The model

The Sunshine Coast ED Access Trial was an initiative of the Australian Department of Health who provided funding to the Butterfly Foundation in partnership with a Primary Health Network (PHN) in regional Queensland to improve access to ED treatment. The aim was to investigate what effect introduction of expanded MBS items would have, and hence the model was embedded in the context of the MBS (https://www.anzaed.org.au/newmbsitems/) which specifies: that referral and review be provided by a medical practitioner; the inclusion of severe EDs only; the approved treatment modalities; and the number of treatment sessions provided. MBS items were not only available for psychological therapy but also sessions with a dietitian.

The Sunshine Coast is a periurban region of 1633 square kilometres with mixed urban and rural characteristics. The population of 346,648 is clustered in small communities, primarily along the coastal fringe of the region. The coastal area has a strong economy, particularly in health care, education, and tourism and an unemployment rate comparable to the national average at 4% in March 2022. The more rural hinterland has pockets of relative poverty with a higher rate of unemployment at 6.3% in March 2022. Public transport in the region is limited and the need for reliable transport is identified as a critical development issue.

The data examined in the current report were collected between February 2019 and 18 August 2021, a period of 30 months. The research was approved by the Bellberry Human Research Ethics Committee, Application No: 2018-09-728-FR-1. From November 2019, when MBS became available for patients with severe presentations of ED, the model focused on ED not meeting the MBS criteria, including earlier intervention for people and who were seeking treatment for their ED for the first time or were seeking treatment to manage the risk of relapse after previous treatment. This latter group tended to have lower levels of symptoms. We therefore refer throughout this report to two groups: one with clinically significant symptoms (baseline mean ED-15 cognitive scores ≥ 3.38; more than 1 SD above UK norms [13]), and/or at least one ED behaviour endorsed on the ED-15) and one with normative symptoms who did not meet this threshold.

Care coordination was a critical and unique aspect of the model, supporting the development of a cohort of trained and networked service providers as well as assisting patients to navigate the care system. This required proactive engagement of general practitioners who wished to refer people with EDs for psychological assessment and treatment, and service providers who had skills in delivery of ED assessment and treatment. A Care Coordinator with a nursing background was employed to liaise between the two groups, as well as to work with service providers to triage referrals, support the formation of treatment teams, match services to patient needs, and support practical problem solving as difficulties arose. The role also maintained contact with patients and was available to help resolve barriers that could lead to disengagement from treatment.

Service providers were required to undertake free training, comprising a 3-h introductory session with an emphasis on assessment and diagnosis and at least 10 h of training in a relevant specific treatment modality. Of the ten treatment modalities approved by the MBS, the model supported delivery of three: Cognitive Behaviour Therapy-Enhanced (CBT-E), Family Based Therapy, and Specialist Supportive Clinical Management. The latter two modalities became irrelevant from November 2019, when the MBS was launched for patients with severe presentations of ED, including all presentations of anorexia nervosa. This means our group with normative symptoms could have been treated with CBT-E or other EBI in which service providers had pre-existing expertise, including guided self-help CBT, Dialectical Behaviour Therapy, or Interpersonal Psychotherapy. Trainers from the local region and nearby metropolitan services provided the training, which was assessed for suitability in terms of the alignment of the National Eating Disorder Collaboration “National Framework for Eating Disorder Training—A guide for training providers”. Service providers were required to use key measures of patient progress which enabled ongoing professional development targeted to specific gaps in delivery of effective treatment delivered in real time.

For non-specialist clinicians, integrating a new treatment modality plus a new team-oriented way of working (which at the least involved the general practitioner and could involve a dietitian) into an already established and full schedule of work can be overwhelming. Therefore, the model also advocated four anchor points to support ongoing engagement of the clinician and the patient. Two related to increasing the likelihood of basic evidence-based practice, namely sessional weighing and progress monitoring, in this case using the ED-15 [13]. Two related to creating a supportive team environment, namely dietitian involvement and case conferencing. Service providers were paid a sessional fee when they participated in case conference meetings and the treatment team submitted a summary of team discussion as evidence that the meeting occurred.

ParticipantsService providers

Service providers (n = 72) included the following disciplines: Clinical Psychologists (n = 20), Registered Psychologists (n = 47), and Mental Health Social Workers (n = 5). On average, practitioners had 9.75 years of experience (SD 5.3; range: < 1–32) in mental health. Data were not collected on years of experience treating EDs.

