Textbook of Psychiatry/Eating Disorders

Eating disorders are serious mental illnesses that include potentially life-threatening behavioral, psychological, and physiological disturbances. Walsh and Fairburn define an eating disorder as "a persistent disturbance of eating behavior or behavior intended to control weight, which significantly impairs physical health or psychosocial functioning," and are not secondary to any recognized general medical or other psychiatric disorder (Walsh & Fairburn, 2002). While eating disorders are illnesses that primarily occur in parts of the world where food is plentiful, they have been reported on every continent, including in developing countries (Nobakht & Dezhkam, 2000; Pike & Mizushima, 2005). While more serious eating disorder cases, and those that present within geographic proximity to tertiary care medical centers, may be referred to specialists for management, most cases are initially identified, and frequently managed by pediatricians, internists, and other primary care clinicians. Eating disorders include anorexia nervosa (AN), bulimia nervosa (BN), and other conditions that in the current diagnostic system are categorized together as Eating Disorders Not Otherwise Specified (EDNOS). These EDNOS conditions include binge eating disorder (BED), night eating syndrome (NES), and sub-threshold syndromes in which some, but not all of the symptoms of the more formally defined eating disorders are present. This chapter will review the clinical manifestations, general epidemiology, and treatment options for the major eating disorders, including AN, BN, and BED.

Clinical features and epidemiology
Anorexia nervosa (AN) is a serious psychiatric illness characterized by failure to maintain a minimally normal weight, intense fear of gaining weight or becoming fat, and preoccupations about body shape and weight. AN has a lifetime prevalence of approximately 0.5%-1% among women, and is estimated to affect one-tenth as many men (Hoek & van Hoeken, 2003). The onset of AN typically occurs in middle to late adolescence, the disorder being significantly more common in industrialized societies such as the United States and Europe than non-Western countries (Cummins, Simmons, & Zane, 2005; Eddy, Hennessey, & Thompson-Brenner, 2007; Hoek & van Hoeken, 2003; Pike & Mizushima, 2005). While the disorder has gained more public attention in recent decades, some version of AN can actually be traced back to the seventeenth century (Bell, 1987; Pearce, 2004).

The defining psychological feature of AN is the relentless pursuit of thinness, which is often manifested by extreme weight control behaviors such as caloric restriction and excessive exercise. Associated with this severe dietary restraint, nearly 50% of individuals with AN also eventually develop episodic "loss of control" eating—that is, the aversive feeling that one is unable to stop or resist eating (Wilson, Grilo, & Vitousek, 2007). In AN, these loss of control episodes may be subjective, including small amounts more than what the individual intended to eat, or objective binges, including irrefutably large amounts of food consumed with discrete periods of time. Regardless of episode size, loss of control episodes may trigger purging behaviors, including self-induced vomiting and the abuse of laxatives or diuretics. However, these compensatory behaviors also may occur in the absence of loss of control eating. As a result of these extreme weight control behaviors, patients with AN maintain a body weight well below that which is minimally medically acceptable. Despite their low weight, patients often experience their bodies, or certain parts of their bodies, as too fat. This intense dissatisfaction with body shape and weight fuels a vicious cycle of weight loss and abnormal eating behavior that it is extremely difficult for individuals with AN to interrupt. The current edition of the Diagnostic Statistical Manual of Mental Disorders (DSM-IV) has characterized AN using criteria that include the maintenance of low weight, the presence of cognitive distortions about body shape and weight, and the presence of amenorrhea for post-menarcheal females. There are two sub-types: restricting type, characterized by those who maintain low weight without any binge eating or purging behaviors, and binge-eating/purging type, characterized by the presence of binge eating or purging behaviors (see Table 1). While no specific weight threshold is identified for the AN diagnosis, DSM-IV includes an example of < 85% of recommended weight for height, and National Institute for Health and Clinical Excellence (NICE) guidelines suggest that body mass index (BMI) < 17.5 kg/m2 may indicate the presence of AN (American Psychiatric Association, 1994). In an attempt to improve diagnostic accuracy and inclusivity, draft criteria for AN proposed for DSM-5 include several changes (American Psychiatric Association, 1994). For example, the proposed criteria for DSM-5 eliminate amenorrhea as a requirement for the AN diagnosis, as evidence suggests that menstruation, while a general indicator of nutritional status, does not provide meaningful clinical distinction among individuals with AN. Furthermore the amenorrhea criterion is not useful for important sub-groups of individuals with the disorder, such as women taking oral contraceptive pills, adolescent patients with primary amenorrhea, and men (Attia & Roberto, 2009).

Individuals with AN, regardless of subtype, often suffer from numerous medical complications consistent with the hypometabolic and malnourished state, including bradycardia (sometimes with prolonged QTc interval), hypotension, hypothermia, and leukopenia (Attia, 2010). Common signs present in patients with AN include hair loss, the development of a downy hair growth on the face, neck and extremities (lanugo), salivary gland enlargement, indigestion, and constipation (Walsh, 2008). Electrolyte abnormalities, such as hypokalemia and hyponatremia, may also present, especially in individuals whose symptomatology includes vomiting or laxative abuse. Even in the presence of significant starvation, it is possible for individuals with AN to display normal laboratory values. Therefore, clinicians should not rely solely on laboratory results to assess acuity of illness. In addition to abnormal laboratory values, individuals with AN commonly display low levels of estrogen and testosterone. These hormonal changes often result in decreased libido, amenorrhea among females, and decreased bone density, eventually leading to potentially irreversible osteopenia and/or osteoporosis. Perhaps the most serious physiological change associated with AN is a prolonged QTc interval, which can lead to cardiac arrhythmia and/or sudden death.

