You should actually share the papers, or the best ones, not just state that they exist.

The reason is that some of them start off with a strong selection bias because they haven't been carefully following a set of people for years; rather, they say, "Hm, let's find some transgender people and see if they are transgender!" Really. That's about the logic of it. Papers with this structure can be used to make conclusions about the experiences of people who identify as transgender (and they're very useful for that!), but they're not very useful in measuring regrets, detransitions, and the like, because the initial step already tends to exclude those people. Consistent with this, if you pre-select a group who identifies as detransitioned, their set of concerns is quite different than the people who mention detransition in the first group.

Now, of the papers that don't have that flaw (or at least have it less), rates of regret also come back relatively low. For instance, the Amsterdam Gender Identity Clinic reports rates of a phenomenally tiny under-1% for gonadectomy (DOI: 10.1016/j.jsxm.2018.01.016). However, they also do a great deal of work to ensure that this is the right thing for the patient (as you can tell by the rates of progression dropping as the number of people applying for treatment increases). Unfortunately, this doesn't extrapolate very well to other situations unless they follow the same protocol with the same skill to determine who to treat in which ways.

But even that is somewhat flawed, because you also have to consider that some people might also have had equal or lesser regret, at this point, had they not transitioned. To really do a careful study, you need some sort of control group that you can generate through ethical means. For instance, of those people who didn't transition after, say, four years of considering it, what fraction claim to regret it?

So it's really a good idea to share the citations you're thinking of, if you're thinking of any. The details can and usually do matter with regards to the conclusions that are reasonable to draw. If you don't have anything in mind, I recommend not specifying any particular number but using language like "the overwhelming majority" so as not to convey more confidence in the exact value than you're prepared to back up.