Eating disorders (ED) in adults are notoriously difficult to treat. Although guidelines recommend several well-established treatments [1], it remains unclear which of these is most effective, and no treatment for anorexia nervosa (AN) currently meets the highest level of evidence [2] (ie, level I evidence according to the Oxford Centre for Evidence-Based Medicine criteria [3]). Currently, only about 50% of patients with bulimia nervosa (BN) and 40% of individuals with AN fully recover after 5 years [4], with relapse rates between 22% and 51% [5]. The highest relapse risk occurs 4 to 6 months after treatment [4]. The debilitating nature of ED makes these high relapse and low recovery rates particularly concerning [6].

At the same time, there is increasing recognition that EDs involve more than eating and weight-related symptoms alone. Evidence suggests that disturbances in self-functioning and interpersonal functioning play an important role in the development and maintenance of these disorders [7]. Historically, EDs have been closely linked to disturbances in identity and the self [8]. Difficulties such as low awareness of needs [9] and poor self-concept clarity [10], paired with high self-criticism and self-objectification, may increase vulnerability to eating pathology by fostering reliance on external validation in defining identity and worth [9,11] and the need to control food and the body as a way to experience autonomy, competence, and identity [8].

Thus, improving treatment to aid transdiagnostic recovery, societal rehabilitation, and relapse prevention is crucial. Given the associations between autonomy-connectedness and ED symptoms and comorbidity, autonomy-connectedness could be a useful addition to therapy [12-14]. Autonomy-connectedness refers to the capacity to identify, express, and pursue one’s own needs, wishes, and values while maintaining meaningful and reciprocal relationships with others. It reflects both self-direction and interpersonal connectedness, capturing the degree to which individuals can identify and strive for their own needs, considering those of others. Autonomy-connectedness consists of 3 components: sensitivity to others’ wishes, needs, and opinions; self-awareness, the awareness of one’s own wishes, opinions, and needs and the capacity to express these in social interactions; and the capacity for managing new situations, understood as the inclination to explore, and feeling at ease in new situations [15]. Autonomy-connectedness is informed by attachment theory and neoanalytical object-relation theory on the development of gender identity [16], and it understands that healthy psychological functioning implies a strong self as well as strong connectedness to others [15]. Autonomy-connectedness deficits are an important underlying vulnerability factor for developing mental health problems [17].

Several considerations support the potential value of autonomy-connectedness: (1) autonomy-connectedness predicts ED recovery [12]; in a longitudinal study, recovered patients showed decreases in sensitivity to others and increases in the capacity for managing new situations compared with patients who did not recover; (2) autonomy-connectedness is transdiagnostic, which is relevant considering the high comorbidity in ED [18]; (3) autonomy-connectedness partly explains sex differences in ED symptom severity [13] and develops during the same life stages (puberty and young adulthood) in which EDs typically emerge [19], with the shift toward independence and separation contributing to sex-specific prevalence differences in mental health issues; (4) autonomy-connectedness promotes a healthy identity by addressing the ego-syntonic nature of EDs, which can contribute to high relapse rates due to strong identification with the disorder [20]; and (5) a well-founded autonomy-connectedness therapy protocol, autonomy-enhancing therapy, exists [21]. Each of these will be elaborated on in the following sections.

The first reason to include autonomy-connectedness in adult ED treatment is that it has been identified as one of the strongest predictors of recovery from ED after 1 year of treatment [12]. Kuijpers et al [12] found that patients who recovered initially had higher sensitivity to others and showed a significant decrease in sensitivity to others along with an increased capacity for managing new situations after treatment when compared with patients who did not recover. These findings indicate that recovered patients benefited from their ED treatment by gaining autonomy.

Second, autonomy-connectedness is a transdiagnostic construct associated with all major comorbidities typical of ED [18]. For example, autonomy-connectedness is linked to anxiety [14,22,23] and depression [24-29], which are characterized by high sensitivity to others paired with low self-awareness. Additionally, excessive sensitivity to others is linked to internalizing personality disorders, while low sensitivity to others and a high capacity for managing new situations are associated with antisocial personality disorder characteristics [30]. Comorbid diagnoses, particularly mood and anxiety disorders [31], may reinforce ED symptoms and can affect treatment efficacy [32]. Importantly, autonomy-enhancing therapy has already been shown to be similarly effective as group cognitive behavioral therapy in treating anxiety disorders and related comorbid symptoms [23]. Incorporating a transdiagnostic construct such as autonomy-connectedness in ED treatment may therefore not only support ED treatment but also enhance the likelihood of sustainable recovery through the concurrent treatment of comorbid conditions.