'''Table 1. Diagnostic Criteria for Anorexia Nervosa, DSM-IV (American Psychiatric Association, 1994)'''

Psychological symptoms associated with AN include distractibility, agitation, and sleep disturbance, along with increased depression, anxiety, obsessionality, and compulsivity (Attia, 2010). In addition to these often transient symptoms thought to be associated with the state of nutritional compromise, many individuals with AN also suffer from comorbid psychiatric diagnoses. In a meta-analysis of the outcome of AN in the twentieth century, Steinhausen found that the majority of patients suffered from one or more additional mental illnesses at follow-up, most commonly anxiety and mood disorders, personality disorders, and obsessive-compulsive tendencies (Steinhausen, 2002). However, it is important for clinicians to bear in mind that many symptoms of these comorbid disorders are exacerbated by the underweight condition, and thus may improve or even remit completely with the restoration and maintenance of normal body weight. In fact, studies of starvation in the absence of AN have been useful in identifying the myriad of ways that psychological symptoms develop and worsen in the context of malnutrition (Keys, Brozek, Henschel, Mickelsen, & Taylor, 1950).

Historically, it has been very difficult to track the long-term outcome of AN for reasons including the lengthy course of the disorder and significant relapse rate among those who undergo acute weight restoration. The limited research that has examined longitudinal outcome of AN has revealed high relapse rates and only modest recovery rates. In a trichotomized system classifying outcome of illness into good, fair, and poor, 46.9% of patients reached full recovery from AN, 33.5% improved, and 20.8% displayed a chronic course of the disorder (Steinhausen, 2002). Patients with AN also displayed significant crossover to other eating disorders, most frequently BN and EDNOS (Walsh, 2008). Furthermore, the crude mortality rate amongst patients with AN has been estimated at 5.0% per decade of illness, a rate as high as that seen in any psychiatric illness (Sullivan, 1995). Among a large epidemiological sample of individuals with AN followed in Sweden, the overall mortality rate was 6.2%, with the most common causes of death for the sample being suicide, substance abuse, and eating disorder-related complications (Papadopoulos, Ekbom, Brandt, & Ekselius, 2009). While there are some limitations to the available data regarding longitudinal outcome in AN, there is strong empirically based consensus that AN is a serious mental illness associated with significant morbidity and mortality.

Etiology
Just as it has been difficult to track the long-term outcome of AN, the task of identifying the causal underpinnings of the disorder has proven to be equally challenging. The relatively low incidence and limited ethnic and gendered scope of the disorder, as well as the complicating medical and psychological consequences of semi-starvation, render AN a particularly difficult disorder to study in a controlled setting (Walsh & Devlin, 1998). Eating disorders have been proven to be more prevalent in women than men. For example Striegel-Moore et al. (2009) examined 3,714 women and 1,808 men and found that men were more likely to overeat than women. They found that around 1 in 5 women versus 1 in 10 men check their body size "very often". While the prevalence of eating disorders in cultures that idealize thinness suggest that social environment may play a causal role the development of the disorder, striking patterns of biological and psychological abnormalities in patients with AN suggest that multiple other factors also contribute (Attia & Walsh, 2009).

Research has provided significant evidence for the role of genetic factors in the etiology of AN (Wade, Tiggemann, Bulik, Fairburn, Wray, & Martin, 2008; Bulik, Sullivan, Carter, McIntosh, & Joyce, 1999). Though specific genes have not been identified, the incidence of AN is greater in families with one affected member, and the disorder has higher rates of concordance in monozygotic than dizygotic twins. Heritable factors may also contribute to the development of AN more distally, through temperamental variables associated with the illness, such as perfectionism, obsessionality, compulsivity, and, particularly in the binge/purge subgroup, emotional lability (Walsh, 2008).

In addition, numerous physiological disturbances, including abnormalities in the gastrointestinal tract and various hormonal and neurotransmitter systems, have been considered as possible risk factors for AN (Walsh & Devlin, 1998). Researchers have speculated that leptin, a hormone that regulates appetite and metabolism, may play a role in the perpetuation of the illness, as leptin levels are typically low in underweight individuals with AN (Grinspoon, Gulick, Askari, Landt, Lee, Anderson, Ma, Vignati, Bowsher, & Herzog, 1996). Increased serotonergic activity has also been implicated in various behavioral and psychological characteristics of AN, including reduced food intake, perfectionism, and rigidity. Such evidence suggests that premorbid disturbances in this neurotransmitter might be a risk factor in the development of AN (Kaye, 2008). However, the influence of nutritional status on physiological processes requires caution in the interpretation of changes identified in AN, as these disturbances may be secondary to semi-starvation rather than contributing factors to the disorder’s development (Walsh & Devlin, 1998). While specific causal factors have not been identified, investigators have made progress regarding some of the factors that may contribute to the perpetuation of the illness or the risk of relapse. Mayer et al. found that among inpatients who had fully restored their BMI to > 20kg/m2, those with a higher percentage of body fat were less likely to relapse in the year following hospitalization (Mayer, Roberto, Glasofer, Etu, Gallagher, Wang, Heymsfield, Pierson, Attia, & Devlin, 2007). Furthermore, Schebendach et al. examined patients’ food records obtained prior to hospital discharge and found that those whose diets had greater variety and higher energy density were more likely to have a good clinical outcome during the year following hospital discharge (Schebendach, Mayer, Devlin, Attia, Contento, Wolf, & Walsh, 2008).

