@Sloejelly
You could only say that some appeared harmed less than others. You could not tell if they are in a better position, despite the harms they have suffered, than they would have been had they received alternative treatment (eg psychotherapy) or no treatment.
My hypothetical was tracking the whole group so that you might find subgroups who might potentially benefit from a treatment, e.g., in the hypothetical, 100% of that subgroup of autistic boys who had gender dysphoria and incongruence before age 4 continued to have gender dysphoria and incongruence as adults, Assuming it's a large enough group, once you have identified that subgroup, you can try various treatments on those with the characteristics.
I don't understand what harms you mean since you wouldn't have used any particular treatments on them yet. You're looking at the data you presumably will get from the clinics in order to try to find subgroups that you hypothesize might benefit from future treatment, and only at that point trying treatments on them (e.g., cross sex hormones, psychotherapy, both, neither). That is the point I would look to see if a treatment helped or harmed.