Intended for healthcare professionals

Letters

Onset of adolescent eating disorders

BMJ 1999; 318 doi: https://doi.org/10.1136/bmj.318.7200.1761 (Published 26 June 1999) Cite this as: BMJ 1999;318:1761

Dieting may be an early sign, rather than a cause, of eating disorder

  1. Nicolas Stettler, Nutrition fellowa,
  2. Andrew M Tershakovec, Associate professor of paediatrics,
  3. Mary B Leonard, Assistant professor of paediatrics and epidemiology
  1. Department of Gastroenterology and Nutrition, Children's Hospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, PA 19104-4399, USA
  2. Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, 423 Guardian Drive, Philadelphia, PA 19104-6021, USA
  3. Centre for Adolescent Health, Department of Paediatrics, University of Melbourne, Parkville Victoria 3052, Australia

    EDITOR—Patton et al reported a positive association between dieting and the development of eating disorders in adolescents.1 A causal effect of dieting on the development of these serious psychiatric conditions was implied in the conclusion. We believe, however, that the data presented did not support the suggested causal effect.2

    Unlike in a randomised clinical trial, in a cohort study the participants select themselves; inthis case the exposure of interest was dieting. Participants at risk of developing eating disorders may have been more likely to expose themselves to dieting. To avoid bias the exposed and unexposed participants in a cohort study should be similar for other important determinants of outcome.3 There is no evidence that this was true in this study. In fact, after adjustment for psychiatric disorders at baseline the hazard ratio of severe dieting decreased threefold, which suggests that the exposed and unexposed groups differed for other, unadjusted, factors.

    In a cohort study the participants must be free of the disease at inclusion.3 Although the 37 incident cases may not have displayed, at baseline, all the signs necessary for the diagnosis of an eating disorder, some of those who were on a diet at the beginning of the study may actually have been in the early stages of an eating disorder; dieting might be an early, non-specific sign of later development of eating disorders and not a cause.

    Although some of these points were raised in the discussion, the conclusion and key messages emphasised that adolescents who diet are at increased risk of developing an eating disorder. We agree that inappropriate dietary behaviours should be discouraged; given the global epidemic of obesity, the importance of promoting a healthy lifestyle, including weight control by a reasonable diet and exercise, should be emphasised for all adolescents.

    Footnotes

    • nstettle{at}cceb.med.upenn.edu

    References

    1. 1.
    2. 2.
    3. 3.

    Author's reply

    1. G C Patton, Professor (patton{at}cryptic)
    1. Department of Gastroenterology and Nutrition, Children's Hospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, PA 19104-4399, USA
    2. Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, 423 Guardian Drive, Philadelphia, PA 19104-6021, USA
    3. Centre for Adolescent Health, Department of Paediatrics, University of Melbourne, Parkville Victoria 3052, Australia

      EDITOR—Despite the inevitable problems of measuring and adjusting for confounding the cohort study is often the only epidemiological tool available for the study of causal associations. Our analyses adjusted for a broad range of confounding factors that we believe are the most relevant to the outcome of eating disorder. These included antecedent weight, mental health state, exercise, and demographic characteristics. Mental health state is particularly important given the view that self concept and self esteem are important antecedents of dieting. Adjusting for these confounders did reduce the level of association with dieting, but it remained strong.

      In some instances antecedent dieting may indeed represent the early stages of an eating disorder. But arguing that dieting represents an early stage of eating disorder is untenable as it would mean classifying close to 70% of female adolescents as having incipient eating disorder. We would agree, though, that further work is needed to delineate the characteristics of dieters and dietaryregimens that most predispose to eating disorders.

      In an earlier paper we have described some features of the epidemiology of adolescent dieting in an Australian community sample.1 Severe dieters were characterised by higher levels of mental health problems, a feature they share with young women with eating disorders. This was not true for the intermediate dieting group, among whom most of the new cases of eating disorder in our recent study arose and for whom dieting was largely determined by weight—even though the body mass index of most young female dieters was in fact within the normal range.

      The broader context that Stettler et al raise—the epidemic of obesity—is crucially important. The message we seek to give is similar: that lifestyle changes that include regular exercise and maintaining a reasonable diet are the least risky options. The challenge is to present aclear and coherent view of what constitutes a reasonable weight and a reasonable diet, in the face of a continuing media barrage of bodily images and dietary programmes that are unsustainable for most adolescents.

      References

      1. 1.
      View Abstract