A Longitudinal Study of Severe and Enduring Eating Disorders

BACKGROUND, AIMS AND HYPOTHESES

Eating disorders (EDs) are psychiatric conditions characterized by a loss of control over food intake. The prevalence of anorexia nervosa (AN) is estimated at approximately 1-2%, while bulimia nervosa (BN) affects 2-3% of the population. EDs significantly impair functioning, have serious health consequences, and are associated with high mortality rates. Around 20% of patients with AN and 10% of those with BN develop a long-lasting illness, often referred to as Severe and Enduring Eating Disorders (SEED). While there is no scientific consensus on the definition of SEED, it is frequently defined as a duration of illness lasting seven years or more. Research suggests that factors maintaining EDs may differ from those that trigger them.

The underlying causes of EDs remain largely unknown, though their origins are considered multifactorial. Psychiatric comorbidities are highly prevalent in EDs, significantly influencing their course and outcomes. Personality disorders (PDs) are associated with poorer treatment outcomes for EDs, but longitudinal studies examining the trajectory of PDs in EDs are limited, and findings are inconsistent.

Dysfunctional emotion regulation (ER) has been identified as a transdiagnostic psychological risk factor for many psychiatric disorders, including EDs. ER difficulties can manifest as undercontrol, characterized by personality traits such as impulsivity and insufficient self-control, or overcontrol, characterized by emotional inhibition and excessive self-control och cognitive inflexibility. Some models propose that undercontrol is central to BN, while overcontrol are core features of AN. However, the course and stability of ER in EDs remain poorly understood due to a lack of longitudinal studies.

Another understudied factor in EDs is loneliness, encompassing perceived social isolation and a lack of connectedness. Loneliness has been linked to ER strategies, such as excessive self-control and emotional avoidance, which in youth can contribute to social isolation, reduced life satisfaction, and a higher risk of enduring mental health problems. The role of loneliness in SEED, however, is not well understood.

Further, research on EDs has proposed that biological factors, including dysregulation of the immune system, plays a role in the development and maintenance of the EDs. Studies indicate a pro-inflammatory state in AN, though it remains unclear whether this is a state or trait marker. The role of inflammation in BN is even less understood, and studies present mixed findings. Some evidence suggests an increased risk of BN in individuals with autoimmune or inflammatory diseases, highlighting the need for further investigation into inflammatory markers over the course of the illness.

In summary, clinical factors such as psychiatric comorbidity and personality disorders; psychological factors including personality characteristics, dysfunctional emotion regulation, overcontrol/undercontrol, cognitive inflexibility, loneliness, and severe eating disorder (ED) symptoms; and biological factors such as immune system dysregulation may all play a role in the development, maintenance, and relapse of EDs. However, research in this field is scarce.

This projects aims is to increase our knowledge about risk factors for a severe course in EDs. Our overarching research question is: What are the key clinical, psychological, and biological risk factors of a severe and enduring course of an eating disorder?

It is hypothesized that a chronic course of ED is related to severe ED symptoms, personality traits related to maladaptive over- and undercontrol, difficulties with emotion regulation (ER), and increased systemic inflammation.

Primary outcomes are

• eating disorder diagnosis at follow-up
• severity of eating disorder symptoms

Secondary outcomes are

• psychosocial impairment
• quality of life
• loneliness at follow-up
• comorbidity at follow-up
• systemic inflammatory activity at follow-up
• Physical status at follow-up

Predictors, moderators and mediators

• emotion regulation
• personality
• cognitive flexibility
• loneliness at baseline
• comorbidity at baseline
• systemic inflammatory activity at baseline
• duration of illness
• time in treatment
• motivation for change
• treatment completion/treatment dropout
• Physical status at baseline

PROCEDURE

Two prospective cohorts of patients with eating disorders (EDs)-one adolescent and one adult-will be followed longitudinally to examine changes in and the impact of clinical factors, personality traits, emotion regulation (ER) difficulties, loneliness, and biomarkers. Data will be collected at baseline, post-treatment, two years post-baseline, and at five, 10, and 20 years. Participants will be recruited consecutively from autumn 2024 onwards from the Eating Disorder Unit at the Uppsala Department of Child and Adolescent Psychiatry (ED-CAP) and the eating disorder unit for adults (ED-P).

Child and adolescent psychiatry (ED-CAP)

The eating disorder unit at the Department of Child and Adolescent Psychiatry at Uppsala University Hospital (CAP) treats patients with AN, BN, Other Specified Feeding or Eating Disorder (OSFED), and Avoidant Restrictive Food Intake Disorder (ARFID) up to 18 years. Patients are referred to the unit either by other healthcare professionals or by themselves and/or their parents.

All patients are systematically assessed using a structure implemented by Swedish ED clinics that provides valid data to the national quality register. The structure consists of an assessment of clinical history, diagnostic evaluation with the child and adolescent version of the MINI International Neuropsychiatric Interview for Children and Adolescents (MINI-KID) or Electronic Psychiatric Screening Interview for children (EPSI-C), the Eating Disorder Examination interview (EDE-I), and clinician rating of symptoms and functioning (C-GAS) and the Clinical Global Impressions scale (CGI). In addition, weight and height are measured, and self-ratings with the Eating Disorder Examination Questionnaire (EDE-Q), Clinical Impairment Assessment Questionnaire (CIA), and Montgomery-Åsberg Depression Rating Scale Self-Assessment (MADRS-S) are collected.

After this initial assessment, all patients will be invited to participate in research. A research nurse will invite all patients and perform the initial somatic examination as well as sampling of blood. The patients who accept participation will sign informed consent to be included in the study. Since the youths are at least 14 years old, their parents/caregivers will also be required to accept participation and sign informed consent. Participants will receive questionnaires sent to them digitally.

