In this study, we aimed to assess the longitudinal course of anorexia nervosa and bulimia nervosa with regard to diagnostic crossover, which may have implications for the validity of the current DSM classification system and thus might be used to inform its revision. Indeed, our findings provide partial validation for the current classification schema: the longitudinal data generally support the distinctiveness of the diagnostic categories of anorexia nervosa and bulimia nervosa. However, less support was provided for the current anorexia nervosa subtyping system.
During 7 years of follow-up, nearly three-quarters of women with an intake diagnosis of anorexia nervosa experienced diagnostic crossover. Approximately half crossed over between the anorexia nervosa subtypes, and this crossover was bidirectional, recurrent, and probable throughout the follow-up period. One-third of those with an intake diagnosis of anorexia nervosa experienced crossover to bulimia nervosa; while crossover from restricting-type anorexia nervosa to bulimia nervosa was unlikely, just over one-half of those with an intake diagnosis of binge eating/purging-type anorexia nervosa experienced crossover to bulimia nervosa. Notably, approximately half of those with anorexia nervosa who crossed over to bulimia nervosa did so in the course of progressing to partial or full recovery, whereas the other half who experienced crossover to bulimia nervosa were likely to cross back over into anorexia nervosa. Women with bulimia nervosa were unlikely to develop anorexia nervosa during follow-up.
The relatively lower frequency of crossover between anorexia nervosa and bulimia nervosa during follow-up as well as the differential rates of full recovery reported here and elsewhere (3
) support the distinctiveness of these two eating disorders. Although one-third of those with anorexia nervosa at intake prospectively developed bulimia nervosa, many of these women were likely to cross back (i.e., relapse) into anorexia nervosa. These data raise the possibility that the transition from anorexia nervosa (particularly the binge eating/purging type) to bulimia nervosa may not represent a change in disorder but rather a change in stage of illness: in practice, the primary difference between binge eating/purging-type anorexia nervosa and bulimia nervosa is weight and the associated amenorrhea criterion. The long-term risk of relapse into anorexia nervosa suggests that a lifetime history of anorexia nervosa may carry important prognostic information; even after crossing over to bulimia nervosa, these women remain vulnerable to relapsing into anorexia nervosa. This finding supports the validity of distinguishing between anorexia nervosa and bulimia nervosa and suggests the clinical relevance of noting a lifetime history of anorexia nervosa in individuals, even when the low weight criterion is no longer met (18
The frequency of crossover between restricting-type and binge eating/purging-type anorexia nervosa suggests that these two subtypes may not be unique diagnostic groups. Previous research has indicated that the restricting type may represent an earlier phase in the course of illness than the binge eating/purging type, as those with the restricting type tend to be younger, with a shorter duration of illness (7
). Yet the data reported here suggest that both the restricting type and binge eating/purging type may be phases (often recurrent) in the illness course of anorexia nervosa, as those with the binge eating/purging type often experienced periods of dietary restriction in the absence of regular binge/purge behaviors throughout their illness, just as those with the restricting-type illness were likely to experience periods of regular binge/purge behaviors. While acknowledging that the presence or absence of regular binge/purge behavior may be clinically useful, the finding that these behaviors come and go during the course of illness in women with anorexia nervosa suggests that the subtypes are not distinctive disorders.
Yet the question of whether anorexia nervosa can and should be meaningfully subtyped warrants further consideration. Individuals with restricting anorexia nervosa who have no history of regular binge/purge symptoms and are unlikely to develop these symptoms, irrespective of follow-up duration, may constitute a small subset of those with anorexia nervosa. The nature of any meaningful differences (e.g., prognostic, genetic, and so on) between this group and those with restricting anorexia nervosa who do develop regular binge/purge symptoms during their course of illness is an area in need of further study.
Strengths and limitations of this study warrant acknowledgment. A clear strength is the comprehensive assessment of eating disorder symptoms collected over a longitudinal period of follow-up in a large sample of women with eating disorders. A unique strength is the fact that weekly symptom data were available, which allowed us to assign eating disorder diagnoses to these women and carefully examine diagnostic crossover. An additional strength is the high retention rate during the 7-year follow-up period. Among the limitations of this study was that bulimia nervosa subtypes were not examined because of the low frequency of nonpurging bulimia. Additionally, data collection for this study began in 1986, prior to the establishment of the eating disorder not otherwise specified diagnostic category. Thus the sample included only women with an intake diagnosis of anorexia nervosa or bulimia nervosa, which precluded longitudinal examination of women presenting for the treatment of eating disorder not otherwise specified. A careful examination of eating disorder not otherwise specified and its distinctiveness from partial recovery in women with a lifetime history of anorexia nervosa or bulimia nervosa is planned as a future investigation. Additional limitations were that all participants in this study were patients who sought treatment and that the mean duration of illness was approximately 6 years. Our findings may not generalize to community samples or to samples of individuals who have been ill for shorter periods. Participants received a wide range of treatment during the follow-up period (19
); the possible impact of treatment on cross-over would be of interest, although it was outside the scope of this naturalistic study.
As the diagnostic criteria of eating disorders come under increasing scrutiny in preparation for the next revision of DSM, careful examination of the current diagnostic entities is needed. The longitudinal data we present here support the diagnostic distinction between anorexia nervosa and bulimia nervosa, but they do not validate the current anorexia nervosa subtyping schema. Future research might continue to explore the longitudinal validity of anorexia nervosa, bulimia nervosa, and the anorexia nervosa diagnostic subtypes (perhaps employing different viable definitions of the subtypes) and extend these studies to include more heterogeneous samples of women, including those with a diagnosis of eating disorder not otherwise specified.