Factorial integrity and validation of the Eating Pathology Symptoms Inventory (EPSI)

Abstract

The Eating Pathology Symptoms Inventory (EPSI) is a 45-item self-report measure of eating pathology designed to be sensitive in assessing symptoms among diverse populations of individuals with disordered eating. The current study represents the first external validation of the EPSI as well as the first to examine the factor structure in an outpatient eating disorder clinic sample. We conducted an exploratory factor analysis in three separate samples: an outpatient clinic sample (n = 284), a college sample (n = 296), and a community sample (n = 341) and compared the observed factor structures to the original 8-factor solution proposed by Forbush et al. (2013). We also investigated whether the subscales correlated with the Eating Disorder Examination Questionnaire (EDE-Q) and a clinical impairment measure among the outpatient clinic sample. Results suggested between 7 and 8 factors for each of the three samples. Our findings largely replicated those of the original EPSI development study, excepting some deviations in the Muscle Building, Cognitive Restraint, and Excessive Exercise subscales. However, confirmatory factor analysis and exploratory structural equation modeling produced poor model fit, which may be related to the item heterogeneity within many of the subscales. Finally, eating disorder attitudes and behaviors assessed by the EPSI were significantly correlated with the EDE-Q and with clinical impairment. Overall, our results highlight both strengths and limitations of the EPSI. Findings provide preliminary support for the use of the EPSI among research with diverse populations, and present several avenues for future research for enhancing clinical utility.