Psychopharmacology of Anorexia Nervosa, Bulimia Nervosa, and Binge Eating

B. Timothy Walsh and Michael J. Devlin

Disturbances of human eating behavior have been recognized as an appropriate focus of clinical intervention for centuries. Anorexia nervosa was clearly described as a syndrome in the late 19th century, and, in the over 100 years since, an impressive variety of behavioral and pharmacological interventions have been suggested, ranging from psychoanalysis to lobotomy, and including most of the pharmacological agents employed in psychiatry. In 1980, with the promulgation of DSM-III, an additional eating disorder—that of bulimia, now known as bulimia nervosa—was recognized. This recognition helped prepare the way for a number of important treatment studies, including trials of pharmacological interventions, in the last 15 years. With the advent of DSM-IV, another eating disorder, binge eating disorder, has been tentatively recognized and is likely to be a focus of psychological and pharmacological treatment research in the next decade.

This chapter will review our knowledge of the effectiveness of pharmacological interventions for these three disturbances of human eating behavior, and it will emphasize studies which have appeared since the publication of the Psychopharmacology: Third Generation of Progress. Much of the work in this time has focused on the treatment of bulimia nervosa and has clearly established the short-term efficacy of antidepressant medication in the treatment of this syndrome. Unfortunately, progress has not been as great in the treatment of anorexia nervosa. Work concerning binge eating disorder is just beginning.

In conceptualizing the psychopharmacological treatment of anorexia nervosa, it is useful to keep in mind the chronicity of its course in many patients (30). Treatment could theoretically include interventions at a number of different stages of the illness: preemptive treatment aimed at preventing disease onset in high-risk individuals, acute treatment emphasizing weight gain, treatment immediately following weight gain aimed at preventing relapse, and long-term maintenance treatment.

The Third Generation of Progress reviewed our knowledge of the efficacy of pharmacological interventions for anorexia nervosa in 1987 (40). At that time, virtually all published studies concerned treatment in the acute weight gain phase, usually given on an inpatient basis. While anecdotal case reports enthusiastically described impressive responses to a variety of pharmacological interventions, only a small number of controlled trials had been carried out, and these generally yielded negative or equivocal results. Unfortunately, in the ensuing 7 years, little additional information has become available. There are probably several reasons for the dearth of studies. First, over the last quarter century, a number of specialized eating disorder units have been established, and the experience of these units indicates that most patients with anorexia nervosa can achieve adequate weight gain during inpatient behavioral treatment. Most of the controlled trials reviewed in the Third Generation of Progress attempted to evaluate whether the addition of medication to structured inpatient treatment conferred any additional benefit. The fact that patients with anorexia nervosa can be, perhaps not easily, but nonetheless successfully, treated acutely on specialized units without the use of medication and the failure of controlled trials to suggest impressive additional benefit from the use of medication have tended to focus the attention of investigators away from inpatient studies of acute weight gain treatment. Instead, interest is growing in the efficacy of psychopharmacological treatment in the prevention of relapse among weight-restored patients. There have as yet been no studies concerning the role of psychopharmacology in prevention of onset of the illness or in long-term maintenance of remission.

Studies of the use of antipsychotic medications in anorexia nervosa were based theoretically on evidence that feeding is, in part, mediated by dopaminergic neurons (5), and they were based clinically on the observation that these medications tend to produce weight gain. A series of early studies (11, 13) reported that chlorpromazine, sometimes in combination with insulin to further stimulate hunger, produced more rapid weight gain and earlier hospital discharge. However, unwanted side effects included the onset of binge eating and seizures, and follow-up studies showed no benefit (14). Placebo-controlled studies of the higher-potency antipsychotics pimozide (55) and sulpiride (54!popup(ch149ref54)) provided little evidence of clinical utility. Therefore, although particular patients may benefit from treatment with antipsychotic medications, existing studies do not support their routine use. As one might expect, long-term antipsychotic therapy in patients with anorexia nervosa, as in other patients, can lead to the development of tardive dyskinesia (10!popup(ch149ref10)).

