Psychometric evaluation of disordered eating measures in bariatric surgery candidates

Introduction: Assessment of disordered eating is common in bariatric surgery candidates, yet psychometric properties of disordered eating measures in this population are largely unknown. Methods: Measures were completed by 405 adult bariatric surgery candidates at pre-surgical consultation. Fit of the original scale structures was tested using confirmatory factor analysis (CFA) and alternative factor solutions were generated using exploratory factor analysis (EFA). Reliability (internal consistency), construct validity (convergent and divergent) and criterion validity (with the EDE as criterion) were assessed. Materials: The measures prioritised for evaluation are the following: Eating Disorder Examination Questionnaire (EDE-Q; n = 405), Three-Factor EatingQuestionnaire (TFEQ; n = 405), Questionnaire of Eating and Weight Patterns Revised (QEWP-R; n = 204), Clinical Impairment Assessment (CIA; n = 204) and the Eating Disorder Examination clinical interview (EDE; n = 131). Results: CFA revealed adequate fit for only the CIA in its current form (CFI = 0.925, RMSEA = 0.096). EFA produced revised scales with improved reliability for the EDE, EDE-Q and TFEQ. Reliability of revised subscales was improved (original scales α = 0.43–0.82; revised scales α = 0.67–0.93). Correlational analyses of the CIA and revised versions of remaining scales with measures of psychological wellbeing and impairment revealed adequate convergent validity. All measures differentiated an EDE-classified disordered eating group from a non-disordered eating group (criterion validity). Diagnostic concordance between the EDE, EDE-Q and QEWP-R was low, and identification of disordered eating behaviours was inconsistent across measures. Conclusions: Findings highlight the limitations of existing disordered eating questionnaires in bariatric surgery candidates. Results suggest revised assessments are required to overcome these limitations and ensure that measures informing clinical recommendations regarding patient care are reliable and valid