Family-based treatment takes longer for adolescents with mental health comorbidities: findings from a community mental health service

ABSTRACT

Children and adolescents diagnosed with an eating disorder often meet the diagnosis of another mental health disorder. In addition to eating disorders, individuals with comorbid disorders have higher suicide rates and more severe and chronic eating disorder symptoms. The present research aimed to investigate the influence of comorbid conditions on the treatment outcomes of children and adolescents that attended a public community mental health service. It was hypothesised that the patients with comorbidities would have a more extended treatment duration, slower rates of weight restoration, more hospital admissions for medical compromise, and poorer functioning than those without comorbidities. Data from 78 past patients at the Eating Disorder Program in Queensland, Australia, were analysed. Patients with comorbidities demonstrated similar recovery rates to those without comorbidities. However, those with comorbid conditions had longer episodes of treatment. The study’s results support using Family Based Treatment for patients with and without comorbidities. The implications of the findings for public mental health services and directions for future research are discussed.

Clinical Implications

• Adolescents with eating disorders often present with comorbid mental health disorders. These comorbid disorders may impact treatment outcomes and require more prolonged treatment durations.

• The findings of this study support the assertion that Family Based Treatment (FBT) is an effective intervention for this population. Clinicians should persist with using FBT for patients with Anorexia Nervosa (AN) and comorbid disorders, despite potential delays in treatment gains.

• The use of adjunct interventions to FBT will require consideration of the nature and severity of comorbid conditions, the presence of psychosocial stressors and the presence of available healthcare resources. Further research on these factors' influence on treatment delays in FBT is required.

The peak age of onset for eating disorders ranges between 15 and 19 years (Hay et al., 2014), with epidemiological research finding an increasing prevalence of eating disorders with age (Lawrence et al., 2015). Studies have consistently found anxiety and mood disorders to be the most common comorbid conditions in patients with eating disorders (Holm Denoma et al., 2014; Hughes et al., 2013). Approximately 60% of adolescent patients with Anorexia Nervosa (AN) also have a mood disorder diagnosis (Bühren et al., 2014; Fennig & Hadas, 2010). The secondary effects of starvation may contribute to a depressive-like state or even to the development of a major depressive disorder (Hughes et al., 2013), with lack of appetite, lethargy, poor concentration, emotional lability, guilt, and social withdrawal shared characteristics of both eating disorders, starvation syndrome, and depression (American Psychiatric Association, 2013; Kim et al., 2010). Additionally, it was found that 25% of adolescents with an eating disorder also meet the criteria for an anxiety disorder during treatment (Swanson et al., 2011).

Family Based Treatment (FBT) utilises the family system to encourage behavioural change in the patient (I. Wagner et al., 2016; Lock et al., 2010). The treatment empowers parents to assume authority over the eating disorder until the child or adolescent can reassume control over their recovery (Mairs & Nicholls, 2016). FBT comprises 15 to 20 sessions divided into three phases, typically delivered over 12 months (I. Wagner et al., 2016; Lucka, 2006). The course of therapy can be extended past 12 months for patients with highly rigid thoughts and behaviours or those with comorbid disorders (Mairs & Nicholls, 2016). FBT has had promising short-term and long-term treatment outcomes for children and adolescents with AN, with approximately 50–70% of patients returning to a healthy weight by the end of treatment (Mairs & Nicholls, 2016; Rosen, 2010). It has also been found to have low relapse rates (Herpertz et al., 2011). Whilst research strongly supports the efficacy of FBT for treating adolescents with AN and BN, a significant proportion of patients drop out prematurely or do not respond to treatment (Lock et al., 2010). Research has sought to further understand this cohort, with comorbid diagnosis thought to be a variable that may explain treatment non-response.

Mental health comorbidities are essential to address during treatment. Studies have found that up to half of all children and adolescents seeking FBT have other mental health conditions (e.g. Le Grange et al., 2008). Research has shown that treatment outcome is predicted by fewer comorbidities and lower levels of depression (Silén et al., 2015; Vall & Wade, 2015). Furthermore, compared to adolescents with AN and no comorbid diagnosis, adolescents with AN and mental health comorbidity were found to have longer treatment duration, higher rates of hospitalisations, and more costly treatments (Blinder et al., 2006; Silén et al., 2015). One possible explanation is that comorbidities interfere with the treatment process. For example, social anxiety may prevent engagement in treatment due to fear of negative evaluation and avoidance of interaction with others (Gowers et al., 1999). Alternatively, comorbidities may be a marker of greater severity of illness and complexity in presentation, which may independently predict poorer outcomes (Lock et al., 2006). There is evidence that individuals with eating disorders and comorbid depression or anxiety are more likely to have chronic eating disorder symptoms and poorer health than those without comorbidities. They also have poorer financial functioning and higher mortality rates than those without comorbid conditions (Keski‐rahkonen et al., 2014).

