Hey there, below is a list of comments based on yours. I currently work on clinical trials and I hope that this helps clear up anything. All together some of your comments are on point with only a subtle lack of nuance. Hope this helps.
* Your first point on the trial registry is certainly correct. You wouldn't want multiple N of 100 trials. The point of that slide is exactly that, we need higher quality data to make better decisions. If we can combine the larger data sets together it is an untapped resource. Notice they don't advocate doing this. Instead, they are using this slide to advocate for their position that a large adaptive trial should be done. That is the TOGETHER trial.
*You mention a problem with the word "complications," it is grammatically consistent and implied that the complications are from COVID. The trialists are the ones who make that determination, hence the need for blinding. Some cancer or kidney disease patients may have complications and that would be measured in the adverse events. See the comment below as to what I expect to see in the peer-reviewed paper
*You mention that the randomization was not stratified between groups based on risk factors. Great comment, it probably should have been, every trial has its flaws and this could be a big one. What I expect to see in the peer-reviewed paper is a table detailing how many and what kind of comorbidities were in each group. That will help us compare "apples to apples."
*Regarding the primary care/ER/hospital recruitment it is incorrect to assume that primary care or ER patients were hospitalized at some point. I'm a former paramedic and someone who works on clinical trials. We often have patients visit a primary care doc or ER and leave the same day. That's what they are talking about here. Now, some may have been inpatient (in the hospital) and we will have to wait for the peer-reviewed paper to see.
*Regarding hospitalization it is an extension of hospitalization beyond the 5 days that would be included as the progression of COVID-19 would lead to that (see the protocol for more). Definitely hard to determine in the slides so I don't think this is your fault at all
