State-Wide University Implementation of an Online Platform for Eating Disorders Screening and Intervention

Abstract

The Internet-based Healthy Body Image (HBI) Program platform uses online screening to identify individuals at low risk for, high risk for, or with an eating disorder (ED) and then directs users to tailored, evidence-based online/mobile interventions or referral to in-person care to address individuals’ risk/clinical status. We examined findings from the first state-wide deployment of HBI over the course of three years in Missouri public universities, sponsored by the Missouri Eating Disorders Council and the Missouri Mental Health Foundation. First, the screen was completed 2,454 times, with an average of 2.5% of the undergraduate student body on each campus taking the screen. Second, ED risk level in the participating students was high—over 56% of students screened were identified as being at high risk for ED onset or having a clinical/subclinical ED. Third, uptake for the HBI online/mobile interventions ranged from 44–51%, with higher rates of uptake in the high-risk compared to low-risk group. Fourth, results showed that, for students with a clinical/subclinical ED, use of the clinical mobile application Student Bodies-Eating Disorders intervention resulted in significantly decreased restrictive eating and binge eating. Neither vomiting nor diet pill/laxative use was found to decrease, but reports of these behaviors were very low. This is the first deployment of a comprehensive online platform for screening and delivering tailored interventions to a population of individuals with varying ED risk and symptom profiles in an organized care setting. Implications for future research and sustaining and broadening the reach of HBI are discussed.

Eating disorders (EDs) are serious mental illnesses associated with high morbidity and mortality, clinical impairment, and comorbid psychopathology (Klump, Bulik, Kaye, Treasure, & Tyson, 2009). EDs are particularly common in college-age women, yet less than 20% of students who screen positive for an ED report receiving treatment (Eisenberg, Nicklett, Roeder, & Kirz, 2011). Delays in treatment result in prolonged illness, disease progression, poorer prognosis, and greater likelihood of relapse. In addition, many ED cases go undetected. A majority of individuals who receive treatment for an ED are first seen by their primary care physician; however, one study indicated that 92% of frontline medical providers believed they had missed an ED diagnosis (Linville, Benton, O’Neil, & Sturm, 2010). Additionally, screenings for EDs on college campuses are lacking, with one survey indicating that only 22% of colleges offer year-round ED screening opportunities and less than half (45%) offer ED screenings once per year or semester (National Eating Disorders Association, 2013). Even when mental health screening is offered on college campuses, available data suggest that reach is limited (e.g., completed by 8% of students invited by email in one study; Haas et al., 2008). Thus, opportunities for improved screening, prevention, and treatment of EDs on college campuses are greatly needed.

Cognitive-behavioral therapy (CBT) is the most well-established treatment option for adults with EDs (Hay, 2013; Linardon, Wade, de la Piedad Garcia, & Brennan, 2017) and involves addressing the processes that are maintaining the ED, including concerns about shape and weight, extreme dietary restraint, and the individual’s ability to handle day-to-day events and moods. CBT for EDs was first developed as a traditional face-to-face treatment but has evolved over time to include guided self-help and Internet-based versions as ways to broaden access to treatment (Agras, Fitzsimmons-Craft, & Wilfley, 2017).

Telehealth and mobile and Internet-based technologies have the potential to improve care for EDs on college campuses by overcoming barriers to treatment and increasing efficiency. Such interventions can offset in-person clinical demands, increase reach of screening and access to care, and reduce costs (e.g., Kazdin, Fitzsimmons-Craft, & Wilfley, 2017). Mobile technologies have been successfully used for screening, prevention, and treatment of EDs, including among college students (e.g., Beintner, Jacobi, & Taylor, 2012; Jones, Kass, Trockel, Glass, Wilfley, & Taylor, 2014), with high acceptability given their convenient and anonymous format. Moreover, estimates suggest that digital ED intervention costs less and results in fewer people who need to utilize in-person treatment than standard care (Kass et al., 2017a). A final benefit of offering mobile interventions is that one can ensure that the treatment that is being provided is evidence-based. Currently, when individuals receive care, it is typically not an evidence-based treatment (Kazdin et al., 2017). A key challenge is delivering these promising technologies effectively to populations, including linking screening directly with prevention and treatment.

