INTRODUCTION

Outpatient treatments for adolescent restrictive eating disorders have a substantial evidence base, with family-based treatment (FBT) generally supported as a front-line treatment for anorexia nervosa. Alternatives, such as cognitive behaviour therapy modified for eating disorders (CBT-e), also display effectiveness among adolescent populations (Dalle Grave et al., 2019; Lock & Le Grange, 2019). However, even with well-implemented evidenced-based therapies, more than half of young people do not achieve full recovery (Byrne et al., 2017; Dalle Grave et al., 2019; Lock et al., 2010). Those who do not recover tend to require multiple inpatient admissions and readmissions, experience significant health complications as a result of starvation, are at increased risk of engaging in self-injury and/or suicide and place increased demands upon family and health services over time (van Hoeken & Hoek, 2020). Hospitalisation has historically been used as a primary option to increase treatment intensity with this group. However, hospitalisation is costly, and outcomes are mixed, with evidence suggesting that there are limited long-term benefits of inpatient admission beyond medical stabilisation and containing the acute risk (e.g. Gowers et al., 2000; Strober et al., 1997).

Day programme treatment

Over the last decade, a range of intensive community treatment options have emerged for those who do not respond to existing outpatient treatments. Typically described as day programmes (DPs) or partial-hospitalisation programmes, these serve as an alternative to hospital or other outpatient treatments. A recent systematic scoping review of the literature investigated outcomes from adolescent DPs (Baudinet & Simic, 2021). The review found that although there is wide variation in DP models of care, similarities include increased intensity of treatment (several hours a day, multiple days per week), supervised/supported meals, and group, family and individual therapies. The review found that DP treatments are associated with weight gain and reductions in both eating disorder and comorbid psychopathology and that these gains were sustained across follow-up periods.

While the review indicated the effectiveness of DP treatment as an alternative to hospitalisation or traditional outpatient treatment, it also highlighted that the programmes varied significantly in treatment length, treatment volume, population treated, programme aims and theoretical frameworks guiding treatment, making it difficult to compare outcomes or investigate mediators and/or moderators to effective treatment. The current study aimed to investigate initial outcomes for adolescents with restrictive eating disorders who had not progressed to recovery in outpatient FBT.

Family factors in day programme treatment

The majority of DP treatments are predominantly informed by FBT (Baudinet & Simic, 2021). FBT assumes that the ability of the family and patient's parents to assume control over eating disorder behaviour is the central mechanism of change in recovery (Ellison et al., 2012). An aim of FBT is to increase parental self-efficacy, which in turn is thought to increase the ability of parents to take control of eating disorder behaviour. However, there have been limited studies investigating parental variables related to this hypothesis in the context of DP treatment (Homan et al., 2021).

Two studies have investigated parental self-efficacy utilising the parent versus anorexia scale (PvA; Rhodes et al., 2005) across two separate FBT-informed DPs (Girz et al., 2013; Hoste, 2015). Both studies found that parental self-efficacy improved by the end of treatment and continued to increase post-treatment at three- or six-month follow-up. Similarly, Robinson et al. (2013) investigated the influence of parental self-efficacy on treatment outcome. They found that parental self-efficacy improved over the course of treatment and that this improvement predicted improvements in young person eating disorder symptoms. However, this was a mixed sample of both outpatient and DP participants, and as such, the influence of different treatment modalities on these data is not clear. Martin-Wagar et al. (2019) investigated parental empowerment (related to self-efficacy) as a predictor of treatment outcome in DP treatment for adolescents. They found that low levels of parental empowerment at entry to treatment predicted weight restoration, which was the opposite of their hypothesis. They did not measure change in empowerment over the course of treatment.

Given the common use of FBT as the framework informing DPs and the focus on parent self-efficacy within this treatment framework, the limited extant studies highlight that research in this area is warranted to clarify the role of parental self-efficacy in DP treatment.

