Bedtime for Snake Oil

COVID-19: Bedtime for Ivermectin

And Hydroxychloroquine, Too

There are still bots on X promoting the idea of ivermectin and Hydroxychloroquine for COVID-19. For all practical purposes, the COVID-19 pandemic is not entirely over. But let’s look back at it. A lot of people thought there was some nefarious thing going on when doctors wouldn’t prescribe ivermectin or Hydroxychloroquine for COVID-19.

I am a physician. I can tell you there is no reason to give these drugs to somebody with Covid. If the pandemic is over, this is only relevant for future pandemics. If you don’t believe me, then read on. I will explain why this is not a nefarious action by doctors or pharmaceutical companies to deprive you of effective therapy. It is simply snake oil that I am trying to protect you from.

Now, if you do the right things AND take Hydroxychloroquine and ivermectin, I’m OK with that. They are ineffective, but at least you are protecting yourself from dying by doing the right things. But, if you do not want to do the right things (vaccinate, wear masks, socially distance, and wash hands), I hope I can change your mind or, at the very least, give you something to think about.  

Now, another note. Ivermectin and Hydroxychloroquine are equally ineffective against COVID-19, and these arguments apply to both.

Another principle is that one does not need to know the mechanism of action of a therapeutic, only the effect. For example, no known mechanism of action of Hydroxychloroquine or Ivermectin would indicate their purpose in treating COVID-19, but that doesn’t mean there wouldn’t be an effect. The question I will answer is this: What is the effect of treating COVID-19 with either drug?

COVID-19 Treatment: Desired Effect

Well, there’s one of the bugaboos, too, now, isn’t it? What effect do we want the drug to have? The answer to this question is the hardest thing to get from advocates of these drugs: What’s the effect? In other words, what will change? Would it be the survival rate, duration of symptoms, and/or communicability? I’ve never had anybody answer that question until recently, and the vague answer they gave, wisely, I might add, was “mitigation of symptoms.”  

Either way, the problem will be designing a prospective randomized, double-blind, placebo-controlled trial to prove any of these things. I take that back: The most significant issue will be getting an institutional review board (IRB) to approve a study of this type. That is unless You can provide a reason to think these drugs would have an essential effect.

And, of course, that would first require you to decide what the effect is going to be and the magnitude of that effect. And you will justify that with a reason to think there would be that effect based on the mechanism of action. So, since there is no mechanism of action of these drugs known to affect viral illnesses, you will likely not get past this step in your desire to prove that these drugs should be taken for COVID-19.

Of note, this is an ethical problem. You want to put humans at risk of dying from a disease to test a therapy for which there is no known mechanism for which you would expect an effect, and you cannot define the effect you expect. There is no way this gets IRB approval. Remember, there is a vaccine that was adequately tested and proven to reduce the chances of severe COVID-19 by 95%.

So, you would have to prove that these drugs would not only equal that but beat it so severely that the existence of a proven preventative therapy wouldn’t matter. Again, there is no way you will get this past IRB approval, considering the facts in the case.

Problems Conducting “The Science”

But you are an influential scientist and can conduct this study with IRB approval to prove that these drugs will 100% prevent you from dying of COVID-19. Let’s design this study.

It will be a prospective, placebo-controlled, randomized, double-blind study. Let’s go through each of these terms.

Prospective: This means you determine what you will accept as the interpretation of the data before you collect it. Doing so prevents you from collecting the data, analyzing it, and subsequently hypothesizing something completely different. This practice is considered invalid in science. Many call these practices p hacking or HARKing. (Hypothesis after results are known.) Valid scientists will even preregister their studies to prove they did not commit this crime against science.

Placebo-controlled: In this case, you would divide your randomized population into two groups. One group will get the study drug, and the other will get the placebo. That way, once the study is complete, you will know if the study drug had an effect, i.e., a detectable benefit compared to the placebo. If the results are the same in each group, then the study drug had no effect. If there is a difference between the two groups, you must ensure this wasn’t just due to chance.

In other words, you want the results to reach “statistical significance.”

Apples to Apples
You must compare apples to apples when testing treatments for COVID-19.

Randomized: This refers to ensuring that one subset of your study population does not overrepresent any factor, such as age, gender, general health, etc. You want to compare apples to apples between your study and placebo groups. For example, if you wanted to determine the average height of Americans, you wouldn’t just measure the Americans on the basketball court of an NBA game. Your results would be skewed in this situation. It’s 6 feet, 6.5 inches in this example.

Your results are skewed toward taller people. This is an error in randomization. When studying our drugs, we want to make sure one group isn’t skewed toward younger, healthier people. So, with proper randomization, both groups would have equal amounts of younger, healthier people, equal men and women, for example.  

Double-blind: This means that neither the subject nor the scientist knows beforehand whether any particular subject received the study drug or the placebo. This practice prevents the introduction of bias. For example, if you knew you got the drug and hoped it worked, you might attribute more effect to the drug than what you would have if you didn’t know if you got it or knew you got the placebo. It would also prevent the scientist from splitting one way or the other when attributing the drug’s effects.

The Math of Bad Science

OK, so we know the study type we are going to do. The most challenging part of pulling this off, even for this influential scientist who got through IRB, will be determining the population size. In other words, how many test subjects do I need to show a difference between the study drugs and placebo?

What effect do we want to study, and how pronounced of an effect do we want? Let’s say we want 100% survival of COVID-19 as the effect. Now we know what we need. We know that COVID-19 kills 1.5% of the people who get it. So, these drugs will save that 1.5%. Of course, we understand that if they get COVID-19, each study participant has a 98.5% chance of surviving without treatment. And as we know, survivors were anywhere from asymptomatic to intubated and on mechanical ventilators in the ICU but survived. It doesn’t imply that they were completely back to normal, just that they didn’t die.

Also, it’s essential to know that in most clinical science realms, you want the likelihood that the results were due to chance (i.e., the results of your study are not real) to be less than 5%—in other words, less than 1 in 20. This is typically expressed as the p-value of the results. Most clinical scientists accept a p-value of <.05, indicating that the study population was adequate to prevent randomly getting the wrong results more than once in 20 times.

Feasibility of the Study

Knowing this, we want to know how many study subjects are needed to demonstrate a 1.5% improvement in survival for a disease that 98.5% of all people survive. We also want less than a 1 in 20 chance of getting wrong results (results due to chance).

Now, keep in mind: I am not a professional statistician, so if you are and you have anything to add, please contact me!

To meet the above requirements, the study size must be 4,547,575 people. I did this calculation using chatGPT. I had always spitballed it at 5 million, but this is a more accurate calculation. Either way, it is probably not a feasible study to conduct.

So, this study is not feasible, so let’s change the desired effect that we want. Let’s say instead of a 100% improvement in survival, you want only a 1% improvement in survival. In this case, you need over 10 million subjects. Even more, it is not feasible.

Science 1, Snake Oil 0

If you want to “mitigate symptoms” or shorten the duration of illness, you will probably find it impossible to design a study to prove this. What would you mean by “shorten,” and how would you prove it? I suggest that there is no way to do that scientifically.

In summary, even if you were this influential scientist, you could not prove any benefit of these drugs. There is no reason to think they will work and no safe or practical way to demonstrate a benefit. You can continue believing that the drug companies and doctors are in cahoots, but hopefully, this has given you something to consider.

Copyright © 2024 William E. Franklin, DO, MBA