Thank you all for your good wishes! It's a big day. I may even snap a pic.
To answer Bruce's question, once treatment is the standard, the control group will be the current accepted treatment. Researchers won't include a non-treated group for ethical reasons if it's been concluded that treatment is a benefit. You could follow a non-treated group if you got the cooperation of people who refused standard treatment but were willing to otherwise be followed up for the same outcomes as the treated group. No one will do that study when treatment is already felt to be a benefit (although Dean Ornish managed to get control groups of coronary patients who didn't qualify for surgical treatment. It doesn't seem to be done in breast cancer research, though).
The original studies for tamoxifen and the aromatase inhibitors are what I need - how great the actual benefits are. But the control groups wouldn't be broken down into sub-groups that were careful to modify what risk factors they could; the study design would need both groups to be as homogeneous as possible.
I'd like to see how a group would do who refused the anti-estrogens but exercised, lost weight, didn't drink, followed a very low-fat whole food plant-based diet, drank green tea, included curries in their diets (curcumin is important but only works when combined with pepper and other curry spices - capsules don't work), etc. Like the Ornish studies, it would be nice to have one group follow an entire program and follow another group not on the meds but who declined (or were randomized into a control group, ideally) and followed their preferred lifestyle. The 'treatment' group would end up sub-classified according to how much of the program they followed, but also analyzed overall.
It'd be difficult to find subjects - they usually start on meds and then quit due to side effects. The fact they'd been on different meds for different times would be a confounder. It would be very very difficult to get approval for such a study. And funding - even more difficult. But I can dream ...