In my humble opinion, what you are missing is actual communication with your doctor.
In this digital age of everything happens in an instant, and conversation doesn’t really happen, this is further exacerbated by overbooked and overburdened doctors offices.
It seems as if they are pretty much giving you everything from soup to nuts in your treatment. This makes sense since there are multiple ways to target this cancer. If this is true, I might ask why, it might be that there is something in your cancer that is a little worrying.
I took a boatload of endocrine therapy and I don’t view as debilitating.
I am not crippled in any way. Yes there are some arthritis issues but I am have always been very active and it would be expected. I am still very active and ride horses and take care of my property.
My thoughts are that I would not risk going through all of that again, not even by 1.6% based on someone else’s experience with a drug.
Did you talk to your oncology team about this worry?

I don't think you are missing anything in her argument; other than AI is what we currently have for treatment - and recommend. All the drugs recommended, or in use, have gone through clinical trials and deemed safe for use by our Food and Drug Administration (FDA) - BUT we don't have full data on long term effects for 10-20 years out from initial use. So folks, we are de-facto building new information for the next generation of BC patients as we go through our individual journeys. I am grateful for the survivors ahead of me because their info helps me make an informed decision about choices (drugs, treatment plans, BMX or not). BC is so individualized, and LCIS being only 10-15% of BC, so under-studied we are almost shooting in the dark about what type of effects any treatment will have on our overall physical, and mental, health.

It is a good thing you did a deep dive and found AI drops occurrence by whopping (LOL) 1.6%. I agree with you that that is not much for any associated future side-effects. The 1.6% is probably an average which means some people had a lower probability of recurrence and others had a higher. Nonetheless, at 1.6% I would probably decline and live my best life without continued treatment.

BUT keep in mind that the computer program used (or software calc ran), or any research article you read on longterm prognosis and effects of a drug, is only one of probably several. There could be differences in averages, or percentages, reported because each research project (even if on the same topic) has a different pool of persons/patients participating; and this will shift rates/percentages of recurrence, etc. reported between projects. So one research article, or software program (this to has a margin of error because the program was feed data generated by one, or possibly several, research projects) is not enough to make any decision. There is not typically a wide range of difference between data sets and possible outcomes...otherwise the drug would not be approved by FDA for use. But there is always individual responses to any drug that may or may not contribute to larger issues down the line.

You could look for another software package to run your individual risk for recurrence - or do a deeper dive into research on AI and longterm effects. OR if this is the software of choice believe the stats, know for some recurrence is higher and others lower, and go from there. But you're right 1.6% reduction is not much...and maybe not worth the risk for possible side effects - debilitating or not! Good luck with the decision process...