Development and usability testing of a cognitive-behavioral therapy-guided self-help mobile app and social media group for the post-acute treatment of anorexia nervosa

Abstract

Objective:

Anorexia nervosa (AN) is often treated in the acute setting, but relapse after treatment is common. Cognitive-behavioral therapy (CBT) is useful in the post-acute period, but access to trained providers is limited. Social support is also critical during this period. This study utilized a user-centered design approach to develop and evaluate the usability of a CBT-based mobile app and social networking component for post-acute AN support.

Method:

Participants (N = 19) were recently discharged from acute treatment for AN. Usability testing of the intervention was conducted over three cycles; assessments included the System Usability Scale (SUS), the Usefulness, Satisfaction, and Ease of Use Questionnaire (USE), the Mobile Application Rating Scale (MARS), a social media questionnaire, and a semi-structured interview.

Results:

Interview feedback detailed aspects of the app that participants enjoyed and those needing improvement. Feedback converged on three themes: Logistical App Feedback, boosting recovery, and Real-World App/Social Media Use. USE and MARS scores were above average and SUS scores were “good” to “excellent” across cycles.

Conclusion:

This study provides evidence of feasibility and acceptability of an app and social networking feature for post-acute care of AN. The intervention has potential for offering scalable support for individuals with AN in the high-risk period following discharge from acute care.

1. Introduction

Anorexia nervosa (AN) is a highly debilitating, costly, and lifethreatening psychiatric disorder that affects up to 4 % of women in their lifetime (Keski-Rahkonen & Mustelin, 2016; Mustelin et al., 2016; Smink et al., 2013; Udo & Grilo, 2019). Mortality from AN is extremely high, with a meta-analysis indicating a weighted mortality rate (i.e., deaths per 1000 person-years) of 5.1. This rate is 6 times that of the general population and is one of the highest mortality rates of any psychiatric disorder (Arcelus et al., 2011).

Patients with severe AN are often treated in the acute setting (e.g., inpatient), but 31–52 % of patients relapse after treatment, with the greatest risk of relapse occurring in the two months immediately following discharge (Berends et al., 2016; Berends et al., 2018; Khalsa et al., 2017). Recent systematic reviews have highlighted the need to support transitions from intensive treatment for adults with AN (Bryan et al., 2021; Giel et al., 2021); cognitive-behavioral therapy (CBT) may be particularly helpful in this endeavor. For example, in one study, adults with AN receiving nutritional counseling after acute treatment relapsed significantly earlier and at a higher rate than adults receiving CBT (53 % vs. 22 %), and the overall treatment failure rate (i.e., relapse and drop out combined) was significantly lower for those receiving CBT (22 %) compared to nutritional counseling (73 %) (Pike et al., 2003).

There is also evidence that digital interventions are successful in supporting treatment of mental health disorders generally, and AN specifically. For example, recent systematic reviews have demonstrated the positive effects of digital interventions for anxiety, substance use, and depression (Lattie et al., 2019; Bonfiglio et al., 2022; O’Logbon et al., 2023). Regarding AN specifically, Fichter et al. (2012) compared a digital, CBT-based relapse prevention program to TAU in 258 women with AN in Germany, over a nine-month period after inpatient treatment. Results indicated the intervention group gained weight while the TAU group experienced weight loss, and these differences in weight were statistically significant (Fichter et al., 2012). Furthermore, Neumayr et al. (2019) randomized 45 women with AN in Germany to an 8-week mobile intervention (the German version of “Recovery Record,” an app for self-monitoring of eating as well as ED thoughts, feelings, and behaviors) with therapist feedback as an adjunct to TAU or to TAU alone. Patients reported high acceptance of the app, and there were non-significant small to moderate between-group effect sizes favoring the intervention regarding BMI and ED psychopathology at post-intervention; however, there were no significant differences at 6-month follow-up (Neumayr et al., 2019).

These results suggest there may be potential for a digital aftercare intervention for AN, but there are limitations to the existing work, including: 1) to date, no digital aftercare interventions developed in English or tested in the U.S., which is important given the unique treatment climate in the U.S. (e.g., large country with limited access to qualified providers in many areas, no universal healthcare); 2) use of older technology with limited sophisticated features that users have come to expect (e.g., Fichter et al. (2012)); and 3) even when new technology has been used (i.e., Recovery Record app), it has not included specific features for aftercare support for this group, such as helping users develop internal motivation to change, addressing weight loss during the intervention, or fostering social support (Neumayr et al., 2019). Overall, these findings indicate that a digital, CBT-based intervention for the post-acute treatment of AN that specifically addresses the unique needs of this population could have extremely high potential for increasing accessibility of quality care.

