3) don't trust the title or the abstract. In this case (https://jamanetwork.com/journals/jama/fullarticle/2777389), they had no deaths in the ivermectin group and one death in the placebo group (hidden in "Table 2 Outcomes in the Primary Analysis Population"). So all they had to do to please their sponsors was to make the "time to resolution of symptoms" the most important point and to downplay deaths.