Joachim2 on Natural Selections

[Joachim2]

The study which resulted in dexamethasone becoming part of the officially designated standard of care (reported here: https://www.nejm.org/doi/full/10.1056/NEJMoa2021436) found an a 17% reduction in death, with a 95% confidence interval of 7% to 25% reduction (or as given in the study, 83% of the control-arm deaths in the treatment arm with a 95% confidence interval of 75% to 93%). (The impact was more positive--around 26% reduction--for those on ventilation, about 18% reduction for those on oxygen only, and actually negative--increasing the mortality rate--for those not requiring either.)

The evidence quality in the study was high in some respects, such as in that a large number of patients were enrolled. Dosage was essentially binary, so there is apparently no evidence pointing, even directionally, toward an optimal dosage. And so apparently no evidence of a dosage-dependent effect.

Although the overall numbers treated are lower, the various ivermectin studies, taken together, show dose-dependent positive effects of significantly larger magnitude than in the dexamethasone study, and show positive effects at every stage and severity of disease (including for prevention as discussed here), and under varied conditions and against various controls.

Statistically speaking, larger effects such as those appearing from ivermectin can be detected by smaller studies. (When the effect is large [and dose dependent] you don't require as much data to conclude that the results are not likely the product of chance.)