Treatment of AN
Despite the fact that AN is an old illness, effective treatments continue to elude clinicians. Studies in this area are few, and none have identified clear empirical support for particular psychotherapeutic or pharmacologic treatments. In part, the challenges to treatment research in AN have resulted from features of the illness itself, including the low prevalence of AN in the general population, and the small percentage of individuals with AN who are treatment-seeking. These characteristics of AN make it difficult to recruit and retain adequate numbers of subjects for clinical studies (Wilson et al. 2007). Additionally, the medical issues present in AN make management difficult and expensive, further complicating efforts to rigorously study psychiatric interventions for this disorder.

Treatment has evolved using various settings, including outpatient, day treatment, and hospital-based programs. It is generally accepted that treatment for AN needs to emphasize weight restoration. Behavioral management programs, aimed at normalizing weight and eating behavior, reinforce healthy behaviors and overall clinical progress with the use of consistently applied contingencies. While it is always desirable to utilize the least restrictive treatment setting in order to facilitate recovery, structured treatment programs such as inpatient, residential, and partial hospital programs may be necessary when outpatient efforts are unsuccessful or unavailable, or when medical or psychiatric status requires a higher level of care to assure safety. Inpatient treatment is often recommended for individuals who have rapidly lost a substantial amount of weight (usually defined by a weight below 75% ideal body weight for one’s height, or a BMI of 16.5kg/m2). Voluntary treatment is also highly preferable, but involuntary arrangements may, at times, be appropriate, especially when a patient’s weight falls into a medically dangerous range (Attia & Walsh, 2009).

Treatment for AN should target full restoration of normal weight with associated resolution of physiological changes that may have developed in the context of acute starvation. Healthy weight ranges are usually defined as being at least 90% of weight recommended for given height, but should consider pre-illness weight and weights at which normal physiological functioning such as normal menstrual activity is known to occur (Attia & Walsh, 2009). Treatment plans may include specific behavioral expectations (e.g., eat 100% of food that is prescribed by a treatment program, utilize staff observation and other treatment interventions aimed to help interrupt purging behaviors, achieve recommended weight gain, etc.). Failure to make clinical progress may be met with additional interventions aimed at increasing caloric intake and decreasing behaviors of illness. Examples of these interventions may include prescription of additional food or nutritional supplements, increase of supervision, and decrease of prescribed activity. Such interventions are not intended to be punishment for "bad behavior," but rather flexible changes to facilitate the ultimate goals of weight gain and recovery.

In behavioral programs, patients are prescribed a diet of an adequate number of calories to achieve weight restoration. Approximately 3500 kcal over maintenance requirements are needed for every pound gained; therefore, sizeable numbers of calories are needed to achieve consistent weight gain. Treatment programs for AN differ in prescription of nutritional plan. Programs that achieve the typical weight gain rates of 2-4 lbs/week generally prescribe 3500-4000 kcal/day to patients at the peak of their weight gain needs. However, calories are typically started at a lower level and increased in a step-wise fashion in order to avoid refeeding syndrome, a syndrome consisting of serious, potentially life-threatening medical symptoms that may occur as refeeding begins and limited nutrient stores are tapped for catabolic processes (Attia & Walsh, 2009). Symptoms of refeeding syndrome may include hypophosphatemia, hypomagnesemia, significant fluid retention (both peripheral and visceral including risk of congestive heart failure (Mehanna, Moledina, & Travis, 2008). Risk factors for the development of refeeding syndrome include seriously low weight (e.g., BMI < 16.5 kg/m2), recent precipitous weight loss, and metabolic disturbance upon presentation, including significant hypokalemia. It is recommended that physical status, including weight, presence of edema, and electrolyte levels be evaluated closely during the first two weeks of acute refeeding.

The Columbia Center for Eating Disorders begins weight restoration treatment with a caloric prescription of 1800 kcal/day. Pending medical stability (generally determined within the first week of hospitalization), the daily diet is increased by 400 kcal every 2 to 3 days until it reaches 3800 kcal, 3000 kcal of which are provided in solid food and the remaining 800 kcal of which are provided in liquid supplement. Because a large part of treating AN emphasizes restoring not only weight, but normal eating behaviors, oral feedings are preferable to nasogastric ones whenever possible. However, this option may be considered for resistant patients, or those who do not enter treatment voluntarily.

