In April 2019 The Sun newspaper ran the headline: ‘Pack of bacon a week increases risk of bowel cancer by a fifth, study suggests’.

What type of statistical measure is being reported?

In 2016 the BMJ sent out a press release announcing that ‘High levels of education associated with heightened brain tumour risk’. The research was based on the health and education records of more than 4 million Swedes. One newspaper then reported the research with the headline: ‘Why going to university could give you a brain tumour’.

Why is the newspaper’s headline misleading?

You are reporting on a paper which found that obese woman are 20% more likely to develop breast cancer than normal-weight woman. Based on this, what is the probability of an obese woman developing breast cancer?

A study explored the association between the consumption of ultra-processed foods and the risk of death, reporting a 1.62 hazard ratio for those eating 4 or more portions a day.

Susan, a journalist, wrote in her article: ‘Eating 4 portions of pizzas, pastries and processed meats every day increases risk of death by 62%’.

This number is accurate, but sounds very worrying. What would have better conveyed the magnitude of these results?

According to one study, a new drug halves the risk of a heart attack in the next year.

Normally, 1 in 1,000 people have a heart attack in a year. How many people have to take the drug in order to prevent one heart attack?

A study reports a relative risk increase of 2 of developing heart arrhythmia for those drinking two cups of coffee per day.

The study was performed in 100,000 people, randomised into groups A and B with 50,000 each. Group A had 2 cups of coffee per day and group B didn’t have any coffee at all. In group A, two people developed heart arrhythmia, whereas in group B, one person did.

Which feature of these results is the most problematic when trying to draw conclusions?

A press release reports that women with type 1 diabetes are 33% more likely to give birth prematurely.

Julia, a healthcare journalist, is surprised by this and wants to investigate further, so she contacts the researchers. She finds out that premature births occur in 6% of women without type 1 diabetes.

What percentage of births would you expect to be premature in women with type 1 diabetes?

Odds Ratios are frequently used to report the results of medical research.

For example, a study published in 2013 explored the association between taking statins and reporting muscle pains. They reported an Odds Ratio of 1.18 - meaning that the group taking statins had 1.18 greater odds of reporting muscle pains than the group not taking statins.

In absolute figures, they found 87% of people taking statins reporting muscle pains, compared to 85% of those not taking statins.

Based on this research, a Daily Mail headline claimed that statins ‘raise risk of [muscle] problems by up to 20 per cent.’

What mistake did the journalist make?

You can find out more about the Winton Centre for Risk and Evidence Communication via our website - wintoncentre.maths.cam.ac.uk - or you can contact us at wintoncentre@maths.cam.ac.uk