Forty years of progress in the treatment of the eating disorders

Abstract

This paper reviews progress in the treatment of the eating disorders between 1968 and 2008. In 1968, no evidence-based treatments were available for any of the eating disorders, and binge eating disorder had not yet been described. In 2008, there are still no evidence-based treatments for anorexia nervosa, although a specific form of family therapy (based on the Maudsley model) appears promising. Both psychotherapeutic and psychopharmacological evidence-based treatments are now available for bulimia nervosa and binge eating disorder. Therapist-assisted self-help is a promising approach to treatment because it may reduce the costs of treatment, allow for more adequate dissemination to areas lacking specialty resources and also provide a basis for algorithm-driven approaches to treatment.

• Anorexia nervosa
• Binge eating disorder
• Bulimia nervosa
• Eating disorders
• Evidence-based treatment

Relatively little was known about the eating disorders in 1968. Although anorexia nervosa (AN) had been described for centuries [1], no evidence-based treatments were available. Bulimia nervosa (BN) was scarcely known and binge eating disorder (BED) and subclinical syndromes namely, eating disorder not otherwise specified (EDNOS) had not been described. Gerald Russell was one of the first to describe BN in 1979 [2]. Although some case histories describing aspects of the syndrome were already in the literature, it had taken Russell from 1972 to 1978 to collect 30 cases, something that could be done in a few months today. BED, first called non-purging BN [3], was not formally described until 1992 [4], hence nothing was known about treatment of this condition in 1968, although cases of binge eating had also been recorded through the centuries. Similarly, the concept of EDNOS was introduced with DSM-IV in 1994 [5].

A search of PubMed using the term “eating disorders” revealed a rapid increase in publications related to eating disorders over the past 40 years. As can be seen in Fig. 1, the increase began in the early to mid-1980s and has continued exponentially to the present. This increase coincided with a rapid increase in cases of BN presenting to clinics in the early 1980s [6]. As a result of the increased interest in eating disorders, the International Journal of Eating Disorders was founded in 1982 and undoubtedly contributed to the increase in the number of eating disorder publications.

Fig. 1.  Results of a search of PubMed for papers that include the term “eating disorders” in the title or abstract between 1968 and 2008.

Treatment for the eating disorders in 1986

Table 1 summarizes the state of treatment in 1986 for the two extant eating disorders: AN and BN. As can be seen, there were no evidence-based treatments for either disorder, although a few small-scale treatment trials had been published for both disorders. In the case of AN, most of the trials were small-scale and had shown negative results. For BN, positive results were beginning to emerge in small-scale placebo-controlled trials for antidepressant medication. However, no controlled trials of a psychotherapeutic intervention for BN had been published.

Table 1.  The state of treatment for eating disorders, 1968.

Disorder	Evidence-based treatments available (one or more large-scale controlled trials)	One or more controlled trials
Anorexia nervosa	None	Few relatively small-scale trials mostly negative results
Bulimia nervosa	None	Few small-scale trials of antidepressants
Binge eating disorder	N/A	N/A

In the next sections, progress in treatment research over the last 40 years will be briefly reviewed for each of the three eating disorders. It should be noted that much of the progress in eating disorders has been in areas other than treatment such as nosology, epidemiology, psychopathology, medical complications, assessment, prevention, neurobiology and theory, which are outside the scope of this paper.

Progress in the treatment of AN

There are several major problems that militate against the success of treatment outcome studies in AN. The first is the relative rarity of the disorder and hence the difficulty in recruiting participants, often resulting in inadequate sample sizes. For example, in a multisite study of adults with AN [7], recruitment at each site peaked in the second year of the study and then rapidly fell to close to zero [8], hence four intervention sites were insufficient to produce the required sample size. Thus, in future studies a larger number of sites will be needed in order to obtain the necessary sample size. Recruitment of adolescents appears to be easier [9]. A further problem is the large drop-out rate found in studies of adults with AN, who are presumably more chronically ill than adolescents, because of the reluctance of such patients to engage in treatment that might lead to weight gain. Drop-out rates may reach 40% or more, essentially militating against an outcome analysis because drop-outs are likely to be non-random across treatments. Drop-out rates in child and adolescent populations are lower, probably because of parental insistence on treatment attendance. Another problem, given the severity of AN and the high rates of medical complications and death, is that a no-treatment comparison or a placebo control would be regarded as unethical. Hence, the specificity of any particular treatment is difficult to assess.

