Observations Medicine and the Media

Anorexia in young children: press coverage was flawed

BMJ 2011; 343 doi: https://doi.org/10.1136/bmj.d5072 (Published 17 August 2011) Cite this as: BMJ 2011;343:d5072
  1. Margaret McCartney, general practitioner, Glasgow
  1. margaret{at}margaretmccartney.com

Freedom of information requests by the Telegraph were misinterpreted as showing a threefold rise in UK childhood anorexia, but the methodology of the newspaper’s approach is not strong enough to draw this conclusion, says Margaret McCartney

On 31 July the Sunday Telegraph ran the headline “Hundreds of preteen children treated for eating disorders” (www.telegraph.co.uk/health/children_shealth/8672564/Hundreds-of-preteen-children-treated-for-eating-disorders.html). It followed, “Almost 600 children below the age of 13 have been treated in hospital for eating disorders in the past three years, new figures have revealed. The statistics include 197 children between the ages of five and nine—with cases within this age group almost doubling over the period.”

The article followed with a comment on the “‘pernicious’ celebrity culture” glorifying thinness as a reason for the trend and noted that 35 NHS trusts gave the information but that “even these statistics, disclosed under the Freedom of Information Act, are likely to be an underestimate. Some NHS hospitals treating such patients refused to provide any data, while among the 35 hospitals, some would only disclose the figures for those children admitted to wards after becoming dangerously emaciated.”

Freedom of information requests have become a standard tool in the journalist’s toolbox. They are cheap to do, requiring simply an email asking for relevant information, and relatively quick, with a maximum of 20 days’ response time, and effectively three and a half days for a government officer to obtain the information (more than this and it can be declined on basis of cost). Yet earlier this year the British Journal of Psychiatry published data examining the same question, reporting as part of the British Paediatric Surveillance System, which relies on reporting cards filled in by consultants to monitor rare conditions (2011;198:295-301, doi:10.1192/bjp.bp.110.081356). So is health information like this, obtained by freedom of information requests, accurate and useful?

Rachel Bryant-Waugh is consultant clinical psychologist and joint head of the feeding and eating disorders service at Great Ormond Street Hospital in London. She said, “I think that there is a fundamental difficulty in confusing terms: it’s easy to see how these things get a bit muddled, and it then comes across in a way that is slightly misleading. It is very unlikely indeed that 5–7 year olds present with anorexia nervosa. However, children [in this age group] may have low weight and significant eating difficulties but for many other reasons—for example, a difficult medical history, gastrointestinal problems, other adverse eating related symptoms related to a medical condition, or a range of psychological factors. Such children might present with eating difficulties as the main problem, but they do not have anorexia nervosa or bulimia nervosa.”

In other words, the requests for data using the Freedom of Information Act, which asked for the numbers of children with “eating disorders,” were not specific enough to enable concrete conclusions. “Eating is a very complex behaviour, in terms of all the processes required; many specialties care for children with eating problems. Formal eating disorders—that is, anorexia nervosa, bulimia nervosa, and eating disorder not otherwise specified (EDNOS)—are relatively rare in young children. However, they may be increasing in middle childhood. But the lack of consistent terminology means that there is a large difficulty in drawing conclusions here.” Dr Bryant-Waugh is currently working to tackle this as part of the new Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria.

Mark Berelowitz, consultant child and adolescent psychiatrist at the Royal Free in London would agree. “The DSM puts eating disorders into two categories, one is eating disorders—anorexia, bulimia, EDNOS—and the other is feeding and eating disorders of infancy and early childhood. From the way the freedom of information question was worded it’s unclear if the data related to anorexia nervosa or the vast number of other conditions much more common in young children, like failure to thrive and other feeding problems. And the other thing is that with the best will in the world it’s hard to produce good data quickly during the summer holidays and in a short time scale: data ordinarily need to be scrutinised carefully before they’re submitted, and I would have been astonished if any trust would have high quality data at its fingertips.”

Of the apparent rise in anorexia diagnoses in the very young he said, “When you come across this seeming threefold rise you have to ask yourself how good are the data: if you project the rate forward you would get an absurd number.” The difference with research questions is in their exactness. “Questions are highly specific when primary care trusts, for example, ask us for data, and of course you can go back and ask for details of what exactly they want, whereas raw data, such as that you get from freedom of information requests, risk erroneous conclusions. The data they got was from all feeding disorders, and then in the press this was expressed effectively as anorexia nervosa. I suspect that if the Sunday Telegraph had asked the Royal College of Paediatrics and Child Health or any specific clinical service how many patients had these disorders in these young age groups we’d have very different numbers.”

This is borne out by Neena Modi, a vice president of the Royal College of Paediatrics and Child Health and professor of neonatal medicine at Imperial College London. She has concerns “about the rigorousness of the methodology. A non-scientific journalist might not appreciate the limitations of NHS data and the need for a carefully designed study. For example, NHS hospital statistics are incomplete and not ideally suited for research of this kind. Diagnoses are often poorly defined so that like is neither being counted nor compared with like; data are entered by non-clinical coders so that complex conditions are often miscoded. There are also considerable missing data. If NHS data are to be used for high quality research it is important this is done by experienced researchers who understand the limitations.”

How does she feel about the inference that anorexia has increased among children threefold over the past five years? “I’d be extremely unhappy about drawing any conclusions from a study such as this that is not methodologically rigorous.”

How better to get such data, then? “If a journalist, researcher, member of public, or organisation wants to conduct a reliable survey of childhood conditions, by far the best approach would be to contact an academic group with appropriate expertise. The Royal College of Paediatrics and Child Health studies childhood conditions in the United Kingdom through a number of programmes, including Child Health Reviews-UK, the British Paediatric Surveillance Unit, the National Diabetes Audit, and the National Neonatal Audit Programme. Depending on the complexity of the study there might be research costs involved but there would be confidence in the conclusions.”


Cite this as: BMJ 2011;343:d5072


  • Competing interests: None declared.

  • Provenance and peer review: Commissioned; not externally peer reviewed.

View Abstract

Log in

Log in through your institution


* For online subscription