A survey of anorexia nervosa using the Arabic version of the EAT-26 and "gold standard" interviews among Omani adolescents

Objective: The purpose of this study was to assess the validity of the Eating Attitude Test (EAT) in identifying the presence and severity of eating pathology in male and female Omani urban adolescents and to establish cut-off scores that matched those of anorexia identified by gold standard interviews without fear of fatness criteria. Methods: Both females (n=126) and males (n=136) were screened using the Arabic version of the EAT-26 and interviewed using a semi-structured, Composite International Diagnostic Interview (CIDI) in order to investigate the relationship between false positives and false negatives at various EAT-26 cut-off points. A receiver operating characteristics (ROC) curve was calculated to discriminate the power of the EAT-26 for every possible threshold score. Results: The EAT-26 identified 29% of the subjects as probable anorexic cases as against 9.5% identified during the structured interview based on the anorexia gold standard (32% males and 68% females). The sensitivity and specificity of the EAT-26 were respectively 24% and 69.6%. When using the ROC curve, a cut-off score of 10 gave the best compromise between sensitivity (64%) and specificity (38%). Discussion: Although the EAT-26 is the most widely used screening instrument in cross-cultural studies, it does not appear to be reliable in identifying probable cases of anorexia among Omani adolescents. The use of a gold standard interview without fat phobia criteria indicated that the rate of anorexia nervosa may be more prevalent among males than previously estimated. This intriguingly high preponderance of males is discussed in terms of prevailing demographic trends in Oman.