Introduction

There is a paucity of research investigating body image disturbance and its treatment in young people with anorexia nervosa (AN). As with adults, a core diagnostic feature in AN during adolescence is disturbance in body image, where self-worth is disproportionately based on shape and weight [1]. Body image disturbance is a robust predictor of AN [2, 3], illness relapse [4, 5] and often persists in otherwise recovered patients [6].

Treatment programmes for adolescents with AN often address body image; however, the effectiveness of these body image interventions remains unknown. Research into body image treatments may be limited because this work can be particularly challenging and distressing for a patient, thus leading to patient and/or clinician avoidance. Body image treatments for adolescents may also be given less priority to the core treatment goals of the re-establishment of normal eating and weight restoration. Research is, therefore, needed to establish whether body image interventions among this population are warranted, to enhance treatment outcomes and to reduce risk of relapse.

Extensive research has looked at body image disturbance and its treatment in non-eating disorder adolescent populations (e.g., [7, 8]). Typically, the goal has been to address perceived media pressure to be thin and reduce internalisation of the thin ideal, a known risk factor of body dissatisfaction [9], while enhancing self-esteem [10]. Cognitive-dissonance-based interventions, whereby participants critique the thin ideal through a variety of activities, have been consistently shown to reduce internalisation of the thin ideal and other eating disorder (ED) risk factors in both internet- and group-based formats [11, 12]. Similarly, cognitive-behavioural-based interventions have been found to reduce body dissatisfaction among adolescent girls (e.g. [13]).

Among adult ED populations, studies have demonstrated that cognitive-behavioural approaches improve body image and related ED psychopathology [14,15,16,17]. A common component of these interventions is behavioural experimentation, which focuses on addressing body image disturbance and related body checking and avoidance behaviours, through mechanisms of habituation and desensitisation [14]. However, given the contraindication of habituating to an unhealthy weight, these interventions depend on an individual being at or near a healthy weight. In response to this, ‘BodyWise’ [18] was developed to provide a low-intensity body image intervention for underweight adults with AN. This treatment also aims to serve as a precursor to more intensive and behavioural body image work when a healthy weight is reached. Recent evaluation of BodyWise indicated a significant improvement in body image concerns, body checking and avoidance, perceived communication and understanding about body image, and quality of life [18].

The current study evaluated a revised version of BodyWise that was adapted for adolescents. ‘Teen BodyWise’ comprises the same session structure (eight × 50-min sessions), content, and topics as the adult programme (e.g. ‘Perception and Unhelpful Behaviours’ and ‘What is Normal?’), but utilises different multimedia approaches and modernised materials to aid engagement (e.g., PowerPoint slides, YouTube clips). Session content, language, and tasks were made age appropriate and gender inclusive. Moreover, the wording was adapted in some sessions to ensure that Teen BodyWise met the needs of adolescents at different stages of weight restoration, from those who are very underweight to those who are close to their expected body weight. Consistent with research on effective body image interventions [10], groups were multisession and interactive in format. Like the original, Teen BodyWise draws upon cognitive-behavioural and cognitive-dissonance strategies and aims to prepare a young person for engagement in more intensive body image work later in treatment.

Given the importance of early intervention in AN [19], research into body image interventions in a clinical population of adolescents is of great value. No research to date has solely focused on a younger AN population and it remains unclear if Teen BodyWise could produce similarly positive outcomes to those found with adults [18]. This pilot study evaluated Teen BodyWise among adolescents with AN, specifically exploring participant experience and any changes in body image disturbance after completing the group. It was hypothesised that participants would be satisfied with the group and would show significant changes in shape and weight concern, ability to communicate about body image, sociocultural attitudes, and motivation to improve body image. It was also hypothesised that there would not be any significant change to body checking and/or avoidance given that they are behavioural indices of body image not directly targeted in the group.

Methods

Participants

Fifty-six adolescents (54 female and 2 male) were recruited from two specialist child and adolescent inpatient eating disorder units, Newbridge House, Birmingham (n = 34) and Wisteria Ward, South West London and St George’s NHS Trust, London (n = 22). Participants had a diagnosis of AN, made in accordance with DSM-5 criteria, were aged between 13 and 18, and needed to attend at least six of the eight sessions to be included in the analysis. Of the original sample, two declined to finish the outcome measures, one withdrew consent to participate, and one attended only two sessions. Fifty-two (93%) adolescents (50 female and 2 male) were, therefore, included in the final analysis. All participants were also attending other aspects of their respective treatment programmes, including weekly individual therapy.