Several studies have examined the association of comorbid diagnoses with treatment outcomes in FBT. Lock and colleagues (2005) compared short-term FBT (10 sessions over six months) to long-term FBT (20 sessions over 12 months). Eighty-six adolescents with AN were randomly assigned to either treatment conditions or evaluated at the end of one year (Lock et al., 2005). Results found that participants with more severe obsessive-compulsive presentations required a longer-term FBT program to achieve outcomes similar to those without comorbidities (Lock et al., 2005). Another study found that lower baseline levels of comorbid major depressive disorder were associated with lower remission rates for patients with BN treated with FBT (Le Grange et al., 2016). These results suggest that comorbid conditions add complexity to FBT treatment and lengthen the duration required to restore healthy body weight and reduce eating disorder symptoms. However, discrepant results have been found. Le Grange and colleagues (2012) found no relationship between comorbidities and treatment outcomes in adolescents with AN treated with FBT (52). Trainor et al. (2020) recently investigated the FBT treatment outcomes of 107 adolescents with an eating disorder and comorbid mental health diagnosis. Results showed gains in body weight and reduced eating disorder symptoms amongst those with comorbid conditions. It was also found that rates of comorbid mood and anxiety disorders reduced from 54% to 26% at the end of FBT (Trainor et al., 2020).

These conflicting findings may be attributable to differences in the studies’ sample characteristics (e.g., age and onset of disorders) and methodology (e.g., definitions of treatment outcomes). Research trials in non-naturalistic settings may be skewed due to the participant selection criteria. Such criteria may lead to the recruitment of patients with fewer comorbidities and reduced severity of symptoms (Stiles‐shields et al., 2013; Stirman et al., 2003). It was found that clinic patients in naturalistic, public health settings had higher levels of functional impairment, more depressive symptomatology, and were more likely to have mental health comorbidity than those in research trials (Swanson et al., 2011; Swinbourne et al., 2012). This discrepancy was supported by findings that participants with more severe baseline depression or anxiety, more comorbid conduction and more significant life stressors were more likely to be treated in a clinical setting than a research setting (Aldao et al., 2016; Baker-Ericzén et al., 2010). Studies have also found that patients with eating disorders and mental health comorbidities in clinic-based settings have slower recovery and higher drop-out rates in FBT than in research settings (Forsberg & Lock, 2015; Lock et al., 2006).

Such findings highlight the need for further understanding the efficacy of interventions in naturalistic treatment contexts and clinic-based populations. Specifically, there is a need to understand the influence of clinical characteristics, such as mental health comorbidities, on treatment outcomes for FBT in these settings. Understanding factors impacting treatment outcomes may help identify those at most risk for non-response and improve recovery rates in early intervention to reduce the life-long burden of eating disorders. This is particularly important given the large number of families who seek treatment for eating disorders through these services and the wide dissemination and use of treatments such as FBT amongst this population of children and adolescents.

This study aims to evaluate the impact of comorbidities on FBT among children and adolescents with eating disorders who present to an Australian community mental health service. Specifically, the study aimed to compare patients with and without a comorbid mental health disorder in terms of their profile of presenting symptoms, treatment engagement, and treatment outcome. Based on the literature reviewed, it was hypothesised that compared with those without comorbidity, those with comorbid mental health disorders would have:

• Longer community treatment episodes are measured by the length of treatment (days).

• A higher frequency of hospital admissions during their treatment episode.

• Lower ideal body weight at initial assessment and a slower rate of weight gain, as measured by the patient’s Body Mass Index (BMI).

• Poorer recovery in terms of general mental health and social functioning as measured by scores on the Health of the Nation Scale—Child and Adolescents (HoNOSCA).

• Poorer recovery in terms of daily functioning, as measured by scores on the Children’s Global Assessment Scale (CGAS).