Social support may also promote recovery from AN (Darcy et al., 2010; Dawson et al., 2014; Federici & Kaplan, 2007; Tozzi et al., 2003; Wetzler et al., 2020). Qualitative work reveals that ED recovery is largely influenced by a sense of connection to self and others (Linville et al., 2012). Notably, social media offers a scalable way for individuals with AN to access social support from others with similar experiences (Aardoom et al., 2014). Development of positive, recovery-focused online communities is especially critical in EDs given the existence of many pro-ED online communities that, although connecting people experiencing these problems, encourage disordered behaviors (Rouleau & Von Ranson, 2011). Despite containing many harmful aspects, users perceive social support as one of the key functions of these sites (Bohrer et al., 2020; Rouleau & Von Ranson, 2011) and desire for support, interaction with others, and connecting with others with an ED are reasons individuals report for engaging with them (Csipke & Horne, 2007; Wilson et al., 2006). Individuals with AN who are being discharged from higher levels of care for an ED, where they were surrounded each day by peers with a shared understanding of what they were going through, may experience a high degree of isolation upon returning home. Thus, social support, particularly from peers, is important in the process of recovery from AN, including the need for a positive social networking outlet, where one could connect with others in recovery. However, support from peers also needs to be moderated given evidence that online interactions with peers with EDs can turn negative (Fitzsimmons-Craft et al., 2019).

Given the effectiveness of CBT in the post-acute care of AN, early successes of digital interventions in improving AN outcomes, and the ability of social media to promote social connection, a digital CBT-based intervention for the post-acute intervention of AN, paired with a social networking component, has high potential for increasing access to quality care in the post-acute period. Indeed, it is during this period that individuals with AN may be more physically and cognitively capable of engaging in approaches targeting psychological symptoms (Bardone-Cone et al., 2010; De Young et al., 2020), but patients encounter many barriers to treatment (e.g., long waiting lists, cost), and even when they do receive care, it is often not evidence-based (Fairburn et al., 2013). Not only does this approach address such barriers to treatment, but it can also be used as in-the-moment support (e.g., at times when patients are not able to see a member of their treatment team), offering a critical new model of treatment delivery to support individuals with AN in the post-acute intervention period.

Recent research has highlighted the importance of user-centered design (also known as human-centered design or design thinking), which entails centralizing the design of such technological tools on those who will be using or impacted by them and the contexts in which they will be implemented (Graham et al., 2023). User-centered design promotes greater engagement, acceptability, and effectiveness of digital interventions. Critical to an intervention’s success is refining and developing the intervention, which involves performing usability testing to assess whether the product functions as intended (Graham et al., 2023). In line with a user-centered design approach, this study developed and evaluated the usability of a CBT-based, mobile app, alongside a social networking component, to address the post-acute care and relapse prevention of AN. We detail the process of developing the app and social networking component using user-centered design and three iterative cycles of usability testing.

2. Method

2.1. Participants and recruitment

A total of 19 participants were enrolled (n = 5, Cycle 1; n = 5, Cycle 2; and n = 9, Cycle 3). Participants were primarily recruited in ED treatment centers via flyers, emails to those discharged from intensive treatment, and social media posts shared by treatment centers. Eligibility criteria required that participants were at least 18 years old, identified as a cisgender woman, reported being discharged from acute treatment (i.e., inpatient, residential, partial hospitalization program, intensive outpatient) for AN within the past two months, endorsed having a physician monitoring their medical safety, owned a mobile phone, spoke English, and were a U.S. resident. Although individuals of all genders suffer from EDs, we limited enrollment to females as rates of AN are much higher in females than males (Hudson et al., 2007), and in studies reporting on outcomes of residential treatment for EDs (where participants in the current study were recruited from), samples were overwhelmingly female (Peckmezian & Paxton, 2020).

2.2. Mobile app design

The mobile app, Helping HAND: Healing Anorexia Nervosa Digitally, was developed based on an existing CBT-based mobile app for college women with binge-type EDs, Student Bodies-Eating Disorders (SB-ED) (Fitzsimmons-Craft et al., 2020). Content was adapted to be more relevant to women with AN being discharged from acute treatment (e.g., addressing weight gain in recovery), and these refinements were informed by Pike et al.’s (2003) in-person, CBT-based treatment protocol for women with AN transitioning out of inpatient care. The Helping HAND program was hosted on the platform of an industry partner, SilverCloud Health. The SilverCloud Health platform facilitates symptom monitoring by offering users the option to complete questionnaires assessing their ED-related thoughts and behaviors and also allows users to access support of a trained coach, with whom users can message weekly and video call monthly. Notifications were not possible through the app during usability testing, but during the subsequent randomized controlled trial, email notifications will be enabled. Additionally, for usability testing, participants were not able to try the app with the symptom monitoring feature or coached support, but they were able to provide feedback on these proposed features.

2.3. Mobile app modules

The Helping HAND app comprises eleven modules: Getting Started, Motivation, Addressing Weight, Addressing Food, Coping Well, Thinking Well, Addressing Core Beliefs, Body Wellness, Media Wellness, Relationship Wellness, and Relapse Prevention. Each module includes a brief introduction, a personal story section, content related to the module’s topic, and a space for participants to outline their goals. See Fig. 1 for examples of Helping HAND components and Supplementary Materials for a breakdown of each of the modules. Adaptations to the app, detailed in Table 1 of Supplementary Materials, were made over the course of usability testing, but the breakdown of the modules remained the same.

Fig. 1.

