Commentary: Stability of eating disorder diagnoses

Abstract

In conclusion, these data have several implications for nosologic systems and for future research directions. First, it will be important to examine temporal stability in greater detail. Specifically, it would be useful to investigate the temporal stability of the constituent symptoms (in addition to diagnoses) to determine whether certain symptoms show greater stability over time relative to other symptoms. It will be particularly important to do this in community-recruited nontreatment seeking samples, because it is difficult to separate the effects of syndromal temporal instability from response to treatment with patient samples. It would also be useful if studies adopted a standard reporting convention to facilitate comparison across studies and potentially meta-analytic reviews. The focus on diagnostic stability at the symptom level may also reveal somewhat artificial instabilities of certain diagnoses. For instance, minor changes in weight or eating behavior can result in a person receiving an entirely different eating disorder diagnosis (which may explain some diagnostic cross-over). Furthermore, there is a lack of universally accepted definitions of constructs used in research on eating disorders. For example, recovery has been defined either as the absence of any eating disorder symptoms or the absence of an eating disorder diagnosis. In addition, EDNOS is referred to as partial recovery or a "subthreshold" eating disorder, by some investigators, while others have emphasized that most cases of EDNOS reflect "mixedeating disorder symptoms, rather than just falling short on a particular criterion.25 Second, it might be useful for the diagnostic system to reflect the course of eating disorders and a history of crossover from one eating disorder diagnosis to another, as this information appears to have prognostic significance. It would be useful for future studies, particularly those that use large samples, to attempt to replicate the findings regarding the predictive effects of diagnostic crossover. Third, an increased emphasis on predictors of a chronic course and crossover may identify additional symptoms that should be included in the current eating disorder syndromes. Elucidation of such outcome predictors may allow the identification of subtypes that show a particularly pernicious course. Although some predictors have emerged for AN,9 identification of the predictors for the course or outcome of BN and EDNOS has been elusive. 6,9 It will also be important for researchers to conduct additional studies on the prospective course of eating disorders over various time windows. In addition, relatively few prospective studies have examined the peak periods of risk for onset of eating disorder symptoms and development of the syndromes. It will be vital to determine whether participant age moderates diagnostic stability, as certain developmental periods may be more condusive to stability, such as adolescence, versus others, such as young adulthood, wherein emerging adult rolls (e.g., parenthood) may disrupt diagnostic stability. © 2007 Wiley Periodicals, Inc.