Method

Community treatment setting

The Eating Disorders Program (EDP), based in Queensland, Australia, is a publicly funded specialist eating disorder service for children and adolescents. The EDP provides assessment and treatment for children and adolescents between zero and 18 years with eating disorders and other comorbidities (Queensland Government, 2019). The EDP is staffed by a multidisciplinary team who offers medical and psychological healthcare services. Clinicians at the EDP are trained in various psychotherapies (e.g. FBT, Cognitive Behaviour Therapy—Enhanced (CBT-E), Dialectical Behaviour Therapy for Adolescents (DBT-A), Emotion Focused Therapy (EFT)). FBT is the most utilised treatment modality at the clinic (Queensland Government, 2019). Individual therapy in other modalities were provided as part of Phase 2 interventions in FBT to address comorbid concerns. The EDP’s clinical process entails routine physical and mental health monitoring. The psychiatrists and nurse practitioners in the service prescribed and monitored psychiatric medications for comorbid mental health concerns. The psychiatrists conduct assessments with all the clients at the service and confirm psychiatric diagnoses in consultation with the treating clinicians at the EDP. Diagnoses are periodically reviewed as part of a three monthly case-review meeting with all EDP clinicians. The staff at the EDP use a secure, electronic database to record patients’ demographic and clinical information as part of routine service provision. The electronic database provides clinicians with information regarding hospitalisations and clinical outcomes relating to BMI and client and clinician-rated measures such as the CGAS and HONOSCA. Each time a patient is referred to the EDP, a new treatment episode is created in the electronic database. The treatment episode is closed when the patient is discharged from the EDP.

Participants

Archival data over two years was accessed and de-identified from 78 past patients aged between eight and 18 years (M = 14.21 years, SD = 1.89) who were at the EDP service between March 2017 to March 2019. Much of the sample was female (84.6% female, 15.4% male). Patients included in the study were those offered and agreed to participate in FBT as their primary treatment at the EDP. Those patients who declined FBT or were not provided FBT and/or participated in another evidence-based treatment (e.g., enhanced Cognitive Behaviour Therapy: CBT-E) were not included in the sample. Table 1 lists the sample’s frequency of eating disorder diagnoses (AN, BN, Other Specified Feeding and Eating Disorders [OSFED], atypical anorexia). Of the 78 patients, 46 (60.3%) had a comorbid diagnosis. This rate of comorbidity appears to be higher than other samples reported in the literature. For example, Le Grange et al. (2008) found that 45.1% of children and adolescents presenting with comorbid psychiatric disorders in outpatient services. As displayed in Table 2, the comorbid diagnoses were categorised into main disorder categories as per the DSM-5 (American Psychiatric Association, 2013). Most patients presented with anxiety-related disorders. Please note that patients with multiple comorbid conditions (i.e. more than one comorbid disorder) were transferred to another specialist mental health services designed for severe and complex presentation. As such, these patients were not part of the current sample.

Table 1. Frequency of eating disorder diagnoses.

	Frequency	Percentage of Sample (%)
Anorexia nervosa	64	82.1
Bulimia nervosa	4	5.1
OSFED	7	9.0
Atypical anorexia	3	3.8

Note. N = 78. OSFED = Other Specified Feeding and Eating Disorder.

Table 2. Frequency of comorbid diagnoses categorised into main disorder types.

	Frequency	Percentage of Sample (%)
Anxiety related disorders	27	34.6
OCD	10	12.8
Depression related disorders	5	6.4
ASD	2	2.6
Mixed depressive and anxiety disorder	2	2.6
Attention deficit and hyperactivity disorder	1	1.3

Note. N = 47 cases who presented with a comorbid diagnosis.

Measures

Sociodemographic information and clinical characteristics

Sociodemographic information included age at the start of the patient’s service episode and gender. The patients’ primary eating disorder diagnoses were collected, and if present, diagnosed comorbid mental health disorders were also collected.

Service use

Information regarding service use collected included: the start and end date of each EDP service episode, duration of each service episode, psychological treatment type (all participants in the study were provided FBT), and the number of hospitalisations. This information was measured, collected, and reviewed throughout the treatment process at the beginning, every three months, and at discharge. Regarding the duration of service episodes, all patients maintained a service episode with the EDP even if they were hospitalised. None of the patients in the sample re-presented to the EDP throughout the data collection period.

