Lies, damn lies and statistics!

Posted: 01/02/2020

This morning I read the following headline: “Dementia Risk Linked to Some Medicines’.

The article on the BBC website cites a newly published study which hints at a potential link between commonly prescribed medications (in this case anticholinergics) and dementia. The study was very large and appears to be well designed, but was ‘retrospective’, meaning it looked back over data trying to identify outcomes that were not considered initially.

The headline grabbing conclusion of the study is that these drugs are linked with a higher risk of dementia. There is however a problem with such conclusions. ‘Link’ is often misinterpreted as ‘x causes y’. It was found that more people developed dementia in the group of patients taking these medicines than in a matched group who were not. The increased risk has been suggested as high as 30%.

Stats like this are alarming and capture attention, but it must be understood that an individual’s absolute risk is not 30% higher, it is their relative risk. When reading a headline that says ‘increases risk by 30%’ it is easy to add 30% to whatever we think our risk was – lets say 10%, and we end up concluding that we now have a 40%, or 4 in 10 chance of developing Alzheimers.

This is not the case!

A 30% increased risk moves our baseline 10% up to 13% (an increase of 3% points, 30% of 10%). So in all we had a 1 in 10 chance, and if we take the meds it is now 1.3 in 10. Not quite as alarming, but no headline grabber.
The other thing to bear in mind is that a ‘link’ DOES NOT mean x causes y. What are the possibilities here?

1. Anticholinergic medications increase your risk of Dementia Or perhaps…

2. People who develop dementia have symptoms early on which are often treated with anticholinergics Or maybe…

3. Taking anticholinergics is more common in people who do a particular thing, and it is that thing which raises the risk of dementia.

These possibilities are well recognised and must always be considered where studies identify links rather than causation. We describe them as confounders and confounding they certainly are.

So what should I do?
It is difficult to give blanket advice when news such as this study breaks. The best approach is to discuss your particular risks with your GP, who will balance benefit against harm. No doctor wants their patient’s to develop dementia, however statistics must always be taken with a pinch of salt. (not literally, as this could increase your risk of hypertension by many % points!).

If you would like to read more about the study in question follow the link below:

Latest News

View all news

© Mayfield Clinic Oxford 2024

Powered by Hero.