The whole "study" is post hoc ergo proper hoc. They didn't measure PCR rates, the key indicator of whether the drug works or not. (See the left axis of Raoult's graph here.) They only measured outcomes which could be influenced by dozens of variables.

It's likely (almost certain) that the drug was working perfectly when the patient died of co-morbidities or because they started giving the drug too late.

The left wants us dead.