and focuses largely on the patient's eating and weight. Here the parents are helped to take control over the way the patient eats. In the second phase, once the patient's weight is under control, responsibility for weight management is transferred to the patient and treatment focuses on more general family and individual concerns.

The comparison individual treatment used by Russell et al. (1987) was devised specifically for this study. In content, it was based on conventional posthospitalization follow-up appointments (as practiced at the Maudsley Hospital), but the sessions were more frequent and lasted longer than usual, to match the intensity of the family treatment. The individual treatment is probably best characterized as a form of supportive psychotherapy that encourages patients to eat healthily and maintain an appropriate weight.

The results of the Russell et al. (1987) study favored the family therapy approach. At the end of treatment (1 year after discharge from the hospital), 6 out of 10 patients who received family therapy were judged to have a good outcome, as assessed via the Morgan-Russell scales (Morgan & Russell, 1975), compared with 1 out of 11 patients who received the control treatment (p < .02). Also, the family therapy patients had a better outcome in terms of weight regain (from their prehospitalization weight); the percentage weight regain in the two treatment conditions was 25.5% and 15.5%, respectively (p < .01). Similarly, the family therapy patients were better at maintaining their new higher weight, with 5 out of 10 patients keeping their weight above 85% average body weight, compared with 1 out of 11 patients who received the control individual psychotherapy (p < .05). There was also a lower dropout rate among the patients who received family therapy (1/10, vs. 7/11 in the control condition, p = .024).

A major strength of this study was that the patients were followed up 5 years after the completion of treatment (Eisler et al., 1997). At this time point, both patient groups had done well; the mean percentage of average body weight was 103.4 ± 13.2% in the group treated with family therapy and 94.4 ± 16.8% in the control group (p = ns). In terms of overall outcome on the Morgan-Russell scales, the results continued to favor the family therapy group, with 9 out of 10 patients having a good outcome, vs. 4 out of 11 patients in the control group (p = .024). This study is limited by its modest scale and the post-hospitalization design, but has had a major influence on the design of more recent investigations and on current treatment recommendations.

The later studies from the Maudsley group have focused on family therapy alone, the premise being that family therapy is “established as an effective treatment for anorexia nervosa in adolescence” (Dare & Eisler, 2002, p. 317). The next investigation (Le Grange, Eisler, Dare, & Russell, 1992) was a pilot study for a subsequent trial (Eisler et al., 2000). The goal was to compare two different ways of delivering the Maudsley method of family therapy: one involved all the family being seen together (subsequently termed “conjoint family therapy,” the original method), the other consisted of separate treatment sessions for the patient and the parents (subsequently termed “separated family therapy”). In contrast with the Russell et al. (1987) study, family therapy was provided from the outset of treatment rather than after a period of hospitalization. This study is of greater relevance to the routine management of patients with anorexia nervosa, most of whom are not admitted to a hospital.

This pilot study involved just 18 patients (mean age 15.3 ± 1.8 years; mean duration of illness 13.7 ± 8.4 months; mean percentage of average body weight 77.9 ± 7.6%). Both groups responded well, despite receiving a modest number of treatment sessions (8.9 ± 4.1 sessions over 6 months), with substantial weight regain and improvement on various measures of psychopathology. Not surprisingly, given the small sample size, there were no statistically significant differences in outcome between the two groups.

The third study in the Maudsley series (Eisler et al., 2000) followed from the Le Grange et al. (1992) pilot study. It involved a comparison of the same two family-based treatments, but on a larger scale. Forty patients were randomized to the two treatments, their mean age being 15.5 ± 1.6 years and mean duration of disorder being 12.9 ± 9.4 months. Treatment took place over 1 year and involved on average 16 sessions. The

conjoint sessions lasted 1 hour, whereas the separated sessions lasted 90 minutes.

Once again, both groups of patients improved markedly; their percentage average body weight increased from 74.3 ± 9.8% to 87.0 ± 13.0% (p = .001), and the equivalent figures for body mass index (BMI) increased from 15.4 ± 2.0 kg/m² to 18.5 ± 3.6 kg/m² (p < .001). There was a substantial decrease in eating disorder features—for example, the total score on the Eating Attitudes Test (Garner & Garfinkel, 1979) decreased from 47.7 ± 25.7 to 19.7 ± 16.1 (p < .001), and there was improvement on the Morgan-Russell scales. The analysis of differences in outcome between the two treatments revealed few statistically significant findings, although the pattern of the findings suggested that separated family therapy might be more potent at addressing the specific psychopathology of eating disorders, whereas conjoint family therapy might be more effective at addressing general psychopathology, such as depressive and obsessional features. With only 20 patients per treatment condition, however, the study did not have sufficient power for comparisons of this type. Nevertheless, the findings do suggest that family meetings of the type required for conjoint family therapy may not be needed for the Maudsley method to achieve its effects.

