Can CBT-E be delivered in an online group format? A pilot study of the Body Image module in a child and adolescent eating disorder service

ABSTRACT The increased prevalence of eating disorders during the COVID-19 pandemic has led to long waiting lists in child and adolescent services. A pilot study was conducted to evaluate the feasibility and acceptability of providing the Body Image module, from the enhanced cognitive behavioral therapy for eating disorders (CBT-E), in a virtual group setting. Primary outcomes were acceptance rates, completion rates, qualitative feedback and quantitative data from routine questionnaires. From 22 eligible referrals, 12 participants accepted and enrolled in therapy. Eight completed all six sessions. Qualitative feedback was positive, with both the content and group nature of the intervention being described as helpful. There was an reduction in scores in the Clinical Impairment Assessment and all subscales of the Eating Disorder Examination for Adolescents, suggesting this was a feasible method of providing psychological therapy within the service. A larger trial is recommended to robustly test the effectiveness of the intervention compared to one-to-one in-person CBT-E, and whether the full CBT-E protocol can be effectively delivered in the same format.


Introduction
Since the COVID-19 pandemic began, rates of eating disorders among children and adolescents have increased (Solmi et al., 2021;Taquet et al., 2021) and publicly funded services in the UK have come under immense pressure (Shaw et al., 2021). An increasing number of urgent new assessments and young people at high medical risk has led to increased waiting times for those young people who need a psychological intervention but are medically stable. It is therefore important to investigate whether the delivery of interventions can be modified to reduce this wait (Bowen et al., 2009). One option for increasing the provision of psychological therapy to children and adolescents is to offer treatment in a group setting.
Many national guidelines indicate that children and adolescents with a diagnosis of anorexia nervosa should be offered an eating disorder-based family intervention as a first-line treatment (e.g., "Dutch Foundation," Hay et al., 2014;Couturier et al, 2020;NICE, 2017) in the treatment of adolescent anorexia nervosa, although the terminology used and the specific form of family intervention recommended does vary (Hamadi & Holliday, 2020). However, treatment outcomes are generally agreed to be modest, with two recent trials of family-based therapy reporting full remission rates at the end of treatment of 34% (Lock et al., 2010) and 22% (Le Grange et al., 2016). Enhanced cognitive behavioural therapy (CBT-E; Fairburn, 2008) is another approach for treating eating disorders. It is a protocol that was designed to be delivered on a one-to-one basis by a clinician and has been adapted for children and adolescents . It treats a range of eating disorder presentations within the transdiagnostic model . The core CBT-E protocol, for people who are not underweight, is 20 sessions in length, and divided into several discrete modules.
In the transdiagnostic model, overvaluation of shape and weight is considered to be the core psychopathology of an eating disorder . This concept is introduced in the Body Image Module in the CBT-E protocol and defined as the assignment of excessive importance to one's body weight and shape, when evaluating one's self-worth (Mitchison et al. 2017). Overevaluation of weight and shape often leads to behaviours such as body checking and comparison (Fairburn, 2008) involving the repeated scrutiny of disliked body parts. Scrutiny tends to magnify apparent or perceived defects (Dalle Grave & Calugi, 2018) which, in turn, maintains body image concerns. Furthermore, body image concerns often focus on the difference between current and ideal body weight or shape (Neighbors & Sobal, 2007).
While the literature on group treatment is limited, positive outcomes have been reported in the delivery of in-person interventions to adults whose worries about body image do not reach the cut-off for clinical severity (Cash & Hrabosky, 2003;Strachan & Cash, 2002), and those with diagnosed eating disorders (Hilbert & Tuschen-Caffier, 2004;Mountford et al., 2015). Chen et al. (2003) compared group and individual CBT for adults with a diagnosis of bulimia nervosa. The majority (73.3%) of participants completed treatment and outcomes were similar in both arms of the trial. Regarding the CBT-E protocol, Dalle Grave et al. (2008) reported on the provision of group CBT-E and made recommendations based on these experiences but did not provide quantitative outcome data. They suggested that both the content and home tasks in group sessions should remain the same as with individual CBT-E, but that sessions should last 90 minutes, rather than 50. They additionally explain that a key barrier to group therapy provision was that it was very difficult to arrange, because of patient availability. This implies significant challenges for longer-term group therapy, where all members of a group would need to be able to commit to every session over an extended period of time. Wade et al. (2017) conducted a randomised control trial, in an adult population, evaluating the effectiveness of group CBT-E for eating disorders by comparing outcomes of an immediate start with an eight-week delayedstart wait-list group. The first eight weeks of group CBT-E led to a reduction of eating disorder psychopathology, whereas no significant reduction was observed across the eight weeks for those randomised to the waiting list. The paper reported that 70% of those who entered the trial completed treatment (attending an average of 18 sessions out of the scheduled 20), and 67.9% of those completers achieved a good outcome, concluding that a group version of the treatment was a valid method of reducing eating disorder psychopathology.
During the COVID-19 pandemic, measures to combat the spread of the virus have meant that in many locations, in-person therapy has not been possible. Online group therapy may have value even outside of a pandemic setting as distance and access to services have long been recognised as barriers to psychological therapy (Reardon et al., 2017). Therefore, one positive effect of the pandemic response is that interest in, and availability of, virtual treatment has increased.
Within the field of eating disorders, several services have reported on their move to online working, mostly within high-intensity treatment. Plumley et al. (2021) described the online continuation of an adult eating disorders day programme, noting that emotive topics could be challenging to manage in a virtual setting and suggesting that a useful strategy is to set an activity that can be completed individually and then reflected on as a group.
One pilot study delivered a virtual intensive outpatient program for adults with eating disorders. They were able to recruit participants during COVID-19 and there was a good level of patient adherence to the programme. Responses from participants and clinical outcome data suggested it was an acceptable intervention for patients (Blalock et al., 2020) An example from child and adolescent services examined the experiences of young people who attended an intensive day-treatment programme for eating disorders online (Brothwood et al., 2021). Although a high proportion (71%) rated each component of treatment as either somewhat or very helpful, the paper noted that therapeutic alliance was affected, and several technical problems were reported.
In summary, there is some evidence that in-person group provision of eating disorder therapy is effective, including CBT-E, in an adult population. There is also emerging evidence that it is possible to deliver eating disorder treatments online. To the authors' knowledge, there are no trials evaluating group CBT-E for children and adolescents, either in person or online. To address the pressure on services, a pilot study was conducted to test whether it was feasible and acceptable to deliver CBT-E in a group setting to children and adolescents in an eating disorder service. To manage the restrictions of COVID-19, the intervention was delivered virtually.
While feasibility studies are limited in the conclusions that can be drawn, they provide a time-and cost-effective means of testing whether an intervention could work (Bowen et al., 2009). Out of a clinical need, this study evaluated the group and the virtual nature of the intervention simultaneously. As the logistical challenges of arranging longer-term therapy in a group setting have been identified, the study selected a module from CBT-E to pilot. The Body Image module from the CBT-E protocol was chosen because thoughts and behaviours that were driven by over-evaluation of shape and weight, were the most common presenting problem in patients on the waiting list. This pilot was conducted with a view to determining whether an online group approach could be used more widely by the service.

