Ugh. PCR can have false negatives, but in a clinical setting they can do this really crazy thing called "looking at the patient" or "asking them how they feel." So, if they were clearly sick, as they would be before getting to the hospital and into Chloroquine treatment, and they got qualitatively better, that is a good indication that the PCR result is correct.

It's difficult to due true double blind random studies in these cases because doctors are there to give everyone the best chance at recovery, not to pick half of a group to potentially do worse.

Hopefully we can ethically do well design studies on controlled groups with moderate symptoms, but until then this imperfect clinical data is very interesting.