To meet this need, we developed an online platform for screening and delivering tailored interventions to a population of individuals with varying ED risk and symptom profiles via system-level implementation on college campuses. Specifically, the digital Healthy Body Image (HBI) Program—available by computer or smartphone—uses an evidence-based online screening tool to identify individuals at low risk for, high risk for, or with a clinical/subclinical ED, and offers tailored, evidence-based online/mobile interventions or referral to in-person care to address students’ risk/clinical status (see Figure 1). HBI was developed based on a systematic program of research evaluating tailored online preventive interventions for reducing ED risk and onset (Beintner et al., 2012; Wilfley, Agras, & Taylor, 2013). All of the online interventions offered are based on CBT principles and have been established in prior research (see Table 1). In brief, students at low risk for an ED are offered the StayingFit online, universal health education intervention, which has been shown to result in increased fruit and vegetable consumption, lower weight/shape concerns, weight stabilization in normal weight students, and weight loss in overweight students (e.g., Taylor et al., 2012). Students at high risk for an ED are offered the Student Bodies-Classic online, targeted ED prevention intervention, which has been shown to reduce ED risk and symptom progression (e.g., Beintner et al., 2012). Students with a clinical/subclinical ED other than anorexia nervosa (AN) are offered the Student Bodies-Eating Disorders (SB-ED) mobile, guided self-help intervention (see Figure 2), which contains units covering the core components of CBT for EDs and has been shown to reduce ED symptoms and symptom progression (e.g., Saekow et al., 2015). Students with possible AN are referred to in-person services on their campus and are not offered an online intervention, given the medical monitoring warranted for this disorder. We worked with campus stakeholders, including administrators and counselors in the campus counseling centers, to garner support and resources for pilot testing HBI on two universities, the results of which demonstrated the feasibility of system-wide implementation of HBI (Jones et al., 2014).

Figure 1.

Suite of tailored, evidence–based interventions, available on the Healthy Body Image Program online platform. Figure adapted from Wilfley et al. (2013).

Table 1.

Online Screening and Evidence-Based Interventions for the Identification, Prevention, and Treatment of Eating Disorders included in the Healthy Body Image Program

Component	Features	Main Findings
Stanford- Washington University Eating Disorder Screen	Online, categorizes individuals into risk (low, high) and symptom (screen positive for anorexia nervosa, screen positive for other eating disorder [ED]) categories Assesses factors that confer increased risk (e.g., weight/shape concerns, dieting, impairment, and body mass index)	Sensitivity and specificity are high for cases of DSM-5 EDs (≥.80 for anorexia nervosa, bulimia nervosa, or binge eating disorder) (Graham et al., 2018; Wilfley et al., 2013)
StayingFit	Online, for individuals at low risk for an ED Cognitive-behavioral-based intervention with an online, asynchronous, guided discussion group Encourages healthy weight regulation via self-monitoring, goal setting, stimulus control, and appetite awareness	Results in increased fruit and vegetable consumption, lower weight/shape concerns, weight stabilization in normal weight students, and weight loss in overweight students (Jones et al., 2008; Taylor et al., 2012)
Student Bodies- Classic	Online, for individuals at high risk for an ED Cognitive-behavioral-based intervention with an online, asynchronous, guided discussion group Focuses on reducing weight/shape concerns, enhancing body esteem, and reducing disordered eating behaviors (e.g., binge eating)	Reduces ED risk and symptom progression (e.g., Beintner et al., 2012; Kass et al., 2014; Taylor et al., 2016)
Student Bodies- Eating Disorders	Mobile, cognitive-behavioral therapy, guided self-help intervention for students with a clinical or subclinical ED other than anorexia nervosa Promotes healthy eating patterns, positive self-talk, appropriate coping strategies for negative emotions, enhanced social support, and relapse prevention strategies	Reduces ED symptoms and symptom progression (Jacobi et al., 2012; Saekow et al., 2015)

Figure 2.

Sample screenshots of the Student Bodies-Eating Disorders mobile intervention highlighting its key features.

This paper describes findings from a state-wide deployment of HBI over the course of three years in Missouri public universities. Specifically, the goal of this paper is to report on: 1) reach of HBI in the Missouri public universities over the course of three years (i.e., number of online screens completed and proportion of undergraduate student body reached); 2) results of the screen (i.e., percent low risk for an ED, high risk for an ED, positive screen for AN, positive screen for other clinical/subclinical ED) and differences in risk/clinical status groups on demographic variables; 3) uptake of the online/mobile interventions; and 4) results regarding the effectiveness of the mobile SB-ED intervention for students who screen positive for a clinical/subclinical ED other than AN. We focus on the results of SB-ED given that outcome data were not available for the other groups and because we wished to focus on reporting on outcomes for the newest intervention in our suite of programs now delivered as a mobile application, SB-ED. To our knowledge, this is the first deployment of a comprehensive online platform for screening and delivering tailored interventions to a population of individuals with varying ED risk and symptom profiles, including deployment of a clinical mobile application (“app”) for students screening positive for an ED, in an organized care setting—Missouri public universities.