Method
This study was approved as a clinical service evaluation by the Oxford Health NHS Foundation Trust.

Participants
The service is a regional mental health service in the south-east of England offering Assessment and treatment for? people with eating disorders who are under the age of 18. In 2021, the service received 220 referrals. Of these, 93.2% were female and 6.2% were male. The majority of referrals described themselves as white (84.5%) and 9.5% described themselves as having a multiple ethnic group. The remainder of referrals described themselves as Asian/Asian British (3.2%), Black/African/Caribbean (0.5%) and other (1.4%). Two referrals (0.9%) had an unknown ethnic group.
All clinicians within the Oxford service were invited to refer patients who met the eligibility criteria. The referral period was two weeks. Patients were eligible for referral if they were currently open to the eating disorder service, had an allocated care coordinator and were eating regularly. Patients had to be medically stable and be on the waiting list for psychological therapy due (or partly due) to body image concerns.
Referrals were screened by the lead author and those who met the criteria were invited to take part in the six-session intervention. Based on patient feedback conducted in advance, it was agreed the maximum number of participants in any group would be eight. Parents and other family members did not attend the presentations. This was to encourage young people to share their experiences with others with similar difficulties. The participant's care coordinator remained responsible for monitoring physical observations including weight and speaking with the participant about regular eating.

