analyze the treatment options available for that disorder as well as the rationale for their use based on the current understanding of the biological aspects of these conditions.

Psychological Treatment of Eating Disorders

G. Terence Wilson Rutgers, The State University of New Jersey Carlos M. Grilo Yale University School of Medicine

Kelly M. Vitousek University of Hawaii

Significant progress has been achieved in the development and evaluation of evidence-based psychological treatments for eating disorders over the past 25 years. Cognitive behavioral therapy is currently the treatment of choice for bulimia nervosa and binge-eating disorder, and existing evidence supports the use of a specific form of family therapy for adolescents with anorexia nervosa. Important challenges remain. Even the most effective interventions for bulimia nervosa and binge-eating disorder fail to help a substantial number of patients. A priority must be the extension and adaptation of these treatments to a broader range of eating disorders (eating disorder not otherwise specified), to adolescents, who have been largely over- looked in clinical research, and to chronic, treatment- resistant cases of anorexia nervosa. The article highlights current conceptual and clinical innovations designed to improve on existing therapeutic efficacy. The problems of increasing the dissemination of evidence-based treatments that are unavailable in most clinical service settings are discussed.

Keywords: anorexia nervosa, bulimia nervosa, binge-eating disorder, cognitive behavioral therapy, dissemination

According to the American Psychiatric Associa-tion’s (1994) Diagnostic and Statistical Manualof Mental Disorders (DSM–IV), anorexia nervosa and bulimia nervosa are the two best characterized eating disorders. Patients who do not meet criteria for either anorexia nervosa or bulimia nervosa may be diagnosed as “eating disorder not otherwise specified” in the DSM–IV classification system. Binge-eating disorder, for which there are provisional diagnostic criteria, has been the most intensively researched disorder within the category of eat- ing disorder not otherwise specified. In this article we summarize the evidence on the efficacy of current psycho- logical treatments for eating disorders and discuss ways in which their application and effectiveness in clinical prac- tice might be enhanced.

ANOREXIA NERVOSA Anorexia nervosa is defined by the successful pursuit of thinness through dietary restriction and other measures, resulting in body weight below the normal range (usually operationalized as � 85% of expected weight or a body mass index [BMI] � 17.5 kg/m2). Patients’ views of their symptoms are complex and variable, often combining feel- ings of being “too fat” with pride in the achievement of

thinness and restraint. Patients are intensely fearful of losing control and becoming overweight; over time, nearly half succumb to binge eating. Semi-starvation brings about profound and predictable changes in mood, behavior, and physiology. These include depression, social withdrawal, food preoccupation, altered hormone secretion, amenor- rhea, and decreased metabolic rate. Anorexia nervosa typ- ically begins during adolescence and principally affects girls and young women; its prevalence rate among females is 0.3% (Hoek & van Hoeken, 2003). Aggregate results from long-term follow-up studies indicate that nearly 50% of patients eventually make a full recovery, 20%–30% show residual symptoms, 10%–20% remain severely ill, and 5%–10% die of related causes (Steinhausen, 2002).

Treatment Efficacy The most salient fact about psychotherapy research on anorexia nervosa is that there is remarkably little evidence to review. Over the past 20 years, only 15 comparative trials have been completed and published. The persistent deficit of controlled treatment research in anorexia nervosa is attributable to distinctive features of the disorder, includ- ing its rarity, the presence of medical complications that sometimes require inpatient management, and the extended period of treatment necessary for full symptom remission in established cases. Patients’ ambivalent attitudes about recovery compound these challenges at every phase of research, making it more difficult to recruit samples, pre- vent attrition, and secure participation in follow-up assess- ments (Agras et al., 2004).

Family Therapy

Family therapy is the most extensively researched treat- ment for anorexia nervosa, contributing at least one cell to more than half of all randomized controlled trials. In gen- eral, the results have been encouraging; unfortunately, they

G. Terence Wilson, Rutgers Eating Disorders Clinic, Rutgers, The State University of New Jersey; Carlos M. Grilo, Department of Psychiatry, Yale University School of Medicine; Kelly M. Vitousek, Department of Psychology, University of Hawaii.

Preparation of this article was supported in part by Grant RO1 MH63862 to G. Terence Wilson and Grants R01 DK49587 and K24 DK070052 to Carlos M. Grilo. We are grateful to Tanya Schlam and Robyn Sysko for their helpful comments on the manuscript.

Correspondence concerning this article should be addressed to G. Terence Wilson, Eating Disorders Clinic, Rutgers University, Piscataway, NJ 08854. E-mail: tewilson@rci.rutgers.edu

199April 2007 ● American Psychologist Copyright 2007 by the American Psychological Association 0003-066X/07/$12.00 Vol. 62, No. 3, 199–216 DOI: 10.1037/0003-066X.62.3.199

are widely misunderstood (Fairburn, 2005; Vitousek & Gray, 2005).

