Recovery from Eating Disorders

Abstract

The article by Von Holle et al. in this issue is one of a series of reports emerging from a large and impressive group of investigators assembled to collect information on the genetics of eating disorders [1]. Having rigorously obtained an extensive array of information from a large number of probands and family members, the investigators are appropriately mining the data in numerous ways. In this paper they turn their attention to temporal patterns of recovery among 901 women with anorexia nervosa and bulimia nervosa. The retrospective data were collected from subjects at the time of their assessment and entry into the genetics studies. The investigators report what appears to be, at first blush, a remarkably low rate of improvement: fewer than 20% of participants reported recovery at the time of assessment. However, this observation must be interpreted in light of the study’s definitions and methods. The investigators chose to rate as recovered only those individuals who denied being below normal weight, as well as reporting no behavioural symptoms suggestive of an eating disorder, for a full 3 years. While the investigators’ rationale for this very conservative criterion is solid, it obviously reduced the fraction of subjects who had recovered. For example, because the average time between reported onset of the disorder and the assessment was 12 years, individuals, on average, had to begin their full recovery no more than 9 years after onset. Other studies of recovery from eating disorders have generally utilized less stringent criteria, making comparisons across studies difficult. In addition, as noted by the authors, most reports of recovery are derived from clinical populations, and examine time to recovery from entry into treatment. In the current report, duration is measured from the time of reported onset. Furthermore, individuals were recruited expressly to participate in a genetics study, and were at various stages of illness and recovery at the time of their assessment. Conceivably, individuals with more chronic courses may have been more motivated to participate. In any case, it is very difficult to apply these estimates of time to recovery to patients presenting for treatment. As the authors emphasize, both the conduct of their study and the interpretation of their results are limited by the lack of a consensus in the eating disorders field on precisely what constitutes recovery. Unfortunately, this is not a newly recognized problem, nor is it confined to eating disorders [2]. Because of the multifaceted nature of psychiatric disorders, and the paucity of objective, easily obtained markers of illness, it has proven difficult for consensus to form around definitions of relapse, recovery and remission. In the eating disorders field, the wide range of definitions limits the ability to pool data across studies, much less to have confidence about comparisons between studies. However, within the Von Holle et al. study the interpretation of different temporal patterns of recovery between eating disorders is more straightforward and of interest. In the early years after onset, the odds of recovering from anorexia nervosa were greater than those of recovering from bulimia nervosa but, over time, the probabilities of recovery shifted to favour bulimia nervosa. This pattern is consistent with the results of other types of studies, such as those based on follow up of clinical samples, and echo the anecdotal reports of many clinicians. These varied sources of information suggest that an adolescent B. Timothy Walsh, Professor of Psychiatry, College of Physicians & Surgeons, Columbia University, Director, Division of Clinical Therapeutics, New York State Psychiatric Institute