Method

State-wide deployment of HBI was sponsored by the Missouri Eating Disorders Council and the Missouri Mental Health Foundation. The Missouri Eating Disorders Council is a state-mandated council within the Missouri Department of Mental Health and was passed into law in 2010 to improve access to treatment, raise awareness, and provide education related to EDs. After members of the Missouri Eating Disorders Council learned about the promising pilot results of HBI (Jones et al., 2014), discussions began in April 2013 to implement this program more broadly, given the potential for this program to help bridge the large treatment gap for EDs (Kazdin et al., 2017). Over the next month, we developed and presented a proposed scope of work to the Missouri Eating Disorders Council, who subsequently voted to deploy this program in the Missouri public universities as a state-wide initiative. We also note that following the launch of the program, HBI was discussed at all meetings of the Missouri Eating Disorders Council, which occurred every other month. At these meetings, we were able to provide updates on the progress of the initiative as well as answer any questions Council members had, facilitating ongoing collaboration between the Missouri Eating Disorders Council and our team.

Participants

University enrollment

We approached all of the Missouri public universities (N = 13) to enroll in this initiative, which was described as a comprehensive, evidence-based online platform for EDs screening, prevention, and treatment that would be made freely available on their campus. With support from members of the Missouri Eating Disorders Council, HBI was presented to the Council on Public Higher Education in Missouri, comprised of the presidents and chancellors of the 13 public four-year universities in Missouri, who voted and provided approval of this initiative. Webinars on HBI were then presented to the directors of Student Health and Counseling Centers, again in collaboration with members of the Missouri Eating Disorders Council. Four universities participated in Year 1 of the initiative (2013–2014), 6 universities (4 continuing and 2 new) participated in Year 2 (2014–2015), and 8 universities (6 continuing and 2 new) participated in Year 3 (2015–2016). All participating universities were required to identify an on-campus clinical representative for HBI, typically a counselor in the student counseling or health center, to maximize student uptake and safety. Schools choosing not to participate did so for a variety of reasons (e.g., lack of identified on-campus clinical representative, student counseling or health centers focused on other ongoing initiatives and reported being too busy at the time to participate).¹ This project received institutional review board approval at the coordinating university.

Student enrollment

All students on campus ages 18 years or older were eligible for participation in HBI. Students signed up by accessing the HBI website, providing their informed consent, and completing the online screen. Recruitment efforts primarily targeted undergraduate students, but graduate students and postdoctoral fellows were also eligible for the program. The only exclusion criteria were being <18 years or not being a current student at one of the participating universities.

In preparing to enroll students, we met with stakeholders to learn about standard approaches for engaging students in campus-wide initiatives, and we collaboratively developed campus-specific recruitment strategies designed to recruit a diverse sample (see Table 2 for an overview of recruitment strategies utilized). In addition to working with an on-campus clinical representative, partnerships were established with departments, faculty, and student groups to build a cadre of invested individuals who could help promote HBI to students.

Table 2.

Recruitment Strategies Utilized for Student Uptake into the Healthy Body Image Program

Domain	Strategy
Electronic Media	Universal Email Blasts to university-wide subpopulations (e.g., to all first- year students; to all undergraduate students) or to the entire student body
	Emails to Faculty Members to send program information to other faculty members and their students
	Emails to Student Group Leaders to send program information to group members and peers on campus
	Program Links on Websites (e.g., Student Health Services webpage, Departmental webpages)
Social Media	Postings on Facebook about the program and relevant body image issues in the media
	Tweets on Twitter about the program and relevant body image issues in the media
	Features in E-Newsletters and University Health Magazines
Printed Materials	Standard one-page flyers
	One-page flyers with tear-off tabs
	Quarter-sized flyers
	Features in Printed University Publications (e.g., student newspapers)
	Paraphernalia (e.g., stress balls, pens, water bottles) with the program logo and website
Presentations	In-Classroom Presentations (e.g., relevant Psychology or Women and Gender Studies lectures)
	Presentations to Student Groups (e.g., health-oriented organizations, sororities)
	Presentations to Residential Advisors to share information about the program with their residents
	Presentations at a Table in Student Centers or at Events (e.g., at campus orientation, activities fairs), during which students had the opportunity to complete the screen on-site
Other	Offered as a mental health resource by peer health educators/counselors to individuals in need

Materials and Procedure

HBI platform

As detailed in Figure 1 and Table 1, the HBI platform, developed by researchers at Washington University in St. Louis and Stanford University and hosted by a technology partner, Lantern, uses online screening to identify individuals at low risk for, high risk for, or with an ED, and offers tailored, evidence-based online/mobile or in-person interventions to address students’ risk/clinical status.