The fourth RCT was by Robin and colleagues (Robin, Siegel, Koepke, Moye, & Tice, 1994; Robin et al., 1999). They compared “behavioral family systems therapy,” a treatment similar to the original Maudsley method, with “ego-oriented individual therapy,” a psychodynamically oriented treatment in which patients are seen individually, with occasional supportive sessions for their parents. In the latter condition there was little or no direct emphasis on changing eating habits or increasing body weight.

Thirty-seven patients were randomized to the two treatments, the mean ages of the family therapy and individual therapy groups being 14.9 years and 13.4 years, respectively (p < .05), and their baseline BMI being 15.2 ± 1.8 kg/m² and 16.6 ± 2.1 kg/m², respectively (p = .038). All the patients had developed anorexia nervosa within the previous 12 months. The two treatments were more intensive and multifaceted than those provided by the Maudsley group, and involved 12 to 18 months of treatment with weekly sessions for half of the treatment and sessions every other week thereafter. In addition, both groups of patients saw a dietician who prescribed a diet designed to restore body weight at a rate of 1 lb/week. Furthermore, those patients whose weight was below 75% of ideal or who had significant cardiac problems were hospitalized at the outset and received a structured refeeding program until they reached 80% of their target weight and were medically stable. This applied to 58% of the family therapy group and 28% of those receiving individual therapy (p = .099). While in the hospital, the patients also received their assigned form of psychotherapy.

The outcome of both groups was positive, both at the end of treatment and 1 year later. There was one statistically significant difference between the groups: patients in the family therapy group had a greater increase in BMI. The mean BMI posttreatment and at 1 year follow-up was 19.9 ± 1.9 kg/m² and 20.7 ± 2.7 kg/m² in the family therapy group, and 18.9 ± 1.9 kg/m² and 19.8 ± 3.1 kg/m² for those receiving individual therapy. On all other measures of outcome there were no statistically significant differences between the two groups. At the end of treatment, two thirds of the patients had reached their target weight. By the end of 1-year follow-up, 80% of those who had received family therapy had reached their target weight; there was no such increase for those who had received individual therapy. In considering these findings, it is important to note that it is not possible to attribute the changes observed specifically to the two psychotherapies received. All the patients also received extensive dietary advice, and many were hospitalized during the initial stages of treatment (especially those receiving family therapy).

The most recent of the five RCTs (Geist, Heinmaa, Stephens, Davis, & Katzman, 2000) compared two 16-week family-based interventions. The interventions occurred in the context of considerable additional treatment, including an initial period of inpatient treatment (lasting on average 6 weeks) that involved an assertive refeeding program, as well as milieu therapy, and individual and group psychotherapy. In addition, following discharge from the hospital pa

tients had continuing medical and nursing contact that was designed to encourage further weight regain. A further complicating factor is that many patients and their families declined to take part in the study: 59% of the eligible patients refused, and only 29 patients entered the trial. The two treatment groups did not differ in their outcome.

There has been little research on the treatment of adolescents with anorexia nervosa. The studies that have been conducted are small, with the largest study including just 20 patients per treatment condition. Thus the power of these studies to detect differences between treatments is minimal, and it is difficult to evaluate the findings of these studies. Three larger studies (I. Eisler, personal communication, 2003; S. G. Gowers, personal communication, 2002; J. Lock, personal communication, 2002) are currently under way, and may provide additional information about the efficacy of treatments for adolescents with anorexia nervosa.

Another limitation of this research is the quality of the assessment measures used. None of the studies has employed standardized and psychometrically sound instruments of the type routinely used in adult eating disorder treatment trials. As a result, it is difficult to gauge the true extent of the patients' improvement.

In addition, limited data are available on the longer-term effects of treatment. Such data are important, because not only is relapse into anorexia nervosa common but the eating disorder may evolve into bulimia nervosa or an eating disorder not otherwise specified (EDNOS) (Fairburn & Harrison, 2003). To assess the frequency of diagnostic migration and determine whether treatments differ in their ability to influence the long-term course of the disorder, repeated assessments are required and measures capable of characterizing any form of eating disorder must be used.

None of the studies in this area has included a delayed-treatment (“waiting list”) control condition. It is conceivable that once adolescent patients and their parents request help, changes have already begun to take place that will lead to symptomatic improvement. And even if this were not true of the majority of patients, it might be true for a significant minority. However, the serious psychological and medical morbidity of anorexia nervosa makes the employment of a waiting-list condition ethically problematic, and there has been very little discussion of what might constitute appropriate control conditions against which new interventions for anorexia nervosa should be compared.