Screens illustrate Helping HAND components. From left to right: app home page with module overview, personal stories, meal planning tool, and interactive CBT tools.

2.4. Social media development

Prior to usability testing, study team members drafted sample social media posts to be reviewed by usability testing participants. Posts were specifically created to increase engagement and were categorized into four groups–informational posts, coping strategy posts, posts that encourage participant interaction, and inspirational material–that have been found to facilitate online prosocial communication and recovery-focused support among individuals with AN (Eichhorn, 2008). The study team also explored various social media platforms that could offer interactive group spaces (e.g., Facebook, Instagram, Discord, Reddit), to determine their fit with the study’s aim of creating a private, recovery-focused group for solely app users to connect. Ultimately, Facebook and Instagram were deemed by the study team to be the two most feasible options, and over the course of usability testing, the study team solicited feedback from study participants on their preferred platform.

2.5. Procedure

Usability testing was conducted with three cycles which has been deemed optimal when developing mHealth interventions using user-centered design (Molina-Recio et al., 2020). Between each cycle, refinements were made to the app content and functionality, as well as the social media page, based on the feedback gathered. All procedures occurred remotely (i.e., not in person) and were approved by the Institutional Review Board.

2.5.1. Cycles 1 & 2

After eligible participants (five different participants for each of the two cycles) provided informed consent, participants viewed the existing SB-ED SilverCloud program, a content outline for the proposed mobile intervention, and a Google document version of adapted modules. A think-aloud protocol was utilized to elicit participants’ instantaneous reactions, thoughts, and opinions via an interview (Jaspers et al., 2004) which was audio-recorded. Post-interview, participants completed an online questionnaire, including the measures described below, as well as specific questions about the proposed social media networking feature (e.g., platform they would prefer, sample posts) and potential content interests. Participants were compensated with a $50 Amazon electronic gift card.

2.5.2. Cycle 3

Cycle 3 differed from Cycles 1 and 2, entailing remote, unsupervised usability testing during which participants had access to the app and social networking component for two weeks. Upon joining the Facebook group, participants were encouraged to enable push notifications. Notably, participants did not have the option of engaging with a coach or the symptom monitoring feature during this study period. After two weeks, participants completed a 60-minute semi-structured interview call. Interviews, conducted by a research team member, were audio recorded. After the interview, participants completed an online questionnaire (the same used for Cycles 1 and 2) and were compensated with a $75 Amazon electronic gift card.

2.6. Measures

2.6.1. Quantitative data

Demographic information, including race, ethnicity, sexual orientation, household income, insurance status, and living region, was collected.

The System Usability Scale (SUS) was administered post-engagement with the mobile app to assess its usability. The scale consists of a 10-item questionnaire along a 5-point Likert scale, with response options ranging from strongly disagree (1) to strongly agree (5). Scores are scaled from 0 to 100, with a SUS score over 68 delineating “above average” usability. The scale is validated for use in small sample sizes (Bangor et al., 2008; Lewis & Sauro, 2009).

Participants also completed the Usefulness, Satisfaction, and Ease of Use (USE) Questionnaire Short-Form, which measures four dimensions: usefulness, ease of use, ease of learning, and satisfaction of users (Lund, 2001). Items are scored along a 7-point Likert scale, with response options ranging from strongly disagree (1) to strongly agree (7). For each subscale, items were averaged to generate or score.

Participants additionally completed the Mobile Application Rating Scale (MARS), a 23-item validated scale that assesses the quality of health apps and includes one subjective quality rating—a reflection of the user’s overall satisfaction—and four objective quality subscales: engagement, functionality, aesthetics, and information (Stoyanov et al., 2015; Stoyanov et al., 2016). All four dimensions are scored on a 5-point Likert scale from inadequate (1) to excellent (5) and scores for these dimensions are averaged to generate a score.

Finally, participants completed a study-specific social media survey that assessed how often they would like to see posts, which social media platform they would prefer (i.e., Facebook, Instagram), social media content with which they would engage (i.e., posts, stories, reels, Q&As, polls), and the extent to which they would enjoy interacting with others going through a similar experience on a social media platform. This survey can be found in Supplementary Materials.

2.6.2. Qualitative feedback

Semi-structured interviews were conducted to solicit participant feedback on the individual mobile app modules (e.g., Getting Started, Motivation, Addressing Weight), the utility and design of the mobile app, and overall positive and negative experiences. The interview guides for each cycle can be found in Supplementary Materials.

2.7. Analytic strategy

2.7.1. Quantitative analysis

Means and standard deviations for the quantitative measures were recorded across cycles for descriptive purposes. Inferential statistics were not used given the small sample size.

2.7.2. Qualitative analysis

2.7.2.1. Iterative development.

The initial versions of the Helping HAND mobile app and social networking component were tested by Cycle 1 participants. Interviews were transcribed and qualitative feedback was assessed by frequency (i.e., how often certain suggestions were made) and feasibility to inform refinements to the app and social networking component between cycles. After each cycle was completed, participant feedback was compiled and discussed in a study team meeting to evaluate proposed program alterations; feasible changes were applied prior to the following testing cycle. Some suggestions, such as including the social networking feature within the app, were not feasible given limitations to the SilverCloud Health platform on which the app was housed.

