Nice!
The HCQ looks definitively like report+publication bias, but it would be nice to see an analysis by an statistic expert.
The Ivermectin looks more interesting. I read the first study on the list Chowdhury et al. https://c19ivermectin.com/chowdhury.html
* If I understand correctly it's no double blind, but as far as I can see both group get the same treatment. Also the criteria are hard (like death, instead of how much pain do you feel in the scale from 1 to 10), so my guess is that it's not a big problem here.
* It's strange that one group got Ivermectin and the other got HCQ. So why is this study not in the other site?
* I think it's not preregistered, so it's not clear how many similar experiments were made.
* The main result is that 0/60 patients with Ivermectin get hospitalized, but 2/56 patients with HCQ. To simplify the discussion, can I round it to 0/60 patients with Ivermectin get hospitalized, but 2/60 patients with HCQ?
* This looks like 1±1/60 in both case to me. I'm a Mathematician but I'm not an expert in statistic. Anyway I took to many laboratory classes for Physics, and it definitively looks like 1±1/60 that is very unconvincing. I don't know how they get 80.6% from that, but I can ask one of my friends.
* Anyway, let's use the null hypothesis and suppose that have Ivermectin and HCQ.no effects. And for simplicity, let's assume that the hospitalization rate of no treatment is 1/60. So if we repeat these kind of experiments with groups of 60 patients we must use a binomial distribution. From https://www.di-mgt.com.au/binomial-calculator.html with n=60 and p=0.01666666666
Probability of 0 cases: 36.5%
Probability of 1 case: 37.1%
Probability of 2 cases: 18.6%
Probability of more than 2 cases: 7.8%
So if we make many experiments, and each of them has 60 persons with Ivermectin and 60 persons with HCQ, then in 36.5%18.6%=6.8% of them the we will get only by a fluke a result like the reported. So if there are 16 not pre-registered studies, on of them may get a result like this. Pre-registration is very important.
Looking at the other studies, in early interventions there are too many that have 0 cases in the treatment branch. For late intervention treatments, most have not 0 cases in both groups. It's suspicious, and my guess is that is inclining the result.