Adult Eating Disorder Assessment Study

Overview

The hypothesis of the study is that low-cost self-completion questionnaires relating to eating disorder symptoms will predict the subsequent results of a detailed, semi-structured interview assessment of eating disorder symptoms that has been calibrated according to expert clinical diagnostic case thresholds. Eating disorders are recognised as a research priority among healthcare professionals, adults with lived experience, and their carers alike. There is a need for measurement methods that can reliably and systematically identify symptoms of common forms of eating disorder, including those fulfilling agreed diagnostic criteria for anorexia nervosa, bulimia nervosa, and binge-eating disorder. Effective diagnostic interview tools can facilitate early detection of eating disorders. To develop rules for determining whether diagnostic criteria for eating disorders are met, a study involving adults referred to specialist eating disorder services is required. In this study, assessments by eating disorder clinicians will be compared with researcher assessments using a semi-structured interview assessment, the Schedules for Clinical Assessment in Neuropsychiatry version 3 section 9 (SCANv3s9). The study population will consist of 100 adult patients referred to specialist eating disorder services, including patients whose referrals are accepted by these services, as well as those who are not. Clinical assessments and structured interview assessment findings will also be compared with those from widely used screening tools for eating disorders, the Eating Disorder Examination Questionnaire short-form and the SCOFF (Sick, Control, One, Fat, Food) questionnaire. The aim of this study is to evaluate the accuracy of the Eating Disorder Examination Questionnaire Short-Form (EDE-QS) and the SCOFF (Sick, Control, One, Fat, Food) questionnaire in adults referred to specialist eating disorder services, in a comparison with the Schedules for Clinical Assessment in Neuropsychiatry subsection on eating disorders and specialist NHS clinician assessments of the probability of eating disorder. This will help develop appropriate and accurate benchmarks for estimating the prevalence of eating disorder symptoms and clinical diagnoses, in adults referred to specialist eating disorder services, as well as the wider population (through combining the findings from this study with those of the community 2023 Adult Psychiatric Morbidity Survey).

Study Analysis Plan Work Package 1 (WP1) relates to assessment of the entire study clinical population (n = 100). For WP1, the threshold for diagnosis of common forms of eating disorders (anorexia nervosa, bulimia nervosa, binge-eating disorder, and any eating disorder) by clinicians will be used as the reference (criterion) standard, and 95% confidence intervals will be performed for all estimates. Receiver-operating characteristic (ROC) curves and areas under the curve (AUC) will be calculated for all measurement tools (SCANv3s9, EDE-QS, and SCOFF), with reference to the criterion standard, as well as sensitivity, specificity, positive predictive value, and negative predictive values for a range of cut-off scores for these tools. Additionally, correlation coefficients will be calculated with respect to SCANv3s9, EDE-QS, and SCOFF findings, relative to those of the reference standard. To calibrate the SCANv3s9 interview, symptom findings from the researcher-assessed SCANv3s9 will be compared with NHS clinician's diagnostic thresholds. However, the research team may modify the SCANv3s9 diagnostic algorithms in the event that they diverge substantially from clinician judgment, an approach that has previously been employed when calibrating the Autism Diagnostic Observation Schedule (ADOS) against clinical opinion. Work Package 2 (WP2) relates to retesting of a subsample of the WP1 population (n = 25). For WP2, test-retest reliability will be assessed using Cohen's kappa coefficient (κ), as well as simple percentage agreement calculations.