I suppose the 3-month thing is because the men in the trials were recently diagnosed. As you say, the cancer could have been there, unsuspected and untreated, for some time before diagnosis, but the time of diagnosis is the first measuring point they can start at. It's a bit arbitrary, but it's all they can do.
The 22-month (median) extra life expectancy was for men with advanced bone mets; the advantage is probably less for a cancer that is less advanced. I think, too, that the men in the trials were under 70, and I know that an 82-year-old who was diagnosed by the same consultant at the same time as Tony was not offered chemo. That's why I'm not sure that demand would be high - when you narrow down the criteria to those who would probably benefit most (advanced Pca, under 70, otherwise fit), and then allow for the fact that many men would choose not to have chemo, I don't think the numbers would be overwheming.
Marje