Body image measures

Perception of understanding and ability to communicate about body image (BICU)

The BICU [18] is a five-item questionnaire designed to assess perception of ability to communicate about body image and understanding of body image. It was developed for the original BodyWise study. Respondents indicate their agreement with statements (e.g. ‘I feel able to talk about my body image’; ‘I feel listened to about my body image’) on a 5-point Likert-type scale, ranging from 1 (‘definitely disagree’) to 5 (‘definitely agree’). The original BodyWise study reported high levels of internal reliability for this measure.

Motivation to Change Questionnaire

Motivation to change was measured through responding to three questions related to changing body image: ‘How important is it to make changes to improve my body image’, ‘How confident am I that I can make changes to improve my body image’ and ‘How ready am I to make changes to improve my body image’. This is a common and readily used brief measure of motivation to change adapted to fit the research question [e.g., 20]. Respondents completed a visual analogue scale ranging from 0 to 10 (0 = ‘not at all’ and 10 = ‘very’).

The Sociocultural Attitudes towards Appearance Questionnaire-3 (SATAQ-3)

The SATAQ-3 [21] measures awareness and internalisation of societal standards of appearance. There are three subscales which assess internalisation (general and athlete), pressures, and information. Respondents indicate their agreement with various statements on a 5-point likert scale, ranging from 1 (‘definitely disagree’) to 5 (‘definitely agree’). This measure has been assessed with adolescents and has shown high levels of internal reliability [22].

Eating Disorders Examination-Questionnaire (EDE-Q)

The EDE-Q [23] is a routine measure of ED psychopathology and behaviours. In this study, the two subscales ‘shape concern’ and ‘weight concern’ were used to assess body image disturbance. Participants respond to 28 items using a 7-point response format ranging from 0 (‘not at all’) to 6 (‘markedly’). The EDE-Q is routinely used to assess ED psychopathology among adolescents [24, 25] and scores from this measure demonstrate good reliability [26].

Body Image Avoidance Questionnaire (BIAQ)

The BIAQ [27] assesses level of avoidance due to concern about physical appearance (e.g. ‘I wear baggy clothes’; ‘I avoid going clothes shopping’). Participants respond using a 6-point response format ranging from 0 (‘never’) to 5 (‘always’). Consistent with previous research [14, 18], a shortened 15-item questionnaire was used. Higher scores indicate greater avoidance. This scale has demonstrated modest to acceptable reliability in adolescent populations [28].

Body Checking Questionnaire (BCQ)

The BCQ [29] is a 23-item questionnaire measuring checking behaviours related to overall appearance, specific body parts, and idiosyncratic checking rituals. Participants respond using a 5-point format ranging from 1 (‘never’) to 5 (‘very often’). This scale has demonstrated good reliability for adolescents [30].

Satisfaction measure

Amended Client Satisfaction Questionnaire-8 (CSQ-8)

The CSQ-8 [31] is an 8-item self-report questionnaire designed to measure client satisfaction with treatment services and has a maximum score of 32. The wording was amended in the original BodyWise study and in the current study to refer specifically to the group [18]. After attending the group, participants were asked to fill in a feedback form which included questions concerning perceived effectiveness of the intervention, quality of therapy received, and improvement in symptoms. Questions included ‘How satisfied are you with the amount of help you have received?’ and ‘If a friend were in need of similar help, would you recommend the group to him or her?’ Participants are also provided blank spaces for written feedback. They were asked what they found helpful about the group, what they found unhelpful about the group and how they thought the group could be improved. The CSQ-8 has demonstrated good reliability in research with adolescents [32].

Procedure

Teen BodyWise was being facilitated at both study sites prior to study commencement. Only those who had not yet participated in the group were recruited to this study. While each unit had the expectation that the adolescents would participate in Teen BodyWise as part of the therapeutic programme, participation in the study (i.e. the completion of questionnaires for research purposes) was voluntary. All adolescents attending the group were told about the study and given a Participant Information Sheet, emphasising that to decline research participation would have no impact on their treatment. Informed consent was obtained from each adolescent and, if under 16, a person with parental responsibility. An uncontrolled, before and after study design was employed with questionnaires administered at the beginning and at the end of the group. Weight, height, and duration of illness were obtained from routine unit documentation. This study was granted ethical approval by the Harrow Research Ethics Committee (14/LO/0224). See Mountford et al. [18] for a description of the original BodyWise programme.

