Dr. Kaye and Colleagues Reply

To the Editor:

We appreciate the interest and comments of Dr. Godart et al., Drs. Micali and Heyman, and Drs. Södersten and Bergh. We regret that space limitations preclude a comprehensive review of all literature relating to anxiety and eating disorders. Nonetheless, we note that there are more similarities than differences between our results and those of Dr. Godart et al. once we account for differences in sampling and possible cultural differences.

There are some notable differences in the types of anxiety disorders diagnosed between our two studies, although the overall prevalence of anxiety disorders was similar. The most notable differences rest with the observed prevalence of OCD in our group and the absence of significant differences in the prevalence of OCD between participants with anorexia and patients with bulimia nervosa—both of which surprised us as well. It is important to note that our study group was collected from nine different sites in North America and Europe and thus was unlikely to reflect a single-site bias. In an earlier study (1), we found that the form of OCD most commonly observed in individuals with eating disorders is characterized by a focus on symmetry, exactness, and order. It is likely that researchers who do not routinely assess this variant of OCD may detect fewer cases of OCD. We acknowledge that particular factors may have enriched our study group for certain traits. For example, it is possible that families with more obsessional traits may have been more successful in completing the extensive diagnostic and blood collection protocols of our genetic study or that certain disorders may have been more common in families that have two or more members with an eating disorder. Regardless, both our study and that of Dr. Godart et al. highlight that there is a strong relationship between eating disorders and anxiety disorders, although much remains to be learned about shared phenomenology or biology.

In answer to Drs. Micali and Heyman, the diagnosis of OCD was made with the Structured Clinical Interview for DSM-IV (SCID), not with the Yale-Brown Obsessive Compulsive Scale. The latter was used to identify specific obsessions and compulsions and to rate the severity of the disorder. We stipulated that by using the Yale-Brown Obsessive Compulsive Scale first, we were likely to identify more cases by reminding subjects of symptoms, such as symmetry/exactness, which might be overlooked in response to a general screen, such as the SCID’s. The method of ascertainment of OCD in the comparison subjects was different so that a direct comparison is not possible. However, an earlier study that used similar methods (reference 4 of the letter of Dr. Godart et al.) found two women with OCD in a group of 44 women in the community.

The SCID assesses current (in the past 6 months) generalized anxiety disorder only. Furthermore, the screen asks whether subjects have been “particularly nervous or anxious” in the past 6 months. People who are always nervous or anxious often deny being “particularly” anxious in the past 6 months (there is no difference from their usual elevated state of anxiety). Thus the SCID may miss the diagnosis in subjects having the disorder whose current level of anxiety is not higher than their usual level. That is a shortcoming of the instrument. Furthermore, our study group contained many recovered subjects, whereas the group of Dr. Godart et al. consisted of those with current eating disorders. Because anxiety may be elevated in those who are currently ill with eating disorders, this could account for a significant amount of the discrepancy.

Observed comorbidity in a group can reflect several underlying processes:

Our study group was not appropriate to tease out the underlying cause of comorbidity in eating and anxiety disorders, nor did we claim to do so. In our genetic study, we were less interested in addressing the underlying process of comorbidity that can be addressed more effectively in population-based samples, and more interested in identifying patterns of comorbidity in families enriched for anorexia nervosa that may assist us with the refinement of phenotypes for genetic studies.

As for the comments of Drs. Södersten and Bergh, our article did not break down by subgroups the rates of childhood onset of OCD before the onset of eating disorders. Our calculations show that when the entire group of subjects with anorexia nervosa is taken into consideration, at least 14% of the subjects with anorexia nervosa had an onset of OCD before they developed anorexia nervosa. In comparison, 24% of the subjects with anorexia nervosa and bulimia nervosa and 22% of those with bulimia nervosa had an onset of OCD before they developed an eating disorder when the entire study group is studied. We should have noted in the article that the focus of the study was on relative pairs affected by bulimia nervosa and that the anorexia nervosa group was relatively small and might have been skewed because of how the subjects were selected. In a more recent study (unpublished data), we collected 744 subjects with anorexia nervosa and bulimia nervosa in which the entry criterion was a diagnosis of anorexia nervosa. Fifty-seven percent of the entire anorexia nervosa/anorexia nervosa and bulimia nervosa group had a lifetime diagnosis of OCD, and 36% of the entire group had an onset of OCD before the onset of anorexia nervosa or anorexia nervosa and bulimia nervosa. These data support our contention that a premorbid onset of OCD is common in people who later develop anorexia nervosa.

The point of this article was to identify susceptibility factors that may precede the onset of anorexia nervosa. As noted in our article, other recent studies clearly show that anxious, obsessional, and perfectionistic traits commonly occur in childhood and predate the onset of anorexia nervosa. It is also important to recognize that obsessional traits that are known to be common in anorexia nervosa might be described as OCD, obsessive-compulsive personality disorder, a rigid perfectionistic temperament, or other diagnoses depending on the assessment instrument and perhaps the bias of the investigator. It is well recognized by experts in the field of behavior that a DSM diagnosis, while a necessary attempt to categorize illness, may not necessarily reflect how behavior is coded in the brain. We need to continue to better characterize these traits and understand their pathophysiology rather than get caught up in unanswerable arguments about whether anorexia nervosa is a variant of another DSM diagnostic category.