2.7.2.2. Thematic analysis.

To examine participant opinions on the mobile intervention, the study team transcribed the interview recordings directly, then analyzed these transcripts using qualitative inductive thematic analysis with a realist lens focused on understanding the realities and experiences of the participants (Braun & Clarke, 2006). Thematic analysis is structured to identify repeating patterns and contexts of participant feedback and fit our analysis goals of assessing this feedback through an inductive, realist lens. The study team took the following steps of analysis, in line with Braun and Clarke (2006): 1) read the transcripts to create a framework of understanding for participant feedback, 2) created codes and groupings of codes, 3) organized codes into initial themes, 4) reviewed themes, 5) defined and named themes, and 6) wrote a report on the data (i.e., this manuscript).

2.7.2.3. Coding process.

Before the coding process began, coders discussed their positionality and salient identities to begin reflecting on how these would impact the analytic process.¹ Next, each coder read and reviewed the transcripts to begin crafting an initial code book. This initial code book was then discussed among all coders and was used to code 1–2 transcripts each. After each coder had used the initial code book to code at least one transcript, the group met several times to discuss alterations to the proposed codebook and generate a final version. Each transcript was coded twice, by two different coders, and then coding disparities were identified and discussed as a team until a consensus was reached. After all transcripts were coded with the final set of codes, codes were first grouped into subthemes based on their interconnected meaning and context within the transcripts. Subthemes were then grouped into overall themes, and groupings were reevaluated as the transcripts were reviewed again. Finally, coders named and defined the themes and subthemes. Coding was done utilizing the software Dedoose and theme development was done using post-it notes while cross-referencing coded transcripts on Dedoose (Dedoose version 7.0.23, web application for managing, analyzing, and presenting qualitative and mixed method research data, 2016).

3. Results

3.1. Participants

Sample sizes above nine often achieve coding saturation and sample sizes between 16 and 24 achieve meaning saturation (Creswell & Creswell, 2018; Hennink et al., 2016). Therefore, we expected that the 19 participants enrolled sufficed for the purposes of qualitative analyses. All participants were cisgender female, and participants largely identified as White and Non-Latina (n = 18, 94.7 %). Most participants (84 %) reported a household income of at least $50,000 and most participants had private insurance and lived in suburban areas. See Table 1 for full details on study participants.

Table 1.

Table of participant characteristics by usability testing cycle.

Participant #	Age (in years)	Race and ethnicity	Sex	Sexual orientation	Household income (in thousands)	Insurance status	Living region
Cycle 1 (n = 5)
1	23–30	White, non-Latina	Female	Lesbian/gay	$20–35	Private	Suburban
2	31–40	White, non-Latina	Female	Heterosexual/straight	$50–75	Private	Suburban
3	31–40	White, non-Latina	Female	Heterosexual/straight	$75–100	Private	Urban
4	41–50	White, non-Latina	Female	Heterosexual/straight	$100–150	Private	Suburban
5	31–40	White, non-Latina	Female	Heterosexual/straight	$75–100	Private	Suburban
Cycle 2 (n = 5)
6	18–22	White, non-Latina	Female	Bisexual	$100–150	Private	Urban
7	31–40	White, non-Latina	Female	Questioning	$20–35	Private	Urban
8	23–30	White, non-Latina	Female	Heterosexual/straight	$100–150	Private	Rural
9	41–50	White, non-Latina	Female	Heterosexual/straight	$50–75	Private	Suburban
10	41–50	White, non-Latina	Female	Heterosexual/straight	$100–150	Private	Suburban
Cycle 3 (n = 9)
11	18–22	Bi/multiracial, non-Latina	Female	Lesbian/gay	$50–75	Private	Suburban
12	31–40	White, non-Latina	Female	Queer	$20–35	Public	Suburban
13	18–22	White, non-Latina	Female	Bisexual	$200+	Private	Suburban
14	18–22	White, non-Latina	Female	Heterosexual/straight and questioning	$150–200	Private	Suburban
15	31–40	White, non-Latina	Female	Lesbian/gay	$100–150	Private	Suburban
16	23–30	White, non-Latina	Female	Heterosexual/straight	$20–35	Private	Suburban
17	41–50	White, non-Latina	Female	Heterosexual/straight	$100–150	Private	Suburban
18	18–22	White, non-Latina	Female	Heterosexual/straight	$100–150	Private	Suburban
19	18–22	White, non-Latina	Female	Lesbian/gay	$75–100	Public	Suburban
Total (N = 19)

3.2. Scale scores

The average SUS, USE, and MARS scores from Cycle 1 (Table 2) indicated the initial intervention had above average usability when compared to the standards of the measures and demonstrable user acceptance even prior to any refinements. Moving to Cycle 2, the app demonstrated improved SUS, USE, and MARS scores from the Cycle 1 baseline in all but USE Usefulness, USE Ease of Learning, MARS Information Quality, and MARS Subjective Quality subscales. By Cycle 3, the app showed an increase from Cycle 1 in all measured usability scores apart from the MARS Engagement score, which dropped by only 0.01 points. Overall, Helping HAND was rated positively by users, achieving SUS ratings of “excellent,” USE ratings of “good,” and MARS ratings of “excellent” (Lewis & Sauro, 2009; Lund, 2001; Stoyanov et al., 2016).