Patients in behavioral management programs are also offered a variety of therapeutic interventions from a multidisciplinary team of clinicians, commonly composed of physicians, psychologists, nurses, nutritionists, clinical social workers, and occupational therapists (Attia & Walsh, 2009). Treatment is aimed at confronting disordered eating behaviors and "feared" foods, as well as practicing normal eating. Meal and post-meal supervision, food shopping and cooking groups, and outings to local restaurants may be included in treatment programs to help patients process normal food selection and behaviors around eating with support from staff and peers. Additionally, patients generally participate in regular therapy and discharge planning sessions with psychologists, psychiatric residents, and social workers in individual, group, and family settings.

Such structured and comprehensive behavioral programs are largely effective in helping patients with AN to normalize weight. Yet additional outpatient treatment dedicated to relapse prevention is generally necessary in order to maintain healthy eating and weight. Specific outpatient psychotherapies have been examined in AN, and preliminary evidence supports one of these approaches—a family-based therapy (FBT) for adolescents with AN—as a potentially helpful intervention for outpatient weight restoration and maintenance. FBT, also called, the "Maudsley" method, named for an approach that was developed at London’s Maudsley Hospital, assigns parents the responsibility of refeeding their child, utilizing many of the same treatment principles and reinforcements of structured behavioral programs. For adults, however, results with FBT have been less successful (Lock, 2001; Russell, Szmukler, Dare, & Eisler, 1987). Cognitive behavioral therapy (CBT) has also been studied in outpatients with AN, with more mixed results than that shown for FBT. CBT for AN was first described by Garner, Vitousek, and Pike in 1982 (Garner, Vitousek, & Pike, 1997). The regimen shares many basic therapeutic strategies with Fairburn’s CBT model of BN (1985), but emphasizes the AN-specific issues of enhancing motivation, recognizing the problems associated with semi-starvation, and encouraging weight gain. In a small randomized controlled trial of 33 weight-restored women with AN, Pike et al. found CBT to be more helpful than nutritional counseling at preventing relapse (53% relapse among those receiving nutritional counseling versus 22% relapse among those receiving CBT) (Pike, Walsh, Vitousek, Wilson, & Bauer, 2003). However, a larger study using CBT together with fluoxetine vs. placebo for relapse prevention found that the entire sample had relapse rates of greater than 40% (Walsh et al. 2006). Additional studies have identified clinical improvement in those receiving CBT; however, it is unclear whether CBT fares any better than other specific psychotherapies for individuals with AN (McIntosh, Jordan, Carter, Luty, McKenzie, Bulik, Frampton, & Joyce, 2005; Ball & Mitchell, 2004; Channon, De Silva, Hemsley, & Perkins, 1989).

Because individuals with AN commonly suffer from anxiety, depression, and obsessionality, medications that are generally helpful for these symptoms in other clinical populations have been studied in AN. Overall, results from these studies, most of which have examined antidepressant medications in small samples, have been disappointing (Attia & Schroeder, 2005). Preliminary evidence is more promising regarding the potential utility of olanzapine, an atypical antipsychotic medication, in the treatment of AN. Case reports, open treatment trials, and one small randomized controlled trial (Bissada, Tasca, Barber, & Bradwejn, 2008) suggest that olanzapine may help with both weight gain and alleviating psychological symptoms, namely obsessionality, for individuals with AN (McKnight & Park, 2010).

The generally negative findings from studies of antidepressants in acute AN have led some to posit that the poor response to medication may result from the state of malnutrition and its influence on factors that affect medication response, such as neurotransmitter activity. As a result, medication studies in acutely weight-restored individuals have been conducted in order to examine the possible utility of medication at preventing relapse. Unfortunately, these studies have also failed to identify an effective pharmacologic treatment for AN. For example, a large, placebo-controlled medication trial conducted by Walsh et al. did not demonstrate any benefit from fluoxetine in the treatment of weight-restored patients with AN (Walsh et al. 2006).

AN remains a challenging psychiatric condition for which a single clear, empirically-validated standard of care does not exist. There is no question, however, that weight restoration is an essential first step in treating this disorder, and that behavioral methods are often effective in facilitating this process. Medications alone are used less frequently in the management of AN, but are commonly incorporated as part of a multimodal approach to taking care of patients with this complex disorder.

Clinical Features and Epidemiology
In contrast to AN, bulimia nervosa (BN) is a relatively new eating disorder, first described in the 1970s (Russell, 1979) and officially recognized by the American Psychiatric Association in 1980 (Walsh & Devlin, 1998). BN is defined by the frequent episodic consumption of objectively large amounts of food and the use of inappropriate compensatory behaviors to avoid weight gain. DSM-IV specifies that these episodes must occur at least twice weekly for three months in order to meet full diagnostic criteria for BN. A DSM-IV diagnosis of BN also requires overconcern with body shape and weight (American Psychiatric Association, 1994) (see Table 2). BN is more common than AN; an epidemiological review by Hoek and van Hoeken report a 1.0% aggregated prevalence rate for BN (Hoek & van Hoeken, 2003). Additionally, the gender ratio for BN reflects higher percentages of men affected than that seen in AN: 1.5% for females and 0.5% for males (Hudson, Hiripi, Pope Jr, & Kessler, 2007). Other studies have indicated even higher prevalence rates of 5-10% for more broadly defined manifestations of the disorder characterized by occasional binging and purging (Walsh, 2008). New diagnostic criteria being proposed for DSM-5 would lessen the frequency criteria of binge and purge episodes for BN to once weekly for three months, making it more likely that some of these sub-threshold cases currently classified as EDNOS would be considered full syndrome BN according to the proposed diagnostic system (American Psychiatric Association, 2010). Like AN, BN affects predominantly adolescents and young adults in industrialized societies (Fairburn & Harrison, 2003), however, the disorder has been described in a variety of non-Western cultures as well (Nobakht & Dezhkam, 2000; Pike & Mizushima, 2005).