Physical health

Body mass index (BMI)

The BMI measures an individual’s weight relative to their height. BMI is calculated by dividing the individual’s weight (in kilograms) by the square of their height (in metres). BMI weight ranges determine if the individual is underweight, normal weight, overweight, or obese for their age and sex.

Psychological health

Routine Outcome Measures (ROM) were assessed following the National Outcomes and Casemix Collection recommendations for child and youth mental health services (AMHOCN, 2009), which requires regular scoring of the Health of the Nation Outcome Scales Child and Adolescent Mental Health (HoNOSCA), Children’s Global Assessment Scale (CGAS), and Strengths and Difficulties Questionnaire (SDQ). Clinician-rated ROM (CGAS and HoNOSCA) were completed by the health care professional responsible for the child, including child and adolescent psychiatrists, psychologists, social workers, or occupational therapists. All clinicians were offered training to rate these measures. Data from client reported SDQs were not utilised as response rates were poor and the number of valid completed items on the measure was low. This represents a limitation of the study and is discussed further in the Discussion section.

Health of the nation outcome scales – child and adolescent (HoNOSCA)

The HoNOSCA is a 15-item clinician-rated measurement designed to assess child and adolescent outcomes in mental health services. It includes four subscales: behavioural (BEH), symptomatic (SYP), social problems (SOC), and impairment (IMP) and a total score based on questions 1 to 13 (Gowers et al., 1999). Questions 14 and 15 were excluded because they pertained to parental knowledge about the children’s difficulties and available services (Pirkis, Burgess, Kirk, Dodson, Coombs, & Williamson, 2005). Overall mental health and social functioning are calculated by the sum of the 13 scores (Gowers et al., 1999). Questions 14 and 15 were excluded as they did not assess the child’s functioning. The HoNOSCA is a well-validated measure with acceptable reliability (α = 0.65) and is sensitive to change (Gowers et al., 1999; Yuan, 2015).

Children’s global assessment scale (CGAS)

The CGAS is a clinician-rated scale that provides an overview of the patient’s global level of adjustment (Shaffer et al., 1983). Patients’ daily functioning in various areas is assessed at home, school, or peers (Shaffer et al., 1983). The clinician provides a numeric rating from 0–100 based on their observations from the past month (Shaffer et al., 1983). Higher scores suggest a higher level of functioning, with scores greater than 70 indicating no significant functional impairment (A. Wagner et al., 2007; Shaffer et al., 1983). The CGAS is a reliable and valid measure sensitive to change across time (Steinhausen, 1987).

Data collection

Ethical approval for the study was sought from the University Ethics Committee (HREC: H20REA019) and the Queensland Health Ethics Committee (Ethical approval ID: HREC/17/QRCH/321). Data for the present research was extracted from a larger data set of Child and Youth Mental Health patients across the health district. This larger pool of data was collated from an electronic records database—the Consumer Integrative Mental Health Application (CIMHA) - that contains routine service provision and outcome data about clients accessing public mental health services. A chart auditing tool was developed to standardise data collection across different researchers in the team. The final data set generated was cross-checked and verified by the senior clinicians and researchers within the research team.

Statistical analyses

All analyses were conducted with SPSS version 25 (IBM, 2017). Independent samples t-tests were used to examine significant differences between service usage outcomes for those with and without comorbidities. Post-hoc analyses (Hedges g) were used to determine the effect size. Two-way mixed ANOVAs were used to analyse each group’s change in BMI, CGAS, and HoNOSCA scores across the two-time points: assessment and discharge. Post-hoc analyses (partial eta squared) were then conducted to explore the impact of comorbidity on physical and psychological outcome scores.

Results

Table 3 displays the means and standard deviations for the two groups for BMI and scores on the CGAS and HoNOSCA at admission and discharge.

Table 3. Descriptive statistics for body mass index (BMI) scores at assessment and discharge.