Table 2.

Descriptive statistics for SUS, USE, and MARS questionnaires by usability testing cycle.

	Cycle	Descriptive
SUS Total		Mean	SD	Min.	Max.
	1	82.50	15.50	57.50	100.00
	2	87.50	12.60	72.50	100.00
	3	88.30	10.50	72.50	100.00
USE Usefulness	1	5.56	0.70	4.80	6.67
	2	5.13	0.78	4.13	6.86
	3	5.65	0.84	4.25	6.75
USE Ease of Use	1	5.54	1.28	3.60	7.00
	2	5.62	1.18	4.20	7.00
	3	6.01	0.81	4.18	6.64
USE Ease of Learning	1	6.27	1.46	3.67	7.00
	2	6.20	1.02	4.50	7.00
	3	6.31	0.75	4.75	7.00
USE Satisfaction	1	5.34	0.79	4.40	6.40
	2	5.87	1.06	4.50	7.00
	3	5.70	1.05	3.57	6.71
MARS Engagement	1	4.00	0.51	3.40	4.60
	2	4.20	0.55	3.40	4.80
	3	3.99	0.44	3.40	4.80
MARS Functionality	1	4.40	0.58	3.50	5.00
	2	4.40	0.63	3.40	5.00
	3	4.64	0.53	3.50	5.00
MARS Aesthetics	1	3.86	0.51	3.33	4.33
	2	4.20	0.97	3.00	5.00
	3	4.22	0.60	3.33	5
MARS Information Quality	1	4.29	0.68	3.50	5.00
	2	4.02	0.73	3.00	5.00
	3	4.40	0.47	3.86	5.00
MARS Subjective Quality	1	4.05	0.41	3.50	4.50
	2	3.80	0.97	2.25	4.75
	3	4.17	0.68	2.27	5.00

Note. SUS = System Usability Scale, USE = Usefulness, Satisfaction, Ease of Use Questionnaire, MARS = Mobile App Rating Scale. Possible ranges of scores were as follows: SUS - 0 to 100, USE - 1 to 7, and MARS - 1 to 5. Higher score reflected greater levels of the constructs (e.g., greater ease of use, greater engagement). Cycle 1 (n = 5), Cycle 2 (n = 5), and Cycle 3 (n = 9).

3.3. Social media questionnaire

Most participants reported having social media accounts (13/19, 68 %), Instagram accounts (12/19, 74 %), and Facebook accounts (12/19, 74 %). Most participants (11/19, 58 %) thought that Facebook would be a better platform for the social media component, sharing that it better facilitates commenting, is more interactive, and is easier for groups. Participants also endorsed preferences for interactive, motivational, and relatable content for the social media group and expressed enthusiasm about having the opportunity to engage with such a group (see Table 3).

Table 3.

Social media survey results.

How much do you agree or disagree with the following statements, regarding your preferences for the type of content this social media page could provide. (1 = Strongly Disagree to 5 = Strongly Agree)	Mean (SD)
I would like posts that have themes (e.g., Motivation Monday, Well-Being Wednesday, Fun Fact Friday)	4.15 (1.11)
I would like posts that ask me questions and encourage me to engage (e.g., by leaving a comment).	4.42 (0.77)
I would be looking for positive content (e.g. inspirational quotes).	4.42 (0.77)
I would be looking for content to acknowledge how difficult recovery can be.	4.52 (0.90)
I would be enthusiastic about having the opportunity to interact with others who have also recently.	4.35 (1.11)

3.4. Thematic analysis

Feedback converged on three main themes (see Table 4); each theme is described and illustrated below. See Table 1 of Supplementary Materials for examples of feedback integrated into subsequent cycles organized by themes and subthemes.

Table 4.

Main themes and subthemes of participant feedback.

Theme 1: Logistical App Feedback	Theme 2: Increasing Success in Recovery	Theme 3: Real-World App/Social Media Use Implications
Program-Specific Input	Reasons for Recovery	Motivations to Use App in Real World
Input Applicable to Other Apps	Connection	Goals for Social Media
	App Increasing Motivation	Unsupervised App Use
	App Helping Overcome Barriers	App Accessibility

3.4.1. Logistical App Feedback

Participants commented on app logistics, such as the amount of text, specific content, colors, functionality, and privacy. In general, participants gave feedback on app logistics that was specific to Helping HAND or applicable to other apps.

3.4.1.1. Program-Specific Feedback.

Relevant to Helping HAND, participants did not report issues navigating the app, and reported enjoying the autonomy they experienced while using the app. For example, one participant shared, “I like how you can really see everything laid out in front of you and you can choose where to go. That’s nice versus being locked in and scrolling through screens” (P7). Participants also commented on the aesthetics of the app, reporting that the general layout was “Very visually appealing and clear” (P6). Furthermore, participants gave feedback on specific content, generally finding it to be comprehensive and useful.