Intervention
Detailed slides were created on Microsoft PowerPoint for all six sessions, mirroring the CBT-E Stage Three Body Image module from Fairburn's (2008) manual. This included exploring the young people's system of selfevaluation, which was drawn out as a pie chart. Following this, the treatment covered the extended formulation of overevaluation of shape and weight, and the roles that body checking, body avoidance, comparison making and "feeling fat" play in maintaining it. As with individual CBT-E, each session provided psychoeducation, set a home task and reviewed the previous week's home task. Tasks included completing self-monitoring forms and conducting behavioral experiments, as outlined in the protocol. Each week during the sessions there were questions and activities for the participants to consider or complete individually, which were then fed back or reflected on as a group. When participants were asked questions during sessions, they had options of answering verbally or via the chat function.
Each session was delivered in Microsoft Teams and two groups ran concurrently, each containing the same six sessions. These were facilitated by the same qualified clinical psychologist each time, who was experienced in delivering CBT-E, and supported by one of five assistant psychologists, who took turns to co-facilitate. Therapy sessions took place weekly with a two-hour slot allocated for each. The first 30 minutes was reserved to welcome participants and address technical issues. All participants and facilitators were required to have their cameras on during sessions, except when completing individual tasks. At the end of the session, participants booked a follow-up call with an assistant psychologist.
During this weekly 15-minute call with an assistant psychologist, participants were asked about the preceding session to check their understanding of the topic. They reviewed the home task and participants were invited to ask questions or provide feedback. Written records of home tasks were to be sent to the facilitator, via email, 24 hours prior to the next session. This was to allow the intervention slides to be updated and personalized, to ensure the experiences of all group members were acknowledged and covered.
An a priori decision was made that participants who missed a session were not to continue with the intervention. This was due to the nature of CBT-E where sessions, psychoeducation and home tasks all build on those of the preceding session.

Feasibility
Feasibility was evaluated by the demand, practicality, and ability to implement the intervention. This was measured in several ways. It was not known whether patients would wish to take part in a group therapy and, therefore, demand was determined by whether the young people in the service wished to be referred for the intervention. It was measured by reviewing the number of referrals and the acceptance rate by those invited to take part.
Practicality and ability to implement the intervention were measured in many ways. These included whether the protocol could be delivered in a virtual format and in the time available, and whether participants attended the sessions and completed the homework. It also included whether the online group format appeared to support participants sharing experiences and ask questions. The service also monitored technical and admin issues which affected the delivery of the intervention.

Acceptability
Acceptability was evaluated by how participants responded to the intervention. It was primarily measured by completion rates and qualitative feedback. The feedback was sought in writing via email at the end of the intervention and participants were invited to comment on any aspect of the process and to make recommendations for the future. Providing feedback was optional and was done via an e-mail to the assistant psychologist supporting the final session. Any verbal feedback from participants during group sessions and follow-up calls was also recorded by the assistant psychologist.

Outcomes
Clinical outcomes were measured using scores from two routinely used outcome questionnaires (see below). Those who took part in the study filled in the following measures before treatment began, at the end of treatment, and three months after the end of treatment. In all cases, the questionnaires were sent to participants in Microsoft Forms format. These were competed by the participant and scored by one of the co-authors. The scored results were sent to the lead author. In line with guidelines for pilot studies (Lancaster et al., 2004) descriptive statistics were used to present the data. The Eating Disorder Examination for Adolescents (EDE-A; Carter et al., 2001).
The EDE-A is the adolescent version of the Eating Disorder Examination Questionnaire (EDE-Q; Fairburn & Beglin, 1994). It is a 36-item self-report questionnaire using a seven-point rating scale, with a total raw score range of 0-138, where higher scores indicate more problematic eating behaviors and attitudes. It measures four subscales of eating disorder psychopathology (Restraint, Eating Concern, Shape Concern, and Weight Concern) and a global score. It differs from the EDE-Q in focusing on the past 14, rather than 28, days and using simplified language. It is used as standard in child and adolescent eating disorder services in the United Kingdom, however limited research on the adolescent version means there is less evidence for its psychometric properties (Mantilla et al., 2017) compared to the EDE-Q. The Clinical Impairment Assessment (CIA; Bohn et al., 2008).
The CIA is a 16-item self-report questionnaire testing the severity of psychosocial impairment due to an eating disorder. Questions cover a range of different areas of life and each item is rated on a four-point Likert scale. Scores range from 0 to 48, with higher scores indicating higher impairment. The clinical cutoff score is 16 and it has demonstrated satisfactory psychometric properties among a sample of young women (Reas et al., 2010).