Service users

The model relied on local health service providers to identify people with EDs, refer and/or provide treatment. Referrals were accepted from February 2019 to June 2021. Eligibility criteria for service users were (1) meeting the DSM-5 [10] diagnostic criteria for an ED, (2) residing in the health network area; (3) aged over 14 years; (4) referred by a GP or other recognised health professional; and (5) assessed as safe to receive community-based treatment by a medical professional and a mental health professional.

Design

The study design was a case series (no comparison group), where the primary outcome for patients was a short self-report measure of ED cognitions administered on a session-by-session basis. Secondary outcomes included body mass index (BMI) at baseline and end of treatment, and remission. Diagnosis by treating mental health practitioner, uptake of appointments with a dietitian, regularity of case conferencing by the treatment team, and use of in-session weighing in treatment sessions were also recorded. Data were drawn from clinician records which were deidentified for the evaluation component.

MeasuresBaseline demographics

Clinician diagnosis was supplied by mental health practitioners after assessment (usually over two sessions). Age, duration of illness, presence of comorbidities, socioeconomic status and previous psychological treatment for ED were drawn from patient data records.

Primary outcomes: eating disorder cognitions and behaviours

We used the ED-15 [13] as our primary outcome. The first ten items from the ED-15 assess ED cognitions over the preceding week (e.g., “Compared my body negatively with others”), rated on a 7-point Likert scale from 0 “not at all” to 6 “all the time” [6]. The last 5 items from the ED-15 assess ED behaviours using count responses, with participants reporting the frequency of objective binge eating, vomiting, laxative use, dietary restriction and driven exercise. Clinicians were instructed to collect the ED-15 measure at each session from service users, both to discuss individual results in session and to record outcome data. The ED-15 has been validated in clinical and non-clinical samples and demonstrated acceptable concurrent and convergent validity [14]. Cronbach’s alpha for the 10-item cognitive measure in the present study was 0.91 (item-total correlations > 0.41).

Secondary outcomes

BMI Height and weight were measured by the clinician at baseline; clinicians were asked to continue weighing at each session and this was examined as an outcome for patients who were underweight (BMI < 18.5).

Remission Remission status was based on attaining normative ED-15 cognitive scores (mean ≤ 3.38; within 1 SD of UK norms [13]) and absence of ED-15 behaviours. Last available observation was used if participants did not complete therapy.

Predictors of dropout and outcome

Fixed baseline moderators included: duration of illness, previous psychological treatment for an ED, presence of comorbidities, presence of a normative level of disordered eating and presence of any ED behaviours. Practice adherence predictors, recorded in service user data records, included regular weighing in treatment, regular case conferencing (meetings between treatment team members), and number of sessions provided by the dietitian. Case conferencing was eligible for rebates and suggested monthly, but frequency was at the discretion of the treatment team.

Statistical analysesDropout: percentage and predictors

Completion was defined as either (1) Patient and service provider(s) agreed that patient was ready for discharge from ED treatment, or (2) Patient discontinued treatment after ≥ 10 sessions with a decrease in ED reported in records. This combination represents those considered to have received a sufficient dose of treatment. Logistic regression was used to test predictors of treatment completion.

Sessional changes over treatment and predictors

Multilevel modelling (MLM; statistical software R version 4.0.3) was used to analyse change over time in sessional ED-15 scores (10 cognitive items). This approach estimates trajectories of change for all participants, and accommodates incomplete data, together with data collected at differing time points for individuals. Time was modelled as weeks from baseline assessment. The following sequence of models was used. Step 1: Null model fitted to calculate the intraclass correlation coefficient (ICC) and assess whether significant variance was present in the outcome. Step 2: Growth modelling to assess whether the average variance changed over time. Both linear and quadratic components were fitted. Step 3: Random slope modelling to test whether trajectories of change varied across individuals. Step 4: Testing whether an autoregressive error structure improved model fit. Step 5: Testing interactions to see if any of the following variables predicted the trajectories of change: weight classification (underweight = BMI < 18.5), duration of illness, previous psychological treatment for ED, presence of comorbidities (Yes/No), compliance with evidence-based practice (team-based approach—use of ≥ 3 sessions provided by a dietitian; use of ≥ 2 team case conferences; weighing at ≥ 50% of visits with mental health practitioner). Post hoc, mean ED-15 scores (10 cognitive items) were used to calculate within-group effect sizes (Cohen’s d) to allow comparison to clinical trials [15]. Analyses were undertaken for both treatment completers and intention-to-treat (ITT) samples.