Screening

The online Stanford-Washington University Eating Disorder Screen (SWED) is a brief screening tool that assesses ED behaviors, pathology, and impairment to detect individuals at high risk for or with an ED. The SWED includes questions from the Weight Concerns Scale (WCS; Killen et al., 1994) and items adapted from the Eating Disorder Examination-Questionnaire (EDE-Q; Fairburn & Beglin, 1994) and the Eating Disorder Diagnostic Scale (EDDS; Stice, Telch, & Rizvi, 2000). Responses are used to categorize individuals into one of four categories: 1) possible AN, based on body mass index (BMI) and elevated weight and shape concerns; 2) possible clinical or subclinical ED other than AN, based on binge eating and/or purging behaviors in the past 3 months; 3) high risk for an ED, based on elevated weight and shape concerns; and 4) low risk for an ED, based on not screening into one of the above categories. The SWED screening algorithm has been validated and used in past research (e.g., Graham et al., 2018; Taylor et al., 2016). Participants also self-reported their age, gender, race, and ethnicity.

Interventions

As detailed in Figure 1 and Table 1, students at low risk for an ED were offered StayingFit. Students at high risk for an ED were offered Student Bodies-Classic. Students who screened positive for a clinical or subclinical ED (except for AN) were offered SB-ED. Individuals with possible AN were referred to in-person services on their campus and were not offered an online intervention, given that more intensive medical monitoring is warranted for this disorder.

SB-ED

Specific details on the SB-ED intervention are described below, as the format of the intervention changed during the course of the trial. These changes were designed to facilitate quality improvements for the user experience while the core intervention principles remained the same. The version of SB-ED offered in Year 1 of the initiative was a web-based intervention only. This version of the intervention was in a more traditional, online intervention format, utilizing longer, weekly sessions. In Years 2 and 3 of the initiative, the intervention was offered via iPhone mobile app (with computer and mobile web browser access available as well). Additionally, only 12 accounts for the SB-ED intervention were made in Year 1, which was not enough data to provide a meaningful analysis of effects for this version of the intervention. Mobile delivery via an app is particularly desirable over traditional web-based activities, given that as of 2015, 90% of time on smartphones is spent using apps vs. 10% in a web browser (Khalaf, 2015). Given that the long-term goal of HBI is to be delivered as a mobile platform, we focus the current paper on results of the intervention from Years 2 and 3 so that the findings can be informative for future delivery efforts.

As mentioned above, the mobile SB-ED intervention was designed to be appealing to users. The intervention is offered in 5–10 minute sessions, with 40 core sessions total. These 40 core sessions were supplemented by an additional 23 sessions addressing comorbidities and relapse prevention. This brief session delivery option was chosen given research suggesting that users want to use programs and apps in very short bursts of time (e.g., Nitsch et al., 2016). Users were provided access to the intervention for 8 months, and each user was assigned a personal coach to guide them through the intervention (see next section for more information). The intervention contains units covering the core components of CBT for EDs including: reducing disordered eating behaviors (e.g., via self-monitoring, psychoeducation, monitoring weight, monitoring triggers, conducting behavioral chain analyses, and utilizing self-care and social support); improving body image (e.g., decreasing importance of body image on self-esteem); changing behavior (e.g., re-introducing forbidden foods); regulating emotions; addressing shape checking and avoidance; challenging negative automatic thoughts; and preventing relapse. The intervention was delivered on a platform hosted by the technology company, Lantern. See Figure 2 for sample screenshots of the SB-ED intervention, highlighting its key features, including the ability for coach and user to message within the mobile app environment.

SB-ED Coaching

Coaches for the intervention were graduate students in clinical psychology or postdoctoral fellows, working at the university deploying HBI. Coaches used a clinical management “dashboard” to efficiently monitor multiple users at one time. The dashboard provides information on users’ goals, progress, and intervention use, as well as the ability to message users. Coaches underwent extensive training prior to working with users, including training in CBT for EDs, motivational interviewing, key tenets of effective online coaching, and use of the coach management dashboard. Coaches were instructed that they were responsible for: providing timely messages to users; supporting users in improving their eating habits and changing their relationships with their bodies; and providing ongoing feedback on participant progress in the intervention and symptom changes. Coaches received ongoing in-person supervision from a licensed clinical psychologist.

