By TWD Australia

February 22, 2016 | News

Lies, Damn Lies and Statistics

Dr Joe Kosterich

A Facebook post that received a lot of likes, looked at rates of heart attacks in different countries and how it related to their habits. It noted that the Japanese ate little fat whilst the French ate a lot and both had fewer heart attacks than the British and Americans. There were a series of other comparisons.

No matter how it was sliced or diced, there were more heart attacks in English speaking countries. The conclusion was “eat and drink what you like. Speaking English is apparently what kills you”.

Whilst we can all have a laugh at this, there is a serious point. Not a week goes by without some sort of food scare. A few weeks ago it was bacon. And then vegetable oils, which we have been told, are good for us, also cause cancer.

Some 43 out of 50 ingredients selected at random for a cookbook were found to have an equal number of studies showing an increased and decreased risk of cancer.

We can laugh at the link between speaking English and heart disease. Yet there are truckloads of studies, which claim a “link” between something we eat or do and cancer or heart disease or some other illness. Statisticians have shown “links” between the star sign you were born under and certain conditions.

Statistical correlation does not equal causation.

The other statistical quirk is relative risk. Reducing the risk of some event from 0.5% to 0.25% will be promoted as “halving the risk”. In real terms it is reduced by 0.25% but this does not sound as impressive. Thus relative risk is used to promote pharmaceuticals and other medical interventions. Drugs for osteoporosis were sold on the basis of halving the risk of a fracture – from 2% to 1%.

So what do we do? It is best to ignore any talk of medical breakthroughs and to ignore any study, which claims to “link” something with illness. By next week the breakthrough will be forgotten and what caused cancer will now reduce it.

Words by TWD Australia.