Thoughts On: Ivermectin

Whether or not Ivermectin is effective against COVID is, in many ways, the Story of the Summer, so I must cover it. We expect the ongoing Oxford study will bring some clarity to the pharmacological question.

  1. The extant studies are unclear. There is not a single effective experimental study. There isn’t even data on usage as a prophylactic. There is evidence that there are regions where a majority of the population is using Ivermectin as a prophylactic against COVID (as opposed to river blindness). An otherwise-mediocre Argentine study suggests 90% of the population could be doing so.

  2. The studies that show Ivermectin has a 90% effect all have methodological and credibility issues (ref. The Guardian). Yuri Deigin describes one of these (Carvallo et al.) as “obvious shit”.

  3. The studies that claim Ivermectin is ineffective all have issues with a misunderstanding of concepts like the null hypothesis and p-values. In other words: they take a small study that shows a 25% effect, claim that a 0% effect has p>0.05 and a 95% effect has p<0.05, so therefore it is ineffective.

In conclusion, we have failures of measurement.

  • It should not be hard to measure population-level consumption of Ivermectin, but I am utterly unable to do so.

  • It is hard to assess which journal articles are good unless you do the assessment yourself; the existing journal system certainly is not doing a good enough job. Nobody appears to be trying to do so through AI analysis (or at least not well enough to matter). Nobody appears to be trying to do so by out-sourcing to a panel of their researchers.

  • It should not be hard to do statistical analysis of patient data records, yet the small studies that are done are often analyzed incorrectly.

Our long-form thoughts are at Yevaud.