Participant characteristics
Twenty-five referrals were made, of whom 22 (88%) were accepted as eligible. The lead author contacted all eligible referrals via email and/or telephone to discuss the intervention, provide an information sheet, and ask whether they would like to take part. Four (18.1%) could not be reached or did not reply within the time limit. Six (27.3%) declined for a combination of reasons including: changing their mind about wanting psychological therapy (n = 4; 18.1%), not being able to commit to the scheduled appointments (n = 1; 4.5%) and not wanting a group intervention (n = 1; 4.5%). Twelve participants (54.5%) accepted and confirmed they could commit to all the scheduled sessions. Eight of the 12 participants (66.7%) completed all six sessions. Of the four who were withdrawn after having missed a session, one attended five sessions, one attended four sessions, one attended two sessions and one attended one session. Of the eight who completed the course, all participants (100%) returned end of treatment data and five (62.5%) additionally returned follow-up data, three months after the end of treatment. See, Figure 1 for a flow chart of the recruitment process.
All 12 participants were female and white British, and the average age was 15.79 years (range, 13.19-19.69; standard deviation 1.47). Seven (58%) had a diagnosis of anorexia nervosa and four (33%) had a diagnosis of atypical anorexia nervosa. One participant (8%) did not have a formal diagnosis as it was felt that the participant would find it unhelpful, however they met the diagnostic criteria for atypical anorexia nervosa. Diagnoses were made by the lead clinician who assessed the person on their introduction to the service. This was prior to their inclusion in the current study. Diagnoses were based on criteria from the diagnostic and statistical manual of mental disorders (American Psychiatric Association, 2013). Lead clinicians conducted a full clinical interview and incorporated results from a range of psychometric measures used by the service before making a diagnosis.
Weight for height was calculated as a percentage of the 50th centile value for height, age and sex for each participant. The mean average weight for height was 93.54% (range, 78.31-94.74; standard deviation 9.43). The mean average time that participants had spent in the service was 6.75 months, but this varied significantly 25 referrals 2 not accepted as receiving other individual therapy 1 not accepted as not medically stable 22 invited to therapy 6 declined 4 did not respond 12 enrolled in therapy 8 attended all 6 sessions 1 attended 5 sessions and was withdrawn after missing a session 1 attended 4 sessions and was withdrawn after missing a session 1 attended 2 sessions and was withdrawn after missing a session 1 attended 1 session and did not wish to continue (range, 2-43, standard deviation 11.66). Eleven (91.7%) had previously received anorexia nervosa-focussed family therapy or was currently engaging in this treatment, and one had received no prior treatment. All were eating regularly and therefore had addressed key elements of stage 1 of the CBT-E protocol, and all had identified body image concerns as the main barrier to their recovery.

Clinical outcome data
The end-of-treatment questionnaires were filled in by all eight completers. There was a reduction in the means of: the EDE-A global score; all EDE-A subscale scores; and the CIA score (see , Table 1). This indicated a reduction in eating disorder psychopathology and psychosocial impairment. Five of the eight completers also filled in the three-month follow-up questionnaires. There were further reductions in scores from the Weight Concern subscale, the Shape Concern subscale, the Global EDE-score and the CIA.
All eight completers had a baseline CIA score in the clinical range. At follow up, two people no longer scored in the clinical range. Similarly, one person moved from above clinical cutoff to below, on the EDE-A global score, and two moved from above clinical cutoff to below on the Shape Concern subscale. Figure 2 shows the changes in both individual and mean scores. Effect sizes have been reported as Cohen's d (see, Table 2). The largest effect size at follow up was in the CIA, followed by the Shape Concern subscale. However, effect sizes from pilot studies are not considered to be reliable due to the very small sample sizes (Leon et al., 2011).

Qualitative feedback
All eight completers sent written qualitative feedback. The following is a summary of the topics discussed.

