsIncreased Self-Awareness and Reflection

Some users reported that the intervention resulted in positive changes in their ability and frequency of self-awareness, self-realization, and self-reflection. One study suggested that the intervention helped most users by promoting new knowledge and understanding of their ED [].

Four studies reported that the majority of users noticed increased capability and frequency of self-reflection and self-awareness regarding their eating behaviors and beliefs about eating [,,,]. Users identified these positive impacts and attributed them to the intervention:

Participants described the overall positive impact of “holding off on an immediate desire” and learning to reflect on one’s feelings and actions.
[ 65 ]Improvements In ED Behaviors

The majority of users from two studies were able to use their new skills and understanding of ED and apply them when they were about to eat [,]. In one study [], most par