Permanently Deleted

  • DetroitLolcat [he/him]
    ·
    edit-2
    3 years ago

    Tl;dr: that "meta-analysis" linked above is a bunch of bunk science. This journal entry breaks it down well, but it's written in really academic terms and might not be super understandable without a public health/statistics background: https://ebm.bmj.com/content/early/2021/05/26/bmjebm-2021-111678

    Disclaimer: I have a background in statistics and public health. I have no background in virology/immunology/anything of the like. I can tell you why these statistics are crap, I can't tell you how ivermectin or covid-19 works.

    So that story is a meta-analysis of studies of ivermectin as a COVID treatment. In layman's terms, a meta-analysis is like a combination of a bunch of studies together. For example, if 10 universities conducted independent studies of smoking and lung cancer, another university could do a study where they look at the populations from all 10 of those studies combined and see if the same conclusions hold. A meta-analysis is typically a very powerful tool to confirm a hypothesis since it has a giant sample size and a very diverse, generalizable population. But a meta-analysis is only as good as the data fed to it, it doesn't generate new data.

    Now, a little background on the ivermectin craze: Ivermectin started getting traction because a few computer simulations and very early test-tube studies showed it might be able to fight the virus. Cranks and some doctors -particularly in Latin America - took this as evidence that ivermectin should be good for clinical use despite absolutely no clinical trials or clinical evidence for it working. The most-cited test-tube study used a dose of 5 micromole concentration before ivermectin had positive effects on fighting the virus. That's over 100-times the concentration of ivermectin that you would get from an FDA standard dose. In layman's terms, that means "so we dumped 100 times the normal amount of horse dewormer in a test tube full of covid and it had some effect." Not exactly airtight science.

    Now, let's look at the meta-analysis. If there were really that many good trials on ivermectin as a COVID cure, that would be pretty impressive! But it's an example of "garbage in, garbage out." For example, none of these "studies" in the linked article tell you how people were registered for the trials, how they got their control groups, the strategies they used to search for appropriate test subjects, or the level of certainty of their estimates.

    Let's dive into their data a little bit. So, right off the bat, if you go to the "All Studies" tab at the top of their restults section, you can look at the "Improvement, RR [CI]" header. That means "Improvement, Relative Risk [Confidence Interval]". This is a pretty normal metric in epidemiology, basically the risk of an outcome of people who got the treatment vs people who didn't. The confidence interval is like the margin of error. For results to be "statistically significant", the confidence interval shouldn't contain zero (i.e., you're confident the treatment's effect is not zero!). The first four studies here all happen to have a lower confidence interval of 0.01 or 0.02. So it just happened to be a wonderful coincidence that these studies were just barely statistically significant. I'm pretty skeptical of that.

    So, these ivermectin studies. What were they actually studying? Was it to prevent death? Hospitalization? Symptoms at all? Well...turns out they were just combining any study they can get their hands on to make the data look significant. Stay in the "All Studies" section. The first study was against "hosp." (hospitalization), the second study was testing for increased viral load, the third study against death, the fourth one against death, etc. So all of these studies were studying different things. You can't do a meta-analysis for completely different outcomes! Furthermore, look at the ivermectin doses on the right column. First study used a dose of 14mg, then 12mg, then 36mg, etc.

    So bottom line: there's no standard dose and they weren't even looking for a specific outcome. They just cherry-picked any data from any study that looked significant and stuck it in their "meta-analysis." None of this would pass peer review.

    I can't tell you the methodological flaws with every study in this "meta analysis" because it would take forever, but hell I guarantee you the people who made this shit didn't read the studies either.

    • Dirtbag [they/them]
      ·
      3 years ago

      “so we dumped 100 times the normal amount of horse dewormer in a test tube full of covid and it had some effect"

      Jesus. Christ.

      🤦

    • CrimsonSage [any]
      ·
      edit-2
      3 years ago

      Lol they just poured horse dewormer on viral particles and cells and then were suprised when shit broke. Fucking brain geniuses we got here. Tell ya what I'll take some covid positive cells and pour 100g of salt on em, bet that'll work. Get these chuds drinking salt water by next week, I'll sell crimson Saltwater for 20$a bottle.

        • MemesAreTheory [he/him, any]
          ·
          3 years ago

          And they'd deserve the kidney stones for being chuds. I draw the line at shitting out their own intestinal lining, that's a fate too cruel for even chuds.

    • Dirtbag [they/them]
      ·
      3 years ago

      Well…turns out they were just combining any study they can get their hands on to make the data look significant.

      What ths fuck is wrong with these people? Seriously deranged shit.

    • Dirtbag [they/them]
      ·
      edit-2
      3 years ago

      there’s no standard dose and they weren’t even looking for a specific outcome. They just cherry-picked any data from any study that looked significant and stuck it in their “meta-analysis.” None of this would pass peer review.

      Fucking clowns.

      :funny-clown-hammer:

      • Woly [any]
        ·
        3 years ago

        Friendly reminder that bleach, hydrochloric acid, and fire will all eliminate the presence of covid-19 in an in vitro study.

          • Woly [any]
            ·
            edit-2
            3 years ago

            That's true, but the petri dish won't scream while it dies, so you can pretend that nothing went wrong!

      • Dirtbag [they/them]
        ·
        3 years ago

        BMF noooo. not this hill bud, it's covered in layers of intestinal linings and shit.