Owen Yang
Have you thought about what risk is for a real person?
When we say 10% of pre-diabetes can move to proper (or ‘nominal’) diabetes every year, there is a 10% risk of progression every year. But what does it really mean?
I would say doctors can sometimes get confused, but statisticians or data scientists tend to be more ignorant.
Some reason of this ignorance is how statistics works. One fundamental theory it operates on is this random reality. Therefore, it is very easy for statisticians to think there is 1 in 10 ‘chance’ that someone may progress to diabetes.
This is true but not entirely true.
What may be ‘more true’ is that out of 100 people form the same population, 10 (or 10%) may be expected to progress to diabetes in the coming year.
However, there is an assumption of randomness in the 1 in 10 chance statement for an individual. Out of the 100 people, some are more likely to progress to diabetes for many reasons that may or may not be known, and therefore the true chance for that individual is not really 10%. The random assumption works when the different risks in the 100 people evens out.
I am not that great with fancy physics, but I guess in a model of reality in which ‘future has already happened,’ or in which ‘future has been determined,’ you could say that the 10 people are determined to have diabetes. The fact that we do not know the they are yet does not mean it could be any 10 people. So in this extreme scenario, it is more a proportion than a ‘risk’.
I do understand from mathematical and Bayesian point of view, one could argue 10% is true until proven otherwise. If you think in this way, you might have missed the point. I certainly would not say you are wrong. The point is that when facing a patient, we are not having an academic argument. We are trying to help.
When we are with a patient, it is not our place to argue with them what probability means. It is our place to deliver information in a way that is understandable and true at the same time. This will never be perfect.
In this case I would probably say ‘as far as we know, for every 100 patients at this pre-diabetes stage, 10 patients continue to have proper diabetes in the following year.’
I guess this does not change a thing, but I hope if would be a good mind exercise.
