The DSM-IV diagnostic criteria for eating disorders are useful but imperfect. Perhaps the most glaring problem is that the current criteria do not provide a category, beyond the nonspecific eating disorder not otherwise specified, for a substantial fraction of the individuals who present to clinicians for evaluation and treatment. Eating disturbances that do not meet the full DSM-IV criteria for anorexia nervosa or bulimia ner-vosa are inadequately described, and it is unclear how clinically significant eating problems are to be differentiated from other eating pathologies, and whether individuals classified as having eating disorder not otherwise specified will develop full-blown disorders. Several promising approaches to these problems have been developed in recent years, such as the Great Ormond Street criteria (Lask & Bryant-Waugh, 2000) and the categories of the Diagnostic and Statistical Manual for Primary Care (DSM-PC), and these deserve further examination. Longitudinal examinations of the course of eating disorder symptoms during adolescence and the course of associated psychological and physical problems (e.g., obesity) would also be very valuable in defining the evolution and characteristics of adolescent eating disorders.

In adults, anorexia nervosa and bulimia nervosa affect approximately 1% and 3% of women, re spectively, with rates among men estimated at one tenth of those observed in women (Hoek, 2002). The small number of methodologically rigorous epidemiological studies leads to significant uncertainty about the prevalence and incidence of anorexia nervosa and bulimia nervosa among adolescents. The peak incidence (number of new cases per year) of anorexia nervosa appears to occur in late adolescence (Hoek & van Hoeken, 2003), but the combined prevalence (number of current cases) of anorexia nervosa and bulimia nervosa appears to be somewhat less among adolescents than among adults. Methodologically rigorous studies may find that the published rates do not adequately reflect the true prevalence of eating disorders in adolescents. Epidemiological research on eating disorders among adolescents is limited in several important ways. Nationally representative samples are needed to determine more precisely how common eating disorders are among American youth. The extant (and limited) evidence suggests that ethnic minority children need to be included to gain a more accurate understanding of risk for eating disorders in non-white youth.

The epidemiological literature has, for the most part, used clearly articulated diagnostic criteria for anorexia nervosa and bulimia nervosa, and the difficulties in defining the spectrum of other eating disturbances have presented a significant barrier to describing the prevalence of other potentially important but less well–defined conditions. For example, future studies should include criteria for binge eating disorder to permit estimates of the prevalence of this syndrome. A uniform instrument for measuring eating disorder symptoms that is efficient and accurate in detecting anorexia nervosa, bulimia nervosa, and binge eating disorder and is standardized across studies would be of great value. Psychiatric interviews that “skip out” of the eating disorder questions after a negative answer may underestimate eating pathology, especially in young samples with atypical clinical presentations of anorexia nervosa or bulimia nervosa (Kreipe et al., 1995). Finally, parent reports of eating behavior might improve the detection of anorexia nervosa, a disorder in which denial is a hallmark.

Future studies should assess both current and

past eating disorders. Several population-based studies have shown that eating disorders may be transient (albeit in many cases recurrent) and point prevalence rates may therefore not fully reflect the extent of eating pathology in adolescents (Patton, Coffey, & Sawyer, 2003). Even when the adolescent's eating disorder is time limited and nonrecurring, it may represent a marker for psychopathology that conveys important clinical information.

Psychiatric comorbidities are common among both adolescents and adults with anorexia nervosa and bulimia nervosa, and include mood disorders, anxiety disorders, and substance use disorders. However, the relationship between anorexia nervosa, bulimia nervosa, and comorbid disorders is unclear. The outcome of adolescent patients with anorexia nervosa who receive early treatment appears better than that of patients who do not; however, those patients who remain ill have high rates of psychiatric comorbidity and are at risk for premature death. Most adolescent anorexia nervosa patients improve or get well, but a substantial percentage remains permanently symptomatic. The data on the course and outcome of adolescent bulimia nervosa are very limited. Diagnostic migration occurs frequently from anorexia nervosa-restricting subtype (AN-R) to anorexia nervosa-binge purge subtype (AN-B/P), and from AN-B/P to bulimia nervosa. It is currently not possible to identify those patients likely to migrate.