Accommodating parental behaviour

A central tenet of FBT is that increased parental self-efficacy decreases the degree to which parents accommodate and enable eating disorder behaviours in their family (Rienecke & Le Grange, 2022). While several studies investigating accommodating and enabling behaviours have occurred in adult outpatient treatment settings, very few studies have focused on adolescent DP settings (Goddard et al., 2011). There is some evidence to suggest that higher parental accommodation and enabling behaviours are associated with worse treatment outcome for adolescents and that these behaviours decrease after treatment as usual (TAU) (Salerno et al., 2016). An outpatient study of FBT for older adolescents and young adults (16–25 years) found that accommodating behaviour from parents decreased across treatment; however, this did not predict changes in eating disorder behaviour or weight gain (Dimitropoulos et al., 2018). A study investigating FBT-informed DP treatment found that parent scores on the accommodating and enabling scale (AESED; Sepulveda et al., 2009) decreased from intake to discharge in carers of adolescents with anorexia nervosa (AN) (Wagner et al., 2020). In this study, decreases in AESED scores were related to decreased eating disorder symptomology for the young person. The relatively few adolescent DP studies and mixed results highlight the importance of further research to clarify the relationship between accommodating behaviour, parental self-efficacy and treatment outcomes in DP settings (Dimitropoulos et al., 2018).

Parental mental health

The literature investigating the role of the adolescent's comorbid psychiatric diagnoses has largely focused on treatment outcome (e.g. Homan et al., 2021). Despite the central role of the parent in adolescent eating disorder treatment, there has been far less focus on the relationship between parental mental health and treatment outcome. In a study of outpatient adolescent eating disorder treatment, parental distress decreased across treatment (Salerno et al., 2016). A secondary analysis of the same data found that parental depression was significantly associated with the maintenance of the eating disorder psychopathology and negatively predicted body mass index (BMI) at follow-up (Monteleone et al., 2022). Together, these results support the interrelationship between parent and adolescent mental health in the maintenance of AN and suggest that parental mental health could be a critical treatment target in family interventions. While there is evidence showing that DP interventions reduce carer burden, there is no research on the impact of DP engagement on parental depression and anxiety (Girz et al., 2013).

The current pilot study

This study served as a pilot study for a larger study that will further investigate the variables of focus. The current study aimed to investigate the effectiveness of a newly implemented DP on the weight, eating disorder, depression and anxiety symptoms of adolescents with restrictive- eating disorders who had demonstrated poor response to FBT-AN in an outpatient setting. Additionally, the study aimed to investigate changes to parental self-efficacy, accommodating and enabling behaviour, depression and anxiety symptoms in the parents of the adolescents.

Informed by existing data on similar DPs, the first hypothesis was that the young people would show improvement in core eating disorder factors (i.e. increased weight and decreased eating disorder psychopathology) and reduced comorbid depression and anxiety post-treatment compared to pre-treatment (Baudinet & Simic, 2021). Informed largely by investigations in outpatient FBT and adult eating disorder treatment, the second hypothesis was that parents would show increased self-efficacy and reduced accommodation and enabling behaviours post-treatment. Further, informed by evidence that eating disorder interventions significantly reduce carer burden, the third hypothesis was that parental anxiety and depression would be reduced post-treatment (Girz et al., 2013).

METHODS

Participants

Participants were referred to the eating disorders day programme (EDDP) from the Children's Health Queensland Child and Youth Mental Health Service network. The study was approved by Children's Health Queensland's research ethics committee (reference: HREC/20/QCHQ/67708). All the participating families provided informed consent for their data to be included in publication. Eligibility criteria were that participants had been engaged in community-based FBT-AN for no less than three months and had stalled in treatment progress, struggled to remain in community treatment versus inpatient treatment, and/or demonstrated comorbid diagnoses that impacted treatment and could not be resolved by pharmacological intervention alone. Being stalled in treatment was based on clinical judgement at case review by the multidisciplinary team. Factors that were indicative of stalled progress included weight loss or lack of expected weight gain, ongoing medical compromise, lack of success empowering parents to support regular eating with their young person, lack of reduction of eating disordered behaviour (e.g. purging) or escalating emotional distress within the family.