	Assessment	Discharge
Group with comorbidities (N = 46)	17.69 (3.19)	21.80 (2.81)
Group without comorbidities (N = 31)	17.92 (2.82)	21.30 (2.99)
Total (N = 77)	17.78 (3.03)	21.69 (2.88)

The results of the two-way mixed ANOVA showed that there was no significant main effect for Comorbidity [F(1,75) = 0.04, p = .84, ηp2 = .001] on BMI scores at assessment, with the group with comorbidities and the group without having similar baseline BMI scores (Table 3). Similarly, no main effects were observed for comorbidity for scores on the CGAS [F(1, 67) = 0.02, p = .88, ηp2 < 0.001] and all the HoNOSCA sub-scales. A significant main effect of time was found on BMI scores [F(1,75) = 225.36, p = < 0.001, ηp2 = 0.75], with patients showing a significant difference in BMI scores between assessment and discharge (Table 3). Main effect of Time was also found on CGAS scores [F(1,67) = 68.28, p < .001, ηp2 = 0.51], with patients showing a significantly higher CGAS score at discharge (M = 64.00) compared to assessment (M = 48.58; Table 3). Similarly, gains were observed across all the HoNOSCA sub-scale scores, with a main effect for Time on Total Score [F(1, 29) = 42.34, p < .001, ηp2 = 0.59]; behavioural problems [F(1, 56) = 11.15, p = .001, ηp2 = 0.17]; impairment [F(1, 62) = 34.87, p < .001, ηp2 = 0.36]; symptomatic problems [F(1, 39) = 16.80, p < .001, ηp2 = 0.30] and social problems [F(1, 62) = 50.38, p < .001, ηp2 = 0.45]. No significant interactions between Comorbidity and Time were found on BMI, CGAS or HoNOSCA scores.

The average treatment episode length for the overall sample was 382 days (SD = 223). There was a high rate of variance among the sample. The minimum treatment episode length was 41 days, with the longest treatment episode length being 1050 days. An independent samples t-test was conducted to compare the average treatment episode length in the group without comorbidities (N = 32) and those with comorbidities (N = 46). There was a significant difference in the scores for the group without comorbidities (M = 288 days; SD = 180) and the group with comorbidities (M = 446; SD = 220), t(75) = −3.23, p < .01, with a medium effect size (Hedges g = 0.75). Given the large range in lengths of the treatment episodes, the t-tests were conducted on the sample without outliers (N = 3; treatment episode length over 900 days). The t-tests found a significant difference between the groups in this sample t(72) = −1.18, p < .01. The average number of hospitalisations during treatment for the overall sample was one admission (SD = 0.77). Table 4 displays the frequency and descriptive statistics related to the number of hospitalisations for the overall sample and each group. An independent samples t-test found no significant difference between the group with comorbidity and the group without comorbidity’s number of hospital admissions, t (76) = −1.11, p = .27.

Table 4. Descriptive statistics for children’s global assessment scale (CGAS) scores at assessment and discharge.

	Assessment	Discharge
Group with comorbidities (N = 39)	49.08 (9.30)	63.21 (14.36)
Group without comorbidities (N = 30)	47.93 (9.41)	65.03 (14.61)
Total (N = 69)	48.58 (9.36)	64.00 (14.39)

Discussion

The current study aimed to compare patients with and without mental health comorbidities with regard to their severity of ED symptoms, treatment engagement, and treatment outcomes. The group with comorbidities were found to have a longer duration of treatment in comparison to the group without comorbidities. This hypothesis was supported, as it was found that the group with comorbidities had a significantly longer treatment period when compared to the group without comorbidities. The prolonged treatment duration for patients with comorbidities may be attributed to the longer time these patients may spend in each phase of FBT. This outcome is consistent with previous literature, which found that patients with comorbidities have improved treatment outcomes when participating in FBT delivered over a longer period (Lock et al., 2005). Such findings suggest that these patients require a more prolonged treatment duration for remission (Lock et al., 2005; Mairs & Nicholls, 2016).

Overall, the treatment length for both groups was longer than the duration of FBT for individuals without comorbidities reported in research trials. As the EDP is a tertiary mental health service, referrals are often more complex and severe than clients with less severe presentations (Aldao et al., 2016; Baker-Ericzén et al., 2010). This complexity may also predispose the patient group to have higher subthreshold levels of comorbid disorders, as it is common for patients with eating disorders to have subthreshold anxious and depressive symptomology (N. Godart et al., 2015; N. T. Godart et al., 2000). While the patients with subthreshold comorbidities may not meet the criteria for clinical diagnosis, these patients’ treatment durations may be longer than the standard treatment duration for the group without comorbidities in this study. This finding may further support that patients without formally diagnosed comorbidities in community treatment settings have more severe presentations than in research trials. It is also possible that such differences may be attributable to variances in treatment delivery and fidelity. This may need to be further investigated in future studies. In addition, future research may investigate differences in the number of treatment sessions provided. The time linked to other forms of service provision (e.g. phone calls to the family, meetings with school staff) linked to each group.

