How often? Quite a bit, I bet, and abandonment both physically (therapist terminates) and emotionally (therapist withdraws from client).

Why no statistical data? It's not to the advantage of the profession. Also it would be rather hard to collect unless you want to do something like keep a database of all therapy clients and check in with them OR get someone to fund a long-term, sizable study of therapy clients.

Safeguards? See not in the interest of the profession above. Therapy is a business. You could force mental health checks as part of the licensure process, a la the police or the armed forces, but therapy schools claim to do that, don't they? The guards will guard themselves, to alter the old quote.

Why can they claim legitimacy? Because studies have been funded to argue that therapy is beneficial to most clients. Because too many people (not just clients) grant therapists authority - mental, emotional, guruishness, and so on. Because the business polices itself.

You know, I have not been harmed by therapy, but I'm a cynic about it. And whenever I find them helpful, I assume it was an accident.