The best studied approach is a specific form of family therapy known as the Maudsley model (Dare & Eisler, 1997; Lock & le Grange, 2005). A published manual out- lines treatment procedures in detail (Lock, le Grange, Agras, & Dare, 2001). As applied to adolescent patients, the intervention involves 10–20 family sessions spaced over 6–12 months. The recommended “conjoint” format specifies that all family members should be seen together. In the first phase of treatment, parents are directed to take complete control over their anorexic child’s eating and weight and are coached to find effective means of doing so. Once the child begins to comply with parental authority, external control is gradually faded. In the later stages of therapy, the adolescent’s right to age-appropriate autonomy is explicitly linked to the resolution of her eating disorder.

The Maudsley model was first tested as a means of preventing posthospitalization weight loss in different sub- groups of anorexia nervosa patients in a study by Russell, Szmukler, Dare, and Eisler (1987). The study yielded sev- eral striking results. In the subset of younger patients with more recent onset, conjoint family therapy produced an impressive rate of recovery (90% symptom-free at 5 years) and was far more effective than a dynamically oriented individual approach (Eisler et al., 1997; Russell et al., 1987). For patients with an older age at onset or a longer history of illness, neither treatment appeared beneficial.

Two of the three conclusions suggested by this small study have been supported by subsequent research. The higher-than-expected rate of recovery has also been evident in case series (e.g., le Grange, Binford, & Loeb, 2005) and randomized controlled trials (e.g., Eisler et al., 2000; le Grange, Eisler, Dare, & Russell, 1992; Lock, Agras, Bry-

son, & Kraemer, 2005) of family therapy for adolescents with anorexia nervosa. Such favorable results, however, may simply reflect the characteristics of the samples to which this approach has been delivered (Fairburn, 2005; Vitousek & Gray, 2005). In both controlled trials and naturalistic catchment-area studies, outcomes for young adolescents are much more encouraging than the aggregate 50% recovery rate cited for all patients with anorexia nervosa (e.g., Nilsson & Hagglof, 2005; Steinhausen, 2002).

The second confirmed finding of the Russell et al. (1987) study is that symptom duration is a strong predictor of response to family therapy (as it is for other modes of treatment). In a trial of family therapy for adolescents with relatively recent onset, patients who attained a good out- come had been symptomatic for just 8 months at the start of treatment, compared with 16 months for those with intermediate or poor outcomes (Eisler et al., 2000). At the other end of the prognostic spectrum, use of the Maudsley model in an adult sample with an average duration of 6 years yielded minimal clinical improvement in the majority of patients (Dare, Eisler, Russell, Treasure, & Dodge, 2001).

The third notable finding of the Russell et al. (1987) study—that the Maudsley model was much more effective than individual treatment for adolescent patients—has little support. Subsequent research by the same group of inves- tigators has focused on examining different formats and intensities of the Maudsley approach (Eisler et al., 2000; le Grange et al., 1992; Lock et al., 2005) rather than testing it against alternative models of treatment. Two other teams did compare a similar version of family therapy with indi- vidual treatment, finding it slightly more effective than ego-oriented psychotherapy in an adolescent sample (Robin, Siegel, Koepke, Moye, & Tice, 1994) and equiv- alent to cognitive behavioral therapy in a mixed sample of adolescents and young adults (Ball & Mitchell, 2004). At present, then, there is little basis for the widespread belief that family therapy is specifically efficacious for adoles- cents with anorexia nervosa (Fairburn, 2005). There are, however, other sound reasons for adopting the approach. The Maudsley model has been examined more often than any other psychological treatment and is readily dissem- inable.

The National Institute for Clinical Excellence (NICE, 2004) in the United Kingdom has conducted arguably the most comprehensive and rigorous evaluation of the avail- able evidence on the treatment of eating disorders. The NICE evaluation process includes professionals from dif- ferent disciplines and applies consistent standards across medical specialty areas. Recommendations of best clinical practice are assigned a grade from A (reflecting strong empirical data) to C (expert opinion in the absence of strong data). For anorexia nervosa, NICE (2004) specified that family interventions directly addressing the eating disorder should be offered to younger patients (although not necessarily in place of individual therapy). This rec- ommendation was awarded a grade of B for the strength of the supporting evidence; reflecting the general paucity of

G. Terence Wilson

200 April 2007 ● American Psychologist

treatment research, all other suggested guidelines for an- orexia nervosa were given a grade of C. Few clinicians would disagree that parents should be included in the treatment of young patients. Optimal means of doing so, however, have yet to be determined.

