The DSM-IV-TR does not currently provide guidelines on how to classify persons initially diagnosed with anorexia nervosa (AN) or bulimia nervosa (BN) who still have symptoms but no longer meet the full criteria for these disorders. Should they retain the original diagnosis of AN or BN or be assigned to a separate category of eating disorders not otherwise specified (EDNOS)? Is symptomatic improvement or change part of the natural course of these two disorders or does it constitute another disorder, such as EDNOS?

Little long-term research has been done on EDNOS

There is little longitudinal research on the course of EDNOS. In one early study, 82% of a small cohort of women with EDNOS either endorsed a history of BN or AN or developed one of these two disorders during a mean follow-up of 40 months (Int J Eat Disord. 1993; 40:S89).

In an attempt to clarify the issue, K.T. Eddy, PhD and colleagues examined subthreshold presentations in 246 women who had initially been diagnosed with BN or AN (Psychol Med 2010.printed online before publication). The women were followed for a median of 9 years, and weekly symptom data were collected at frequent intervals with the Longitudinal Interval Follow-up Evaluations of Eating Disorders (LIFE-EAT-II). Five subtheshold eating disorder presentations were examined, and recovery was defined by minimal or no symptoms; that is, failure to be classified as AN, BN, or any of the subthreshold presentations during 13 consecutive weeks.

Results

During the follow-up, 191 of the women, or 77.6%, experienced a period of at least 3 months in which they had symptoms that were subthreshold for AN or BN. This included 64.7% of those who had an intake diagnosis of restricting type AN (ANR), 67.1% of those who had an intake diagnosis of binge-purge subtype AN (ANBP), and 91.8% of those with an intake diagnosis of BN. Among the women who had an intake diagnosis of ANBP, transitions to subthreshold ANR (40%; 34 women) or subthreshold ANBP (35.3%; 30 women) were most common, followed by transitions to subthreshold BN (23.5%; 20 women). Transition to purging disorder involved 23.5% of women. Transition to BED occurred in only 8 women, (or 9.4%).

The authors’ findings indicate that during follow-up, most of these women developed subthreshold presentations. Just as they had hypothesized, subthreshold presentations tended to resemble intake diagnoses. Thus, those with AN were more likely to develop subthreshold AN, while those with BN were most likely to develop subthreshold BN. Furthermore, those with ANR were unlikely to develop subthreshold presentations resembling BN spectrum disorders, whereas those with ANBP or BN developed a range of presentations characterized by binge-eating and purging.

For women with either type of AN, transition to subthreshold ANR or ANBP followed weight gain. For women with ANR who moved to subthreshold ANR, the transition meant the onset of binge-eating and/or purging; for those who moved to subthreshold ANR, this meant a decrease in binge-eating and purging.

During the follow-up, women with lifetime diagnoses of AN and BN developed multiple suthreshold presentations over time. Although the transitions were highly variable and lasted from 3 months to nearly 7 years, the median time of any subthreshold episode, regardless of the initial diagnosis, was less than 1 year. Finally, most women with an initial diagnosis of AN did not recover during follow-up, whereas those with BN did.

Conclusion: subthreshold symptoms are part of the course of illness

The authors concluded that for most individuals with a lifetime AN and/ or BN, the course of illness will include symptom fluctuations that could be clinically described as EDNOS. However, their study results suggest that these subthreshold presentations are part of the course of illness for individuals with AN and BN, rather than different disorders, or types of EDNOS. Eddy and colleagues state that their findings raise questions about when and if it is ever appropriate to diagnose EDNOS in individuals with lifetime BN or AN. It might be more clinically sound to diagnose patients with modifiers such as “anorexia nervosa,” restricter-subthreshold type,” bulimia nervosa, subthreshold type,” for example. Well-designed longitudinal studies may help clarify these boundaries.