Though there is considerable overlap in the characteristics of AN and BN, these eating disorders are distinct from each other in several important respects. First, the characteristic feature of BN is that attempts to restrict food intake are interrupted by episodes of binge eating, during which individuals consume an unambiguously large amounts of food within discrete periods of time, and experience a sense of loss of control (Fairburn & Harrison, 2003). Binges are usually followed by purging, the most common methods being self-induced vomiting and the abuse of laxatives and diuretics. The frequency and intensity of these binge/purge episodes tend to escalate over time, enough so that many patients develop the ability to induce vomiting without mechanically triggering the gag reflex (Walsh, 2008). A small subset of patients with BN do not purge, but rather compensate for binge episodes with fasting and exercise. The combination of binge eating with extreme weight-control behavior generally places individuals with BN at a normal body weight, distinguishing them from those with AN, who are, by definition, underweight (Fairburn & Harrison, 2003). Because patients with BN typically present with a healthy body weight, the disorder has historically been associated with a significantly lower crude mortality rate (0.32%) than that described in AN (Steinhausen & Weber, 2009). However, results from a recent study by Crow et al. have called this into question, describing mortality rates for BN as high as 3.9% (Crow et al. 2009).

'''Table 2. Diagnostic Criteria for Bulimia Nervosa, DSM-IV (American Psychiatric Association, 1994)'''

While there may be fewer medical risks of BN as compared with AN, BN also is associated with several potentially serious physiological complications. The most common medical problems in BN are dehydration and hypokalemia due to vomiting or the consistent over-ingestion of laxatives or diuretics. Hyponatremia and hypochloremia may be present, especially in individuals who replenish their fluid requirements with large amounts of plain water. Metabolic alkalosis may occur in individuals who purge by vomiting with the consequent loss of gastric acid. Conversely, acidosis may be present in those who abuse laxatives (Walsh, 2008). Individuals with BN may also develop a callus or scar on the dorsal side of the hand, commonly named "Russell’s sign," due to the repeated impact of the teeth rubbing on the hand that may result from manual stimulation of the gag reflex (Halmi, 1987; Russell, 1979). In addition, the recurrent exposure to acid from vomitus may contribute to the gradual weakening of tooth enamel and eventual dental erosion causing a "moth-eaten" appearance of teeth. More serious consequences of bulimic symptoms such as esophageal tearing or gastric rupture have also been described (Walsh, 2008).

As in AN, individuals with BN tend to present with psychiatric comorbidities (Steinhausen & Weber, 2009). In his meta-analysis of 79 studies of BN, Steinhausen identified mood and anxiety disorders, personality disorders, and impulsive behavioral disturbances, including substance use disorders and self-injurious behaviors, as pathologies frequently seen among patients with this disorder. In addition, Steinhausen found that patients with BN often cross over to develop other eating disorders (up to 32% of individuals studied). The most common crossover was to EDNOS, followed by AN, and finally BED, though it is suspected that crossover to this diagnosis is underreported, as BED was not yet officially recognized as a disorder at the beginning of the time period studied. Perhaps even more striking is the fact that one-third of patients who present for treatment for BN have past histories of AN (Walsh & Devlin, 1998). The prognosis for individuals with BN is better than that seen in AN, although outcome studies suggest that significant numbers continue to struggle with symptoms of illness years after initial presentation (Steinhausen & Weber, 2009).

Etiology
Researchers have identified a range of biological, psychological, and environmental factors that might predict risk for the development of BN. First, there is good evidence that genetics may predispose certain individuals to BN. In fact, one study showed heritability to account for as much as 60% of variance in liability to the disorder (Bulik, Sullivan, & Kendler, 1998). Neurochemical abnormalities have also been described in BN, including those that, on a theoretical basis, offer support for several hypotheses regarding possible mechanisms responsible for the initiation and perpetuation of BN symptoms. Both the serotonergic system, involved in control of hunger and satiety, and dopaminergic system, central to the function of driven behaviors and reward, have been implicated in BN (Kaye, 2008). Additionally, research has shown the release of cholecystokinin, a gut hormone responsible for signaling post-meal satiety, to be blunted in BN (Devlin, Walsh, Guss, Kissileff, Liddle, & Petkova, 1997). A stomach relaxation reflex following the ingesting of a small liquid meal also appears to be lessened in BN, as compared with the normal gastric response seen in control subjects (Walsh, Zimmerli, Devlin, Guss, & Kissileff, 2003).

