Prevention of eating disorders: where do we start?


This systematic review was conducted to assess the state of play in eating disorder prevention research. The prevention of eating disorders is both a broad and controversial area of investigation. The breadth of prevention activities around eating disorders arises from a need for different activities to address the risk and prodromal factors covering anorexia nervosa (AN), bulimia nervosa (BN), binge eating disorder (BED) and other/unspecified eating disorders (OS/UFED). A controversy arises from disputes about what can be considered to be mutable precursors to eating disorders and which immutable risk factors (like gender) can and/or should be used as guides for targeting preventive interventions.

Le et al (2017) argue that the considerable financial costs of eating disorders treatment make the case for looking into the possible benefits of preventive interventions, and they present an overview of studies indicating some promise for prevention in this field.

It is worth mentioning here how challenging it is to research preventive interventions. Often it is not possible to follow through a sample for long enough to see if the intended preventive effect is observed at the time it is expected to occur. It is also rarely practical to gather a large enough sample of people to enable observation of potentially small effects in a reliable way, or to control for all the other variables that may influence outcomes.

However, even allowing for these challenges, Le et al (2017) make it clear that the existing literature on preventive interventions for eating disorders is not of a high quality and that, whilst not of a low quality, the handful of meta-analyses and systematic reviews on preventive interventions also have their own shortcomings. What a previous meta-analysis by Stice et al (2007) had indicated was that intervention factors (e.g. having an interactive, multi-format, professional-delivered package targeted at the highest risk individuals) seemed to improve prevention efficacy. Watson et al’s (2016) review similarly found that ‘selective’ interventions (e.g. those targeting at-risk population groups) had the best efficacy outcomes.

As is hinted at in the paper, Le et al (2017) approached this review task at the same time as another research team was undertaking a very similar review (Watson et al., 2016). Whilst the knowledge of each other’s work was probably quite frustrating, it actually seems to have been beneficial for Le et al (2017) and the field as a whole, as Le et al (2017) have added not only to the scope but also the approach of Watson et al (2016).

It's notoriously difficult to carry out reliable prevention research, because studies need to be large, long and carefully conducted if they are to find accurate results.

It’s notoriously difficult to carry out reliable prevention research, because studies need to be large, long and carefully conducted if they are to find accurate results.


It would be nigh on impossible to provide a summary of the review approach, so thorough and exacting as it is. Suffice to say that it sailed through the categories of the CASP systematic review appraisal checklist (PDF) and that as a reader I felt highly assured of the scope and rigour of the review and analysis. The whole process, selection of statistics, consideration of biases: it’s all there and could be replicated, if you had two years to spare.

Articles were sourced using the existing systematic reviews and this paper added 34 studies to Watson et al’s (2016) work. Interventions were classified as:

  1. Universal
  2. Selective
  3. Indicated (targeted at people who have some eating disorders symptoms; and it therefore could be argued that these are not strictly preventive).

Within each of these three groupings were a variety of different approaches, intervention types and delivery styles, age groups, countries and settings.


101 trials were investigated in this review: 18 universal, 79 selective and 4 indicated. The vast majority were aimed at people under 24 years old, with most delivered in a group format and/or a classroom setting.

Le et al (2017) present their findings by therapeutic type/intervention approach (for example, cognitive dissonance, media literacy, CBT) under the three global at-risk population groupings of universal, selective or indicated. They consider comparisons between the preventive interventions and active controls, and then preventive interventions and wait-list or no intervention conditions.

This systematic review is therefore a collection of mini-reviews, and the small size of these starts to raise questions about their individual quality. Having said this, my sense is that Le et al (2017) have made the most sensible classification approach they can and have to work with what is available. The reality is that this is a field where people are trying out ideas and inevitably this means some compromises in reviewing. Le et al (2017) appear aware of this, providing sufficient detail in each mini-review for the reader to make their own decisions on interpreting the analysis.

Universal prevention

  • ‘Media literacy’ approaches appear to help reduce eating disorders risk factors for adolescents for over two years post-intervention. However, some effects became insignificant with the removal of one study (illustrating the value of Le et al’s thorough approach)

Selective prevention

  • Cognitive dissonance interventions were found to reduce eating disorders risk factors for selective populations, as were CBT and media literacy approaches, and there was good evidence to suggest multi-session interventions were superior to one session approaches. Again, sensitivity analyses then challenged the apparent benefits of the interventions
  • Le et al (2017) advise ‘real-world’ testing for selective prevention interventions, believing that the evidence base is now sufficient to indicate that this would be worthwhile, but I would suggest that researchers would be better advised to focus on research quality.

Indicated prevention

  • There were not enough studies to draw conclusions about the indicated preventive interventions.

These limited findings sit against a backdrop of publication bias and research biases were identified in well over half of the papers. On top of these problems is the very important caveat that only 8 of the papers looked at risk of eating disorders onset as an outcome. This is important, and potentially could get lost or misunderstood, but the vast majority of the papers reviewed were looking at the impact of preventive interventions on ED risk factors, not on ED outcomes.

This well conducted systematic review highlights the lack of relevant and reliable prevention research for eating disorders.

This well conducted systematic review highlights the lack of relevant and reliable prevention research for eating disorders.


Every outcome from this review needs to be handled delicately, being based on a small number of studies of sometimes questionable quality. Le et al (2017) acknowledge this openly, and perhaps it is safest to see this review as a really valuable foundation stone for the more robust systematic review that can hopefully be carried out in five or ten years’ time.

Coming back to some of the problems of universal prevention interventions for eating disorders, the paper does not reflect on the possibility that an intervention actually raises risks inadvertently beyond the measures taken (which is slightly different to saying that the risks measured did not show any signs of increase). For example, in obesity programmes, it is possible that these might increase anorexia nervosa risks in a universal population.

There is also a question of the amount of benefit that can be achieved when the mutable risk factors for eating disorders only account for a small proportion of the known risk factors. This is not to be pessimistic about the possibility for prevention, but we need to know more about the costs of providing multi-session interventions, and the costs of any iatrogenic outcomes, before moving forwards with any commitments to intervention implementation.


Despite the number of papers reviewed, the diversity of approaches and populations covered, and the varying quality of the studies included means that this systematic review of eating disorders prevention interventions has relatively limited findings. One of the things it does provide is an illustration of the rigour with which a systematic review can be conducted. This is great, on the one hand (why do a systematic review if you are not going to be thorough?), but the sheer number of tests, checks and individual study exclusions in this review results in something rather like p-hacking in reverse, in that, if you scrutinise everything, you are bound to find multiple reasons for having ambivalence towards any results.

As a field, eating disorders prevention intervention work has a long way to come, and from this review, it would seem that endeavours to further the field need to focus on methodological rigour and quality.

We have a long way to go before we can reliably prevent eating disorders.

We have a long way to go before we can reliably prevent eating disorders.


Primary paper

Le LKD, Barendregt JJ, Hay P, Mihalopoulos C. (2017) Prevention of eating disorders: A systematic review and meta-analysis, Clinical Psychology Review, Volume 53, April 2017, Pages 46-58, ISSN 0272-7358,

Photo credits

Share on Facebook Tweet this on Twitter Share on LinkedIn Share on Google+