What is the EPSI?

Current measures of eating-disorder psychopathology are narrow in scope (e.g., assessing only one aspect of disordered eating, such as disinhibited eating) or have serious limitations (e.g. gender biases, inconsistent factor structures, and poor discriminant validity). To address these problems, our team developed the EPSI (Eating Pathology Symptoms Inventory; pronounced ‘ep-see’). A multidimensional self-report measure was developed to assess eating-disorder pathology among a wide variety of populations. It can be used in clinical settings as well as in research. The EPSI is a 45-item questionnaire that has eight scales, including body dissatisfaction, binge eating, cognitive restraint, purging, restricting, excessive exercise, and negative attitudes towards obesity and muscle building. 

The EPSI has been tested in > 30,000 individuals. Our recent work also found that the EPSI scales are invariant in a large sample of adolescent (n=5,250) and adult (n=24,571) patients with an ED (Richson et al., 2021), making our system ideal for assessing eating-disorder psychopathology across the lifespan.



The Importance of the EPSI

The EPSI has demonstrated strong evidence for test-retest reliability, internal consistency, and convergent/discriminant validity across a range of samples and most scales (except Muscle Building) are invariant across cisgender men and cisgender women (Forbush et al., 2014).  


The EPSI was designed to be a more inclusive eating pathology measure that is able to assess eating disorders across a range of populations accurately. The EPSI has less gender and weight category bias (Forbush et al., 2013; Forbush et al., 2020), a more replicable factor structure than other traditional eating-disorder measures, and stronger criterion validity for distinguishing among individuals with different types of eating disorders compared to other established measures of eating pathology (Forbush et al., 2013; Forbush et al., 2014). The factor structure also replicates in non-Western cultures (Tang et al., 2015; Sahlan et al., 2022).