Participant safety is of utmost import, and several safety precautions and procedures were implemented. First, the intervention was monitored daily by coaches to assess for safety concerns (e.g., endorsement of suicidality) or significant changes in symptoms (e.g., increased purging). Participants endorsing suicidality were provided information on the National Suicide Prevention Lifeline. Furthermore, any participant deemed unsafe or needing more intensive intervention at any time was referred to on-campus resources at the student counseling and/or health centers by their coach. This protocol was designed to ensure that a higher-level of care was recommended when it may be warranted. Students were also allowed to continue use of the SB-ED intervention so as not to terminate access to a resource that may also have benefit given the ability to access it at all times, which is in contrast to in-person care that is only available at designated times. Second, as mentioned, SB-ED was not offered to students who screened with possible AN, for whom more intensive medical monitoring is warranted. These students were provided an in-person referral only. Finally, at the beginning of the online screen, all participants were provided with information on how to make an appointment at their respective counseling/health center and were also instructed to contact 911 in case of an emergency.

SB-ED Ongoing Symptom Monitoring

At the beginning of each SB-ED session (for the version of the intervention offered in Years 2 and 3 of the initiative), participants reported on ED symptom levels over the past 24 hours. Specifically, participants were asked the following 4 questions: 1) How many of your main meals did you restrict in the past 24 hours?; 2) How many times have you binged in the past 24 hours?; 3) How many times have you vomited in the past 24 hours?; and 4) How many times have you taken diet pills or laxatives in the past 24 hours?

Analytic Strategy

First, reach of HBI in the Missouri public universities over the course of three years was examined descriptively (i.e., number of screens completed). Given that recruitment efforts primarily targeted undergraduate students, we used data obtained from the participating universities’ Office of Institutional Research (or similar office) to examine reach of HBI to undergraduate students relative to the size of the undergraduate student body in the participating year to generate an estimate of the proportion of the undergraduate student body reached by the program. For schools that participated in the initiative for more than one year, undergraduate enrollment numbers were adjusted to account for one (for schools that participated two years) or two additional incoming freshman classes (for schools that participated three years) (using freshman enrollment data for those years from the participating universities). Second, results of the screen regarding ED risk/diagnostic breakdown over the course of three years were examined descriptively. Differences in risk/clinical status groups on demographics variables (i.e., age, gender, race, ethnicity, BMI) were examined using analyses of variance (ANOVAs) for continuous measures and chi-square tests for categorical variables. Third, we examined uptake of the online/mobile interventions over the course of three years. Enrollment in the different interventions was operationalized as a binary measure of whether students signed up for their assigned intervention. For the AN referral group, uptake of suggested resources (the closest equivalent of enrollment for this group) is unknown; thus, these students were excluded from analyses regarding enrollment status. We used chi-square tests to compare enrollment across groups. Finally, we examined results on the effectiveness of the SB-ED mobile app used in Years 2 and 3 of the initiative. To compare multiple users with different durations of intervention use, we first normalized users’ time in the intervention, such that their first time stamp of an ED symptom report was coded 0 and their final time stamp of an ED symptom report was coded 1; as such, normalized time for each intervention touch point for each user ranged from 0–1. This method of normalizing time allows us to examine user changes in ED symptoms over the course of their own personal time spent in the program (e.g., whether that was 1 week or 8 months). Next, we analyzed changes in restrictive eating, binge eating, vomiting, and diet pill/laxative use over the intervention. We ran four separate ordered regression models with normalized time in the intervention as the independent variable (ranging from 0 to 1) and 24-hour ED symptom reports (i.e., ordered response category of ED symptom count) as each of the dependent variables to demonstrate whether time significantly predicted ED symptom counts and thus whether ED symptoms significantly changed over the course of the intervention. Users with at least 3 ED symptom reports for a particular outcome were included in analysis so that a meaningful symptom trajectory could be analyzed (Laurenceau, Hayes, & Feldman, 2007). We also calculated the correlation between each user’s number of sessions completed and their slope of change for each outcome variable.

Results

Screen Reach and Descriptive Statistics

Over the course of 3 years, the screen was completed 2,454 times: 323 times in Year 1 of the initiative; 622 times in Year 2; and 1,509 times in Year 3. Participants ranged in age from 18 to 66 years, with a mean age of 22.89 years (SD = 6.59). Participants were primarily female (82.4%), with fewer identifying as male (16.1%), transgender (0.4%), gender non-conforming (0.8%), or preferring not to answer/feeling as though none of the above applied (0.3%). Most participants identified as White (78.1%), 7.8% as Black or African American, 5.9% as Asian or South Asian, 0.2% as Native Hawaiian or Pacific Islander, 0.9% as American Indian or Alaskan Native, 4.4% as multiracial, and 2.7% as other races. Regarding ethnicity, 4.0% identified as Hispanic. Mean current BMI, based on self-reported height and weight, was 26.85 kg/m² (SD = 7.18).

