I think 3% is easily justifiable. There is a paper which argues risk is cumulative and is 33 per 1000 patient years or something like that. My experience confirms the cumulative nature of risk as I didn’t get PFS for 8 years on finasteride but got it eventually upon resumption after a hiatus. If you take finasteride long enough you are more likely to create the circumstances eventually that would lead to PFS.
If you think about it, there are many people on this site who got PFS after more than 5 year on finasteride. None of these people would be captured in the longer clinical trial of finasteride lasting 5 years. So by definition, lifetime risk of side effects is under-estimated in the literature.
Further, Traish is talking about persistent side effects, not PFS per se, I think. Any sort of lasting effect due to finasteride is a persistent effect. I have argued and I think this would be easy to justify that 100% of people who take finasteride would have penile tissue changes. In a lot more than 5% this would be outright and visible shrinkage. So as you can see it is in fact very easy to defend this number depending on how you define persistent side effects.
Also, Traish is arguing that side effects are significantly under-reported in the clinical studies due to bias, which is true as we know. Belknap wrote a paper about this.