Third, autonomy-connectedness is an important factor in explaining observed sex differences in ED symptom severity. In a large, representative community sample, autonomy-connectedness differences between the sexes partially explained sex differences in ED symptom severity [13]. Men and women consistently differ in sensitivity to others, with women reporting much higher sensitivity to others than men [33]. These differences may stem from gendered attachment experiences, which in turn shape distinct adult identities and psychological vulnerabilities. Importantly, the development of autonomy-connectedness coincides with the life stage in which EDs typically emerge: puberty and young adulthood [19]. Adolescence marks the move toward changes in relationships with parents (growing independence), while adolescents’ relationships with peers become more and more important, which indicates that the “self” is still being developed. This is when the prevalence of anxiety, depression, and ED starts spiking in girls compared with boys [34], and the abovementioned differences in autonomy levels also emerge [33]. Interestingly, observed sex differences in ED symptom severity are substantially reduced when controlling for sex differences in autonomy-connectedness [13]. This suggests that autonomy-connectedness may partly account for sex differences in ED symptoms. These sex differences in autonomy-connectedness may arise from differing gendered attachment experiences for both sexes [8]. This finding suggests that it might not be solely “femaleness” that increases the likelihood of developing ED, but rather feminine attachment experiences.

Fourth, autonomy-connectedness relates to self-awareness and identity. ED, particularly AN, are often described as ego-syntonic phenomena (ie, the ED is seen as part of one’s true self, not an external illness that requires treatment) [20]. This ego-syntonic nature is suggested to contribute to high relapse rates [20]. Even if the current treatment effectively removes most ED symptoms, patients often struggle with identity and low empowerment afterwards, as the ED was intertwined with their sense of self [35]. Therefore, additional efforts are needed to rebuild a healthy self.

Finally, although a well-founded autonomy-connectedness therapy protocol, autonomy-enhancing therapy, exists [21], little is known about its effectiveness in ED treatment. Addressing this gap constitutes the need for the current study. For more details on autonomy-enhancing therapy, refer to the Methods section.

In summary, autonomy-connectedness appears to provide new insights into ED, and it is a promising and likely useful addition to ED treatment. Therefore, the study described in this protocol is aimed at testing the effectiveness and mechanisms of autonomy-enhancing therapy in the last phase of ED treatment for AN and BN.

Throughout autonomy-enhancing therapy, we expect (1) ED symptoms to decrease, (2) autonomy-connectedness to improve (ie, self-awareness and the capacity for managing new situations to increase, and sensitivity to others to decrease), and (3) autonomy-connectedness to mediate treatment success.

We will explore the differences between ED diagnoses (AN and BN) in relation to hypotheses 1 to 3.

Additionally, with respect to the secondary outcomes, we expect increased self-esteem and quality of life and reduced depression and anxiety at the posttreatment assessment compared with the pretreatment assessment.

Ethics approval for the study protocol was obtained from the Medical Ethics Review Committee of VU Amsterdam (NL83503.018.23) and the internal research ethics committee of GGZ Oost Brabant. All participants received written and verbal information about the study and provided informed consent to participate. All participant data are being deidentified. Data collection takes place through ResearchManager EDC (ResearchManager), a secure program through which participants are assigned a unique participant number. Only the principal researcher has access to the key linking participant numbers to names and email addresses.

Two designs are combined: a single-case experimental design (SCED) and a repeated measures (RM) design. Given the vulnerable and highly specific population of patients with severe EDs, a SCED was considered more feasible and ethically appropriate than, for example, a randomized controlled trial, as it reduces the number of participants required and avoids withholding treatment from a control group.

A SCED multiple baseline design differentiates between time and intervention effects, with participants serving as their own controls. Experimental control is demonstrated when changes in outcome variables occur consistently after the introduction of the intervention across staggered baselines. Replication of this pattern across participants strengthens causal inference regarding intervention effects. The number of baseline measurements is randomized (ie, the number of measurements before the start of the intervention), varying from 5 to 8 weeks (ie, 5‐8 measuring points). Weekly measurements are taken for the duration of the assigned baseline plus the length of the intervention. The Eating Disorder Examination Questionnaire–Short (EDE-QS) [36] and the ACS-30-ST measure short-term ED symptoms and autonomy-connectedness characteristics, respectively.