Future studies of adolescents with eating disorders should include individuals with comorbidities, such as substance use disorders, to aid in developing treatment strategies for these dual-diagnosis conditions. Studies of the course and outcome of adolescent bulimia nervosa are needed, and early identification and intervention strategies need to be developed.

While most medical complications associated with eating disorders are reversible with nutri tional rehabilitation and cessation of the binge–purge cycle, there are indications that growth retardation, osteopenia, and, possibly, structural brain changes are not entirely reversible. Studies probing structural brain changes in anorexia nervosa and their relationship to neuropsychological changes are needed. In addition, there is a pressing need to develop efficacious treatments for osteopenia among adolescents with anorexia nervosa.

There is only one evidence-based treatment for adolescents with anorexia nervosa, the Mauds-ley method of family therapy, whether delivered in a conjoint or separated format. The empirical evidence supporting the treatment is limited, however. Whereas a subgroup of anorexia nervosa patients may have an inherently good prognosis, it is clear that significant numbers of these patients do not do well. For example, Eisler et al. (2000) found in their study (in which all the patients received the Maudsley method of family therapy) that 15 of 40 patients were judged to have a “poor” outcome on the Morgan-Russell scales and 4 patients had to be admitted to the hospital because of continuing weight loss. At the end of the study by Robin et al. (1999), the authors note that “even with comprehensive, multidisciplinary interventions such as those evaluated in this study, not all adolescents with anorexia nervosa will improve. Twenty to 30 percent of the patients did not reach their target weights, and 40% to 50% did not reach the 50th percentile of BMI by 1-year follow-up. Clinicians and researchers will need to continue to develop

innovative approaches to helping these more resistant patients” (p. 1489).

A potentially promising treatment for adolescent anorexia nervosa patients is cognitive-behavioral therapy (CBT). This therapy is the leading evidence-based treatment for bulimia nervosa (National Institute for Clinical Excellence, 2004; Wilson & Fairburn, 2002), a disorder that has much of the same psychopathology as that of anorexia nervosa (Fairburn & Harrison, 2003; Fairburn, Cooper, & Shafran, 2003). Additionally, CBT is already used with adults with anorexia nervosa (e.g., Garner, Vitousek, & Pike, 1997), and a recent cognitive-behavioral conceptualization of anorexia nervosa may pertain to adolescent patients, given its emphasis on the early stages in the evolution of the disorder (Fairburn, Shafran, & Cooper, 1999). Finally, CBT has been successfully used to treat other psychiatric disorders in adolescence (Kazdin, 2003; Kendall, 2000).

If CBT were to be developed as a treatment for these patients, it would need to be adapted in certain ways. It would need to be based on a model of the maintenance of anorexia nervosa, focusing on the processes involved in recent-onset cases (e.g., Fairburn, Shafran, et al., 1999). The therapy would need to take account of the developmental psychology of adolescence and the specific concerns of adolescents, and it would need to be adjusted to accommodate the developmental variability seen among adolescents (Holmbeck et al., 2000; Weisz & Hawley, 2002). It would also need to involve the patient's family and possibly the school.

A challenging issue in the study of new treatments for anorexia nervosa is the choice of the comparison, or “control,” treatment. In theory, it might be useful to compare the effect of a new intervention with that of no treatment at all or a waiting-list condition. While such a comparison neatly controls for the effect of time alone, it is difficult to justify a delay of treatment for individuals with a disorder having such serious medical and psychiatric morbidity. Furthermore, documentation that a new treatment is superior to doing nothing does not provide strong evidence of specific clinical utility. A comparison between two interventions likely to be useful (e.g., the Maudsley method and suitably adapted CBT) may be difficult to interpret without controls for the effect of time and for the nonspecific effects provided by any intervention. The notion of a “treatment-as-usual” comparison group has appeal, but the definition and implementation of such an intervention in the context of a research study are far from clear.

