To the Editor: In their letter, Dr. Johnson and his colleagues raise an important issue for discussion, namely, a potential selection bias in the published outcome studies on anorexia nervosa due to referral of the patients to specialized centers and the predominant inclusion of inpatients. However, by referring to their clinical experience, my discussants do not provide convincing arguments that this is necessarily the case.

My review covers a wide range of almost 50 years of internationally published outcome studies. In this period, anorexia nervosa has required specialized treatment so that referral to expert centers has been the rule rather than the exception. Inpatient treatment has been the predominant mode of intervention, and treatment policies favoring day clinics and ambulatory practices have appeared only in the recent past. Currently, it is unclear whether the latter interventions are restricted to less severe and subclinical cases only. Prospective cohort studies, both in a single center (1) and in international multicenter studies (2, 3), have reflected the seriousness of anorexia nervosa by showing that the patients spent 25%–30% of the entire follow-up period in either inpatient or outpatient treatment. A large proportion of these patients required repeated hospitalizations. One of the very rare community-based studies (4) showed that after 10 years, 27% of the patients still suffered from an eating disorder and more than one-third had other psychiatric disorders.

These empirical observations argue strongly against the high remittance rate of at least 50% within 1–3 years. Furthermore, the data in my review showed that, depending on duration of follow-up, not 50%–70% but rather 33%–73% of the patients recovered with only very limited period effects over the past 50 years. The inclusion of the improved cases would not result in an average of 90% but, rather, only 80%. However, this combination of data would imply a problematic underestimation of clinical problems that remain in improved cases, leaving aside other psychiatric disorders, additional psychosocial problems, and the need for further treatment. As stated in my review, anorexia nervosa did not lose its relatively poor prognosis in the 20th century.