Eating disorders are commonly occurring illnesses (1) and have many associated complications, including high rates of suicide (2). The focus of treatment development has been on psychotherapy, particularly the use of cognitive-behavioral therapy (CBT) (3). The use of CBT in bulimia nervosa has received an "A" evidence grade in the National Institute for Health and Clinical Excellence guidelines (4), although CBT has shown less clear evidence of benefit for anorexia nervosa.

The treatment literature has at least two clear limitations. First, treatment development has focused on bulimia nervosa and, to a lesser extent, anorexia nervosa, in spite of the fact that the majority of individuals with eating disorders in clinical settings meet criteria for eating disorder not otherwise specified (5). Although eating disorder not otherwise specified is the most common eating disorder diagnosis, no evidence-based treatments have been established for it (other than for binge eating disorder, a specific form of eating disorder not otherwise specified). Second, while CBT has clearly demonstrated efficacy in bulimia nervosa and leads to significant improvement of symptoms, the majority of patients receiving treatment in most trials are not symptom free at the end treatment (6). Thus, there is room for improvement in efficacy.

In this issue, Fairburn and colleagues (7) report the results of using a newly refined CBT in a diagnostically broad sample of individuals with eating disorders. A total of 154 patients with bulimia nervosa or eating disorder not otherwise specified and a body mass index above 17.5 participated in this trial. Participants were randomly assigned to a waiting list control condition or to one of two treatment conditions: a focused form of CBT that concentrated on eating-related psychopathology or a broader form of CBT that used several treatment sessions to address mood intolerance, clinical perfectionism, low self-esteem, or interpersonal difficulties; the focus of sessions in the broad treatment was determined by individual case formulation. Treatments were provided for 20 weeks with a 60-week follow-up. The results showed substantial effects for the treatments, with slightly over half of treated patients reporting eating disorder symptoms within one standard deviation above the community mean as measured by the Eating Disorders Examination (8). By contrast, patients assigned to the waiting list control condition showed little change in clinical status. While eating disorder diagnosis (bulimia nervosa versus eating disorder not otherwise specified) did not predict response to the two forms of treatment, the broader treatment was more efficacious in patients with mood intolerance, clinical perfectionism, low self-esteem, or interpersonal difficulties.

The results of this trial represent a significant advance in our knowledge about the treatment of eating disorders. Because of the challenges involved in treating these disorders, a new empirically supported treatment is most welcome. Notably, this new treatment provided response rates in the upper range of those reported in previous trials of other treatments (9). The results are especially important given that this study had less strict inclusion and exclusion criteria, and therefore participants might have been inherently less treatment responsive. Also, the indication of some treatment specificity, wherein non-eating disorder clinical issues appeared to be a useful guide for predicting treatment effectiveness, has high potential for clinical utility. Perhaps most important, in a sample that was more heterogeneous and clinically relevant than samples studied in previous trials, this study is among the first to demonstrate the efficacy of a manual-based treatment for eating disorder not otherwise specified. The inclusion of participants with eating disorder not otherwise specified will, we hope, be emulated in future trials. The use of a diagnostically broad sample also dictated the use of alternative outcome measures to those used in the past (for example, in bulimia nervosa trials, changes in frequency of binge eating or purging), which is also a useful step forward.

While the treatment used in this study represents an important advance, it should be noted that not all participants responded, which underscores the need for development of other treatments and perhaps indicates room for further refinement of this treatment. Fairburn et al. have elsewhere proposed the use of a single diagnostic entity for eating disorders (10). One of many lines of evidence that could be used to test this concept is treatment response. To that end, the demonstration of comparable efficacy for this new treatment in individuals with a BMI below 17.5 would provide useful support.