Low levels of interrater reliability in a standard measure of outcome in eating disorders (the modified Morgan-Russell Assessment Schedule).

Abstract

The interrater reliability of the modified version of the widely used Morgan-Russell Assessment Schedule was assessed. Sixty-six female patients presenting for treatment for the first time in a general hospital and a community health center were assessed by a treating clinician and a researcher during sequential interviews. Two extra items were added to the Schedule to allow for the assessment of bulimic behaviors. Ratings were analyzed using the Cohen's weighted kappa. Poor to very poor levels of agreement were found between rater pairs assessing the 66 separate patients. The necessity for a specific, standardized format in using the Morgan-Russell Schedule is discussed. In its current, unstandardized format the Schedule may not be suitable for comparing outcome in eating disorders across different centers.