The RM design assesses long-term ED symptoms and autonomy-connectedness characteristics, as well as depressive and anxiety symptoms, self-esteem, and quality of life, every 3 months at the beginning, middle, and end of the intervention, along with a 3-month follow-up (4 measurements). After treatment, participants will also be interviewed about their experiences with autonomy-enhancing therapy. The interviews are intended primarily to provide a richer understanding of participants’ questionnaire scores and changes over time, including participants’ experiences with the intervention, how they perceived changes in autonomy-connectedness, well-being, and ED recovery, and whether these experiences are reflected in quantitative assessments. In addition, the interviews will be used to explore the acceptability and feasibility of the intervention and to identify factors relevant to future implementation. As the qualitative component is exploratory and descriptive in nature, interview data will be reviewed and summarized to identify recurring experiences and implementation considerations rather than subjected to an in-depth qualitative analysis. For a study procedure flowchart, refer to Figure 1. The data will be analyzed using mixed models with an unstructured covariance matrix.

Participants will be 22 to 30 patients (aged 18‐65 years) at the Center for Eating Disorders, GGZ Oost Brabant, Helmond, the Netherlands, in the last phase of multidisciplinary ED treatment, where treatment focuses on consolidating recovery, facilitating the transition to outpatient care or treatment completion, promoting self-management, and preventing relapse.

Prior to this last phase of treatment, participants will have completed 9 to 12 months of day care treatment for AN or BN, consisting of 2, 3, or 5 days of group therapy every week, including psychotherapy, sociotherapy, psychomotor therapy, drama therapy, and supported meals and snacks. All participants will have a BMI higher than 18.5 kg/m².

All patients moving into the last phase of treatment are eligible for participation in this study. Inclusion and exclusion criteria for this study are therefore identical to the criteria for moving into the last phase of treatment. Eligibility for moving into the last phase of treatment is determined by the multidisciplinary treatment team. Patients enter this phase when ED symptoms are considered to be in remission at a behavioral level, allowing treatment to shift from symptom reduction toward maintaining recovery gains and preventing relapse. Main reasons for not progressing to the last phase include patients finding treatment closer to home, patients requiring specific care for posttraumatic stress disorder, autism, or personality disorders, or patients dropping out of treatment after failing to adhere to day care weight gain requirements.

All patients in day care ED treatment at GGZ Oost Brabant are informed about the study by the main practitioner and the principal investigator, both orally and in written form. Before deciding to participate, patients have opportunities to ask questions and have 2 weeks to consider participation. Those who agree to participate fill out an informed consent form on paper. Each set of 4 participants is randomly allocated without replacement to a baseline length of 5, 6, 7, or 8 weeks, using a predetermined random list of 150 numbers generated using statistical software by an independent person in the presence of the main investigator. To start participants on their baseline measurements on time, main practitioners are asked to estimate when patients are likely to move on to the last phase of treatment. When participants have started baseline measures but need more time in day care treatment, baseline measures are paused until they are close to moving on.

When participants have started baseline measures but choose not to continue into the last phase of treatment, their allocated baseline number is placed back on top of the list.

All questionnaires are answered digitally on any device. Weekly measures will take approximately 10 minutes; 3-monthly measures will take approximately 30 minutes. Questionnaire invitations as well as reminders are sent by email.

Posttreatment interviews will take place at the GGZ facility or online, conducted by the investigator and guided by a topic-list. Interviews are audio-recorded and later transcribed.

Autonomy-enhancing therapy’s goal is to enhance autonomy-connectedness at behavioral, attitudinal, and emotional levels by working on personal, autonomy-related goals.

The investigational treatment, autonomy-enhancing therapy, consists of 15 weekly 2-hour group sessions, following the existing autonomy-enhancing therapy protocol [21]. The autonomy-enhancing therapy protocol prescribes a mindfulness exercise, space to share progress toward personal autonomy goals, and the appointment of a chairperson for each session who, with therapist support, keeps time. Each week features different autonomy-related themes (eg, relationships, boundaries, body and sexuality, and emotions), including psychoeducation, exercises, homework, and discussion of autonomy-related difficulties. Autonomy-enhancing therapy groups typically consist of 8 participants (range 4‐10).

