The value of early detection of a disorder is generally undisputed in medicine. Foremost among the benefits of early detection is the chance for early treatment, enhancing the likelihood for a shorter duration of illness and a full recovery. Commonly, heightened public awareness about medical disorders facilitates early diagnosis; however, so far this has not happened in the eating disorders. Certainly one reason is that the public finds it difficult to distinguish between widespread and perhaps justified dieting efforts that often lead to disordered eating and the restrictive eating habits seen in the eating disorders. Since in the early phase the signs and symptoms of intractable dieting - e.g. refusal to eat, weight loss, sense of loss of control over eating - can overlap with the symptoms in AN or BN, physicians and health care personnel need to be cognizant of the differences in the signs associated with temporary dieting and the symptoms of an eating disorder.
The more accurate the classification and the more concise the diagnostic criteria for a disorder, the easier is its detection. The phenotype of AN has been stable over centuries and can be classified with validity. As stated in ICD-10 , ''the clinical features of the syndrome are easily recognized, so that the diagnosis is reliable with a high level of agreement between clinicians''. Nonetheless, the diagnostic criteria have evolved. Differences between DSM-IV  and ICD-10  indicate that clinicians have not always reduced the complex clinical features in the same way into diagnostic categories. To a substantial extent clinical emphasis and not experimental evidence has introduced changes into the diagnostic criteria.
The DSM-IV comprises three syndromes: AN, BN and eating disorder not otherwise specified, whereas the ICD-10 includes, in addition to AN and BN, atypical AN and atypical BN, overeating associated with other psychological disturbances, vomiting associated with other psychological disturbances, other eating disorders, and eating disorders unspecified. There is close agreement on the criteria for the two major syndromes discussed in this chapter. Regarding AN, the principal conceptual difference is the strong emphasis on an attitudinal dimension as a principal criterion in the DSM-IV, namely the individual's ''refusal'' to maintain a minimally normal body weight as opposed to the assessment of ''body weight loss'' as the primary criterion in the ICD-10. Another difference is the subdivision into the restricting and binge eating/purging subtypes in the DSM-IV. This subdivision is important, because subtype differences go beyond the clinical manifestations and have implications for treatment and ultimately to pinpoint aetiological differences . The purging and nonpurging types of bulimia nervosa in DSM-IV correspond to BN proper and hyperorexia nervosa in ICD-10.
The early detection of AN is hampered first by factors inherent in the disease process, especially the denial of illness. AN patients typically identify with the weight loss and assert that they are well and that all is normal . This denial of illness not infrequently extends to family members. Second, food restriction and the high value placed on exercise for a healthy body fit cultural norms and may not be recognized as prodromes to AN. BN patients also tend to be secretive about their abnormal eating pattern. Moreover, the demarcation between shape concerns of a dieting person and between eating binges and great variations in intake is imprecise, as feelings of remorse and guilt and a sense of loss of control may follow simple overeating.
Essentially, children, adolescents or young women with AN or BN do not wish to come to the attention of physicians. Comparisons between the numbers of BN patients in epidemiological treatment studies and the prevalence of BN in females aged 16-24 suggest that a minority of individuals with BN seek treatment . For the medical practitioner, the appearance of physical symptoms such as amenorrhoea in AN or parotid gland swelling in BN aids in detecting the disorders , albeit not necessarily in the early stages. Clearly, important to consider in the differential diagnosis is the fact that AN occurs primarily in Caucasian women and is virtually non-existent in Afro-American women .
The information that classic AN can occur in childhood may not be common knowledge. Bryant-Waugh and Lask  reported that few medical practitioners in the UK were familiar with AN in childhood. A mere 31% among paediatricians and only 3% of family practitioners in a geographical area mentioned a possible diagnosis of AN when they were asked to evaluate two case vignettes of childhood AN. More recently Nicholls et al.  have pointed out that at most 50% of childhood cases fit the diagnostic criteria of the DSM-IV or ICD-10, with the remainder falling into the category eating disorder not otherwise specified. These authors have reported that the criteria developed by their group at Great Ormond Street Hospital, the GOS criteria, diagnosed most children with high interrater reliability. Interestingly, the primary GOS criterion for the patient's attitude in AN is not ''refusal'' but ''determined weight loss''. The GOS criteria for BN are quite concise: ''recurrent binges and purges; sense of lack of control; morbid preoccupation with weight and shape'', but they do not specify the minimum frequency of the behaviour.
The first step towards early detection of eating disorders would be to train general practitioners in their diagnosis. Such training effectively doubled the incidence rate in the epidemiological survey of Hoek et al.  by comparison with the incidence reported by Turnbull et al.  for AN (8.1 versus 4.2/100,000 population), but not for BN (12.2 versus 11.5/100,000 population).
Screening refers to the performance of a medical evaluation and/or diagnostic test in asymptomatic persons in the hope that early diagnosis may lead to improved outcome. The simplest and shortest screening instrument for AN which gives information about eating concerns and behaviours is the Eating Attitudes Test (EAT 40 or EAT 26) by Garner et al. . The score reflects whether an individual is free of concerns, is dieting preoccupied or has abnormal attitudes or behaviours. The instrument has been widely validated. For identifying BN in young women, two simple questions - ''Are you satisfied with your eating pattern?'' (''no'' response) and ''Do you ever eat in secret?'' (''yes'' response) - had a sensitivity of 1.0 and a specificity of 0.9 for BN and hence had a high validity for detecting BN .
The screening survey conducted by Rathner and Messner  in over 500 schoolgirls confirmed the relevance of using multiple antecedent risk variables for early case detection. Even though the yield of clinical cases was low - a point prevalence of 1.3 % for AN and no cases of BN - all AN cases and subclinical cases were detected in the at-risk groups.
The screening tool developed by the McKnight investigators  for disordered eating contains a weight concern scale, the Rosenberg Self-Esteem Scale, and two depression scales. Interestingly, the test-retest reliabilities for the scales were high for the older girls, but low for the elementary school children, suggesting that the younger group might not have understood some questions or concepts. This survey was not designed to detect eating disorders. In fact, it would require a longitudinal study following the same population to examine whether scores on the various scales or changes in scores can predict the onset of disordered eating. Such a large-scale longitudinal study would need to cover the entire period of high risk for the development of eating disorders, for example 8-18 years for AN and 13-25 years for BN, in order to be useful for early identification of eating disorders.
Was this article helpful?