Adolescents

A total of 13 young people (mean age 14.38 years, SD = 1.19, range = 13–17, female 69.2%) started and completed the DP in the 2020 calendar year, with a primary diagnosis of either AN (62%) or atypical AN according to the International Classification of Diseases' ten criteria. EDDP runs four terms a year alongside the school terms. Six participants completed two terms of DP, with the remainder completing one. All but one participant had at least one secondary psychiatric diagnosis as diagnosed by the assessing psychiatrist. Secondary diagnoses included generalised anxiety disorder (three participants); anxiety disorder, unspecified (three participants); mixed anxiety and depression (two participants); social phobias (two participants); obsessive compulsive disorder (one participant), delusional disorder (one participant); post-traumatic stress disorder (one participant); and attentional deficit hyperactivity disorder (one participant).

Parents

A total of 20 parents (11 female, mean age 46.25 years, SD = 4.99), comprised of seven partnered pairs and six single parents (four female) started and completed the DP and provided pre- and post-intervention data. All were biological parents.

Measures

Demographic and hospital admission data

Participants responded to items relating to their age and sex. Hospital admission data were collected from the patients' medical files.

BMI centiles

BMI centiles were calculated using the individual's weight, height and the Centre for Disease Control and Prevention growth charts (www.cdc.gov/growthcharts).

Eating disorders examination questionnaire (EDE-Q)

The EDE-Q adapted for adolescents was used (Carter et al., 2001). The EDE-Q is a 36-item measure that yields four subscales: restraint, eating concern, shape concern and weight concern. Items are scored on a 7-point Likert scale (0–6), with higher scores indicating greater severity of eating disorder symptoms. Adaptations from the original version (Fairburn & Beglin, 1994) include shortening the time frame of reported symptoms from 28 days to 14 and simplifying the wording and phrases in some items. Mean scores were used for each subscale. A global subscale score was derived by summing the mean scores from the four subscales and dividing by four. The clinical cut-point utilised for this measure was 2.82 (details on calculations are described in the results), with scores above this point indicative of a clinical population. Previous studies have reported acceptable psychometric properties and excellent Cronbach's alphas (.93–.96; Forsén Mantilla et al., 2017).

Clinical impairment assessment (CIA)

The CIA is a 16-item self-reported measure of psychosocial impairment due to eating disorder features (Bohn et al., 2008). Items are scored on a 4-point Likert scale (0–3) and are summed to form a total score, with higher scores indicating higher levels of impairment. The clinical cut-point utilised for this measure was 16, as recommended by Bohn et al. (2008), with scores above this point indicative of a clinical population. Previous studies have reported acceptable psychometric properties and an excellent Cronbach's alpha (.94; Reas et al., 2010).

Patient health questionnaire 9 (PHQ-9)

The PHQ-9 is a 9-item self-reported measure of symptoms of depression (Kroenke et al., 2001). Items are scored on a 4-point Likert scale (0–3) and are summed for a total score, with higher scores indicating greater depressive symptoms. The clinical cut-point utilised for this measure was ten, as recommended by Kroenke et al. (2001), with scores above this point indicative of a clinical population. Previous studies have reported acceptable psychometric properties and Cronbach's alphas of .86–.88 (Wisting et al., 2021).

Child anxiety scale (CAS-8)

The CAS-8 is an 8-item self-reported measure of anxiety (Reardon et al., 2018). It is scored on a 4-point scale, with higher scores indicating greater anxiety symptoms. It has shown good psychometric properties and acceptable internal consistency (Cronbach's alphas of .73–.85; Reardon et al., 2018).

Parent versus anorexia scale (PvA)

The PvA is a 7-item self-reported measure of parental self-efficacy in treating their child's eating disorder (Rhodes et al., 2005). Items are scored on a 4-point Likert scale (0–3) and are summed to form a total score, with higher scores indicating higher levels of impairment. Cronbach's alphas from previous studies have been adequate (.78: Rhodes et al., 2005).

