4 - (SYM 124) Psychometric Performance of the Eating Disorder Flexibility Index in U.S. Nonclinical Student and Clinical Adolescent Samples

Background: Inflexibility, or the inability to react dynamically and adaptively to changing internal states and environmental demands, is a transdiagnostic maintaining mechanism of varied psychopathologies, including eating disorders (ED). The Eating Disorder Flexibility Index (EDFLIX) is designed to assess general cognitive and behavioral flexibility as well as in relation to weight, shape, food, and exercise. The measure is increasingly used in research studies in the United States. To date, the EDFLIX, which was originally developed and validated in Norwegian, has not been formally validated in its English form in clinical or nonclinical samples.

Method: This study examined the factor structure of the English EDFLIX in undergraduates (n=578, 57.6% female, 50.2% White) and adolescent ED patients (n=63, 87.0% female, 91.3% White). Clinical participants’ average length of illness was 13.52 months (SD=16.24), average body mass index fell in the underweight range (M=18.46kg/m², SD=2.10), and mean EDE-Q global score was 2.91 (SD=1.73). We first explored the underlying structure of the 36 EDFLIX items in undergraduate students through exploratory factor analysis (EFA), followed by confirmatory factor analysis (CFA) in a second student sample to validate the model(s) uncovered in EFA. We then sought to validate the model(s) using CFA in the clinical sample.

Results: Parallel and exploratory factor analysis suggested the EDFLIX may comprise two or three underlying factors. The two- and three-factor solutions accounted for 25% [RMSEA=.07, 95% CI(.07,.08); TLI=.64] and 29% of the variance in scores, respectively [RMSEA=.06, 95% CI(.06,.07); TLI=.72]. Confirmatory factor analyses in the nonclinical and clinical samples did not support either model. In the nonclinical sample, indices of absolute fit fell within the acceptable range, but incremental fit indices did not, suggesting overall poor fit with the two-factor structure. Data also fit poorly with the three-factor structure, though a nested model chi-square test suggested significantly better fit compared to the two-factor model. Similarly, data from the clinical sample fit poorly with both models, with indices of absolute and incremental fit falling outside the acceptable range. A nested model chi-square test again suggested significantly better fit of the three-, compared to the two-factor model.

Discussion: Findings indicate that the English version of the EDFLIX may not capture underlying aspects of flexibility in nonclinical and clinical ED samples as previously proposed.