Reflections on the intervention
Seven of the eight participants described finding the content very helpful. The eighth described therapy as a useful refresher, having previously covered some of the topics while in the service. There were a range of reasons given for why treatment had been helpful, including the permission to talk about body image difficulties and the psychoeducation about how certain behaviors are both caused-and maintained by-the eating disorder. All participants who commented on the self-monitoring home tasks and behavioral experiments described them as productive and helpful. However, some people were surprised to discover how frequently they engaged in behaviors such as body checking, which made one person feel sad and another anxious. Two people said that self-monitoring increased their preoccupation with shape and weight in the short term.
Time spent in session as a group reflecting on the home tasks was described as helpful and reassuring. Two people noted that doing this showed that their home efforts and written work were valued and useful.
Four people said the course made them more confident. Two said it had helped them understand why making changes was important. Two commented that the information contained in the PowerPoint slides was useful and accessible.

Group format
When being invited to take part in this intervention, several people reported being anxious that it was taking part in a group setting. In the written qualitative feedback, however, five participants spoke specifically about finding the group nature of the course helpful. It was found by most to be validating and reassuring, with specific comments about it helping people realize that they are not alone in their struggles and that others have similar worries and engage in the same behaviors.
During the intervention, the issue of comparison between group members was raised and addressed. Two participants noted in their written feedback that this was an ongoing but manageable concern. In feedback in individual calls, two other people said they had worried this would be a problem, but had found the environment supportive, rather than competitive.

Online format
When being invited to take part in this intervention, several young people reported being anxious about the prospect of having their cameras on for therapy. This was because they would be able to see themselves, and others would be able to see them, in a way that is not possible in in-person therapy. It was hypothesized that this may act as an exposure exercise, leading to decreasing levels of anxiety over time. The feedback about using cameras was positive in follow-up calls, with several people saying it made the group feel more interactive. There was no further negative feedback on having cameras on once the intervention began. In the written qualitative feedback, one young person commented that although virtual treatment was suboptimal, there was a positive side because they could complete tasks alone which made them feel less judged.

Technology, admin and resources
The content of the sessions was delivered in the time available, and the PowerPoint format appeared to work well. Participants shared experiences and asked questions regularly throughout the sessions. Overall, there were few technological difficulties, however on one occasion, a participant did not have access to a working device with which to join the session. The initial 30 minutes of the session, which was allocated to address technical issues, worked well on several occasions, as people were able to identify and fix connection and computer difficulties, preventing them from missing a session. However, one person fed back that 30 minutes was too long to wait.
The majority of participants attended on time. Additional time and resources were needed where people did not attend sessions on time. Facilitators emailed and telephoned participants, meaning other young people were waiting. Chasing home tasks also took up time. Overall, 40.7% of written home tasks were sent in on time with no prompts; 25.9% were returned late, after one or more prompts, and 33.3% were not sent in. When home tasks were returned, they were deemed by the clinicians to be of a high quality. One person gave feedback that it would be helpful for the service to send a reminder to everyone about sending in the home tasks, the day before they were due. Participants reported several reasons for not returning home tasks, including having COVID-19, forgetting and technical difficulties. The pilot was also run over the summer holidays. Some participants found that the home tasks increased their preoccupation with weight and shape. This may have been a contributing factor in not completing home tasks.
One young person recommended providing a way to answer questions or make comments anonymously, particularly at the beginning of the course due to the challenges of speaking about sensitive topics to people they did not yet know. The available methods (speaking or using the chat function) did not provide anonymity.