Accommodating and enabling scale for eating disorders (AESED)

The AESED is a 33-item measure that yields five subscales: avoidance and modifying routine, reassurance-seeking, meal rituals, control over family and turning a blind eye (Sepulveda et al., 2009). Items are scored on a 5-point Likert scale (0–4), with higher scores indicating higher levels of parental accommodating and enabling behaviours for their child's eating disorder. Scores are summed for each subscale. A total score is derived by summing all scores. Cronbach's alphas from previous studies have been good (.77–.92: Sepulveda et al., 2009).

Generalised anxiety disorder questionnaire (GAD-7)

The GAD-7 is a 7-item self-reported measure of symptoms of generalised anxiety (Spitzer et al., 2006). Items are scored on a 4-point Likert scale (0–3) and are summed for a total score, with higher scores indicating greater anxiety symptoms. The scale has been shown to have good psychometric properties and excellent internal consistency (Cronbach's alpha = .94; Spitzer et al., 2006).

Procedure

The EDDP is a multidisciplinary child and youth mental health team, including psychiatry, social work, nursing, psychology, occupational therapy, dietetics, paediatrics, teachers and music and art therapy. All mental health team members are trained and supervised in paediatric eating disorders and evidenced-based treatment models.

EDDP focuses on identifying and shifting the barriers to treatment through intensifying treatment and ultimately enabling the participant and family to return to community treatment to continue their recovery journey. Key treatment targets for adolescents include weight restoration plus reduced eating disorder, anxiety and depression symptoms. Key treatment targets for parents include an increase in self-efficacy, a decrease in accommodating and enabling behaviours and a reduction in anxiety and depression symptoms. Adolescents and their families attend EDDP for 10–13 weeks per term (matching the school terms) for no more than two successive terms.

Therapeutic model

The EDDP model is built around the principles of FBT-AN (Lock & Le Grange, 2012), emotion-focused skills training (EFST; Lafrance et al., 2020) and supported meal therapy (SMT). FBT-AN is an evidence-based therapy for adolescents with AN that aims to empower the family as the key to helping the young person recover from their eating disorder. EFST is a mode of emotion-focused family therapy, which has emerging evidence as a stand-alone or adjunct treatment for eating disorders (Osoro et al., 2022). EFST is based upon the theory that eating disorder behaviours emerge as maladaptive coping mechanisms for aversive emotions and aims to build parental skills to foster adaptive emotional regulation in their children. SMT is commonly noted as a component of eating disorder DPs and inpatient settings (Baudinet & Simic, 2021); however, descriptive literature and implementation evaluation is limited (Kells et al., 2013). SMT in EDDP is modelled on principles of FBT-AN, aiming to normalise participants' eating behaviours and produce weight gain/weight stability while ensuring the parent remains in control of decisions associated with food and their child.

Parental role in treatment

Parents are active participants in EDDP. They are involved in handover at both the beginning and the end of the day with a focus on food, mood and risk. Consistent with FBT, parents provide the food for the meals completed at EDDP, which is checked in at the start of the day. At the end of the day, clinicians provide the parents with feedback on progress. If a participant struggles to complete a specific meal, their parents are called and asked to problem-solve with the EDDP therapist or attend the EDDP immediately for the next meal to support their child to eat. Clinicians take a coaching role in this situation.

In addition, each participant and their family undergo a minimum of one FBT session a week as well as a family goal-setting and review session. Parents attend an EFST treatment group once a week for the duration of the EDDP engagement. All adolescents have psychiatric reviews to assess their suitability for medical treatment of their ED symptoms or comorbidities.

Adolescent role in treatment

In addition to the family components described above, adolescents engage in at least one group therapy session a day, including a distress tolerance group based on dialectical behaviour therapy principles (Linehan, 1993), a body image group based on the Butterfly Foundation's Body Kind Initiative (https://butterfly.org.au/get-involved/campaigns/bodykindschools/) and the Dove Self-Esteem Project (https://www.dove.com/au/dove-self-esteem-project.html). Individual therapy sessions are provided adjunct to FBT if required and focus on individual therapeutic goals to address barriers to treatment engagement. Participants attend school every day between therapy and meals, with teachers provided by Education Queensland.