Statistical analysis

Data were analysed using SPSS version 23. Means and standard deviations were calculated for the variables of interest, and a paired t test statistic was used to compare variables from pre-(T1) to post-(T2) treatment. Cohen’s d [33, 34] was calculated to provide effect sizes, where 0.2 is small, 0.5 is medium, and 0.8 is a large effect size [33]. Where skewness and kurtosis estimates indicated that continuous variables did not meet requirements for univariate normality (acceptable range between − 1 and + 1 and − 1.5 and + 1.5, respectively) [35], values were transformed using Box–Cox methods to better approximate a normal distribution [36], and t tests and effect sizes calculated using transformed values. Within-subject change scores were generated from the difference in (transformed) values from pre- to post-treatment. Factors associated with significant pre–post changes were evaluated using linear regression models, testing the effects of age, pre-treatment weight, site, and weeks in treatment until commencing the group whilst controlling for baseline score. To control for the possibility of false-positives because of multiple pre- to post-treatment testing across functional measures, the false discovery rate (FDR; [37]) approach was applied to within-group comparisons, with control set to 5% (i.e., a ranked order of ascending p values obtained from t tests were compared to a series of threshold values of significance obtained from an FDR analysis). For all other analyses, a criterion for statistical significance was set at p < 0.05.

Results

Sample characteristics

Table 1 reports the characteristics of the sample (age, % median body mass index (BMI) for age pre- and post-group, duration of illness and admission length). A percentage of the median BMI for age (%mBMI), referenced according to UK norms, was used as a measure of BMI as it takes into account age and gender norms, and is more appropriate for an adolescent population [38,39,40]. There were significant differences between sites for %mBMI at T1 (M = 85.02, SD = 8.76 versus mean = 79.91, SD = 6.43; p = 0.019) and for admission length in weeks (mean = 11.57, SD = 11.29 versus mean = 5.50, SD = 3.60; p = 0.033), but no differences for age, length of illness or %mBMI at T2 (for all comparisons, p > 0.174). In terms of weight across the whole study sample, %mBMI was significantly higher following the group (T1 mean = 81.88, SD = 7.74; T2 mean = 89.84, SD = 7.64; p < 0.001).

Table 1 Demographic characteristics of the sample at pre-treatment
Full size table

Body image measures

Table 2 shows the means of each body image measure at pre- (T1) and post-treatment (T2) and effect sizes. As expected, participants’ understanding and ability to talk about body image was significantly improved at T2 and there were significant improvements in sociocultural attitudes following the group. Among the EDE-Q subscales, significant reductions in both shape concern and weight concern were found. Surprisingly, a significant reduction in body checking behaviours was also reported. These pre- to post-treatment score differences all remained significant after controlling for multiple comparisons. Effect sizes indicated modest effects for shape concern, weight concern, body checking and sociocultural influence, and a more moderate effect for understanding and ability to discuss body image. There were no significant differences in motivation to change body image or body image avoidance.

Table 2 Body image measure scores at pre-treatment (T1) and post-treatment (T2)
Full size table

A series of regressions analyses were carried out to explore potential participant and treatment characteristics (age, %mBMI, site and length of admission) associated with degree of post-intervention change, controlling for baseline score. The change scores of the five measures yielding significant differences post-intervention were used as outcome variables in these analyses. There were no characteristics uniquely predictive of intervention change across any of the measures, with the exception of baseline scores across all variables (for all models p < 0.046; indicating that participants with high levels of body image dysfunction at baseline tended to show larger improvements) and study site for participants’ understanding and ability to talk about body image (B = 2.57; p = 0.004), suggesting improvements were larger in one site than the other, although in both sites participants demonstrated significant within-group changes (Site A: T1 mean = 13.55, SD = 3.32; T2 mean = 15,35, SD = 2.85; p = 0.027); Site B: (T1 mean = 15.44, SD = 4.66; T2 mean = 18.19, SD = 3.87; p < 0.001).

Participant feedback

The mean CSQ-8 score was 23.4 (SD = 4.30), indicating that Teen BodyWise was perceived to be acceptable by participants. Figure 1 shows the mean scores of each CSQ-8 item. Eighteen (35%) participants offered comments about Teen BodyWise. Open ended questions revealed participants mostly found the group helped towards recovery, improved their understanding about body image (e.g., “it was good and helped me to have more of an understanding about weight”), and challenged thinking about shape and weight (e.g., “it helped me rationalise thoughts and see things from different views and perceptions”). One participant expressed the view that a one-to-one format would be preferable, whilst two other participants found the group nature reassuring. Three participants (2%) reported finding some aspects of the group unhelpful to their recovery; in particular viewing weight charts and figures representing the thin media ideal. Last, some participants reported they would find additional group sessions helpful while others suggested covering less psychoeducation material and providing more practical skills for coping with body image disturbance could be beneficial.