The autonomy-enhancing therapy protocol has been slightly adjusted to fit the existing structure of the last phase of treatment in the Center for Eating Disorders, GGZ Oost-Brabant, Helmond. In this phase of treatment, participants are part of a group that follows several treatment modules together. Relapse prevention is covered in standard treatment, so introductory, relapse prevention, and midpoint evaluation sessions are omitted, leaving 12 sessions. Additionally, new members continuously join the group in the last phase of treatment, so patients start with whatever theme is planned for that week, ensuring that all themes are covered, even if the starting point varies. In the last phase of treatment, patients participate for a minimum 3 to maximum 6 months, with continuation decided after an evaluation at 3 months. Continuation or discharge is determined by main practitioners, guided by input from the treatment team. Participation in the intervention stops when individuals are discharged. After finishing their ED treatment, many patients proceed to other forms of treatment (eg, trauma treatment, treatment focused on personality disorders, or general psychological support).

Because autonomy-enhancing therapy is embedded in treatment as usual (TAU) during the last phase of treatment, participants will receive 12 to 24 weeks of autonomy-enhancing therapy (Figure 1), ensuring that each patient covers every theme at least once and at most twice. As this study investigates the real-life application of autonomy-enhancing therapy in ED aftercare, all patients will also continue to receive TAU for the last phase of treatment, consisting of drama therapy, psychomotor therapy, and regular group therapy. TAU focuses on consolidating recovery gains, preventing relapse, and supporting the transition to everyday life.

Therapists providing autonomy-enhancing therapy are trained and certified in the autonomy-enhancing therapy protocol through RINO Amsterdam, a Dutch postgraduate training institute for mental health professionals. As part of maintaining certification, therapists participate in regular intervision meetings in which treatment delivery and adherence to the protocol are discussed. Formal assessment of treatment fidelity is not conducted within the present study. No major risks beyond regular treatment are expected. Any adverse psychological events, treatment deterioration, or participant withdrawal related to study participation will be discussed within the clinical treatment team and documented.

Gender was, in a conventional way, assessed as biological sex as “woman” or “man.”

Sample size and statistical power were approximated using G*Power (version 3.1.9.4). To detect a medium effect size (f²=0.15) in a regression with 7 predictors, 89 independent observations are needed for a power of 0.95 at α=.05. For the SCED analysis, with an intraclass correlation coefficient of 0.2, each participant’s 20 observations yield 4.17 independent observations [45], necessitating at least 22 participants. To ensure robustness, we aim for 30 participants (125 independent observations), achieving a power of 0.99. Considering a 20% dropout rate, we plan to recruit 36 participants. For the RM analysis, assuming 3 observations per participant, we obtain 30×2.14=64 independent observations, resulting in a power of 0.67 to detect a medium effect size (using α=.01).

Differences between individuals in their SCED data will be examined using visual analyses, presenting separate plots for all participants’ measures [46]. Visual analysis will evaluate changes in level, trend, variability, immediacy of effect, overlap between phases, and consistency across participants. Interview data will be used descriptively to better understand participants’ experiences of the intervention and their interpretation of change. The interview data will be reviewed and summarized to identify recurring experiences, perceived benefits and challenges, and considerations relevant to feasibility and future implementation.

The SCED data will be analyzed using a mixed model with 2 treatment periods, incorporating 2 dummy variables (treatment period 1 and treatment period 2, with baseline as the reference). A linear effect of time, centered within each period, represents the average change across periods. A random intercept and an autoregressive error structure will model the covariance structure. The primary hypothesis of the average decrease in ED symptoms is tested with 1 dummy for treatment period (1=treatment period 1 or 2, 0=baseline). This hypothesis will be tested for autonomy-connectedness.

The RM data for all outcome variables will be analyzed using mixed models. Measurement will be modeled with 3 dummy variables (after treatment period 1, after treatment period 2, and after 3 months, all with baseline as the reference category), and an unstructured covariance matrix will model the random effects. The analyses will focus on the effect of dummy “post,” testing the hypothesis that scores improve compared with baseline. Mediation by autonomy-connectedness will be tested in two steps: (1) adding autonomy-connectedness to the model and (2) conducting a dependent-samples t test (1-tailed) comparing autonomy-connectedness after treatment and at the 3-month follow-up with baseline. Mediation will be concluded if (1) autonomy-connectedness affects the outcome variable in the mixed model analysis and (2) autonomy-connectedness improves over time in the t test, indicating a joint-significance test of mediation [47].

All analyses will control for ED diagnosis (dummy AN, with BN as the reference category). We will explore whether the development over time of ED, autonomy-connectedness, and the other outcome variables differ between the two EDs.

Hypotheses on change will be tested 1-tailed, with a significance level of .05 for the multiple baseline data and .01 for the RM data to correct for multiple testing. Missing values on the outcome variable do not affect mixed model analyses [45], whereas participants with missing autonomy-connectedness values will be excluded from the mediation analyses.