What is an eating disorder? Implications for current and future diagnostic criteria

Murray and Touyz (2013) argue that muscle dysphoria more closely resembles an eating disorder than a body dysmorphic disorder (where it is currently categorized) on the basis of the concomitant eating and exercise preoccupation and body image or shape and weight concerns. This is persuasive but raises the question, ‘what is an eating disorder?’ Is an eating disorder primarily a disturbance of eating, or a disorder of body image, or both? In the present conceptualization, how well are eating disorders distinctive from others, e.g. addiction or anxiety disorders? In addressing these questions I will start with an overview of the present diagnostic groups and then argue for a core psychopathology to be recognized as common to all eating disorder diagnostics.

In the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revised [DSM-IV; American Psychiatric Association (APA), 2000], three eating disorders, anorexia nervosa, bulimia nervosa and binge eating disorder, have specific criteria. People who do not meet these criteria but nevertheless have an eating disorder are classified under the category Eating Disorder not Otherwise Specified (EDNOS). Six types of EDNOS are listed in the DSM-IV (APA, 2000). EDNOS is, however, the most common of all the syndromes in both the clinic and community and this is generally agreed to be a major problem with the current diagnostic schemes.

The definition of anorexia nervosa has changed little since Russell’s seminal 1970 paper, which specified that: (1) ‘the patient’s behaviour leads to a marked loss of body weight and malnutrition’; (2) ‘there is an endocrine disorder which manifests itself clinically by cessation of menstruation in those patients who are most commonly afflicted by the illness—adolescent girls or women during the reproductive period of life’; and (3) ‘there are aspects of the psychopathology which are characteristic of anorexia nervosa, irrespective of the patient’s sex. They are essentially manifestations of a morbid fear of becoming fat, which may be expressed by the patient or may be more explicit in her behaviour’. Russell’s criteria also included a description of ‘bouts of overeating’ and extreme weight-control measures such as self-induced vomiting which ‘effectively counteract’ the overeating. (These foreshadowed the key features of the second eating disorder to be recognized, namely bulimia nervosa.) The DSM-III (APA, 1980) did not specify the endocrine disturbance and added a criteria of ‘disturbance of body image’, exemplified by body image distortion. In the DSM-III revision (DSM-III-R; APA, 1987) and in the DSM-IV (APA, 2000), this is described as ‘a disturbance in the way in which one’s body weight or shape is experienced, undue influence of body weight or shape on self-evaluation, or denial of the seriousness of the current low body weight’ (Russell, 1970). The fifth edition of the DSM (DSM-5; APA, 2012) will likely return to Russell’s view, no longer requiring patients to acknowledge their fear of weight gain but allowing that the intense fear of gaining weight or becoming fat may be expressed in ‘persistent behaviour that interferes with weight gain, even though at a significantly low weight’. This also addresses concerns that a fear of fatness is a cultural determinant rather than a core feature. The DSM-5 will also likely remove the DSM-III-R and DSM-IV requirement of amenorrhoea and may not suggest a body mass index or similar threshold for specifying underweight. These most recent revisions, like those of the proposed International Classification of Diseases, 11th revision (ICD-11), are supported by a body of empirical literature (Hebebrand and Bulik, 2011; Uher and Rutter, 2012). They will also have the important impact of reducing the number of patients classified as EDNOS types 1 or 2.

The second eating disorder, bulimia nervosa, as currently defined in the DSM-IV (APA, 2000) and as proposed by Russell in 1983 has a closely similar psychopathology to that of anorexia nervosa, whereby self-evaluation is unduly influenced by weight and shape concerns. Bulimia nervosa thus resembles the binge eating/purging type of anorexia nervosa but without the key criteria of being underweight and in a state of self-starvation. The characteristic behavioural features of bulimia nervosa are those of recurrent episodes of binge eating with extreme compensatory behaviours such as vomiting or laxative misuse, driven exercise, and/or fasting. As with anorexia nervosa, the relaxation of some criteria in the DSM-5 (notably the 3-month duration and twice weekly frequency of behaviours) will further reduce the number of people presently in the EDNOS type 3 category. EDNOS type 4 will, however, remain as there has been no relaxation of the requirement that binge eating be on large amounts of food.