The eating disorder clinic for adults (ED-P)

The eating disorder clinic for adults at the Department of Psychiatry at Uppsala University Hospital (ED-P) treats adults (≥ 18 years) with EDs. The clinic receives referrals from healthcare professionals (mainly physicians or psychologists) from other parts of the regional healthcare system. Patients can also refer themselves.

All patients undergo a systematic evaluation upon arrival at the eating disorder clinic. This consists of a clinical history and a structured diagnostic interview; either the MINI for DSM-5 or the Structured Clinical Interview for DSM-5 Axis I Disorders - Clinical Version (SCID-I CV). Patients also undergo the Eating Disorder Examination interview (EDE-I) and complete the self-report instruments Eating Disorder Examination Questionnaire (EDE-Q), Clinical Impairment Assessment Questionnaire (CIA), and Montgomery-Åsberg Depression Rating Scale Self-assessment (MADRS-S). Weight and height are measured.

All diagnostic assessments at the Department of Psychiatry are conducted by psychologists or physicians who have received training in the procedure. Patients who, during the assessment, are diagnosed with AN, BN, BED, ARFID, or OSFED of moderate to severe intensity, or OSFED with psychiatric comorbidity, and who accept treatment are accepted for treatment at P-ED. Patients who are diagnosed with mild OSFED or moderate OSFED without psychiatric comorbidity are referred to primary care for monitoring or treatment.

After this initial assessment, all patients will be invited to participate in research. An appointed person will invite all patients. If the patient accepts participation, they will sign an informed consent form. Participants will receive questionnaires sent to them digitally.

All Ages: Both ED-CAP and ED-P Participants will be asked about the collection of biomarkers through Uppsala Psychiatric Patient samples (UPP) (ethics approval Dnr 2012/081), which is an infrastructure for the collection of biological materials at the Department of Psychiatry at Uppsala University Hospital. All patients will be invited to participate in UPP, and inflammatory markers will be analyzed in accordance with the informed consent for UPP. Inflammatory markers will be analyzed from venous blood samples drawn from participants before treatment, immediately after treatment or dropout, and at the two-year follow-up.

At baseline, the following data will be collected: Diagnoses, physical examination (BMI, pulse, blood pressure), demographic characteristics, clinical characteristics, personality traits, psychological risk factors (e.g., ER difficulties, including loneliness), and biomarkers (including markers of inflammation). The same procedure will be repeated again after two years. A few questionnaires will be distributed at treatment termination or dropout.

After five, 10, and 20 years, patients will be followed up through National Helath registers and with questionnaires. Weighing will be performed weekly during treatment and at the follow-up two years later. At all time points, questionnaires will be distributed digitally. Process measures, such as the type of received treatment, number of treatment sessions, whether the patient dropped out or completed treatment, will be monitored. At follow-up, participants will be asked to answer a set of questions regarding, for example, living conditions, in addition to the questionnaires assessed at baseline.

QUALITY ASSURANCE PLAN

Interviewers will be trained and quality assured by calculating inter-rater agreement. For self-assessment, well established psychometrically evaluated instruments have been chosen. Special consideration has been given to selecting instruments suitable for both adolescents and adults.

SAMPLE SIZE ASSESSMENT

Sample size calculations were performed to ensure adequate power for the proposed analyses. With a Cohens effect size of d = 0.2, 199 participants will be required in each group, given an alpha of 0.05 and a power of 0.8. Dropout over time is expected, which increases as more time passes, estimating a 50% dropout by the final physical follow-up. Therefore, just under 400 participants need to be included in each group in order to have sufficient data for the last follow-up.

DATA ANALYSES

A range of statistical techniques will be employed to address the hypotheses and research aims, considering the longitudinal design, comparisons between the adolescent and adult cohorts, and diversity of outcome variables:

Descriptive analyses - Descriptive statistics (means, standard deviations, proportions) will summarize demographic, clinical, psychological, and biological variables at baseline and follow-up. Correlation analyses will explore relationships between predictors and primary and secondary outcomes.

Comparisons between subgroups

• Cluster analyses will be performed to identify subgroups of participants with different profiles of personality traits. These cluster will be compared regarding predictors, mediators/moderators, secondary outcomes, and primary outcomes.
• Between-group comparisons will also be conducted using t-tests, ANOVA or nonparametric alternatives (e.g., Mann-Whitney U tests) with regard to predictors and outcomes.

Longitudinal Regression Analyses

• Linear Mixed-Effects Models (LMM): These will evaluate changes in continuous outcomes over time.
• Generalized Estimating Equations (GEE): These models will analyze categorical outcomes, accounting for repeated measures and within-subject correlations.
• Logistic Regression: Logistic models will predict binary outcomes at follow-up.

Mediation and Moderation Analyses

• Mediation Analysis: mediation models will assess how variables mediate the relationship between predictors and primary and secondary outcomes.
• Moderation Analysis: Moderation models will explore how variables influence the relationship between predictors and primary and secondary outcomes.

Survival Analysis

• Cox proportional hazards models will be used to analyze time-to-event data, such as time to dropout and relapse.

Inflammatory Biomarker Analyses

• Levels of inflammatory markers in the ED samples will be compared to samples from individuals with other psychiatric disorders, as well as with healthy controls. Since blood samples will be taken upon repeated times during the study period, comparison of levels of inflammatory markers during disease course will be possible.

Missing data will be addressed as instructed for each instrument, ur by using multiple imputation or mixed-model approaches.

Subgroup Analyses

• Separate analyses will be conducted for adolescent and adult cohorts to examine potential differences in predictors and outcomes.

All statistical analyses will be performed using appropriate software (R, SPSS), and sensitivity analyses will be conducted to assess the robustness of findings.