To the Editor:

We read with interest the report by Dr. Kaye et al. on the rates of comorbid anxiety disorders in individuals with eating disorders. The suggestion that child/adolescent-onset anxiety disorders may be associated with later eating disorders is particularly relevant to child mental health.

In women with eating disorders, the rates of OCD have been estimated at between 3% and 40%, according to diagnosis and group type. Some studies have also investigated the onset of OCD symptoms in women with eating disorders (1, 2).

Dr. Kaye et al. found an OCD lifetime prevalence of 41% in subjects with eating disorders; the onset of OCD was reported as preceding the eating disorders in 23% of the cases. The authors acknowledged that these rates were not comparable with rates found in community samples. They suggested that this might be because of the use of the Yale-Brown Obsessive Compulsive Scale as well as the Structured Clinical Interview for DSM-IV (SCID) to determine lifetime rates of OCD. Dr. Kaye et al. proposed that OCD cases could have been missed had the SCID been used on its own because the subjects did not recognize the nature of their symptoms and did not endorse the SCID screening probe for OCD. Dr. Kaye et al. did not comment on how the Yale-Brown Obsessive Compulsive Scale was used to diagnose OCD. The Yale-Brown Obsessive Compulsive Scale has been validated and is used to determine the severity of current OCD; its scoring is based on current impairment, distress, resistance to obsessions and compulsions, and time spent on these. It is not a diagnostic instrument (3), and to our knowledge, it has not been validated as a lifetime measure. We would be curious to see the rates of OCD diagnosed by this method in the comparison group.

If the average prevalence of eating disorders in females is 0.3% for anorexia nervosa and 1% for bulimia nervosa (4) and 23% (i.e., 0.3% of the females) had OCD before having an eating disorder, it would predict the development of an eating disorder in large numbers of girls with child/adolescent-onset OCD, or about 30%–100%. This would not be consistent with current clinical experience, although prospective studies of children with OCD are few. Dr. Kaye et al. acknowledged that their study group might have been biased because the subjects were recruited for a genetics study and were included if they had a relative with an eating disorder. We would like to highlight a caveat by Godart et al. (5) that "comorbidity studies have to be designed according to their specific goal, rather than being secondary results from other types of studies."