Studies of antidepressants were based on the observed comorbidity of eating and affective disorders and on the observation that semistarvation in previously healthy individuals produces depressive symptoms (23). Controlled trials of antidepressants, including clomipramine in low dosage (12, 36) and amitriptyline (6, 28), failed to provide compelling evidence for their clinical utility during the weight gain phase. Interestingly, weight gain even without pharmacological intervention produces an improvement in depressive symptoms (9, 16), and animal work suggests that starvation may inhibit responsiveness to antidepressants (51). Earlier case reports, summarized in the Third Generation of Progress, suggest that individual patients may benefit from tricyclic antidepressants but tend to tolerate them poorly, and more recent case reports (20, 27) suggest that the selective serotonin reuptake inhibitor fluoxetine may be useful. An open study of the antidepressant levoprotiline suggests improvement in various symptoms characteristic of anorexia nervosa (18). A small controlled study has suggested that lithium may enhance weight gain in inpatients with anorexia nervosa (26), but this has not been replicated.

Other clinical trials have utilized various agents that are thought to directly influence neurotransmitter systems involved in feeding behavior (see Huntgungton's Disease). The antihistaminic antiserotonergic orexigenic agent cyproheptadine has been studied in three controlled trials (25, 28, 56). The results suggest that although most inpatients do not benefit greatly from the addition of cyproheptadine, the drug may have some utility in more severely ill patients (25) and in nonbulimic anorectic patients, whereas in bulimic anorectic patients it may actually detract from clinical response (28). A small controlled crossover trial of the alpha-2 agonist clonidine, which is known to increase feeding behavior in several animal species including primates, showed no clinical benefit in patients with anorexia nervosa (8). Several case and small series reports summarized in the Third Generation of Progress (40) suggest that a variety of agents ranging from adrenocorticotropic hormone (ACTH) to zinc may enhance either the rate or the ease of weight gain, but there have been no controlled studies.

The most innovative approach to the psychopharmacology of anorexia nervosa since the publication of the Third Generation of Progress has been the idea that psychotropic medications might be useful in the post-weight-gain phase—for example, for patients who have gained to their target weight, are about to leave the hospital, and are therefore at risk of relapse. Based on several lines of evidence suggesting the involvement of serotonergic systems in the pathophysiology of eating disorders, investigators, in particular the Pittsburgh group, have studied the selective serotonin reuptake inhibitor fluoxetine in weight-restored patients. Kaye et al. (33) have reported that of 31 anorexia nervosa patients who received fluoxetine for 11 ± 6 months following hospital discharge, 29 maintained their weight at or above 85 percent average body weight. Nonbulimic anorectics appeared to benefit more than bulimic anorectics. Although no data from placebo controlled double-blind trials have yet appeared, the open study data are encouraging and underscore the utility of further investigation of clinical interventions at times other than the acute weight gain phase.

The broadening of the sphere of inquiry to include the post-weight-gain phase of anorexia nervosa and continuing investigations of the biology of anorexia nervosa to provide a rational basis for identifying promising psychopharmacological interventions are encouraging developments. However, despite these new directions, the unfortunate fact is that little concrete progress has been made in the treatment of anorexia nervosa in the last decade. Follow-up studies indicate that for many patients hospitalized with anorexia nervosa, this illness has a chronic course with impressive morbidity and mortality. It would be surprising if biological interventions were eventually not found to be of use in the treatment of this syndrome, which is characterized by manifold biological disturbances. Additional effective interventions, perhaps at different points in the course of the illness, are sorely needed.

Progress in establishing a scientific foundation for the treatment of bulimia nervosa has been much more rapid. In the early 1980s, as clinicians became more familiar with the features of this disorder, it was recognized that many individuals with bulimia nervosa were prone to mood disturbance. This observation led to open trials of antidepressant medication which were encouraging and which were promptly followed by placebo-controlled investigations. At the time of the publication of the Third Generation of Progress, the results of five double-blind, placebo-controlled trials of antidepressants in the treatment of bulimia nervosa were available. In the ensuing years, the results of an additional 13 trials have become available (see Table 1). Members of virtually all available classes of antidepressants have been examined, including tricyclic antidepressants, monoamine oxidase (MAO) inhibitors, and selective serotonin reuptake inhibitors. The patients in these studies have, with few exceptions, been adult women of normal body weight who met DSM-IIIR criteria for bulimia nervosa, who regularly induced vomiting after binge eating, and who were treated on an outpatient basis. The controlled phase of these trials was generally of short duration, most commonly 8 weeks.

Several conclusions can be confidently reached on the basis of this work. First, antidepressant medication is more effective than placebo in the short-term treatment of bulimia nervosa. This finding is supported by virtually all the trials in Table 1. It is of note that most of the studies that failed to find a statistically significant difference between active medication and placebo were ones in which the response to placebo was considerable. In other words, the improvement on antidepressant medication was usually modest and, with the small number of subjects in many of the trials, was clearly distinguishable from improvement on placebo only when the latter was minimal.