Certain psychological characteristics have also been associated with BN. Specifically, depression, anxiety disorders, low self-esteem, and premorbid AN have been described in BN and have been examined as possible predisposing factors (Fairburn, Welch, Doll, Davies, & O'Connor, 1997). Discordant or weight-critical family environment, history of obesity, physical or sexual abuse, and the general cultural pressure to be thin are also thought to contribute to the risk for developing BN (Walsh & Devlin, 1998).

Treatment of BN
In contrast to the status of treatment research for AN, there is a solid evidence base for the treatment of BN. Many randomized controlled trials have been conducted successfully in BN, identifying both pharmacological and psychological interventions that are effective at reducing or eliminating the core symptoms of BN. Systematic reviews of available treatment studies consistently conclude that cognitive behavioral therapy (CBT) as well as antidepressant medications are both effective for short-term management of BN (Shapiro, Berkman, Brownley, Sedway, Lohr, & Bulik, 2007).

Psychological treatments for BN
CBT—specifically, a form of CBT that focuses on identifying, examining, and modifying a cycle of interrelated thoughts and behaviors that maintain the eating disorder—has been found consistently to be helpful for BN (Fairburn, 1985). For example, patients with BN may describe cognitions that support a connection between self-evaluation and body shape and weight. These thoughts and beliefs may lead individuals with BN to excessive dietary restriction and inappropriate weight control behaviors in order to affect self-evaluation. This process of restricting, in turn, predisposes individuals to binge eating and purging, which has the ultimate consequence of worsening self view and reinforcing the eating disordered behaviors. CBT sessions for BN focus on identifying inconsistencies in the patient’s established cognitions and related behaviors in order to help identify alternative thoughts and healthier behaviors. This treatment typically involves 20 sessions that take place over the course of 4-5 months, and has been utilized in both individual and group settings. CBT appears to be associated with lasting cessation of binge eating and purging in 30-50% of BN cases, with many others achieving substantial improvements not only in binge/purge symptoms, but also in self-esteem, social functioning, and general psychological well-being (Wilson et al. 2007).

Another outpatient psychotherapy shown to be helpful in BN is interpersonal therapy (IPT), a short-term, structured treatment originally developed for depression (Klerman, Weissman, Rounsaville, & Chevron, 1984) and adapted by Fairburn for BN (Fairburn, 1993). This intervention does not focus directly on eating disorder symptoms, but rather emphasizes current interpersonal problems that are thought to be maintaining the eating disorder. Though CBT, based on the empirical and clinical evidence, is generally the preferred treatment for BN, the NICE (2004) guidelines have recommended that IPT be considered as an alternative to CBT (Wilson et al. 2007).

While evidence-based psychotherapies, like CBT and IPT, are consistently effective in clinical trials, these specialized treatments have not been successfully disseminated to community practitioners. A shortage of therapists trained to deliver these treatments impedes access to evidence based interventions for many individuals with BN. There has therefore been interest in whether self-help versions of CBT may be more accessible and cost effective options for BN treatment (Wilson et al. 2007). In fact, there is some evidence that Guided Self Help (GSH) options that provide support and information to the user of a self-help manual may be useful for a subset of individuals with BN, and may be an alternative for those who do not have access to specialist treatment options (Grilo, 2000).

Pharmacotherapy for BN
Several medications are helpful in the treatment of BN. The high rates of comorbidity with depression seen in BN led to initial attempts to treat this eating disorder with antidepressant medications. Randomized controlled trials using antidepressant medications, initially examining monoamine oxidase inhibitors (MAOIs) and tricyclic antidepressants (TCAs), began soon after the first published descriptions of BN. These trials consistently found benefit associated with antidepressant compared with placebo at reducing binge eating and purging behaviors in BN samples (Broft, Berner, & Walsh, 2009). The more favorable side effect profile of the selective serotonin reuptake inhibitors (SSRIs) led to more trials using these medications once they became available. In fact, following a large multi-site trial that demonstrated medication efficacy in 387 individuals with BN, fluoxetine became the only medication approved for use in BN by the U.S. Food and Drug Administration (Walsh, 2008). It is of note that the dose of fluoxetine recommended for BN (60 mg) is higher than that typically used in treating depression, and also that antidepressant medications are effective at treating BN irrespective of depression symptoms (Walsh, Gladis, Roose, Stewart, Stetner, & Glassman, 1988).

Other classes of medication have also been studied for possible utility in treating BN. Specifically, topiramate, a medicine used to treat seizures that has been found to suppress appetite and cause weight loss, has been examined in BN and associated with benefit in several studies (Hedges, Reimherr, Hoopes, Rosenthal, Kamin, Karim, & Capece, 2003; Hoopes, Reimherr, Hedges, Rosenthal, Kamin, Karim, Capece, & Karvois, 2003; Nickel, Tritt, Muehlbacher, Gil, Mitterlehner, Kaplan, Lahmann, Leiberich, Krawczyk, & Kettler, 2005). However, cognitive side effects and paresthesias associated with topiramate lessen its acceptability to BN patients compared with SSRIs. Odansetron, a medication used primarily for the treatment of nausea in medical and surgical settings, has been tried in one small trial of BN with some benefit (Faris, Kim, Meller, Goodale, Oakman, Hofbauer, Marshall, Daughters, Banerjee-Stevens, & Eckert, 2000).