In terms of student status, 82.2% were undergraduate students, 15.3% were graduate students, 0.1% were postdoctoral fellows, and 2.3% other. These numbers are consistent with recruitment efforts, which primarily targeted undergraduate students. Over the course of 3 years, an average of 2.5% of the undergraduate student body on each campus took the screen, with the greatest reach being 6.6% at a given college campus and the lowest being 0.5%.

Screen Results

Eating disorder risk level of the participating students was as follows: 43.5% (n = 1,068) screened as low risk for an ED and were offered the StayingFit intervention; 38.9% (n = 955) screened as high risk for an ED and were offered the Student Bodies-Classic intervention; 15.4% (n = 379) screened positive for a clinical or subclinical ED (with the exception of AN) and were offered the SB-ED intervention; and 3.7% (n = 52) screened positive for possible AN and were offered a referral to in-person care.

As depicted in Table 3, ED risk/clinical status groups did not differ in terms of age, race, or ethnicity (ps > .052) but differed in terms of gender and BMI (ps < .001). Follow-up tests revealed that relative to the low-risk group, greater proportions of individuals in the other groups were female. The low-risk group had a significantly lower mean BMI than the high-risk or ED groups, and the high-risk group had a significantly lower mean BMI than the ED group. As anticipated, the possible AN group had a significantly lower mean BMI than the other groups.

Table 3.

Comparison of Demographic Variables Across Eating Disorder Risk/Clinical Status Groups

	Low risk for an eating disorder (LR; n = 1068)	High risk for an eating disorder (HR; n = 955)	Clinical/subclinical eating disorder (with the exception of anorexia nervosa) (ED; n = 379)	Possible anorexia nervosa (pAN; n = 52)	Significance	Pairwise Comparisons
Age	22.64 (6.36)	23.04 (6.59)	23.44 (7.42)	21.29 (3.56)	F(3,2446) = 2.56; p=.053 partial η2 = .003	--
Gender (% Female)	76.3%	86.9%	86.8%	94.2%	Χ2(3,2454) = 50.88; p<.001 Cramer’s V = .08	LR<HR, ED, pAN
Race (% White)	76.9%	78.3%	77.6%	86.5%	Χ2(3,2454) = 2.99; p=.394 Cramer’s V = .02	--
Ethnicity (% non- Hispanic)	96.7%	95.8%	93.9%	98.1%	Χ2(3,2453) = 6.27; p=.099 Cramer’s V = .03	--
Body mass index	25.14 (6.07)	27.99 (7.26)	29.63 (8.39)	20.03 (2.49)	F(3,2421) = 9504.56; p<.001 partial η2 = .08	LR<HR, ED HR<ED pAN<LR, HR, ED

Note. LR = low risk for an ED; HR = high risk for an ED; ED = clinical/subclinical ED (with the exception of AN); pAN = possible anorexia nervosa. Pairwise comparisons listed were significant at least at p < .05.

Online/Mobile Intervention Uptake

Enrollment rates for the HBI interventions were as follows: 43.9% of those assigned to StayingFit enrolled; 50.5% of those assigned to Student Bodies-Classic enrolled; and 47.8% of those assigned to SB-ED enrolled. A chi-square analysis revealed that proportion of students who enrolled in their assigned intervention significantly differed across the three interventions (χ²(2, N = 2402) = 8.77, Cramer’s V = .04, p = .012). Post hoc tests revealed that a significantly greater proportion of those assigned to Student Bodies-Classic enrolled compared to StayingFit (p = .003). There were no statistically significant differences in proportions of those assigned to SB-ED vs. StayingFit (p = .196) or SB-ED vs. Student Bodies-Classic (p = .371) who enrolled.

Results for the SB-ED Mobile App

Finally, we present results on the effectiveness of the SB-ED intervention used in Years 2 and 3 of the initiative. For this version of the intervention, 169 SB-ED accounts were created, with 145 of those individuals (85.8%) engaging with the intervention in some way following enrollment (e.g., completing session content, messaging coach). Users completed an average of 7.48 (SD = 13.59, range = 0–63, median = 2.00) sessions, with 60.4% of users completing ≥ 2 sessions—a statistic of interest given that it demonstrates the percent of users who return to the program to complete more than one session and early engagement. The 145 users who engaged with the intervention generated 55,308 touch points of data with the intervention and sent a total of 826 messages to their coaches. The average length of the user’s time spent in the intervention (i.e., from first to last active day of use) was 47.00 (SD = 74.94, median = 7.58) days.

