Making a pig’s breakfast of research reporting

By David Shaw

On the morning of April 17th, 2019, fear of bacon filled the media: “Even moderate intake of red meat raises cancer risk, study finds” (Guardian); “A rasher of bacon a day ‘ups cancer risk'” (BBC) “Two rashers of bacon a day raises bowel cancer risk by a fifth” (Sky); “ANY amount of ham, bacon and red meat ‘increases bowel cancer risk’” (Wales Online); “Killer Full English: Bacon Ups Cancer Risk” (LBC). All these headlines concerned a study that found new evidence that processed meat increases the risk of bowel cancer.

The headline statistic of a 20% increase in risk was repeated in dozens of newspapers and on most radio news shows and TV bulletins. Yet an actual explanation of what the 20% means was almost entirely lacking, even when the researchers were interviewed directly on Radio 4. Most coverage simply stated that those who ate higher amounts of processed meat were 20% more likely to get bowel cancer than those who ate low amounts of meat. This is technically correct, but woefully uninformative, for two reasons.

First, no baseline rate is given, so readers, listeners and viewers remain unaware of what the risk is increasing from, and what it is increasing to. In the absence of this information, 20% sounds substantial, and no-one wants to increase their risk of bowel cancer by 20%. In fact, the baseline rate given (in the online BBC coverage with the unhelpful headline quoted above) is 40 cases of bowel cancer per 10,000 people who consume low quantities of red meat. That means that each person in this group has a 0.4% risk of developing bowel cancer. This key statistic about the ‘baseline’ risk of developing bowel cancer was absent from all the media coverage. Even if it had been provided, the reported increase in risk of 20% might appear to mean that the study found that those who eat lots of red meat regularly will see their risk of bowel cancer increase to 20.4%!

This leads to the second reason that the “20% risk increase” is uninformative. None of the reporting, not even the more detailed BBC coverage, explained that this is a relative risk increase, not an absolute risk increase. Instead of 40 cases per 10,000, there will be 48 cases in the group that eats the higher amount of red meat. This is indeed 20% more cases (20% of 40 is 8), but in individual terms the absolute risk increase is 0.08%, not 20% – a staggering difference, yet the more informative 0.08% figure was entirely missing from all the media coverage, giving the public an inaccurate picture of the actual risks.

In addition to the extremely misleading misreporting of statistics, the language used in much of the coverage is very leading. For example: “The risk increased by 20% with each extra slice of ham or rasher of bacon (roughly 25g) the study participants ate, and by 19% with each thick slice of roast beef or the edible part of a lamb cutlet (about 50g)”. This implies that every single portion of bacon increases risk by 20%, when in fact the study found that only increased consumption over time increases absolute risk by 0.08%. While not as important as the misrepresentation of statistics, this type of language also leans in the direction of scaremongering.

The researchers who conducted this study also share some of the blame. The research article itself used hazard ratios to describe the increase in risk, but these have been criticised and are not as helpful for the public as absolute risk descriptions. The researchers’ quotes in the media coverage did little to clarify the information provided, and the CRUK press release also did not mention absolute risk. Indeed, the headline on the press release was “Even moderate red and processed meat eaters at risk of bowel cancer”, despite the fact that those who eat little red meat and those who eat none are also at risk of bowel cancer.

It is a damning indictment of science journalism that virtually all coverage of this story in the British media reported the 20% figure, rather than the more accurate 0.08%, thus overstating the individual absolute risk of increased intake of processed meat by over 200%. Of course, had the numbers been reported accurately, the story would not have been so scary, and thus not so interesting for the public. But that is not the point of good journalism, which is meant to inform. The public should be accurately informed about new evidence of risk, and journalists and scientists have a duty to present information in a clear and intelligible way.

Author: David Shaw

Affiliation: Institute for Biomedical Ethics, University of Basel, Switzerland; Department of Health, Ethics and Society, CAPHRI Research Institute, Maastricht University, the Netherlands.

Competing interests: The author eats low levels of processed meat.

(Visited 145 times, 1 visits today)