Although medications (most commonly SSRIs) are useful for short-term management of BN—especially those who have not responded to CBT as a sole intervention—there are few data to support continued efficacy over the long-term (Nakash-Eisikovits, Dierberger, & Westen, 2002). Several studies do support the enhanced effect of combining medication with CBT for treatment of BN, and some authors suggest that multi-modal treatment is the approach most likely to be associated with long-term remission and symptom reduction rates (Wilson et al. 2007).

For patients with BN who do not respond to evidence based outpatient approaches, many of whom present with psychological co-morbidities or multi-impulsive behaviors, more intensive forms of treatment, such as hospital-based and day treatment may be necessary (Walsh, 2008). Inpatient hospitalization programs for BN typically aim to break the cycle of binge eating and purging, and are followed whenever possible by additional treatments that support the continuation of structured eating, strategies to prevent or reduce likelihood of post-meal purging, and weight maintenance. As with AN, inpatient programs for BN generally include several treatment approaches offered by a multi-disciplinary team of clinicians.

Clinical Features and Epidemiology
Binge eating disorder (BED) is characterized by repeated, persistent episodes of binge eating in the absence of the inappropriate compensatory behaviors seen in BN. According to DSM-IV diagnostic criteria, BED is defined by the consumption of unusually large quantities of food in discrete periods of time together with the experience of loss of control over eating. DSM-IV also specifies that binge eating in BED is associated with emotional distress, and must occur at least twice a week for six months to meet full diagnostic criteria (American Psychiatric Association, 1994). Currently a provisional diagnosis under the larger heading of EDNOS in the DSM-IV, BED is being proposed as a distinct eating disorder in the upcoming DSM-5 (American Psychiatric Association, 2010).

In contrast to both AN and BN, BED is seen most frequently in middle-aged individuals, and is evenly distributed across gender and racial demographics, though there is some evidence to suggest that women may be more likely to present for treatment (Tanofsky, Wilfley, Spurrell, Welch, & Brownell, 1997; Wilson et al. 2007). BED is also distinct from the other eating disorders in that binge eating occurs in the context of general overeating as opposed to a setting of dietary restraint (Fairburn & Harrison, 2003). Because of the nature of its symptoms, BED has a strong association with obesity. Although present in approximately 3% of adults, the disorder has a higher prevalence among obese individuals (Grilo, 2002). In fact, BED is present in as many as 5-10% of patients seeking weight control treatment (Fairburn & Harrison, 2003). Because of its links to obesity, BED is associated with increased mortality and risk for numerous medical disorders, including hypertension, diabetes, respiratory illness, cardiac problems, and osteoarthritis (Wonderlich, Gordon, Mitchell, Crosby, & Engel, 2009).

It is important to remember that individuals with obesity do not necessarily suffer from BED. Research has identified a distinct set of psychological characteristics that distinguish patients with BED from obese individuals who are not affected by the disorder. Compared to obese individuals without BED, those with BED report increased rates of depression, anxiety, and dissatisfaction with shape and weight similar to that seen in BN (Walsh, 2008; Yanovski, 1993). Individuals with BED also report a lower health-related quality of life, less life satisfaction, and more functional impairment than obese individuals without eating disorders (Wonderlich et al. 2009). Moreover, eating behavior studies have found that, when instructed to binge eat in a laboratory setting, individuals with BED eat more than their obese counterparts without the disorder (Yanovski, Leet, Yanovski, Flood, Gold, Kissileff, & Walsh, 1992; Goldfein, Walsh, Devlin, Lachaussée, & Kissileff, 1993; Sysko, Devlin, Walsh, Zimmerli, & Kissileff, 2007). Such evidence confirms the meaningful diagnostic and clinical distinction between binge eaters and non-binge eaters within the obese population. In preparation for the upcoming publication of DSM-5, there has been extensive debate about whether the current data available about BED support its identification as an eating disorder separate from the EDNOS category. In their review of BED, Wonderlich et al. argue that the consistent clustering of disturbed eating symptoms in affected individuals, together with the psychological distress, functional impairment, and significant rates of psychiatric comorbidity associated with BED, support its meeting the definition of illness (Wonderlich et al. 2009). In contrast to AN and BN, BED appears to have a shorter and more variable course. BED is associated with both high rates of treatment response and the possibility of spontaneous resolution of symptoms, but also possible symptom recurrence. Patients with BED are less likely to cross over to another active eating disorder than are individuals with AN or BN. Together, this evidence paints a portrait of BED as a variable, yet sometimes chronic disorder that often fluctuates over time.

