Epidemiological study of eating disorders: time for a change of emphasis

Twenty years have passed since epidemiological methods were first applied to the study of eating disorders. Interest in this approach has been intense in part related to speculation that anorexia nervosa has become more common. In spite of considerable effort research workers have encountered a variety of methodological problems which have left fundamental questions unanswered. Difficulties such as low participation rates, questionable methods of case ascertainment and the study of unrepresentative groups were avoidable. Others are intrinsic to this type of research. Anorexia nervosa is an uncommon disorder and achieving adequate and representative samples outside the hospital setting is difficult. Furthermore, in non-clinical settings eating disorders are often not accompanied by severe weight loss or all features of bulimia nervosa. Clinically derived classification systems are thus rendered of limited use.