Viewing AN in more general terms, it seems reasonable to consider adolescent and adult AN separately. The adolescent variant appears easier to treat with lower rates of drop-out from treatment than is the case with adults. It seems likely that duration of illness, a poor prognostic factor, together with increasing cognitive rigidity makes the disorder more difficult to treat in adulthood than in adolescence. Hence, treatment in adolescence may prevent advancement to the more chronic adult state. Table 2 shows the current status of treatment for AN.

Psychotherapy for adolescent AN

Although there has been some progress, especially in the treatment of adolescent AN with family therapy [9–12], most studies have either a small sample size or a high drop-out rate. In a study of adolescents with AN examining the necessary dosage of family therapy, a relatively short course of treatment was found to be as effective as a longer course with double the number of sessions [9], both at the end of treatment and at a mean of 4 years of follow-up [10]. The majority of participants in both groups were doing well at follow-up. However, there has been no comparison of the specific family therapy used (based on the Maudsley model) with another type of family therapy, and the only comparison with individual therapy had a very small sample size finding no differences between family and individual therapies possibly due to lack of power [12]. Two studies in progress are aimed at answering these questions. Until such studies have been completed, it is impossible to know whether the “Maudsley” family therapy is specifically effective, or whether any individual or family psychotherapy would produce similar results.

Psychotherapy for adult AN

In the case of adult AN, the studies have had either a small sample size or large drop-out rates or both, hence not allowing definitive conclusions to be drawn from any published study or from the literature as a whole. Indeed, the authors of one large-scale study with a high drop-out rate argued that no further treatment studies of adult AN should be carried out until the problem of poor treatment acceptance has been solved [7].

Pharmacological treatment of AN

The situation is no better for the pharmacological treatment of AN, despite a fairly large number of controlled trials over the past 40 years, most of which have found no clinically relevant effect for medication [13]. A recent large-scale study comparing fluoxetine and placebo in weight-restored adult patients found no difference between groups in the proportions of patients who maintained their weight over a 1-year trial. However, over half the patients had dropped out of the study by the end of treatment. Similarly, no differences in weight gain were found between fluoxetine and placebo in an inpatient trial in which patients were being treated with other modalities [14]. There have been no controlled studies of pharmacological agents for adolescent AN; hence no medication can be recommended for adolescents.

Summary of treatments for AN

There are no evidence-based psychotherapies or psychopharmacological treatments available for either adolescent or adult patients with AN. Family-based treatment for adolescent AN appears promising, and should probably be regarded as the treatment of choice at this point but awaits the results of further studies.

Progress in the treatment of BN

The situation is somewhat better for the treatment of BN than for AN. BN is more prevalent than AN, making recruitment of larger sample sizes possible. Moreover, adults appear to accept treatment well with drop-out rates in psychotherapy studies between 5% and 27%, lower than the drop-out rate in AN. Similarly, BN patients are more accepting of medication than is the case in AN, although drop-out rates are somewhat higher than for psychotherapy.

Psychotherapy for BN

There are a sizeable number of well-designed outcome studies of psychotherapy in the literature [15]. Overall, CBT has been shown to be superior to no treatment, other psychotherapies such as supportive–expressive therapy [16], interpersonal psychotherapy (IPT) [17] and antidepressant medication [18], [19]. There is some variation in drop-out rates across treatments, with IPT and dialectical behavior therapy (DBT) having rates at the lower end of the range, suggesting that they are more acceptable than CBT. Cognitive–behavioral therapy (CBT) leads to abstinence from binge eating and purging in about 25–35% of cases, with over half of patients no longer meeting criteria for BN. CBT is more effective than IPT for BN at the end of treatment, however; at 1-year follow-up, both treatments appear equally effective [17]. Dialectical behavior therapy appears to be a useful treatment for BN in early studies [20]. However, more research is needed to establish DBT's effectiveness.