Etiology
While researchers speculate that biological, psychological, and environmental factors contribute to the development of BED, as they do other eating disorders, the specifics of these risk factors for BED are still relatively poorly understood. Research does indicate that presence of obesity increases an individual’s likelihood of developing BED (Fairburn, Doll, Welch, Hay, Davies, & O'Connor, 1998). Additionally, there is preliminary data to suggest that BED runs in families, with heritability ranging between 0.39 and 0.57 (Hudson, Lalonde, Berry, Pindyck, Bulik, Crow, McElroy, Laird, Tsuang, & Walsh, 2006; Javaras, Laird, Reichborn-Kjennerud, Bulik, Pope, & Hudson, 2008). These substantial familial associations implicate a potential additive effect of genes on the development of BED. However, research has yielded mixed evidence regarding the preexistence of chemical or neurological abnormalities in individuals with this disorder. Though some early studies have yet to find any distinct neurochemical or hormonal anomalies among binge eaters, others tentatively associate BED with decreased levels of the hunger-stimulating hormone, ghrelin (Geliebter, Gluck, & Hashim, 2005). An individual’s development of BED may also be influenced by numerous non-specific risk factors for psychiatric disorders, such as parental depression and adverse childhood experiences (Fairburn et al. 1998).

Treatment of BED
Treatment of BED usually aims to help patients normalize eating behavior, with any weight loss achieved being secondary to the behavioral improvement. While excess weight clearly represents a significant medical problem for some individuals with BED, it is uncertain whether these individuals receive the same level of benefit (i.e., lasting weight loss) from behavioral weight loss treatment programs as non-binge eaters. Hence treatments that aim to improve binge eating behavior are typically recommended when BED is also present in the overweight or obese state (Wilson et al. 2007).

Specialized psychotherapies have been shown to be helpful in addressing the psychological symptoms of BED. Both CBT and IPT are reliably effective in eliminating binge eating and reducing eating disorder psychopathology both in the short and long term. Although these specialized psychotherapies do not typically yield clinically significant weight loss, they do often prevent further weight gain by reducing or eliminating binge eating behavior (Wilson, Wilfley, Agras, & Bryson, 2010).

Specifically, there is strong empirical support for the use of a form of CBT in BED. Adapted from Fairburn’s CBT for BN, this specialized treatment emphasizes recognizing and modifying unhealthy eating patterns (Fairburn, Marcus, & Wilson, 1993). CBT for BED is associated with a high treatment completion rate (approximately 80%), as well as remission from binge eating in 50% of patients, and general improvements in depression and other psychosocial impairments. IPT has been shown to be about equally as effective as CBT for treating BED in both the short and long term (Wilson et al. 2007). Finally, dialectical behavior therapy (DBT) is a specialized treatment that focuses on emotional regulation and gaining greater awareness of chaotic eating habits that characterize BED. While it does not boast as strong of empirical support as CBT and IPT, DBT seems to be a well-suited and durable alternative treatment for BED (Telch, Agras, & Linehan, 2001).

However, the high cost of these specialized treatments, along with the limited availability of therapists who are adequately trained to perform them, has prompted researchers to investigate more accessible, cost-effective treatment options for BED. Wilson et al. found a mode of guided self help based on CBT (CBTgsh) to be equally as effective as IPT in eliminating binge eating in both the short and long term (Wilson et al. 2010). It is of note, however, that, neither CBTgsh nor IPT produced as much weight loss as did a separate behavioral weight loss treatment.

In addition to psychotherapies, researchers have examined the efficacy of various medications in the treatment of BED (Bodell & Devlin, 2009). Consistent with evidence from combined treatment studies of BN, Devlin et al. found CBT to be more effective than fluoxetine as adjunct treatments to group behavioral weight loss treatment. While the addition of fluoxetine to CBT provided little if any improvement in binge reduction, the medication did decrease psychopathologies often associated with BED, such as anxiety, depression, and dietary restriction (Devlin, Goldfein, Petkova, Jiang, Raizman, Wolk, Mayer, Carino, Bellace, Kamenetz, Dobrow, & Walsh, 2005). Leombruni et al., on the other hand, found sertraline to effect a significant improvement in both binge behavior and binge frequency, as well as clinically significant weight loss for up to 24 weeks in women with BED (Leombruni, Piero, Brustolin, Mondelli, Levi, Campisi, Marozio, Abbate-Daga, & Fassino, 2006). Still other evidence from pharmacotherapy studies of BED suggests high and rapid rates of relapse, as well as simple noncompliance with extended open-label treatments (Wilson et al. 2007).

The numerous characteristics of BED that distinguish it from AN and BN render it a subject of unique interest within the field. Because formal research on BED is in relatively preliminary stages, there is still much to learn about its etiology, symptomotology, and treatment. Future efforts would be well-directed towards studying the longitudinal course of the disorder, untangling the complex issues of its management, and further exploring biological and psychological factors that put individuals at risk for developing BED.

Conclusion
Eating disorders, including AN, BN, and BED are serious conditions in which significant disturbances in eating behavior are accompanied by distressing cognitions about body shape and weight. They are all illnesses with physical, as well as psychological manifestations. Once their symptoms are established, eating disorders may contribute to significant functional impairment, and are often very difficult to interrupt. Evidence-based treatments, including pharmacotherapy and CBT, are helpful in the treatment of BN and BED, but additional research is needed to establish empirically-supported treatments for AN, the eating disorder with the highest rates of morbidity and mortality.