Regarding the Rett trial, one of the primary co-endpoints was CGI-I, apparently added after a consultation with a regulator. It failed, and probably not because there weren't enough enrolled patients. Because of the pandemic, this endpoint was determined remotely and not in person.
CGI-I, never a very sophisticated measurement to begin with, requires the physician's impression of the patient's condition; there is no doubt that using a screen instead of an in-person visit would seriously reduce the information available. Trofinetide's trial, if I recall correctly, used a mix of in-person and remote visits for this endpoint.
What I am describing, originally suggested here by JoeBear, is not a level of noise that could be overcome by higher patient enrollment, but a basic lack of signal (due to pandemic-required methodology) that could never be overcome.
There was no suggestion, by the way, that this endpoint was compromised by a too-high placebo response.
Market Data 
Markets 
AI
