Naturalistic follow-up studies, which do not control for specific treatment effects, estimate a relapse rate of approximately 30% for patients with bulimia nervosa (Herzog et al., 1999; Keel & Mitchell, 1997). None of the published clinical trials evaluating psychological treatments for the acute symptoms of bulimia nervosa specifically focused on relapse prevention for this disorder; instead, relapse prevention was typically an integrated component of the initial intervention. Follow-up data on CBT and IPT, two evidence-based treatments for bulimia nervosa, indicate that the two psychotherapies do not differ in rates of relapse at 1-year follow-up. Some studies suggest that therapeutic changes are well maintained for most individuals who respond well to initial CBT or IPT treatment, with the most enduring recovery being reported by individuals who achieve complete remission of binge eating and purging by the end of treatment (Agras et

al., 2000, Fairburn et al., 1995). However, another study has indicated that as many as 30% of individuals who are abstinent from binge eating and purging at the end of CBT treatment report some resurgence of symptoms during 1-year follow-up (Halmi et al., 2002). It is unclear whether these patients should be classified as “relapsed,” on the basis of the severity of their symptoms. A similar rate of 30% relapse has been reported among individuals who had responded to an eating disorders day program (Olmsted, Kaplan, & Rockert, 1994).

A significant fraction of individuals who receive CBT or IPT fail to achieve full remission of binge eating and purging at the end of treatment. Data suggest that the risk of relapse is greater for this group than for those who achieve complete abstinence of binge eating and purging (Halmi et al., 2002). Therefore, treatment interventions for bulimia nervosa should be targeted not only at those individuals who fail to respond to initial treatment but also those who have a significant but incomplete response to treatment. Maintenance and relapse prevention treatments should promote further recovery and also mitigate against lapses and relapse for these individuals.

As described above, trials evaluating antidepressant medications for underweight patients with anorexia nervosa have failed to show a difference between active medication and placebo for the treatment of eating disorders or mood symptoms. Antidepressant medications may lack efficacy in anorexia nervosa because of the neurochemical disturbances associated with low body weight (Attia et al., 1998). Despite their apparent lack of utility for underweight patients, antidepressant medications may be useful in the prevention of relapse after weight gain.

One placebo-controlled study has addressed this issue (Kaye et al., 2001). Following inpatient

hospitalization, patients were randomized to receive either fluoxetine or placebo in a double-blind fashion. Some patients also received psychological treatment, which was not standardized. Fluoxetine-treated patients were more likely to complete the trial (63% completed) than were patients receiving placebo (16% completed, p = .0001). Patients who completed the trial tended to experience more weight gain and psychological improvement than patients who did not complete the trial. The average age of the patient sample was 22.5 years old, and it is not clear whether any adolescent patients were included.

In the last 15 years, more than a dozen RCTs have demonstrated that antidepressant medications are effective in the treatment of adult bulimia nervosa patients (Zhu & Walsh, 2002). Although antidepressant medications have been shown to reduce bulimic symptoms in the short term, the role of continued pharmacological treatment in sustaining clinical improvement over time is unclear. Several controlled trials have examined the efficacy of antidepressant medications in preventing relapse among bulimia nervosa patients.

Some studies have evaluated continuing treatment with pharmacotherapy to prevent relapse after an initial positive response to medication (Pyle et al., 1990; Romano, Halmi, Sarkar, Koke, & Lee, 2002; Walsh et al., 1991), and an additional study randomized patients to receive either medication or placebo for relapse prevention after receiving a course of inpatient treatment (Fichter, Kruger, Rief, Holland, & Dohne, 1996). A consistent finding across studies has been the significant rate of symptomatic relapse despite continued pharmacological treatment. There is also an indication that TCAs, specifically imipramine and desipramine, and the SSRI fluoxetine may diminish the rate of relapse for patients maintained on antidepressant medications, compared to patients maintained on placebo. These studies suggest that, despite a significant rate of relapse on antidepressant medications, the rate of relapse is greater when medication is discontinued after a few months (Romano et al., 2002). Although the studies evaluating pharmacotherapy as a means to prevent relapse have generated similar results, there have been a relatively limited number of studies in this area, with modest sample sizes and large dropout rates across trials.