Second, despite the number of agents that have been examined, there is no evidence that one antidepressant medication or one class of medication is dramatically more effective than another in the treatment of bulimia nervosa. This conclusion must be regarded with some caution because it is based not on direct comparisons of one agent to another in a single study, but on comparisons between response rates in different studies. It is clear that factors other than which antidepressant agent was employed contribute importantly to response. For example, it can be noted from Table 1 that the change in frequency of binge eating on placebo ranges from 21% deterioration to 52% improvement. Similarly, the response to the same active agent varies substantially across the studies in which it has been used; for example, the improvement in binge frequency on desipramine ranges from 40% to 91%. These observations indicate that factors other than the choice of antidepressant have a major impact on the response of patients with bulimia nervosa. It is not clear what these other factors are, because the studies of antidepressant medication in bulimia nervosa, for the most part, have examined populations that appear comparable in terms of age, duration, and severity of illness.

Although these studies have established beyond question the superiority of antidepressant medication over placebo in the treatment of bulimia nervosa, concerns persist about the place of antidepressant medication in the therapeutic approach to this illness. One concern relates to the degree of improvement commonly achieved during antidepressant treatment. As indicated in Table 1, the average decline in binge frequency achieved during medication treatment was only about 50%, and the highest reported rate of remission was only 35%. Although improved, the average patient was significantly symptomatic at the conclusion of a course of antidepressant medication.

The long-term outcome of antidepressant treatment of bulimia nervosa also remains a significant concern. One of the earliest reports regarding the long-term response to medication was that of Pope et al. (44), who were instrumental in initiating studies of antidepressants in bulimia. In 1985, they described the follow-up of 20 patients an average of 18 months after they had been recruited to participate in a controlled trial of imipramine. Although a majority of patients reported being markedly or more improved, most had required multiple medication trials and most of those who were doing well remained on medication. Similar results were described by Fava et al. (19). In a retrospective follow-up study, they found that improvement in eating behavior symptomatology was generally maintained in 19 patients who had an initial favorable response to fluoxetine and who had been treated with fluoxetine for an average of 10 months. However, 13 of the 19 patients remained on fluoxetine, and all also received psychotherapy. These reports suggested that patients with bulimia nervosa who responded favorably to treatment with antidepressant medication often continued to do well if they remained on medication.

Two controlled studies attempted to determine how frequently bulimia nervosa could be successfully treated with a time-limited course of antidepressant medication, and they were not as encouraging. Pyle et al. (48) examined the maintenance of change in nine patients who had improved during 3 months of treatment with imipramine by randomly assigning them to 4 months of continued treatment with either imipramine or placebo. Five of six patients randomized to placebo and two of three randomized to imipramine relapsed during the succeeding 4 months.

Our own group also attempted to address the issue of the long-term response to antidepressant treatment (59). We carried out a three-phase trial in which patients were initially randomized to either desipramine or placebo and treated for 8 weeks. Patients who responded to desipramine entered an open maintenance phase during which they continued active medication for an additional 4 months or until they relapsed. Patients who continued to do well were then randomly assigned to remain on desipramine for an additional 6 months or to switch to placebo in double-blind fashion. The results of this study were disappointing in that, of 21 patients who entered the maintenance phase, 29% relapsed despite continued desipramine, and only nine patients entered the discontinuation phase. Five were assigned to continue on desipramine, and only one of these patients relapsed during the succeeding 6 months. Two of the four patients assigned to placebo relapsed during maintenance. This study and that of Pyle et al. (48) are too small to permit definitive conclusions about the appropriate duration of antidepressant treatment for bulimia nervosa. However, both studies suggested that symptomatic relapse was frequent even after several months of significant improvement on medication. These studies provide a less optimistic impression of the long-term outcome of a single course of a single medication than do the reports of Pope et al. (44) and of Fava et al. (19), possibly because in the latter studies, patients received trials of additional medications or concurrent treatment with psychotherapy.

In summary, the work conducted to date suggests there are three major clinical problems with the use of antidepressant medication as a primary intervention for bulimia nervosa. First, although a single course of antidepressant medication is superior to treatment with placebo, the response commonly falls short of remission, the desired goal. Second, among patients who initially respond favorably to antidepressant medication, there is a substantial rate of relapse despite continuation of that medication. And, third, even patients who remain improved during long-term medication treatment appear prone to relapse when the medication is discontinued.

