One important point is that the hormones probably add more life than they take. You don't start on hormones AFAIK before they figure they can't cure you anymore. It's then about prolonging life. First to admit that dementia is horrible as well, but for most patients a risk worth taking.
Further reading:
-99% survival in 10 yrs regardless of initial treatment in the dataset (in article)?
-It seems there is a subset where ADT is first treatment. That is at odds with what our oncologist told us iirc. ADT prolonges, but does not cure. Makes some sense to prescribe ADT if alternatives are too heavy, but that is at odds with the 99% 10 yr survival.
As an actuary: only children have 99% 10 yr survival (guesstimate).
Went to the article. Not paywalled (nice). They include common logic comparisons and stuff. Love the approach to text processing. But no basic data tables?
Death should be really common when studying 60-70-80 yr olds. Gleason score mostly 6 or higher. That gives you a 30%+ baseline of dying (ballpark, Google).
Next thought: data gathering conditional on survival. Only if there is a follow-up, you can be included. That still makes sense re the study (relative chance of dementia), but makes interpretation harder. It's an analysis in one branche of the probability tree. ADT might make your dementia chance 100% if nothing else kills you first.
Still a nice study. Just don't think of this if you've just been diagnosed. Draw the entire tree. You need to survive first.