3.4.1.2. Input Applicable to Other Apps.

Participants also gave feedback applicable to other apps. For example, participants reported enjoying interactive activities, sharing “I like anything interactive, anything where…I can push buttons and answer questions or play a game or [watch] a video,” (P4). Furthermore, participants emphasized the value of apps that offer opportunities for self-reflection, suggesting, for example, that “Something that might be good is…every time you log on to the app, it asks you…how are you doing or what do you need today,” (P19). Participants additionally expressed wanting an app that would protect their privacy, and they also endorsed the utility of an app that would allow them to send content of their choosing to their treatment team. Finally, participants were excited about the opportunity to interact with a trained coach through the app, commenting that the extra weekly support would be beneficial.

3.4.2. Increasing Success in Recovery

3.4.2.1. Reasons for Recovery.

Participant feedback also centered around Increasing Success in Recovery. Participants shared their reasons for recovery, reporting that “Something that…propelled me forward was having a sense of independence and…having responsibility for recovery,” (P4), and “I think more, because I have a daughter, I just think more about what kind of a life I want her to have, and how I want to be a part of that life,” (P2). The app offers opportunities to share reasons for recovery, and participants noted that focusing on “whys” for recovery was useful.

3.4.2.2. Connection.

Participants also endorsed the value of connection in recovery and how aspects of the app and social media component could foster connection. One participant shared “[The app] has a sense of connection to it not only through the coaching but also the personal stories…it feels like more than an app that’s going to tell me how to change my thought process,” (P7). Another participant commented on comfort from the social networking component in knowing that she wasn’t pursuing recovery alone. In this way, participants found that the app and social networking component provided opportunities for connection that could help them pursue recovery.

3.4.2.3. App Increasing Motivation.

Participants also reported that the app increased their motivation. One participant shared that “Just logging in and getting on the platform increased my motivation,” (P15). A Cycle 3 participant shared “I feel like it’s had a really big impact on me the last two weeks, and I have…the motivation to wanna log in and work on [recovery] because I really got so much from it” (P18).

3.4.2.4. App Helping Overcome Barriers.

Furthermore, participants reported that the app and social media component could help them overcome barriers to recovery, such as transitioning home after treatment. For example, one participant shared, “I’m trying to find things that I can build my time in now that I’m home…and I think that’s where… SilverCloud Helping HAND came in…I was like OK, well I can log in at this day and time when I don’t have therapy because it’s going to give me time to focus on my recovery and not just feel like I’m just existing here floundering” (P12). Participants also emphasized the benefits of the ability for the app to be used at any time to reinforce lessons learned in treatment. One participant shared, “I think that…when you’re in the thick of it…like my brain didn’t work very well during that time… and it might have been helpful to have…a place that I could touch base on…all the things that I have learned from when I did recovery before the relapse and…a place to remind me what’s important” (P8).

3.4.3. Real-World App/Social Media Use and Implications

Participants also commented on how they could see themselves using the app and social media component in their day-to-day lives. Participants in Cycle 3, who had unsupervised access to the app for two weeks, shared particularly helpful insights relevant to this theme and were able to remark on their actual use of the app.

3.4.3.1. Motivations to Use App in Real World.

Participants reported on what would motivate them to use the app in the real world. These motivations ranged from the app cultivating “A sense of belonging” (P6) to the app including “Relevant and interesting programs” (P13). Overall, participants endorsed that they could see themselves being motivated to use the app in their everyday lives.

3.4.3.2. Goals for Social Media.

Participants also commented on their motivations to engage with the social networking group. Many participants endorsed wanting to feel a sense of belonging with the social media group, sharing, “I think I’d want to get a sense that I was being supported. And that… it was OK to be me… I think just a sense of welcomingness” (P19). Participants also reported that they would want the social media group to be encouraging, including “Quotes or something that keeps you motivated…maybe a reminder of skills” (P17). Overall, participants expressed interest in belonging to and interacting with a recovery-focused social media group, particularly one that fostered connection and offered inspiration.

3.4.3.3. Unsupervised App Use.

Participants in Cycle 3 were also able to report on when and how often they actually used the app, which varied depending on their schedules. Indeed, during the 2-week period of unsupervised use of the app, some Cycle 3 participants completed the entire program, while others made it through a couple of modules. In general, participants enjoyed that they could use the app whenever they had time, whether that be when their family was out of the house or they were in line for the grocery store. One participant reported, “I could just…use it…I didn’t have to sit down and intensely do it…It could be kind of casual,” (P13), highlighting how the app could be used at times of convenience.

3.4.3.4. App Accessibility.

Finally, participants commented on the potential the app has to address systemic barriers such as lack of access to treatment. For example, one participant shared, “I think it will definitely help people. I’m just thinking about people getting out of the [intensive treatment] program and them kind of being on their own again. I think that will definitely help people to bridge that gap,” (P17).

Overall, participants reported that they could see themselves using the app and social networking component in their daily lives, and that the app had large potential for aiding with the transition back to the real world and increasing access to care (Table 5).