More recently, there have been a number of controlled studies of therapist-guided self-help (GSH), usually based on CBT, although many of these studies had relatively small sample sizes [21]. Although there has been considerable variation in the number of sessions, experience of therapists and the self-help manual used, it can be concluded that GSH is superior to a wait-list control group in reducing binge eating and purging, and less certainly may be as effective as full-scale CBT. Abstinence rates range from 9% to over 60% for BN, attesting to the variations in study design and populations studied.

In one of two controlled outcome studies of adolescent bulimics (including those not meeting full criteria for BN), GSH was found to be superior to family therapy [22]. A previous study of adolescents with BN had found that family therapy was superior to supportive–expressive therapy [23]. The advantage of GSH is that it can be provided by less skilled therapists, is shorter in duration and is therefore less costly and easier to disseminate. In addition, GSH is less burdensome for patients because it takes up less of their time and is less expensive. It may also be useful to use self-help as the first step in the treatment of BN, following up with CBT only for those who do not respond to the first step. This may form the beginning of a treatment algorithm for BN if medication was also included. More research is required to test such algorithms.

Pharmacological treatments for BN

The research status of antidepressant medication, both the tricyclic antidepressants and the serotonin reuptake inhibitors, especially fluoxetine, has also changed for the better over the past 40 years. The largest controlled study (398 participants) examined two dosages of fluoxetine (20 and 60 mg) and compared active drug with placebo. The larger dose of medication was more effective than the smaller dose and fluoxetine was superior to placebo [24]. However, a 1-year maintenance study with fluoxetine found that most participants (87%) stopped taking their medication and dropped out of the study [25]. Hence, as in many medication trials, long-term maintenance of efficacy is a problem. Although fluoxetine is the only antidepressant approved by the FDA for use in BN, a number of other antidepressants have also been shown to be effective [26].

Summary of treatments for BN

Effective treatments are now available for BN and some studies suggest that in the near future a treatment algorithm may emerge. CBT, antidepressant medication particularly fluoxetine, and IPT can be recommended as evidence-based treatments, although IPT is slower to act than CBT (Table 2). Because CBT appears to be more effective than medication and acts more quickly than IPT, depending on the patient's preferences, CBT would appear to be the treatment of choice. Smaller scale studies suggest that adding medication to CBT may be marginally beneficial in reducing both depression and bulimic symptoms [27].

There are few studies of adolescents with BN, an area of research that needs attention. In addition, the available treatments have much room for improvement in terms of the proportion of individuals who stop binge eating and purging, and in the maintenance of improvement.

Progress in the treatment of BED

The amount of treatment research on BED is somewhat less than for BN because the syndrome was not well described until the appearance of DSM-IV. However, many of the treatments found effective for BN are also effective for BED [28]. BN and BED differ in their response to placebo conditions, with BED having relatively high rates of such response [29]. The apparently high response rates to treatment for BED are likely due in part to the placebo response. Most patients with BED are also overweight or obese, hence both binge eating and weight are treatment targets.

Psychotherapy treatments for BED

CBT and IPT appear equally effective in reducing binge eating for BED, although they do not lead to clinically significant weight loss. The response to IPT also differs from that of patients with BN, in that studies have shown that IPT is equivalent in its effectiveness in reducing binge eating to CBT both at end-of-treatment and follow-up [30]. CBT was also shown to be superior to medication placebo in one controlled outcome study [31]. This is an important finding because CBT had not been compared with a placebo condition before this study. Hence, it was unclear whether any psychotherapeutic treatment was superior to placebo and therefore had specific therapeutic effects. Fewer studies of self-help have been reported for BED than for BN, although the results should encourage further research.

Pharmacological treatments for BED

The place of antidepressant medication in the treatment of BED is less certain than in BN because there are fewer large-scale controlled studies and most studies are of relatively short duration. Particularly promising is the use of well-established anticonvulsant agents such as topiramate, which lead to reductions in both binge eating and weight. A recent large-scale (n=407) multisite study found binge abstinence rates of 58% in the group treated with topiramate and 29% in the placebo group [32]. Additionally, those receiving topiramate lost an average of 4.5 kg compared with 0.2 kg in the placebo group. It may be that combining medication and one of the effective psychotherapies would be the best approach to the treatment of BED. In one study, the combination of topiramate and CBT led to an abstinence rate of over 80% compared with 61% for placebo [33]; however, the critical comparison with medication alone was not made in this study. A recent large-scale multisite study suggests that sibutramine is an effective treatment for both binge eating and weight loss [32].