Analysis

Data were analysed using SPSS version 28. Paired sample t-tests were used to investigate pre- and post-score differences on outcome variables. One adolescent had not completed the CAS-8, and three parents had not completed the post-treatment AESED. These were deleted pairwise. Mean difference scores were calculated to test for violations of the assumptions of outliers and normality. Three unique outliers were detected for the GAD-7, PHQ-9 and PvA that were more than 1.5 box lengths from the edge of the box in a box plot. Inspection of the values did not reveal them to be extreme, and as such, they were kept in the analysis. Assumptions of normality were violated for the GAD scale as assessed using the Shapiro–Wilk test. Bias-corrected bootstrap confidence intervals (n = 5,000, confidence intervals set at 95 per cent) were used to account for normality violations and assess the significance of t-test comparisons.

RESULTS

Participants

Thirteen participants started and completed one or two terms of the DP and provided pre- and post-intervention data. Hospital admission data showed that in the three years prior to starting the DP, eight of the participants had been admitted to medical or mental health hospital wards, for a total of 23 admissions, totalling 255 nights in hospital (M = 11.30, SD = 7.04). Two participants had admissions during their DP treatment, totalling 14 nights in hospital (M = 7, SD = 8.48). In the period up to 15 months post-completion of the DP, one participant had three further hospital admissions, totalling 22 nights in hospital (M = 7.33, SD = 2.51).

Treatment effect

Outcome variable scores for the adolescents are shown in Table 1. Pre-treatment scores show that the mean participant score was above the clinical cut-points for the EDE-Q, CIA, PHQ-9 and CAS scales. Length of stay was not included as a covariate due to the small sample size. When running the analysis as a two-group ANCOVA (IV = pre/post, Cov = 1 term vs. 2), the effect of treatment held for the main variables (EDE, CIA, depression) with no difference to the interpretation of the results. The effect did not hold for BMI centile with the length of DP (terms) controlled for. This analysis likely needs a bigger sample to determine if this is specific to this sample or a general result across the treatment.

The reliable change index (Jacobson & Truax, 1991) for the EDE-Q and CIA were calculated using Cronbach's alpha of normative data published using adolescent samples (Forsén Mantilla et al., 2017; Mond et al., 2014; Reas et al., 2010; Wade et al., 2008) and the means and standard deviations of the pre-test data from the current sample. Using this method, reliable change (RC) was calculated to be indicated by changes of 0.95 for the EDE-Q and 8.53 for the CIA. For the PHQ-9, a reliable change index of 7, as recommended by Liu and Adrian (2019), was used. The proportions of participants who were unchanged, reliably deteriorated and reliably improved are shown in Table 2.

Using the method recommended by Jacobson and Truax (1991), a clinically significant change (CSC) cut-off score of 2.82 for the EDE-Q was calculated using the weighted midpoint between the means from the sample data from this study and normative nonpatient data (Mond et al., 2014). For the CIA and PHQ-9, the cut-points of 16 and ten, respectively, were used as recommended by Bohn et al. (2008) and Liu and Adrian (2019). CSC was defined as those who had scored above the clinical cut-point before the treatment, achieved reliable change based on RC calculations and finished the treatment with scores below the clinical cut-point (Jacobson & Truax, 1991). The proportions of participants exceeding the cut-points pre-treatment and achieving RC and/or CSC are shown in Table 2. The outcome variable scores for the parents are shown in Table 3.

DISCUSSION

This study investigated the outcomes for both adolescents and parents attending a DP treatment for eating disorders. Consistent with previous studies (Baudinet & Simic, 2021), the cohort exhibited weight gain plus a significant reduction in eating disorder symptoms, eating disorder psychosocial impacts and depression across DP treatment. The reliable/clinically significant change proportions showed that approximately half the adolescent participants demonstrated clinically significant change, which is similar to outpatient FBT outcomes (Lock & Le Grange, 2019). These results indicate that adolescents and families not progressing in outpatient FBT can progress towards recovery in intensive treatment programmes, such as the current study's DP. However, the current study also found that approximately half of the participants showed no reliable change, thereby requiring further research to understand the factors contributing to success or the lack of success in DP. Understanding these factors could inform selected allocation to DP and DP treatment modification. For those who did not respond, it may be that DP treatment did not adequately address their core ED symptoms, comorbidity/ies or the psychosocial factors that maintain their eating disorder. It may be that additional or different methods of addressing these factors may be required for this cohort. Alternatively, it may be that the DP treatment elements could address these factors, but there were barriers to adequate engagement or poor implementation of these strategies, which prevented progress. Examples of potential barriers to engagement or poor implementation could be a lack of motivation to utilise strategies or take an active role in treatment, a lack of understanding/confidence/support in implementing strategies or a failure to apply and generalise strategies in a broader context (i.e. outside the DP environment). Additional ways of addressing barriers to engagement may be required to improve outcomes for this cohort.