In contrast to anorexia nervosa and bulimia nervosa, where there is a core psychopathology, the third eating disorder, binge eating disorder, is conceptualized as primarily a behavioural disorder of recurrent binge eating. A now historic detail is that the DSM-III bulimia [the term nervosa was only added in the 1987 revision (APA, 1980, 1987)] criteria closely resemble those suggested for EDNOS type 6 binge eating disorder (APA, 2000). This reflects the lineage of the term ‘binge eating’, and binge eating as a common problematic behaviour contributing to obesity. In 1955, Stunkard and co-workers termed the uncontrolled overeating of large amounts of food ‘binge eating’, as found in obese people with night eating syndrome (NES). In 1997, he (with others and Wermuth as the first author) described three key features of binge eating: (1) ‘impulsive, episodic, uncontrolled, and rapid ingestion of large quantities of food over a relatively short period of time’; (2) ‘termination of the episode only when a point of physical discomfort has been reached’, ‘self-induced vomiting supports but is not required for the diagnosis’; and (3) ‘subsequent feelings of guilt, remorse, or self-contempt’. With the additions of six or three monthly duration and twice or once weekly frequency and the exclusion of concurrent anorexia nervosa or bulimia, this description is close to that of the DSM-IV (APA, 2000) and DSM-5 binge eating disorder proposed criteria (APA, 2012).

And what about EDNOS, the largest diagnostic category? DSM-IV types 1, 2, 3, 4, and 6 can be matched onto the three main disorders as described above. The remaining type 5 is a disorder of repeatedly chewing and spitting, but not swallowing, large amounts of food. Not specified in the DSM-IV, but much studied since, is a further but probably less common eating disorder, purging disorder. Purging disorder is characterized by recurrent purging behaviours to influence weight or shape in the absence of binge eating (APA, 2012). Like another behaviourally defined eating disorder, NES, purging disorder is, however, deemed not of sufficient validity and empirical study to justify inclusion as yet in the DSM-5 (APA, 2012; Stunkard et al., 1955). Finally, the syndrome ‘feeding disorder of infancy or early childhood’ may be moved from the infancy and childhood section to a renamed parent category of Feeding and Eating Disorders and given a new name of ‘avoidant/restrictive food intake disorder’ (APA, 2012; Wermuth et al., 1977).

Adding to the size of EDNOS is the DSM-IV (APA, 2000) definition of binge eating mandating the size as an unusually large amount of food. When strictly applied, this puts people who binge on normal-sized food portions (subjective binges or subjective bulimic episodes) into the EDNOS category and often out of consideration. For example, a recent epidemiological prospective survey of the incidence of eating disorders in young adults did not include any with EDNOS characterized by subjective binge eating, although seven other full and subthreshold eating disorder categories, including purging disorder, were defined for study (Stice et al., 2009). Not including subjective bingeing in eating disorder groups occurs in spite of evidence that that size of the binge is of less clinical utility, diagnostic validity, and concern to people who binge than is the experience of being out of control when eating (e.g. Latner et al., 2007).

In all of this, the DSM [and equally, if not more so, the ICD, tenth revision (ICD-10) (World Health Organization, 1992)] schemes are remarkably silent on what are the essential or defining features of an eating disorder. All have a disturbance of eating behaviour. In the DSM-III (APA, 1980:67) and DSM-III-R (APA, 1987:65), the section on eating disorders opens with ‘this subclass of disorders is characterized by gross disturbances in eating behavior’. In the DSM-IV, eating disorders are no longer a subclass but a category with their own chapter, and an additional descriptive statement is added, namely that ‘a disturbance in perception of body shape and weight is an essential feature of anorexia nervosa and bulimia nervosa’ (APA, 2000:583). Thus, it is recognized that in anorexia and bulimia nervosa the behaviours are driven by and/or result in a cognitive/emotive psychopathology characterized by body image and eating concerns. However, there is no core psychopathology with delineation of the cognitive, emotive, and/or behavioural diagnostic features common to all eating disorders. Distress regarding binge eating is the sole cognitive/emotive mandatory criterion for binge eating disorder or NES; there are none for the other EDNOS disorders. This is, however, not for the want of trying.