Summary of treatments for BED

Overall, there appear to be effective treatments to reduce binge eating frequency in BED (Table 2). Both CBT and IPT appear equally effective in reducing binge eating and can be recommended as evidence-based treatments, although they do not lead to weight loss [28]. Fewer studies of self-help have been reported for BED than for BN, although the promising preliminary results should encourage further research. Weight loss treatment may also be effective in reducing binge eating, although studies in this area have methodological problems and further research is required. Antidepressants such as fluoxetine and anticonvulsant agents such as topiramate reduce both binge eating and weight, although longer-term studies are needed to ascertain maintenance of effects.

Table 2.  The state of treatment for eating disorders, 2008.

Disorder	Evidence-based treatments available (one or more large-scale controlled trials)	One or more controlled trials
Anorexia nervosa	None	Moderate number of controlled trials, although sample size and drop-out rates in adults present problems. Family therapy for adolescents appears promising
Bulimia nervosa	Cognitive–behavioral therapy (CBT); antidepressant medication (fluoxetine FDA approved); interpersonal therapy	Brief CBT-based therapies appear promising
Binge eating disorder	CBT; interpersonal therapy; sibutramine, topiramate (reduce both binge eating and weight)	Antidepressants appear promising with the added advantage of weight loss; brief CBT-based therapies appear promising

EDNOS

Technically, BED is an example of EDNOS, being recognized in DSM-IV as an entity requiring further research, although it is likely to appear in DSM-V as a full disorder. As such, BED is the only disorder classified as EDNOS for which there has been a reasonable amount of research. There are probably a number of other entities included in EDNOS that will eventually become full disorders. At present, however, most of the disorders within this category are subclinical variants of the three full syndromes. These subclinical variants may have different meanings over the lifespan. For example, in adolescence, subclinical variants of BN and BED may become full syndromes over time, whereas in adults they may represent individuals recovering from a full syndrome disorder. Although in some studies, particularly in adolescents, patients with EDNOS have been included in treatment samples of BN or BED, little is known about their response to treatment compared with the full syndrome. This is unfortunate because EDNOS is probably the most common eating disorder with a prevalence of over 5%. Moreover, the comorbid psychopathology and psychological disturbances associated with EDNOS are similar to those found in the full eating disorder syndromes.

Looking to the Future

Although it is impossible to delineate all the possible variants of treatment research that may be obtained in the near future, some trends are already evident. In the case of AN, increased emphasis should be placed on treatment research in adolescents in order to lower the number of cases of long-term AN in adults with their high mortality and morbidity rates and costs to the healthcare system. Treatments for BN, although effective, are not as effective as they might be. Hence, methods to improve the results of treatment are urgently needed. This may take the form of understanding moderators of outcome, to better match individuals with particular characteristics to particular treatments, or to better understand mediators of outcome so that particular mechanisms of action can be enhanced. All but the smallest outcome studies should study both moderators and mediators of outcome. In addition, as noted above, treatment algorithms may enhance the overall outcome of treatment. The problems in the treatment of BED are similar to those in BN, with the exception that the overweight associated with this disorder also requires treatment. Hence, developing treatments or combinations of treatment that address both binge eating and weight is an important aim of outcome research. The combination of a psychotherapy and medication would appear to be a likely avenue for research in the short term.

In general, more large-scale multisite studies are needed to better inform clinical decision making. Whereas single site studies are useful in the early stages of testing a new therapeutic procedure, their findings can only be cautiously generalized, and all too often they have insufficient power to detect differences between groups. Ultimately the discovery of evidence-based treatments depends on large-scale well-designed outcome studies. Finally, treatment outcome research needs to move toward dissemination of cost-effective therapies that can be administered by masters level therapists. To accomplish this, it is necessary to understand the factors that militate against acceptance of such treatments by clinics, therapists and patients, and to find ways to overcome such factors.

Acknowledgements

Declaration of interest: The author reports no conflicts of interest. The author alone is responsible for the content and writing of the paper.