Controlled trials are needed to identify and evaluate the efficacy of psychological treatments for adolescents with bulimia nervosa. Both CBT and interpersonal psychotherapy (IPT) have been shown to be effective for adult bulimia nervosa patients. These treatments should be adapted and applied to adolescents, because CBT and IPT have been successfully adapted to treating adolescents with other problems. The case is particularly compelling for CBT, as it is the treatment of choice for bulimia nervosa in adults, its adaptation to depressed and anxious adolescents has received the most study and enjoys the most empirical support, and it meshes well with the generic and specific considerations governing psychological treatment of adolescents.

Alternative psychological therapies should also be explored. An adaptation of the family-based treatment developed by Lock, Le Grange, Agras, and Dare (2001) is an obvious candidate for study, given its apparent efficacy with adolescent anorexia nervosa. Research is also needed to assess the comparative efficacy of evidence-based psychological treatments, such as CBT, IPT, or family therapy, with antidepressant medication and their combination (sequencing).

Few controlled studies have evaluated the utility and safety of pharmacological treatments for adolescents with eating disorders, although medications are frequently used in the clinical treatment of these patients. Antidepressant medications have demonstrated efficacy in reducing

binge eating and vomiting behaviors for adults with bulimia nervosa, but additional study will be necessary before any conclusions can be reached about the use of these medications with younger patients. There is currently no evidence for the efficacy of pharmacological treatments in low-weight adults with anorexia nervosa. However, recent reports have suggested a utility of atypical antipsychotic medications, such as olanzapine, with adolescent anorexia nervosa patients, and future research should evaluate these medications in a controlled manner.

For both adolescents and adults with anorexia nervosa, the risk of posthospitalization relapse is approximately 30%–50%. The Maudsley form of family therapy appears to have efficacy for preventing relapse among adolescent anorexia nervosa patients, but family therapy may be most effective for patients who develop anorexia nervosa before age 18 and who have less than 3 years duration of illness. Although there are no data on the efficacy of CBT for relapse prevention among adolescents, this treatment has support for relapse prevention in adult patients. A single controlled study in adults suggests that antidepressant medication may reduce the rate of relapse following weight restoration, but there are no data on the utility of this intervention to prevent relapse among adolescents.

Adolescents with bulimia nervosa have not been the focus of clinical treatment trials, and studies that include adolescent patients are not sufficient to draw conclusions about relapse prevention for these patients. Little is known about the efficacy of psychological or pharmacological treatments for adolescents with bulimia nervosa, both for the acute treatment of the disorder and the prevention of relapse.

Methodological and logistical challenges hamper progress in the development of effective relapse prevention interventions. Operationalized and consistent definitions of treatment response, relapse, remission, and recovery, and a standardized assessment battery would enhance the study of both initial interventions and re lapse prevention. Clinical trials require tremendous resources, and the failure to develop consensus in the field on core terminology and assessment procedures will continue to hinder development of empirically supported treatments. Discussions of relapse and relapse prevention lead to dichotomous distinctions: either an individual has responded to treatment or not; either an individual has relapsed or not. In reality, change in clinical status is continuous, which complicates the establishment of thresholds and standardized classifications. Several authors have attempted to address this important issue (Orimoto & Vitousek, 1992; Pike, 1998).

Intensive but expensive initial treatments for anorexia nervosa, such as inpatient or partial hospitalization, are arguably the most successful in achieving weight restoration, especially among severely and chronically ill patients. Weight restoration is an essential goal in achieving recovery and is the first step in preventing relapse. However, economic pressures on reducing health-care expenditures, at least in the United States, are limiting patients' ability to receive sufficient care to achieve weight restoration. Studies of the short-and long-term costs and benefits of interventions of a range of intensity would be extremely valuable in helping to define which treatments are most cost-effective.

The encouraging work on the utility of the Maudsley method for adolescents with anorexia nervosa should be pursued in studies comparing this intervention to other standard and novel approaches. Empirically supported psychological treatments for adults with eating disorders, including CBT and IPT, should be adapted to the treatment of adolescents. This process should include collaboration between treatment researchers and developmental psychologists who study adolescence. Treatments should be designed to address issues specific to adolescent patients, including motivation, cognitive processing, interpersonal functioning, body image, control, and family issues. The utility and safety of psychopharmacological interventions for adolescents