Most adolescents who recovered directly after treatment with FITNET were still recovered at LTFU.
"Most" is not specific, and only indicates that it's more than 50%. It's not "all" or "nearly all", which seems to suggest that quite a few people went from being recovered after the trial to not recovered at the followup (and vice-versa).
If better numbers are available somewhere, it might indicate if much of the initial recovery was placebo effect, and that the loss of some recovered patients (due to the eventual failure of the placebo effect) was compensated for by the high likelihood of natural recovery in that age group.
If this is the case, it would undermine their claim that CBT should be offered for faster recovery.