Thursday, September 6, 2012

Correlation and Causation

These are not words most of us use every day (unless we are scientists, that is).

I recently read a fascinating news item telling me that people who had cataract operations were less likely to break their pelvis.

Now, one could spend time trying to find a physiological link between the replacement of cloudy lenses in the eye and increased strength of pelvic bone.

But a little thought will show us that once people can see better, they are less likely to trip and fall.

There is a strong correlation between these two events, but one does not physiologically cause the other.

The world is full of apparent links, some of them true, some not.

We were pretty sure that smoking caused lung disease, but because we could not do controlled experiments on people, it was much more difficult to come up with proof that one caused the other (though it accumulated eventually). Note however, that though people are not allowed to smoke on planes, Governments and private investors still reap huge profits from the sale of tobacco. Addictions are extremely difficult to cure. We don’t seem to know what causes gambling addiction, but it is another one that brings huge amounts of money into Government (and private) coffers. Makes one wonder if a real cure would be welcome, or whether enough research is being done into the causes.

We are now optimistic about ASA (acetyl salicylic acid) [sometimes called by the trade name of Aspirin] and its preventive effects on many important threats to our health – heart attacks, colon cancer and others. But we are not sure of these effects, partly because there are not enough truly scientific double blind studies behind the claims, and partly because people’s lifestyles are so diverse that it is almost impossible to unravel the effects of ASA from many other substances they take on board, or many other of their behaviours. Again, this is because we cannot do truly controlled experiments on people (they have to be the worst possible subjects!) and must wait for the truth to gradually become clear after innumerable studies seem to add up, and meta-analyses finally lead us to a reasonable conclusion.

We should always be cautious about accepting supposed causations that may just be correlations.