What makes a good health scare? There are three essential ingredients. First, there has to be something mundane that lots of ordinary people do, and preferably enjoy doing. Second, there has to be a nasty disease, or a frightening health condition.
Third, the word “causes” must appear.
A big scare last week concerned Nutella, with reports that the processed palm oil used in the production of the hazelnut spread might cause cancer; the week before it was busy roads causing dementia; in 2015, readers might recall the scare was bacon consumption leading to bowel cancer.
The problem with such stories is not the underlying science, but that the fact that they generally do a terrible job of conveying a clear sense of the scale of the risk to the typical reader. The palm oil research is at too early a stage for any kind of quantitative headline about risks to health, but we had very specific sounding risk scale claims in relation to the dementia story. Media organisations drew attention to the finding from a Canadian study that the risk of developing dementia could be up to 12 per cent higher for those living within 50 metres of a major road.
This is known as a relative risk. Yet none of the truckload of reports referred to the absolute risk – this is the risk to anyone of contracting the disease in question over their whole lifetime, and it is essential to making sense of such stories. Making a judgement about risk without knowing the absolute risk figure is like setting out on a long and dangerous sea voyage without a compass.
The absolute risk of dementia is certainly calculable. The Canadian study observed 2.2 million people aged 55-85 over a decade. And it found that 244,000 of this sample developed dementia in that time. That implies the absolute risk of developing dementia for people over their lifetime is around 11 per cent regardless of where they live (which is in line with other studies).
So, of 100 people, 11 will typically be afflicted. If living within 50 metres of a busy road increases the risk of dementia by up to 12 per cent, that elevates the 11 per cent absolute risk to around 12 per cent – so, of 100 people living near a busy road, around 12 will be afflicted rather than 11.
That’s one more dementia case per hundred people who live near busy roads all their lives (possibly) attributable to air pollution. Which is not negligible, but probably sounds rather less dramatic to most people than a “12 per cent increase in risk”.
It’s important to spell this out because when people hear a relative risk figure they often, mistakenly but understandably, think they’re hearing the absolute lifetime risk. This was apparent during the 2015 bacon scare, when reports of an 18 per cent increase in the risk of contracting bowel cancer from consuming two rashers of bacon a day were widely interpreted as suggesting an almost one in five chance of getting it.
In fact, the lifetime risk for regular bacon eaters of developing bowel cancer is seven per cent, up from six per cent for those who don’t eat it. Those are surely much more useful statistics to the vast majority of readers than the 18 per cent increase in relative risk in all the headlines. But this isn’t the only piece of context needed to make sense of such stories.
As Jen Rogers of the Royal Statistical Society has pointed out, there are bigger risk factors for dementia than living near a busy road. Smoking, for instance, is reckoned to increase the dementia risk from 11 per cent to 14 per cent and obesity to 17.5 per cent. Old people would be better advised to worry more about these harmful lifestyle factors than where they live.
Similarly, rational people who are worried about cancer should concentrate more on giving up smoking than fretting about their bacon sandwich consumption. The absolute lifetime risk of lung cancer for a man who never smoked is estimated to be about 1.5 per cent. For lifetime smokers it is 17 per cent. That’s a 1,100 per cent increase in risk by the way.
It’s important for people to be able to rank the risks they face. Whatever the palm oil research ultimately concludes, it seems unlikely Nutella will be shown to be as risky as cigarettes. This information – the size of other relevant hazards – is the map for your sea voyage through the oceans of risk assessment.
So there are three basic questions you should ask when you read the next “x gives you y” health story.
What is the absolute lifetime risk? What is the absolute risk when adjusted for the risk from the activity in question? And what are the relative risks from other relevant harmful activities? That is the information you need to decide whether or not to be alarmed.
This article appeared in The Independent on 15/01/17