Table 5.

Themes, descriptions, and feedback from participants across usability testing cycles.

Themes	Description	Illustrative Quotes (P#)
Logistical App Feedback
Program-Specific Input	Feedback from participants that is specific to the Helping HAND app.	“I liked the beauty through the ages thing. And how it mentioned that beauty is not an objective reality and that it’s something that changes. And I thought that was good.” (P19)
		“I’m a big progress tracker, and I love that there’s feedback from… a real person.” (P9)
		“I like the idea of having a social media group like that connectivity piece, yeah a really lovely piece that sets the app apart that kind of brings it to life.” (P7)
Input Applicable to Other Apps	Feedback from participants that is able to be extrapolated to other apps with similar objectives.	“I think something that might be good is a daily like, well, every time you log on to the app it asks you like how are you doing or what do you need today.” (P19)
Increasing Success in Recovery
Reasons for Recovery	What motivated participants to pursue recovery for their eating disorder.	“I also think just the big part for me was also realizing that what life was like without an eating disorder. Like I just had more energy and like I was in a better mood to do things. Where like it was also, it was not just like um. For dealing with my values, there’s also just like I was physically capable of doing more things.” (P13)
Connection	How the program fostered a sense of connection-both in the app itself and in the social networking component.	“The social media aspect of it, like being able to talk to other people without judgment and without really knowing them, I think that’s really going to be more helpful too, because when you go through recovery, you know everyone and they know you, your family and friends know, if you don’t want to be judged you can have a good place to turn to. ” (P4)
		“I think it’s great to have a sense of community and normalization and can be really nice on a hard day.” (P7)
		“When I went in and I read the stories of like you know, the actual people that have experienced the same things as me, I didn’t feel so alone.” (P15)
App Increasing Motivation	Participant feedback on how the app increased their motivation for recovery.	“Just logging in and getting on the platform has increased my motivation. I think just seeing the content and resonating with the part of Addressing Weight and Food and a lot of things that kind of make me feel like, you know, ‘Oh my gosh, I was all alone in this kind of thing now that I’m in outpatient,’ but this has increased my motivation outside of treatment.” (P15)
		“…having a way where like the people could like plan for like things in their life that maybe they couldn’t do during their eating disorder…like the Relapse prevention section…made me more like I guess motivated.” (P13)
		“I think of I think this definitely was really motivating and I think a little more than some of my treatments days because it was so like really focus on the different topics and in detail with it.” (P18)
		“I feel like it’s had a really big impact on me last two weeks and I have like the motivation to wanna log in and work on [recovery]. Because I really got so much from it.” (P18)
App Helping Overcome Barriers	How participants could see this app helping to overcome barriers to recovery.	“I mean, with all of those, with all the content, I think that everyone is going to find something along the way, at least something, probably numerous things that it’s going to help.” (P4)
		“It’s hard to communicate to people with eating disorders I’ve never really had the time, I’ve never really had friends, none of my friends have eating disorders, but having that [social media] component I’ll get that.” (P9)
		“I think having the ability to actually utilize…a coach in a larger platform, like you’re saying. that sounds great, beyond great.” (P12)
		“I think that like when you’re in the thick of it, it might be helpful to have a like my brain didn’t work very well during that time and it might have been helpful to have like a place that I could touch base on like all the things that I have learned from when I did recovery before the relapse and like um a place to remind me like what’s important.” (P8)
Motivations to Use App in the Real World	Participant feedback about what would motivate them to use the program in their real lives.	“I think that’s what I’m trying to find. Things that I can build my time in now that I’m home.” (P12)
		“I just found this app really helpful because it was so interactive and since then it was really just like targeted towards people with anorexia so I could really relate and under like stand it and it was just really helpful because it was so tailored to that and so that was something I looked forward to.” (P18)
		“It really was eye opening because I was like this would be really helpful for me to like keep doing like to have this access to this app down the road.” (P12)
Unsupervised App Use	How Cycle 3 participants used the app in their unsupervised 2-week trial.	“…just setting the time where I’d be able to have space and no distractions so I could really focus on it.” (P18)
		“I found it was really helpful having it like. I actually used it frequently, more frequently when I was at work because it was something I had access to on my work computer and was able to just like complete a reflection or something if I was struggling at work.” (P16)
		“I had a busy work schedule past two weeks and then also I had friends visiting. So there’s kind of just a lot going on. Um, so I just kind of like when I had free time, I just like sat down.” (P13)
		“I also used the app like a self-care thing as well so like whenever I need to do something treatment-esque but on my own terms is something that I could use for myself.” (P11)
Goals for Social Media Use	Participant goals for using the social media group and how the group could meet those goals.	“One of my needs is just that I get lonely. So I think just being in a group in general that I trusted would be meeting my needs.” (P19)
		“Support for each other, like quotes or something that keeps you motivated. Encouragement. You know, maybe a reminder of skills.” (P17)
		“I think just general like encouragement and being able to see that other people and their progress is looking cause I know definitely sometimes whenever I’m in a mindset like feel impossible and being able to see either people like yes I’m struggling and this is what I’ve completed throughout the week or like this was a struggle a month ago and now this is possible, stuff like that I feel like would be really helpful.” (P11)
App Accessibility	How participants see this app fitting in the treatment flow in terms of its accessibility.	“It’s all combined…I think it covers every single piece of the puzzle. In a way that’s like accessible at the palm of our hands, that’s what we do anyway constantly.” (P9)
		“This might be something that would just help you in your recovery even more…like a partly step down from, you know, residential or PHP treatment.” (P10)
		“Actually having something like log in with and like have to help support throughout the weeks and stuff when you’re going through it. I just think it can be really helpful.” (P18)