The number of users with at least 3 ED symptom reports for each outcome variable was as follows: 78 for restrictive eating; 77 for binge eating; 77 for vomiting; and 77 for diet pill/laxative use. Results indicated that normalized time was a significant predictor of decreased restrictive eating (B = −1.14, t = −6.00, p < .001) and decreased binge eating (B = −.74, t = −4.32, p < .001), suggesting that more time in the intervention was associated with greater improvement in these symptoms. The percent of users achieving abstinence (as indicated by reporting no engagement in the ED symptom on the last 2 reports of the symptom) was 53.8% for restrictive eating and 33.8% for binge eating. Based on average symptom change from the first to last report, Cohen’s d was .27 for restrictive eating and .39 for binge eating, representing small to medium effects. Normalized time was not a significant predictor of vomiting (B = .70, t = 1.29, p = .098) or diet pill/laxative use (B = −.22, t = −.64, p = .262). Of note, reports of any vomiting and diet pill/laxative use were very low (i.e., 18 and 22 users, respectively), and 92.2% and 88.3% of users reported no engagement in vomiting and diet pill/laxative use, respectively, on their last 2 reports. Of note, change in symptoms for all four outcome variables was found to be independent of sessions completed (ps > .351).

Discussion

This paper described findings from the first state-wide deployment of an online platform for screening and delivering tailored interventions to a population of individuals with varying ED risk and symptom profiles. Results highlight several key findings, which, together, may suggest that HBI implementation was successful across university campuses. The study is also important given that it provides data on the effectiveness of a program implemented in real-world conditions.

First, we screened nearly 2,500 students for an ED across the state of Missouri in three years, including students of varying ages. Indeed, more than 17% of the sample was over the age of 25. It is possible that the online interventions offered through HBI, which can be accessed anywhere, anytime, may have been particularly appealing to non-traditional students due to competing demands on their time (e.g., caring for children or grandchildren, working a full- or part-time job in addition to taking classes). However, even with using a multi-pronged recruitment approach, only an average of 2.5% of the undergraduate student body on each campus took the screen, with the greatest reach being 6.6% at a given college campus—rates lower than that of college mental health screening that utilized only direct-to-student email recruitment sponsored by the participating universities (Haas et al., 2008). As demonstrated by the increase in screens completed each year of the initiative, deployment of HBI became more successful over time, as the program became further engrained on the campuses. Although a wide variety of recruitment strategies were implemented, it may be useful to identify more targeted strategies that increase uptake, such as the approach used by Haas et al. (2008). In the future, higher-level support, for instance, from university presidents or deans of students, for mandated yearly ED screening or for ED screening for all incoming students at a given university may further increase reach of HBI. Notably, there is no empirical support for the concern that surveys of ED risk factors and behaviors increase risk for EDs (e.g., Celio, Bryson, Killen, & Taylor, 2003) suggesting there are limited downsides to mandated ED screening, with the exception of campuses needing to be prepared to treat cases that are identified and potential costs associated with screening and treatment. Other ways to increase program reach may include garnering university support for multiple entry points into the program (e.g., offered as a resource to all students presenting for treatment at the counseling center, coordinated use of in-person and social media outreach).

Second, we showed that ED risk level in the participating students was high—over 56% of students screened were identified as being at high risk for ED onset or having a clinical/subclinical ED. One population-level study of ED symptoms at 12 U.S. colleges and universities found that prevalence of possible EDs ranged from 12–40%, depending on the definition (e.g., based on ED psychopathology, past month binge eating, or past month compensatory behaviors) (Lipson & Sonneville, 2017). The higher rates of ED pathology seen in our sample demonstrate that students with elevated ED risk are the ones choosing to complete the screen, which was not mandated by the participating universities. The screen thus served as an important method to detect individuals at risk for or with EDs and provide access to care. Findings also highlight demographic differences across ED risk/clinical status groups. Individuals at low risk for an ED had a significantly lower mean BMI than individuals in the high-risk and ED groups (whose mean BMIs were within the overweight range). This aligns with other research suggesting that elevated weight status may increase risk for EDs (Duncan et al., 2017; Kass et al., 2017b). Race and ethnicity did not differ across status groups, providing further research support for the notion that EDs do not discriminate and affect individuals of all racial/ethnic backgrounds (Schaumberg et al., 2017). Known disparities in ED identification and treatment for racial/ethnic minority individuals (e.g., Marques et al., 2011) further highlight the need for accessible screening and intervention, such as that offered through HBI.