On a theoretical level, the studies of antidepressant medications in the treatment of bulimia nervosa leave open a major question about why these drugs are effective in this disorder. The earliest work in this area proceeded on the basis of the assumption that bulimia nervosa could be viewed as a variant of an affective illness and that the pharmacological treatment of bulimia nervosa could follow principles identical to those used in the treatment of mood disturbance. Two findings suggest that this simple model is deficient. First, it has been consistently noted that there is no relationship between the pretreatment level of depression in bulimia nervosa and the response to medication. Several studies have found that the response of patients without sufficient symptoms of mood disturbance to merit the diagnosis of depression respond as well as do patients with concurrent affective illness.

Second, the large study of the Fluoxetine Bulimia Nervosa Collaborative Study Group (22) found a different relationship between fluoxetine dose and clinical response than had been previously identified in studies of depression. In this trial, patients with bulimia nervosa were randomized to receive one of three treatments: placebo, fluoxetine at 20 mg/day, and fluoxetine at 60 mg/day. Only the group receiving 60 mg/day was found to have a consistently superior response to placebo. This contrasts with studies of major depressive disorder in which a dose of 20 mg/day of fluoxetine is consistently superior to placebo. Though only a single study, this work suggests that the pharmacological treatment of bulimia nervosa may not be identical to the pharmacological treatment of affective disorder.

As studies of the antidepressant treatment of bulimia nervosa were progressing, the utility of psychological interventions was also being examined. In the early and mid-1980s, most studies of psychotherapy for bulimia nervosa examined therapies that were based to a considerable degree on cognitive behavioral principles. Studies using waiting list controls convincingly demonstrated the superiority of such psychological treatment. In addition, a few studies suggested that the gains were enduring. Although not directly relevant to the focus of this chapter, it is of note that more recent studies have begun to challenge the notion that cognitive behavioral therapy is particularly advantageous in the treatment of this syndrome. There are now indications that other forms of therapy such as interpersonal psychotherapy and supportive expressive psychotherapy, which do not specifically focus on the cognitive and behavioral disturbances characteristic of bulimia nervosa, are capable of producing comparable degrees of improvement.

The development of effective forms of psychotherapy has led to questions about the relative efficacy of pharmacological and psychological intervention for bulimia nervosa and about the potential advantages of combined treatment. In the last few years, studies have begun to address these issues. Mitchell et al. (43) randomized patients with bulimia nervosa into one of four treatment groups. One group received placebo alone, and another received imipramine alone. The two other groups received an intensive form of group psychotherapy that involved instructional, cognitive-behavioral, and more traditional group psychotherapy elements. Half of the patients in the intensive group psychotherapy received imipramine and half received placebo. Consistent with other reports of the efficacy of medication, the reduction in binge frequency in the group receiving imipramine alone was significantly greater than that in the group receiving placebo alone. However, the reduction in binge frequency in the groups receiving intensive psychotherapy was significantly superior to the reduction in binge frequency in the group which received imipramine alone. The patients receiving psychotherapy attained an impressive degree of improvement, an average reduction in binge eating of 90%. Although there was no difference in the reduction in binge frequency for the patients who had received intensive group psychotherapy with imipramine compared to those who had received the same therapy with placebo, the investigators were able to detect a drug/placebo difference in the reduction in ratings of depression and of anxiety. Thus, even though the results of psychotherapy were substantially superior to those of medication alone, there were hints of a small additional benefit for patients who had received imipramine in combination with the psychotherapy.

A second study comparing the relative efficacy of medication and psychotherapy is that of Agras et al. (2). These investigators examined individual cognitive behavioral therapy (CBT) for 16 weeks, desipramine alone, given for 16 or 24 weeks, or a combination of CBT and desipramine, the latter again given for either 16 or 24 weeks. A clear result of this study, similar to that of the study of Mitchell et al. (43), was that patients treated with medication alone did not fare as well as did patients who received psychological therapy, either alone or in combination with medication. There were hints of a small advantage for the combination of medication and psychotherapy, but these were somewhat difficult to interpret. They were largely confined to the group that received CBT and 24 weeks of desipramine. Yet, the superior outcome of this group was achieved after 16 weeks of treatment when no suggestion of added benefit was seen in the other combined treatment group that received identical treatment up to 16 weeks.

A third trial comparing medication and cognitive behavioral therapy and its combination is that of Leitenberg et al. (37). This study attempted to compare individual CBT to desipramine alone, and to the combination of these two forms of treatment given for 4 months. The investigators terminated this study prematurely, with only seven patients in each treatment group, because of an unexpectedly high dropout rate largely due to side effects among patients receiving desipramine. In the results that were obtained, no evidence of an advantage for combined treatment was apparent.