The only evidence-based argument against using the Maudsley method of conjoint family therapy comes from studies conducted by its proponents. Two randomized con- trolled trials have compared the conjoint format to a “sep- arated” version in which the anorexic child and her parents attend different sessions (Eisler et al., 2000; le Grange et al., 1992). In both trials, there was a trend favoring the theoretically less-preferred “separated” format over the conjoint model, which reached significance for the subset of families rated high in the expression of negative emotion (Eisler et al., 2000). It is not clear either why the published manual strongly recommends the conjoint model despite these findings or why it is being used preferentially in ongoing research.

Cognitive Behavioral Therapy Cognitive behavioral therapy is the most frequently tested individual treatment for anorexia nervosa, having been included in six randomized controlled trial designs. The results are difficult to interpret, however, as four used abbreviated forms of the approach and two could not be analyzed because of attrition from the comparison condi- tions.

A cognitive behavioral therapy framework for concep- tualizing and treating anorexia nervosa was described ini- tially by Garner and Vitousek1 (Garner & Bemis, 1982, 1985; Garner, Vitousek, & Pike, 1997). A number of ex- pansions and alternative perspectives have been presented (e.g., Fairburn, Cooper, & Shafran, 2003; Fairburn, Shaf- ran, & Cooper, 1999; Kleifield, Wagner, & Halmi, 1996; Wolff & Serpell, 1998), most of which are at least broadly consonant with the original proposal.

The model outlined by Garner and Vitousek overlaps substantially with Fairburn’s (1985) analysis of bulimia nervosa, reflecting the shared view that these disorders have core features in common. Many of the same strategies are included in both approaches, with key differences in emphasis for anorexia nervosa being shaped by the impor- tance of motivational issues, the problems associated with semi-starvation, and the need for substantial weight gain (Garner et al., 1997). Considerable attention is allocated to enhancing motivation for change and engaging patients as active collaborators (Vitousek, Watson, & Wilson, 1998). The recommended approach specifies 1–2 years of individ- ual therapy for patients who begin treatment at low weight and approximately 1 year for those who are weight-re- stored.

Three studies have compared a cognitive behavioral therapy condition with one or more alternative psychother- apies (Ball & Mitchell, 2004; Channon, de Silva, Hemsley, & Perkins, 1989; McIntosh et al., 2005). In each, no clear differences were found between cognitive behavioral ther- apy and the comparison conditions. Across trials, the gen- eral pattern was for patients in most conditions to improve

to some degree without achieving full recovery. Unfortu- nately, each of these studies implemented a version of cognitive behavioral therapy for anorexia nervosa that has not been described or recommended in the literature. All offered truncated courses of treatment (18–25 sessions) that differ from those specified by cognitive behavioral therapy experts (Fairburn et al., 2003; Garner et al., 1997).

Interpretation of the other three trials is hampered by the poor showing of the nonpsychological treatments with which cognitive behavioral therapy was compared. Two attempted to examine the effects of cognitive behavioral therapy relative to nutritional counseling. One failed after 100% of participants assigned to nutritional counseling dropped out and refused to participate in follow-up assess- ments; almost all of those receiving cognitive behavioral therapy completed treatment (Serfaty, Turkington, Heap, Ledsham, & Jolley, 1999). In the second, cognitive behav- ioral therapy was superior to nutritional counseling for preventing relapse after inpatient treatment (Pike, Walsh, Vitousek, Wilson, & Bauer, 2003). Compared with patients assigned to nutritional counseling, patients receiving cog- nitive behavioral therapy were less likely to drop out or be withdrawn (22% vs. 73%), slower to relapse, and more likely to achieve a good outcome (44% vs. 7%). The third study was a large multisite trial comparing cognitive be- havioral therapy, fluoxetine, and combined treatment (Halmi et al., 2005). The medication-alone condition was rejected by such a high proportion of patients that it was not possible to analyze the relative effectiveness of treatments.

The strongest conclusion that can be drawn from this second set of studies is that the use of nutritional counsel- ing or medication in the absence of psychotherapy is con- traindicated for anorexia nervosa patients, within or outside the conduct of research. Ironically, the choice of weak comparison conditions made it difficult to gauge the effi- cacy of cognitive behavioral therapy. There were indica- tions in all three trials that cognitive behavioral therapy (or perhaps psychotherapy more generally) does further the crucial objectives of increasing engagement and persis- tence.

There is no empirical basis for the widespread use of antidepressants with this population. Fluoxetine is ineffec- tive with low-weight patients (Attia, Haiman, Walsh, & Flater, 1998), and initial indications that it might support maintenance of gains after inpatient treatment (Kaye et al., 2001) have not been confirmed. A large, well-controlled trial showed no evidence that fluoxetine was superior to placebo or offered any incremental benefit to cognitive behavioral therapy in a sample of weight-restored patients (Walsh, Kaplan, et al., 2006).

Current Challenges and Future Directions Challenges to the identification of evidence-based treat- ments for anorexia nervosa are formidable. The record is discouraging: few comparative trials; inconclusive results;

1 Kelly M. Vitousek’s former name is Kelly M. Bemis.