The second hypothesis predicted that the group with comorbidities would have a higher hospital admissions frequency than those without comorbidities during their treatment episode. Overall, those with comorbidities had more hospitalisations (N = 47) than those without comorbidities (N = 31). However, this difference was not statistically significant. These differences may be statistically significant in a larger sample of participants. This finding does not support previous literature, which has found that patients with comorbidities are more likely to be hospitalised during their treatment (Blinder et al., 2006). It is worth noting that the hospitalisation rates in the sample (61.53%) were higher than those reported in research trials (e.g. 12.2% to 17.9% in Le Grange et al., 2008). Further information regarding the reasons for hospitalisation and the influence of comorbid symptoms (e.g., suicidal ideation linked to low mood) would be worthy of investigation. Research into the cost-effectiveness and burden on families of these hospitalisations would also be helpful.

Patients with and without comorbidities appear to have had similar gains in BMI, general mental health (as measured by the HoNOSCA Total Score) and psychosocial functioning (as measured by the CGAS). These findings do not support the hypotheses regarding those with comorbidities making less progress in FBT. While these findings are similar to findings from research trials (Hay et al., 2014; I. Wagner et al., 2016; Lock et al., 2010; Mairs & Nicholls, 2016; National Health and Medical Research Council, 2009), it is also possible that the comorbidities present in the research sample may not have been significantly interfering with the FBT treatment. Further assessment of the severity and impact of these comorbid conditions may shed light on this hypothesis. Further research is required regarding the implementation of FBT, rates of drop-out and barriers to treatment engagement in such naturalistic settings.

Both groups did not score highly on the HoNOSCA Behavioural Problems subscale. However, they did demonstrate a significant small improvement in these scores between the assessment and discharge. The overall sample’s low scores are most likely attributable to the internalising nature of eating disorders and their commonly associated comorbidities, such as anxiety and depressive disorders (Forbush et al., 2017). While adults with eating disorders have been found to have high rates of self-harm and substance use (Guillaume et al., 2011), research has found that adolescent samples have considerably lower rates than their adult counterparts (Brann et al., 2001). This is attributed to the reduced chronicity of eating disorders during adolescence instead of eating disorders in adulthood (Bühren et al., 2012). Another possible explanation for the overall sample’s low Behavioural Problems subscale score may be the small portion of patients in the sample with BN and binge-eating disorder. Individuals with binging and purging symptoms display higher impulsivity, which has been linked to higher rates of self-harm and substance use (Peebles et al., 2011).

There was a significant, moderate improvement in both groups’ HoNOSCA Impairment, Symptomatic Problems and Social Problems subscale scores, with mean scores on these subscales falling in the non-clinical range at discharge. Similar findings have been reported in studies using the HoNOSCA with adolescents presenting with severe and complex mental health concerns (Shaffer et al., 1983). Once again, the hypothesis regarding differences between the two groups was not supported, with patients reporting similar scores across these subscales. As the HoNOSCA is a clinician-rated measure, there may be measurement issues relating to the scores’ reliability. For example, Hunt & Wheatley (Herpertz Dahlmann, 2009) highlight the risk of ‘drift’ in clinician ratings on the HoNOSCA, where raters tended to underscore items due to sensitisation to severe clients over prolonged treatment periods. Severely unwell patients may be scored at a lower level than is indicated by the scoring criteria. Similarly, raters may be aware of improvements in particular domains and wish to score the item lower than previously or at a level less than the maximum. Such measurement issues and the feasibility and appropriateness of using the HoNOSCA with patients with eating disorders may require further investigation.