Finally, Fichter et al. (21) compared the utility of fluoxetine (60 mg/day) to placebo in the treatment of 40 patients with bulimia nervosa who were hospitalized on a specialized unit for the treatment of eating disorders and receiving an intensive course of psychological and behavioral treatment. Although both fluoxetine- and placebo-treated groups improved considerably over the 5-week trial, there was no indication of an advantage for fluoxetine over placebo.

In summary, there has been substantial progress in the treatment of bulimia nervosa in the last decade. Pharmacological treatments using antidepressant medication have been demonstrated to be efficacious, and this field has now turned to more challenging questions regarding the place of pharmacological treatment in the overall approach to this eating disorder. The available data suggest that, in general, structured psychotherapy is superior to a course of a single antidepressant medication. There are suggestions that the combination of antidepressant medication and psychotherapy may have some utility, but the advantages of a combined treatment approach over psychotherapy alone appear small and often offset by the occurrence of significant side effects, especially with tricyclic antidepressants. Our group is currently conducting a study that attempts to improve on previous designs by combining structured forms of psychotherapy with a sequential medication intervention in which patients are first treated with desipramine and, if this is ineffective or intolerable, are then treated with fluoxetine. This may enable us to determine whether a more complex but clinically realistic approach to pharmacological management is of significant benefit when provided in the context of psychotherapy.

While antidepressant medication has received the most attention, other psychopharmacological interventions for bulimia nervosa have also been explored. Lithium was examined in a controlled trial, but was found to be no more effective than placebo (31).

Three studies have examined the potential utility of fenfluramine, which augments serotonergic activity in several ways, including the stimulation of presynaptic serotonin release and the blockade of reuptake. Because serotonergic activity appears to be involved in the development of satiety, fenfluramine is a theoretically appealing candidate for the treatment of bulimia nervosa. In a small (N = 12) controlled trial lasting 6 weeks, Blouin et al. (7) found that d,l-fenfluramine at a dose of 60 mg/day was superior to placebo. In a larger trial, Russell et al. (49) examined d-fenfluramine, the more active isomer, at a dose of 30 mg/day in 42 patients with bulimia nervosa, 21 of whom were assigned to active drug and 21 to placebo. The results of this study were ambiguous. Using data from all 42 patients who were randomized, the investigators found d-fenfluramine to be superior to placebo in the effect on binge eating. However, 17 of the 42 patients withdrew prematurely, and there was no evidence of a significant drug effect compared to placebo when only data from patients who completed the 12-week trial were considered.

The same group carried out a second trial of d-fenfluramine with two important modifications (17). First, a higher dose (45 mg/day) was used. Second, all patients received an abbreviated 8-week course of CBT in addition to medication or placebo. This study was successful in reducing the dropout rate: Only 4 of 43 patients terminated the study prematurely. Both fenfluramine and placebo groups experienced a decline in binge frequency of about 50%, and there was no evidence that the addition of fenfluramine was of clinical benefit. The results of this study resemble those of studies of combined psychotherapy and antidepressant medication, which have had limited success in demonstrating that medication adds substantially to the effectiveness of psychotherapy.

Another theoretically based pharmacological intervention for bulimia nervosa is the use of opiate antagonists. There is substantial evidence that endogenous opiates are involved in the control of eating behavior (see Huntgungton's Disease), and, in a variety of paradigms, including some forms of stress-induced eating, opiate antagonists reduce food intake. Unfortunately, two placebo-controlled trials of the opiate antagonist naltrexone have failed to provide much support for the efficacy of this agent in bulimia nervosa (3, 41).

A significant development since the publication of Third Generation of Progress has been an increased interest in patients who report binge eating in the absence of attempted methods of compensation such as vomiting, laxative abuse, excessive exercise, or fasting. One formulation of this syndrome is that of Spitzer et al. (52, 53), who have (a) referred to the syndrome as binge eating disorder (BED), (b) developed preliminary criteria, and (c) demonstrated that the syndrome exists in about one-third of obese patients presenting to weight control clinics and in 2–5% of individuals in the community. The challenges of characterizing and treating patients with BED have brought investigators of eating disorders together with obesity researchers, communities that have, until recently, remained surprisingly separate.