Third, uptake for the HBI suite of online/mobile interventions ranged from 44–51%. There were higher rates of enrollment in the high- vs. low-risk group, perhaps due to greater recognition of a need for help, but otherwise, rates of uptake did not differ across groups. For students with EDs, once they enrolled in the SB-ED intervention, the majority completed more than one session (60.4%). They also completed more sessions of the SB-ED intervention (average of 7.48 online sessions completed) than the number of in-person sessions typically completed at college counseling centers (where the modal number of appointments is 1 and the average is 4.7; Center for Collegiate Mental Health, 2016). Thus, our data demonstrate that providing a mobile intervention for college students who screen positive for an ED has great potential in terms of bridging the wide treatment gap for these serious mental illnesses (Kazdin et al., 2017).

Finally, the SB-ED intervention results were encouraging. Indeed, results demonstrated that restrictive eating and binge eating significantly decreased over the course of users’ time in the intervention. While vomiting and diet pill/laxative use were not found to significantly decrease over the course of users’ time in the intervention, this is likely due to low power for these analyses, with only 18 and 22 users reporting any vomiting or diet pill/laxative use, respectively. Future work may also wish to further explore the best ways to assess and target compensatory behaviors in mobile ED interventions, such as with sensing technologies or ecological momentary assessment. While a pilot research study had previously demonstrated the efficacy of SB-ED for individuals with subclinical EDs (Saekow et al., 2015), the current results are notable as they may demonstrate the beneficial effects of SB-ED when deployed at wider scale on college campuses and under less tight controls than are utilized in traditional randomized controlled trials. However, the conclusions that can be drawn from these data are limited by the fact that one cannot know how these individuals would have done if they did not engage in the intervention, as a control group was not utilized in the current project. Furthermore, it is notable that users experienced significant decreases in restrictive eating and binge eating with intervention usage that only lasted a median of 7.58 days (average = 47.00 days). These results thus suggest that the current mobile ED intervention resulted in clinical benefit over a relatively short time period, which may have implications for the development of future mobile apps. Notably, change in symptoms was unrelated to session completion, a finding that may reflect that evaluating intervention usage overall (not just session completion) may provide a better marker for understanding how college students utilize mobile interventions. Merely delivering psychotherapy in a web-based format may not actualize the potential that technological innovations can bring to mental health care (Mohr, Weingardt, Reddy, & Schueller, 2017). As such, researchers and developers should be aware that mobile, guided self-help, CBT-based interventions may be able to deliver clinical impact with relatively small amounts of use, which may inform decision making regarding intervention length. Future work should consider how to further maximize the benefits of mobile program use, including providing program components that target those factors most critical to disorder maintenance—those expected to deliver the greatest impact—early on in the program to ensure that a greater number of users complete, and thus benefit, from this content.

Limitations of the current study include: the SB-ED intervention was only available via iPhone app at the time of this initiative, which may have resulted in lowered rates of engagement than if an Android app had also been available; and lack of long-term follow-up assessment on participating students. Given that HBI was implemented as a state-wide initiative and not as a clinical trial, the collection of follow-up assessment data were outside the scope of the project. However, results of a large-scale, National Institute of Mental Health-funded randomized controlled trial comparing the efficacy of the SB-ED mobile intervention to referral to usual care for college women who screen positive for a clinical/subclinical ED are forthcoming (NCT02076464). Other notable limitations include the lack of information on whether individuals with possible AN follow through with the recommendations provided and the accuracy of self-report assessment of ED symptoms. In the future, accuracy of participant reporting may be increased through use of real-time reporting of behaviors rather than retrospective recall.

Future work will need to address sustaining and broadening the reach of HBI for EDs, for example, through increasing the automaticity of some coaching activities, in order to decrease the time and cost associated with coaching. Prior work has demonstrated the value of guided versus unguided online interventions for EDs (Kass et al., 2014); thus, it will be important for future work to determine how to retain this critical element of provider support within online interventions while also expanding program reach. Future work should also consider the role of telehealth services in HBI’s suite of offerings and assessing the cost-effectiveness of HBI in Missouri, which has demonstrated modest cost savings over standard ED care in initial projections (Kass et al., 2017a), as well as increase the reach of HBI to underrepresented groups and adapt the screening approach and interventions to better address demographic and cultural concerns. Finally, to achieve the ultimate goal of improving students’ overall health and wellbeing, future work should focus on developing and evaluating a comprehensive, online college mental health care platform that can address multiple mental disorders, beyond just EDs. Indeed, we have observed that limited funding and limited person-based resources for ED intervention are organizational-level barriers that hinder widespread adoption and sustainability of our ED-specific platform. Moreover, the prevalence and salience of other mental health issues among college students affects stakeholders’ interest in allocating resources to ED intervention. As such, there may be value in the development of an online, population-based screening and intervention platform for the most common mental health needs of college students (e.g., depression, anxiety, eating disorders, substance use). Students could be offered multiple interventions that are precisely suited to their needs and address the complex symptom presentations with which individuals frequently present.