You bring up a solid point. It would be interesting to see, of the students leaving a PhD program with poor mental health, what fraction already entered the program with poor mental health? My hunch is that the stark majority acquired it while they were in the PhD program, but I have no data obviously.