There have been four controlled studies of the use of medications, antidepressants for the most part, in the treatment of BED or related syndromes. The basic questions posed in these studies include the following: (a) Given the demonstrated efficacy of antidepressants in bulimia nervosa, do these medications produce a reduction in binge frequency in obese patients with BED? (b) Do antidepressant medications facilitate weight loss in obese BED patients? (c) Is there a differential efficacy of medications in obese patients with BED when compared to their efficacy in obese patients without binge eating?

McCann and Agras (39) studied desipramine at an average daily dose of 188 mg in the treatment of "nonpurging bulimics," many of whom would probably meet criteria for BED. Subjects ranged in weight from the upper limits of normal to morbidly obese. Reduction in binge frequency, measured as binge days per week, was substantially greater with desipramine than with placebo (63% decrease versus 16% increase). Neither group had a substantial weight change. Sixteen-week follow-up suggested that the decrease in binge frequency largely disappeared when medication was discontinued.

Marcus et al. (38) compared fluoxetine to placebo in obese binge eaters (many of whom would likely meet criteria for BED) who were receiving standard behavior modification treatment. Fluoxetine (60 mg/day) or placebo was given for 1 year. Patients receiving active medication lost significantly more weight than those who received placebo; however, there was no difference between obese binge eaters and obese non-bingers, suggesting that the drug did not specifically target binge eating. Follow-up data suggested that patients tended to regain weight following discontinuation of fluoxetine. Although binge frequency was not reported, the bulimia subscale of the Eating Disorders Inventory changed equally for both active medication and placebo groups; this suggested that although fluoxetine enhanced weight loss, it did not enhance binge suppression.

Alger et al. (3) studied the effect of imipramine up to 200 mg versus naltrexone up to 150 mg/day versus placebo in normal-weight bulimic patients (see above) and obese binge eaters, many of whom probably met criteria for BED. Both active medications were associated with a significant reduction in binge frequency, but the effect of the medications did not differ significantly from that of the placebo. Imipramine, but not naltrexone, was associated with a reduction in binge duration among obese binge eaters. Subjects were not placed on a reducing diet, and no group showed a substantial change in weight. Although this study did not provide compelling evidence for the utility of either medication, the authors speculate that, given the effect on binge duration, imipramine may enhance compliance in dietary treatment aimed at weight reduction.

De Zwaan et al. (15) conducted a two-by-two combined treatment study in which 22 overweight binge eaters and 42 non-binge eaters received either cognitive behavioral treatment (CBT) or dietary management (DM) and either fluvoxamine (100 mg/day) or placebo. All treatment groups responded similarly with a modest weight loss during treatment and regain at 1 year follow-up. There was a trend toward greater weight regain among binge eaters than among non-binge eaters. The only evidence of efficacy of fluvoxamine was among binge eaters, who showed a greater reduction in Hamilton depression scores with fluvoxamine than with placebo. There was no evidence of a specific effect of medication on binge eating. Limitations of this study included small sample size and possibly inadequate fluvoxamine dosage.

Clearly, the treatment of binge eating among the obese is an area that requires further study. However, the results, to date, are not particularly encouraging. In terms of the original questions, only one study (39) suggested a suppression of binge frequency with medication versus placebo, only one suggested an enhancement of weight loss with medication versus placebo (38), and both of these effects were lost when medication was discontinued. None of these studies provided evidence for a specific effect of medication among obese binge eaters. These results are somewhat surprising in view of the clearly established efficacy of antidepressant medications in reducing binge frequency among patients with bulimia nervosa. However, only two of the four studies reported binge frequency during treatment, and in one of these two studies there was a marked placebo response. Future studies will benefit from (a) more careful assessment of binge frequency along with weight and (b) increased attention to the long-term maintenance of change.

This chapter has reviewed developments in psychopharmacologic treatment of eating disorders since the publication of the Third Generation of Progress. In the case of anorexia nervosa, while the yield from acute treatment studies remains limited, there has been a potentially important refocusing of attention toward the use of medication in the prevention of relapse among weight-restored patients. Studies of antidepressants in the acute treatment of bulimia nervosa have continued to suggest that virtually all classes of antidepressants are more effective than placebo in the short term. However, the available data indicate that a single course of antidepressant medication is insufficient for most patients when they are followed over a longer term. The potential utility of combining antidepressant medication and psychotherapy remains to be clarified. In patients with the newly defined binge eating disorder, the role of medications in weight reduction and binge suppression are under active investigation, but given the preliminary nature of available data, a definitive statement concerning their efficacy must be deferred to the Fifth Generation of Progress.

This work was supported, in part, by grant MH-38355 from the National Institutes of Health.

published 2000