Fairburn and co-workers (2003) have proposed and subsequently tested a ‘transdiagnostic’ model for the treatment of eating disorders. This model has a core psychopathology of overvaluation of body shape, weight, and eating and their control. There is empirical support for body shape or weight over-evaluation to be extended to binge eating disorder, either as a diagnostic requirement (Mond et al., 2007) or as a diagnostic specifier of syndrome severity (Grilo et al., 2008). Grilo and co-workers (2008) found that while obese people with binge eating disorder could be divided into those with and without high weight and shape over(self)-evaluation, the latter still had significantly higher levels of concerns regarding eating, weight, or shape compared to an overweight non-binge eating disorder control group.

In addition to there being no agreed common psychopathology, there is debate over the relevance or not of dividing eating disorder syndromes by behaviours that may fluctuate over time and overlap in clinical significance, and it is recognized that there is ‘substantial diagnostic crossover’ between eating disorders (Gordon et al., 2010). Categorizing syndromes by behaviours causes confusion when people move across diagnoses as they modify or change behaviours. For example, if someone with bulimia nervosa ceases purging but continues to binge, they then have binge eating disorder. A transdiagnostic or single diagnostic category for eating disorders would address this problem.

The main counter to a single diagnosis view is, however, the argument expressed by Birmingham and co-workers (2009) and others that anorexia nervosa differs from other eating disorders in important ways. These differences include a putative endophenotype (Bulik et al., 2007), higher mortality (Arcelus et al., 2011), and other relatively poorer outcomes (Fichter and Quadflieg, 2007; Mitchinson et al., in press). Anorexia nervosa also has the most diagnostic stability compared to other eating disorders (Fichter and Quadflieg, 2007). It thus seems likely to survive over time as a stand-alone diagnostic entity.

Although it is difficult to predict whether new eating disorders will emerge or combine with bulimia nervosa or binge eating disorder, the present trend is to add rather than unite. This is despite empirical support, for example, for subsuming at least some of purging disorder with subjective binge eating as a variant of bulimia nervosa and/or binge eating disorder (Mond and Hay, 2010). Taxometric studies have also tended to combine or find new ways of clustering behavioural and cognitive features and to favour eating disorders being divided into ‘bulimic’ and ‘nonbulimic’ forms (Gordon et al., 2010).

Defining a core psychopathology is not to imply that people who engage in either high-level or disordered eating or who have high levels of body dissatisfaction are not impaired and cannot be helped. These problems can be well conceptualized as subthreshold syndromes that merit further research and targeted intervention strategies. In addition, agreeing on a core psychopathology need not imply that there is only one eating disorder. In my view, such a defining feature that includes, but is not committed to, any single disordered eating behaviour would greatly assist professionals and nonprofessionals alike navigate their way through the ever-increasing number of eating disorder types and subtypes. In other words, it would allow the ‘wood to be seen for the trees’. Such a core psychopathology would include body image, shape and weight, and eating concerns that drive and are driven by disordered eating and weight or body shape control behaviours. It would have the additional benefit of providing a positive definition for EDNOS, which would then not be defined solely on the basis of not meeting the criteria for a specific eating disorder. It would also distinguish an eating disorder from somatoform, addictive, and other anxiety disorders. This is the argument applied in this issue by Murray and Touyz (2013), whereby muscle dysmorphia has primary pathology in body image concerns (but anxiety about inadequate musculature versus fear of fatness) that drive behaviours resembling those seen in anorexia nervosa and bulimia nervosa.

In conclusion, the present state of adding to the number of eating disorders in diagnostic schemes should occur in conjunction with researchers, clinicians, and people who live with eating disorders agreeing on the distinctive and defining features of an eating disorder. Further diagnostic categories can be built and refined from these. These defining features would be found in all eating disorders including EDNOS. Such agreed criteria would greatly aid EDNOS research, as currently EDNOS is either studied as multiple separate subthreshold groupings or researchers have derived their own operational criteria. However, equally if not more important, agreed criteria would reduce diagnostic confusion and create clarity for all.