The comparative utility of statistically derived eating disorder clusters and DSM-IV diagnoses: relationship to symptomatology and psychiatric comorbidity at intake and follow-up

INTRODUCTION: The classification of eating disorders has been a matter of considerable debate. The present paper extends previous work and aimed to compare the utility of statistically derived clusters of eating disorders and conventional diagnoses. METHODS: Adult female eating disorder patients who had previously been classified on the basis of cluster analysis of key diagnostic variables were examined on measures of eating disorder symptomatology and psychiatric comorbidity at intake (N=601) and subsequent follow-up after 6 and 36 months (N=349, N=322, respectively). RESULTS: Compared to DSM-IV diagnoses, clusters demonstrated greater utility in terms of more distinct between-group differences and higher effect sizes in relation to a wide range of variables. The greater utility of clusters was in important respects due to the reallocation of EDNOS patients to more relevant alternative categories and to a greater emphasis on psychological and behavioural features of eating disorders. CONCLUSIONS: In order to achieve a better classification of eating disorders, it will be important to place increased emphasis on common psychological features. There is a need to move away from increased use of subtypes and toward a definition of eating disorder per se.