Further, contrary to the hypothesis, anxiety scores did not significantly change across treatment. It could be that the CAS-8 measures trait rather than state anxiety, which has previously been shown to be associated with eating disorders and does not exhibit change across successful eating disorder treatment (Harrison et al., 2016; Wilson et al., 2019).

Additionally, two participants did not score above the clinical cut-points for eating disorder psychopathology at assessment despite having a diagnosed eating disorder and having been accepted for DP treatment. This could be attributed to minimisation and/or a lack of insight into their condition, which is a common feature of eating disorders (Schoen et al., 2012). Conclusions about the impact of treatment are limited for these participants, and alternate methods may be required to evaluate outcomes (e.g. parent report, qualitative methods).

As hypothesised, the parents demonstrated increased self-efficacy, reduced eating disorder accommodating behaviours and improvements in their own mental health. This is an important and novel finding, as parental factors (especially depression and anxiety) have not been comprehensively studied in DP treatments. While these results demonstrate the impact of treatment on these variables, the temporal associations between them cannot be established. For example, it may be that DP treatment increases self-efficacy, which results in reducing accommodating behaviour. This may result in improved treatment response for the adolescent's eating disorder, which may, in turn, result in reduced parental depression and anxiety. Alternatively, it may be that parental self-efficacy and/or accommodating behaviour is negatively influenced by parental depression and anxiety, which, if addressed through DP treatment, results in increased self-efficacy and/or accommodating behaviour. Such a premise is supported by Monteleone et al.'s study (2022), which highlighted parental depression as a strong maintaining factor in adolescent ED. More research is required to understand the likely complex relationship between parental and adolescent variables and treatment outcomes.

Limitations and future directions for research

This current study is limited by its small sample size and the lack of a control condition or follow-up data. These factors significantly limit the conclusions that can be drawn from this data. Future research could overcome these limitations by replication with a larger sample size and the inclusion of control conditions (e.g. active or wait-list control) would be necessary to conclude that treatment effects were attributable to the DP treatment. Follow-up data would show if the treatment effects were maintained.

This study gives rise to a number of research questions. First, while it is established that DP treatment works, there is little known about what components of the treatment model influence recovery. Future research could investigate the mediators of DP treatment to determine which elements are active and which are potentially redundant. Second, it appears that approximately half of the adolescents did not respond to the DP programme. Little is known about the psychological, demographic and family characteristics of this cohort, which prevents hypothesising about barriers to recovery. Future research could investigate the differences in characteristics (e.g. young people, family environment, parental efficacy) between those who achieve recovery versus those who do not, and/or moderators of treatment outcome and alternative approaches for those who do not benefit from DP treatment (e.g. outpatient or intensive CBT-E).

Finally, while parents and caregivers are critical to recovery from paediatric eating disorders, the specific role they play in recovery is unclear. This small study highlights the positive change in parents engaged in the DP. It could be that parent positive parental change increases the likelihood of the adolescents' recovery in DP treatment; however, further research is required to establish this relationship.

CONCLUSION

The current study supports recent investigations showing DP treatment to effectively improve core eating disorder outcomes and comorbidity. It is the first study to show that parental depression and anxiety scores decrease across family-based DP treatment and extends previous research by showing that parents' self-efficacy, accommodating and enabling behaviours, depression and anxiety are all improved through this approach. Such results provide the rationale for future research to further investigate these variables to learn more about the mediators and moderators of treatment outcomes.