Discussion
This pilot assessed whether it was feasible and acceptable to provide a module from the CBT-E protocol in an online group format to children and adolescents in an eating disorder service. Patients in the service seemed willing to try a group intervention, with just over half of eligible referrals beginning therapy, and two thirds of the participants attending every session. Delivery of the therapy in this way worked well; it was possible to provide the content of the protocol via Microsoft Teams and participants engaged well in the sessions. Providing qualitative feedback was optional, but all eight participants who completed the intervention chose to do so. The qualitative feedback suggested that the intervention was helpful, which is in keeping with other services who reported that it was possible to provide an adolescent eating disorder service, with high satisfaction levels, in a virtual setting during COVID-19 restrictions (Shaw et al., 2021).
The small sample size, due to the pilot nature of the study, is a significant limitation of the study. The results are insufficient to robustly evaluate its efficacy, and a larger sample would be needed to test feasibility and acceptability at a service level. Nevertheless, the reduction seen across all subscales measuring eating disorder psychopathology and psychosocial functioning was encouraging. Calugi and Dalle Grave (2019) highlight the importance of assessing and addressing body image when working with children and adolescents with anorexia nervosa. In all cases, the participants were referred due to body image concerns maintaining an eating disorder. The Shape Concern subscale, which includes items most targeted by the Body Image module, showed the greatest reduction in scores of the EDE-A subscales, suggesting that this was an appropriate intervention for the participants.
It is important to note that 11 of the 12 participants had undergone previous treatment in the service, which may have influenced the outcomes. Furthermore, mean scores on the EDE-A and the CIA remained in the pathological range. This intervention is not considered to be a primary or standalone treatment. It delivers a subsection of one protocol which, as done here, could support those undergoing family based treatment or other treatment, where overevaluation of shape and weight and body image concerns are maintaining the eating disorder. A larger study to determine effectiveness, and compare outcomes with individual CBT-E, would help to determine whether there is an evidence base for this adaptation. Further research could test whether other modules and the full CBT-E protocol can be similarly delivered in this population. This could inform whether, and how far, virtual group CBT-E is an appropriate option within clinical services, and whether specific groups of people are more or less likely to benefit from this mode of delivery. While it is recommended in the CBT-E protocol that the full intervention is delivered, it is common in clinical practice for individual modules and strategies to be used with service-users. Additionally, national guidelines state that future research should determine whether psychological therapies of reduced duration are as effective for patients with eating disorders (NICE, 2017).
Pilot studies have relatively limited external validity (Bowen et al., 2009) and, with all participants being female and White British, outcomes cannot be extrapolated more broadly. However, there is a need for studies to be conducted in real-world clinical settings to establish whether interventions fit the populations that are seen in services (Green & Glasgow, 2006). As the participants were a fairly accurate representation of the people referred to the service, the majority being White and female, the results have practical validity in this setting.
A summary of the challenges and opportunities in delivering CBT-E during COVID-19 (Murphy et al., 2020) states that CBT-E is well suited to virtual therapy due to being a talking treatment which translates well to online videocalls. The summary also highlights the importance of the person in therapy completing tasks and making changes between sessions, which is also described as consistent with working remotely. The results from this study would support these statements.
The greatest drawback of delivering the protocol in a group setting was the policy on missing sessions. As noted earlier, Dalle Grave et al. (2008) highlighted the key logistical challenge of arranging group therapy. The difficult, a priori, decision to remove participants from the course if they missed a session, for any reason, was taken due to the nature of CBT-E. The psychoeducation, home tasks and behavioural interventions all build upon knowledge and exercises from the previous session. During the COVID-19 pandemic, there was no possibility that a clinician could help a participant catch up on material if they missed a session. Permitting people to continue after missing part of the protocol was to risk confusion or poorer outcomes, which the participant may have wrongly attributed to them not being appropriate for therapy or being less likely to recover. Conversely, the policy may have positively contributed to two thirds of people attending every single session.
To provide the full 20-session CBT-E protocol in an online group format, a plan to support participants who miss a session would be helpful. One option may be for services to run several groups, which would allow people to re-join another if they have to miss a session. It may also be helpful for future research to compare different approaches, evaluating overall attendance and outcomes where missing sessions is permitted, and when it is not.
Another limitation of the study was that the qualitative feedback was not anonymised. It was sent via email due to the pandemic as clinicians were unable to gather anonymised feedback on paper. With such a small sample in each group, it was thought that the creation of anonymous online forms would not effectively provide anonymity. This may have influenced the feedback.
The results suggest that provision of online group CBT-E is a possible means of increasing the availability of evidence-based treatments to patients. Delivery of the protocol online appeared to work well, there was a reduction in scores t on all measures of eating disorder psychopathology and impairment, and qualitative feedback was very positive. In a time when waiting lists are increasing, and access to services can be limited for practical reasons, it is useful to know that the content from the CBT-E protocol can be delivered, understood and used in a virtual group setting.