4. Discussion

This paper describes the development and usability testing of a CBT-based mobile app, Helping HAND, in addition to an adjunctive social networking component, for the post-acute treatment of AN. Nineteen adult women with AN recently discharged from intensive treatment provided feedback on the app and social networking component, answering calls to employ user-centered design in the development of digital interventions for EDs (Graham et al., 2023). Feedback guided iterative modifications over three testing cycles and converged on three broad themes: Logistical App Feedback, Increasing Success in Recovery, and Real-World App/Social Media Use and Implications. Ultimately, the app and accompanying social networking feature demonstrated high feasibility and acceptability, demonstrating potential for post-acute support of women with AN.

Even prior to refinements, the Helping HAND mobile app showed excellent usability and acceptability among participants, with highest global usability and acceptability scores reported in the third and final cycle of usability testing. These findings mirror high usability and acceptability reported for mobile apps designed to support recovery from AN (Hamatani et al., 2022) and binge-type EDs (Fitzsimmons-Craft et al., 2020; Fitzsimmons-Craft et al., 2019; Saekow et al., 2015) alike, and suggest that individuals with EDs are open to using apps to support their recovery.

Participants also gave feedback on aspects of the app they liked and disliked, as well as suggestions, and a particularly notable theme of the feedback was how to increase success in recovery. Specifically, participants shared that focusing on their “whys” for recovery was helpful and suggested adding a specific “Recovery Why” tool to reground themselves in their reasons for pursuing recovery. This mirrors past literature that has found that focusing on life beyond the ED facilitates recovery (Cockell et al., 2004).

Participants also reported the value of connection in recovery. Indeed, social support has been found to promote recovery from AN (Darcy et al., 2010; Dawson et al., 2014; Federici & Kaplan, 2007; Tozzi et al., 2003; Wetzler et al., 2020) and a sense of connection to self and others can facilitate ED recovery (Linville et al., 2012). Notably, participants felt that a social networking component could facilitate social connection and consequently aid in recovery.

Regarding social media, most participants indicated that they had Facebook accounts and thought that Facebook would be a better fit for the intended purposes of the group, namely due to its ability to host private (i.e., open to only those who are admitted in), secret (i.e., not visible to anyone outside of the group) groups that foster connection with others. Furthermore, participants were enthusiastic about posts that were inspirational, educational, and interactive, in line with previous research that has found these types of posts to be most helpful to AN recovery (Eichhorn, 2008).

As one strength of this study, this is the first coached mobile app developed specific to the post-acute intervention of AN. Rigorous qualitative analyses, in line with Braun and Clarke (2006), were utilized. Further, developing the app from a user-centered design perspective and engaging individuals most affected by the proposed intervention will facilitate scale-up, and the iterative nature of the study optimized the specificity of feedback provided.

As a limitation, even with intentional efforts to enroll a diverse sample, there was a lack of diversity in the final sample in some respects, including race, ethnicity, geographic region, socioeconomic status, and insurance status. There were also sample restrictions that further limited sample diversity; eligibility criteria required that participants were cisgender females, owned a mobile phone, spoke English, and were U.S. residents. However, it is important to note that our sample is still representative of those treated in higher levels of care (Peckmezian & Paxton, 2020). There remains a need to ensure the acceptability of this app in individuals with different racial, ethnic, socioeconomic, and gender identities, and also expand access to higher levels of care more generally. Furthermore, the current study’s operationalization of acute care, encompassing inpatient, residential, partial hospitalization, and intensive outpatient treatment, is fairly wide-ranging. It is possible that the needs of individuals may differ by most recent level of care. Iterating the app based on feedback from individuals discharging from all these levels of care increases the likelihood that the app will meet diverse needs, but future work should test whether the effects of the app and social media group differ by most recent level of care. Another limitation is that participants were not able to experience the app with the coach support component. As such, no feedback was elicited on the accompanying coach feature beyond whether participants thought they would benefit from such a feature. Finally, not all feedback was feasible within the SilverCloud platform, precluding certain suggestions (e.g., including the social networking feature in the app) from being adopted.

Overall, this study provides preliminary evidence of the feasibility and acceptability of a mobile app and accompanying social networking feature for post-acute support for women with AN, which will subsequently be tested in a pilot randomized controlled trial. If successful, this approach has high potential for offering scalable support for individuals with AN in the high-risk period following discharge from acute care.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.eatbeh.2024.101865.

Data availability

Data will be made available on request.