Editor's note: Only about 15 million people so far have gotten the new bivalent Covid booster that's specifically formulated against the Omega variants that are currently circulating. I got it a few days ago, Walgreen's was offering appointments the same day! I got the flu shot at the same time, both absolutely free. Do it!
So, picking up where we left off, to determine whether an intervention is beneficial, harmful or basically useless you ordinarily need a comparison group. There are actually exceptions. You don't need a randomized controlled trial for parachutes when jumping out of airplanes because you already know what happens just about every time you jump out of an airplane without one. But few medical conditions are like that. Even ordinarily progressive conditions, such as cancer or multiple sclerosis, can spontaneously remit, at least temporarily, and anyway we might like to slow them down even if we can't cure them. But infectious diseases in particular are a battle between the pathogen and the immune system, and the immune system usually wins eventually. So you have to measure the difference between people who get eye of newt and toe of frog, and those who don't.
Unfortunately, it's not that simple. I can round up a bunch of people who have had Creeping Crud and compare the outcomes of those who have taken Eontof and those who haven't, but that's unlikely to prove anything. For starters, the two groups may have been different in all sorts of ways to begin with. Maybe the people who got Eontof were more likely to be insured or otherwise relatively privileged, therefore probably healthier to start with. On the other hand maybe they got it because they were sicker, and therefore their doctor was more likely write a prescription. The latter is called confounding by indication, and it can make an intervention look bad when it's actually helpful.
There are many more problems with this. If I'm asking about the past, it really isn't possible to measure accurately or reliably how sick they were, or how long it took them to recover. As a matter of fact, if I want the intervention to work -- and I probably do -- then consciously or unconsciously I could skew my assessment of original severity and subsequent recovery, even if I'm following both groups from the beginning. This is called ascertainment bias and believe me, it is a very powerful effect. And if the sample is too small, any differences between the two groups could be due purely to chance.
For these and other reasons, it is very commonly the case that positive results in small, preliminary trials aren't confirmed by more rigorous studies. This is what happened with ivermectin and hydroxychloroquine. Some doctors gave them to some patients and had the impression that those patients did better than expected. But people with even minimal expertise in this field knew immediately that the evidence was weak or useless. It is also important to note that there was no biological plausibility for these claims. These are antiparasitic agents that are toxic to metazoans. There is absolutely no plausible argument as to why they would be effective against a virus.
Unfortunately, because claims about these agents became politically controversial, NIH felt it had to fund rigorous clinical trials even though in normal circumstances there would not have been any justification for spending all that money. The results were as expected. Both of them are completely useless against Covid-19 in all stages of the disease and all populations, and they can have serious adverse effects. So, waste of money, harm to population, and yet we still have politicians, their followers, and quack doctors promoting these harmful practices.
Next time, I'll explain the requirements for a properly designed clinical trial, the limitations thereof, and how we can augment well crafted randomized controlled trials with other kinds of evidence to get the most compelling information.
No comments:
Post a Comment