Total Pageviews

Monday, September 3, 2012

Media reporting of food related health claims


In 2009, the World Cancer Research Fund (WCRF)[1] conducted a survey of 2,400 UK subjects to ascertain their views as to the evidence linking diet and physical activity to cancer. WCRF argue that the advice linking diet and physical activity to cancer is both robust and relatively unchanged in the last decade. However, what they found was that in the 55+ group, 60% felt that scientists were always changing their mind and that 30% thought that the best advice was to avoid health advice and eat what you want. The figure for the sample as a whole were marginally lower. A group of London scientists decided to follow this up with a survey of material appearing in the UK press in one week covering food and health and to examine this the media representation to determine the accuracy or otherwise of the coverage. Their paper was published in the journal “Public Understanding of Science”[2].The lead authors were Professor Tom Sanders, a world authority on diet and cardiovascular function and Dr Ben Goldacre, best known for his book “Bad Science”, but he is also a research fellow at the London School of Hygiene and tropical medicine. They were joined by Ben Cooper a medical student and William Lee an MRC Training Fellow in Psychiatry.

Overall, the top 10 best selling newspapers in the UK are read daily by about 10 million people so print media has a very big audience. The study was carried out in the first week of November and the focus was on articles that recorded an actual health claim to a food in some way. Stories about GM or about waste or other issues were excluded - the story had to involve a health claim as defined by the European Food Standards Agency. A total of 111 such stories were reported in the week. The next stage was to subject the reported new story to two grading systems specifically designed to grade scientific evidence. The first of these is called the Scottish Intercollegiate Guidelines Network (SIGN)[3] and that of the WCRF. As the authors point out, each might have its flaws but “..taken together, these tools represent robust and widely recognised measures of relative evidence of quality”.

I have averaged the scores of the two grading systems since they were generally similar. The single most important figure was that only 10% fell into the category “convincing”. All the rest fell into shades of doubt such as 15% “probable and 7% “possible” or 4% “unclassifiable”. By far the biggest figure of 64% is for “insufficient” evidence. In effect the general take home message is that 1 in 10 media stories involving linking some aspect of diet to health is true. For two thirds, the evidence is non-existent and the rest fall in between. Where does the blame lie?

 Journalists rely quite a lot on press releases and the nature of press releases is that the releasing body, be it a company, an NGO, a university or a trade organisation want the media to take up the press release. Ideally, the journalist should use the press release as the starting point to make contact with the researcher and to develop the story from there. But quite often, it is the press release alone which makes it to the print edition. One of the main culprits in this communication change in my view are university communications units. They are constantly looking for press coverage in an ever increasing academic environment and they have a captive audience of academics who enjoy their moment of glory in the media. 

I would go further back in the communication chain to the actual research process and bemoan the growth and dominance of unchallenged data on food and health. Fine, I understand that certain associations between diet and health are not readily amenable to testing in intervention studies. But these are few and far between. What is exasperating is the rush to publicise the relationship observed between serum whatever and some wonderful health attribute in some cross sectional study without any direct evidence from human intervention studies that the relationship stands up to this test. One area which is without doubt the most culpable these days is the great news that “scientists have discovered a link between some nutrient intake profile, a relevant common genetic variation and some disease”. These triangular links of diet, phenotype and disease (e.g. the gene for some lipoprotein, high blood cholesterol and olive oil intake) are ten-a-penny and each worth more or less nothing without some verification with an intervention study. So rapid is the expansion of this unholy triangle that the funding to establish an intervention study is never likely to be extensive. Which of the putative claims do you spend your money on?

An exception is the work done by scientists at the University of Ulster and Trinity College Dublin who showed that if low riboflavin status was corrected in persons on medication for hypertension, those with a common (ca 30%) genetic variation showed a dramatic reduction in blood pressure[4]. So they set out to recruit equal numbers of the three genetic classes (the less common genetic variation being absent or inherited from one parent or both) and they carried out an intervention study (riboflavin supplement versus placebo) which proved that those with the less common genetic variation responded very positively with dramatically reduced blood pressure. They went back then four years later and those that had been randomized to the placebo were now given the riboflavin and vice versa and again they proved the association. This is darn hard work but it is what is needed to take an “association” to a “verification”. Sadly, the moment of glory in the media seems to satisfy most scientists and most universities. It is easy to be critical of journalists for not following press releases in more depth getting a second and their opinion but those in glass houses......


[2] Cooper BEJ et al (2011) The quality of the evidence for dietary advice given in UK national newspapers. Public Understanding of Science, May, 1-10
[4] Wilson et al (2012) Riboflavin offers a targeted strategy for managing hypertension in patients with the MTHFR 677TT genotype: a 4-y follow-up. Am J Clin Nutr.  Mar; 95(3):766-72. Epub 2012 Jan 25.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.