Limitations of the study

Whilst it was beneficial for the study to utilise data from a naturalistic sample, this constraint was the small sample size. With regard to the measure of mental health, it would be beneficial to use standardised instruments to assess EDs and comorbid disorders. In addition to the standard outcome measures utilised in community-based mental health services (e.g., CGAS, HoNOSCA), measures of eating disorder symptoms may be beneficial. For example, the Eating Disorders Examination Questionnaire (EDE-Q; Fairburn & Beglin, 1994; Mond et al., 2016) is a well-validated and reliable self-report measure commonly used to assess eating disorder psychopathology. Data collected from the EDE-Q may provide more detail regarding the severity and chronicity of eating disorder symptoms and how these symptoms present throughout treatment. While the CGAS and HoNOSCA give a general measure of mental health and functioning, using specific criteria of comorbid disorders (e.g., Beck Youth Inventories, Beck et al., 2001) may be helpful. Improving response rates on ROM client-rated measures, such as the Strengths and Difficulties Questionnaire (Goodman, 1997), would have provided insight into the patient’s self-reported strengths and difficulties throughout the treatment process. While the gains in mental health concerns in the participants point to the effective implementation of FBT, measures to ensure the use of FBT to fidelity (such as a fidelity monitoring tool; Thompson-Brenner et al., 2013) may have been beneficial. It is possible that the longer treatment episode length found amongst participants with comorbid conditions may be attributed to clinicians moving their treatment focus away from FBT-related tasks, to focus on comorbid conditions and life challenges instead.

Strengths, implications, and future directions

A major strength of this study was the use of data from patients who had attended an Australian community-based treatment setting. This study is one of the first Australian studies to investigate the efficacy of FBT when applied within a community-based treatment setting and to explore the impacts of added comorbidities on treatment outcomes.

The findings of this study have implications for child and adolescent eating disorder services. In clinical practice, delays in treatment gains may prompt clinicians to review treatment plans and cease FBT to focus on comorbid symptoms instead. For FBT clinicians, the findings of this study, in combination with previous research, support the assertion that FBT is an effective intervention for patients with AN and comorbid disorders. The findings highlight the need for clinicians to persist with the use of FBT given the additional dose of intervention that is required for this population. In addition, previous research suggests that treatment gains in AN during FBT correspond to remediations of symptoms of comorbid disorders, without directly targeting comorbid symptoms (Valenzuela et al., 2017). Given the findings of equivalent outcomes in AN patients with and without comorbidities, FBT may take longer in achieving gains with certain groups of adolescents.

Patients with comorbid disorders often present with psychosocial complexities, such as lack of social support, family members with mental health concerns and poverty. Such factors are likely to contribute to the extended timeframes. It may also require additional clinicians to engage in case management, multi-systemic coordination, and other ecological systems approaches that supports the family, and support the FBT clinicians to persist with treatment for a longer period. Such factors are often captured in the patient’s ICD-11 (Weigel et al., 2019) Z-codes. Z-codes outline family and personal historical or circumstantial factors that may influence a patient’s mental health. An analysis of Z-codes would provide a broader understanding of factors affecting the patient, their presenting problems, and additional stressors that may influence treatment outcomes and duration of treatment.

Along with examining psychosocial factors, it would be important to uncover the length of time patients spend within each phase of FBT and the profile of co-morbidities and Z-codes associated with this. An understanding of these factors could guide recommendations on the use of FBT with priority populations, such as children with comorbid chronic illnesses and single-parent families. Equally, such information could support clinicians to provide more realistic expectations for patients, their families and other relevant stakeholders (e.g school staff) of treatment length and engagement (e.g. helping parents plan for taking leave from work to engage in treatment).

Psychosocial complexities in those with comorbid conditions may require additional healthcare resources to support and scaffold patients and their families—in addition to FBT. The findings of the study indicate that services can anticipate that this may be required for a longer duration than previously proposed. Extended FBT and adjunct services may require additional funding and resources. Given the rising costs of healthcare, further research is required into cost-effective and accessible models of services that can provide extended FBT and adjunct services.

Conclusion

Given the importance of early intervention in mitigating the effects of ED and the rising costs of mental health care, findings such as those from the present study may enable community mental health services to consider more cost-effective, targeted, efficacious ways to provide ED interventions. The current study found that FBT was an effective treatment for adolescent patients with eating disorders, both with and without mental health comorbidities, in a community mental health setting. Both groups were able to restore their weight to a healthy weight range. Notably, the group with comorbid mental health concerns required a significantly more extended treatment period to reach the same treatment outcome than the group without comorbidities. Future research with a larger sample may aim to understand the impact of individual and systemic factors on the duration of treatment and treatment outcomes and the impact of comorbidities on the length of phases of FBT.

Author contributions

JL and GK contributed to the conception and design of the study. JW designed the data mining protocol. JL, TW, SC, DW, PK, BR, CM. and GK wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

Ethical statement

Ethical approval for the study was sought from the University Ethics Committee (HREC: H20REA019) and the Queensland Health Ethics Committee (Ethical approval ID: HREC/17/QRCH/321).

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.