Fig. 1
figure 1

Mean item scores on the CSQ-8 following the group. Error bars represent the standard error of the mean

Full size image

Discussion

The current study reports the findings of an evaluation of Teen BodyWise, a body image group adapted for adolescents with AN. Despite the fact that eating disorder programmes often involve body image treatment, this study is the first to evaluate a body image intervention in adolescents with AN and to elicit participant feedback on the group. There was significant improvement in participant weight after the group which was expected given the therapeutic focus on weight gain during an inpatient admission, via individualised meal plans, meal supervision and limited physical activity.

Similar to the original BodyWise study [18], results indicate small but significant improvements in participant shape and weight concern, suggesting there may be some general impact on body image following completion of the group. Participants also reported an improved understanding and ability to communicate about their body image following the group, with study site being a unique predictor of the magnitude of change. This may be reflective of other treatment differences at the two sites or of the individual style of the therapists. The improvement in understanding and communication of body image is an important finding given the hope that Teen BodyWise, like BodyWise, could reduce fear of discussing body image and ‘set the stage’ for more intensive body image work. However, given the uncontrolled design of this study, it is possible that these improvements could be due to a variety of other influences and highlights the need for a further controlled study to test this potential impact.

The significant improvements in sociocultural attitudes differed from the BodyWise study, which did not find any post-group differences. Mountford and colleagues [18] suggested that thin ideal internalisation might be too entrenched among adults with AN, which could support evaluating this with a younger age. In which case, a controlled study should assess sociocultural attitudes as younger participants may be more amenable to media literacy and challenging unhelpful sociocultural expectations.

Surprisingly, results indicate significant improvement in body checking behaviours following the group. Improvement on a behavioural measure of body image was not expected given the low intensity of the treatment but could reflect a potential benefit of the group. Given the lack of control group, it is unclear whether these behaviours improved due to general body image changes, psychological improvement related to weight gain, or due to the programme message that body checking is unhelpful and perpetuates body image disturbance. However, this finding suggests that a future controlled study of Teen BodyWise should include a body checking measure. As no research to date has looked at body checking and avoidance among adolescents with AN, further research is needed to better understand these body image constructs among this age group. Across all measures, improvements over the course of the intervention were not related to %mBMI, age and length of admission.

Teen BodyWise was considered acceptable and accessible for inpatients at various weights and stages of admission, indicating this group could be a valuable part of treatment. There were no participant characteristics predictive of increased benefit from the group, suggesting Teen BodyWise is appropriate for adolescents with varied weights, ages, and illness duration. Feedback showed this was both a challenging and helpful group, suggesting participants may have experienced some difficulty with discussing their body image. Further exploration of the more challenging aspects of Teen BodyWise could contribute to future research and the refinement of this and similar groups. That this group was considered challenging also suggests it may cover relevant and fundamental material central to the improvement of body image disturbance and distress. The conflicting views of some participants, for example, preference for less psychoeducation and more practical support, could reflect the different stages of treatment and recovery inherent in an inpatient setting. Qualitative data from a larger number of participants could help establish any such patterns.

As mentioned, a limitation of this study is the lack of a control group. It is possible the improvements in body image seen here are due to other aspects of the respective treatment programmes, such as individual therapy, making it impossible to attribute any changes to Teen BodyWise alone. It could be also possible that improvement in body image is due to physical and psychological recovery over the group’s duration, the general experience of being part of a group, or other therapeutic aspects of an admission. Moreover, the stability of post-group improvements in body image is inconclusive given the lack of follow-up data. Future research could compare Teen BodyWise with other body image therapies or treatment as usual to help address this limitation. Research could also explore the adjunctive benefits of more intensive or practical body image work, such as mirror exposure, once a young person has reached a healthy weight as well as identify which aspects or sessions of Teen BodyWise are the most therapeutically potent.

These promising results suggest that Teen BodyWise can be effectively implemented in an inpatient environment and is well accepted. It also offers preliminary support for Teen BodyWise as a beneficial addition to AN treatment; however, future randomised controlled studies, preferably with a follow-up period, are needed to verify the impact of this group on body image disturbance.