That's a bit harsch, I don't know how you could know that someone isn't sincere when they say they want to end their life and are cured four days later. People wanting to end their life because of DPDR isn't absolutely uncommon, apparently, and people having spontaneous sudden remission isn't also completely uncommon, herbs or not. (people wanting attention are not totally uncommon either, but personnally I am less afraid of being tricked by an attention seeker than wrongfully accusing someone who is actually suffering).
It's also possible to feel some effect when one is taking some molecule, and still have the effect dissappear or reduce after a while without being a placebo. It works like that to some extent with opioids, antidepressants, benzodiazepines and so on. And it can be the case with unwanted side effects as well, that are present only in the beginning of the treatment. So, I mean that the observation that the effect disappears after a while should not be enough to conclude that it is a placebo/nocebo effect. The right way to determine that would be to do a controlled experiment, and that's why we do such experiments. But I think that you mean that there is so much placebo stuff everywhere and probably around here for people who deal with a lot of anxiety that it is safer (and often cheaper) to keep the placebo hypothesis as the null hypothesis.
Anyway, placebo or not, if the positive effect doesn't last it's not that interesting anyway, so if it is expensive or very time taking, it's maybe better to wait and see.
The other thing I was wondering about is the meaning we give to placebo or contextual effects (I am not 100% of their definitions). It seems to me that when a molecule is tested, everything that isn't the molecule counts as contextual effect, so any psychological effect that is not being tested in the trial. But what if a person is stuck in a "DPDR symptom obsession --> anxiety --> more DPDR" viscious cycle, one could assume that a short experience of having less symptoms (or less obsessing over symptoms) could be beneficial on the long run, no? And couldn't this be brought by a contextual effect, among other things?