It is widely accepted that family therapy is the treatment of choice for adolescents with anorexia nervosa (e.g., National Institute for Clinical Excellence [NICE], 2004). This is surprising given the modest evidence to support it. Only two studies have compared family therapy to another form of treatment (Robin et al., 1999; Russell et al., 1987), and the findings of the second are difficult to interpret.

The superiority of family therapy over individual therapy has not been clearly established. In the family treatment used in the Russell et al. (1987) trial, great emphasis was placed on getting patients to eat well and maintain a healthy weight; the control treatment did not have the same focus on eating and weight. The same is true of the two treatments studied in the Robin et al. (1999) trial. Thus, in both studies any differences in outcome between the family-based treatment and individual therapy could have been a result of their relative emphasis on eating and weight rather than the modality of the treatment.

As noted earlier, research on the treatment of adolescent anorexia nervosa has concentrated on a very particular form of family therapy. Two ways of delivering this treatment have been compared (Eisler et al., 2000; Le Grange et al., 1992), but no other type of family therapy has been adequately tested. It is important that clinicians be aware of this when deciding how to treat their adolescent patients. It is of note that the Maudsley group has progressively modified their

family-based treatment and now view their original method as “slightly out of date” (Dare & Eisler, 2002; p. 318). They favor treating groups of families at one time (so-called “multi-family group therapy”; Dare & Eisler, 2000), not least because this form of treatment is well received by patients and their families, and are in the process of evaluating this method.

Interpersonal psychotherapy (IPT) was originally developed by Klerman, Weissman, Rounsaville, and Chevron (1984) as a short-term treatment for depression. The primary focus of IPT is to help patients identify and modify current interpersonal problems. As adapted for bulimia nervosa (Fairburn, 1997b), IPT focuses exclusively on interpersonal issues, with little or no attention directed to the modification of binge eating, purging, disturbed eating, or overconcern with body shape and weight. Specific eating problems are viewed as a means of understanding the interpersonal context that is assumed to be maintaining them.

Two major comparative outcome studies of adult bulimia nervosa patients demonstrated that at the end of treatment IPT was significantly less effective than CBT (Agras, Walsh, Fairburn, Wilson, & Kraemer, 2000; Fairburn et al., 1993). At 1-year follow-up, however, the difference between the two treatments was no longer statistically significant. In the absence of a control condition it is not possible to attribute the improvement associated with IPT to any specific treatment effect. However, in the study by Fairburn et al. (1995), both IPT and CBT fared significantly better than a suitable comparison treatment (a form of behavior therapy without the cognitive features of CBT) over the course of follow-up. Given that the behavior therapy treatment was equivalent in the amount of therapist contact and ratings of suitability and expectancy, this single study provides specific evidence of the efficacy of IPT for bulimia nervosa.

A variety of psychological treatments other than CBT or IPT are commonly used to treat bu limia nervosa (e.g., psychodynamic therapy, family therapy; Garner & Garfinkel, 1997), but none has been systematically evaluated in controlled research. None can be considered an evidence-based treatment (NICE, 2004).

It can be predicted that CBT will prove comparably effective for adolescents with bulimia nervosa, as conceptually and procedurally similar forms of CBT that were originally developed as treatments for adults with anxiety disorders and major depression have been readily adapted to adolescent populations. Cognitive-behavioral therapy has been shown to be reliably effective in the treatment of adult anxiety and mood disorders (Barlow, 2002; Hollon, Thase, & Markowitz, 2002; Nathan & Gorman, 2002). Manual-based CBT is as effective as antidepressant medication as an acute treatment of panic disorder and major depression, and has more sustained effects if medication is discontinued (Barlow, 2002). The CBT interventions that have been successfully used in treating adults have been adapted to adolescents.

Manual-based CBT for adolescents with major depression results in greater improvement and faster remission than being assigned to a waiting-list or alternative forms of psychotherapy, including family and supportive therapy, at the end of acute treatment (Curry, 2001). Cognitive-behavioral therapy has also been successfully used to prevent the onset of depression in at-risk adolescents with no prior history of depression (Clarke et al., 2001). There is also evidence of the efficacy of CBT in treating anxiety disorders (Donovan & Spence, 2001); a large RCT found CBT to be significantly more effective than a waiting-list control at post-treatment on a variety of outcome measures. Therapeutic improvement was maintained at 1-year follow-up (Kendall et al., 1997). Replications of this CBT treatment for anxiety disorders in children have shown maintenance of treatment effects for up to 6 years (Kazdin, 2003).

Although not studied as extensively as treatments for adult disorders, psychological therapies for depression and anxiety disorders in ad