All four controlled trials evaluating pharmacotherapy to prevent relapse enrolled only adult bulimia nervosa patients. Therefore, it is unclear whether the results of these studies are applicable to an adolescent population, or if there are special considerations for using antidepressant medications with younger patients for the prevention of relapse.

Pharmacological interventions may be useful in the prevention of relapse of anorexia nervosa following initial treatment. However, replication of the Kaye et al. (2001) results is necessary before firm conclusions can be drawn about the benefits of antidepressant medications for adults with anorexia nervosa. If such a benefit is established for adults, it will be important to extend studies to an adolescent population. Clearly, for some adult patients with bulimia nervosa who initially respond to medication, symptomatic relapse occurs despite continued pharmacotherapy. Therefore, continued treatment with medication after an initial positive response cannot guarantee against relapse. The data from placebo-controlled studies do suggest that when bulimia nervosa patients respond to a medication and are maintained on that medication, they experience lower rates of relapse than those of patients who are switched to placebo (Pyle et al., 1990; Romano et al., 2002; Walsh et al., 1991). Therefore, continuation of an effective pharmacological intervention may reduce the rate of relapse but does not ensure against the return of bulimic symptoms. The question remains as to the optimal length of time to maintain a patient on medication to prevent symptomatic relapse. Additionally, given the absence of data in adolescent samples, it is not known if the pattern of results from the controlled studies of pharmacotherapy for relapse prevention apply to a younger population.

Ideally, patients should be identified at the earliest possible point in the course of the disorder, and treatment should begin as soon as the adolescent, the parent(s), or other professional(s) recognizes a clinically significant eating problem. As implied in the previous section, treatment should frequently be initiated before symptoms have become sufficiently severe to meet full diagnostic criteria for anorexia nervosa or bulimia nervosa. Among the factors to be considered in initiating treatment are the intensity, severity, and duration of symptoms, and the motivation of the adolescent and the family for treatment.

A fundamental and noncontroversial tenet is that treatment should occur in the least restrictive setting in which effective treatment can be provided. A primary real-world consideration is the availability of treatment settings to which the adolescent has access. Larger cities are more likely to have university-based programs, with a full spectrum of treatment options such as outpatient clinics, intensive outpatient and partial hospitalization programs, and inpatient hospitalization. Although referral to more intensive treatment settings, such as residential facilities or inpatient units, may be resisted by adolescents or their parents because of the distance from home, the disruption of family life or schooling, or financial burden, this option may be necessary if other types of treatment are not effective. Treatment options in smaller towns or rural areas are often limited to therapists with varying degrees of interest and expertise in treating adolescents with eating disorders. The skills and interests of the adolescent's treatment providers help to determine where an adolescent will be treated, as some primary care physicians may not feel comfortable monitoring the physical health of adolescents with eating disorders, and some therapists may limit their practice to adults. In these situations, adolescents who might otherwise be treated in their home community may need to be referred to a specialty program. It is best if an appropriate treatment team or program is available locally, allowing an adolescent to live at home and engage in outpatient therapy while also remaining in school and continuing to develop important peer relationships. The challenge for the provider is to determine the balance between ideal treatment and available treatment.

Even when services are available, there are very limited data to guide the practitioner in determining the most appropriate type and duration of clinical services for anorexia nervosa. A recent study suggests that more expensive, intensive inpatient treatment early in the course of anorexia nervosa is associated with reduced relapse and long-term personal, social, and financial costs (Striegel-Moore, Leslie, Petrill, Garvin, & Rosenheck, 2000). For adults with bulimia nervosa, an initial brief and less intensive treatment, followed by more intensive and specialized care for nonresponders within a stepped-care framework, as noted above, might be effective (Garner & Needleman, 1997; Wilson et al., 2000).

Currently, there are no agreed-upon specific treatment protocols for adolescents with eating disorders to guide practitioners in matching the treatment setting and intensity to the patient's clinical status. Instead, adolescents tend to begin in outpatient treatment settings, with visits to medical and mental health services, then progress to more intensive treatment approaches if they do not have a positive response to treatment.