Assessment of eating disorders in patients with anorexia nervosa: Interview versus self -report

Abstract

This study was designed to compare an investigator-based interview, the Eating Disorders Examination (EDE), with a self-report version of that interview, the EDE-Q. Sixty women with anorexia nervosa (AN) completed both instruments. The results of the two measures were then compared. There were significant absolute differences between the EDE and the EDE-Q for the Eating Concern subscale, the Weight Concern subscale, and the Shape Concern subscale, with EDE-Q scores being greater than the EDE scores. The Restraint subscale was also higher on the EDE-Q than on the EDE, but not to a statistically significant degree. Despite these differences, in the majority of cases, the EDE and EDE-Q subscales were within 1 point of each other. For the eating disorder behaviors, the EDE-Q had higher values than the EDE for binge eating, vomiting, and excessive exercise, but only binge eating was significantly greater.;Results indicate that agreement between the EDE and EDE-Q on the individual items informing the overall diagnosis of anorexia nervosa ranges from low to moderate. Agreement for the overall diagnosis of anorexia and for the binge/purge subtype was more impressive. Although correlations between subscale scores as well as eating disorder behaviors were strong, higher levels of disturbance were consistently reported on the EDE-Q than the EDE interview. These differences were statistically significant with regard to the Shape Concern subscale, the Eating Concern subscale, and the Weight Concern subscale, as well as binge eating.;The pattern of findings suggests that the self-report questionnaire, the EDE-Q, may be used in place of the interviewer-based measure when assessing overall diagnosis and subtype, as well as specific, well-defined features. For example, there was good agreement between the EDE and EDE-Q with respect to the assessment of self-induced vomiting and laxative use and the severity of dietary restraint. In contrast, there was a low level of agreement with respect to less-defined features, like binge eating, where significantly higher frequencies were generated by the self-report questionnaire.