Reader Martim sent me a link to this article in The Economist, which is pretty much paywalled but includes a paragraph and two graphs you can see. Click on the screenshot below to see what’s viewable:
And the figures, which buttress the title’s assertion:
Just using Fisher’s Exact Test* on the directionality above (dots to left or right of line), the difference is not statistically significant, but it is in a suggestive direction. We need more data to see if this disparity is statistically significant and thus “real” (i.e. not produced by chance under an equal frequency null hypothesis).
Now this is basically all of the article I’m allowed to read, but it hints at why some studies may show a positive effect of Ivermectin on patients infected with Covid-19. To put it simply, worms are a “comorbidity”, that might be eliminated with ivermectin. If having worms makes raises your chances of dying or hospitalization from having the virus, then taking ivermectin could help save your life not by affecting the virus, but by ramping up your immune system after the worms have gone. Ivermectin would not, then, be of any use in treating patients unless they’re known to be affected with roundworms. (Ivermectin helps get rid of roundworms that cause intestinal strongyloidiasis and onchocerciasis, conditions that weaken your immune system.)
The graphs above suggest what one might predict: ivermectin would be more efficacious against Covid-19 in countries with a higher prevalence of worms, specifically the kind of worms killed by ivermectin. And that’s what the graphs show.
Of course, some of the studies above, both positive and negative, may already have been discredited by subsequent inspection (I haven’t checked), but I’m surprised that nobody has suggested this explanation before. (If they have, I haven’t seen it.)
In a month or two we should know the results of the properly conducted Oxford study on the effects of ivermectin on Covid-19 mortality. I’m almost positive that if the drug does have a positive effect on the disease, it will be minor—certainly much less than that of vaccination or the new Pfizer antiviral pill. But we will wait patiently. I tried to bet my doctor on 3:1 odds (if I won, I’d get $10, while if he won, and Ivermectin had a big positive effect, he’d get $30) that ivermectin wouldn’t show a greater preventive or curative effect than jabs and the new treatments, but he rejected that as a “sucker bet”!
UPDATE: I’ve just managed to subscribe for free and so have seen the rest of the article; it appears that some have considered worms as a comorbidity. A quote (emphasis is mine):
Yet ivermectin’s advocates insist that there is solid science demonstrating the drug’s efficacy. One well-documented website lists and links to 65 different papers on the subject, many of which, on the surface, seem to support this claim. Could this many studies all be wrong? Recent analysis by Avi Bitterman, a dermatologist in New York, and Scott Alexander, a prominent blogger, suggests that the answer is nuanced. Ivermectin probably does help one subset of covid-19 patients: those who are also infected by the worms it was designed to fight.
Wading through the papers whose methodologies appeared sound, Dr Bitterman noticed that the studies that looked best for ivermectin tended to cluster in regions with high rates of infections by strongyloides, a parasitic worm. Common in much of Africa, Asia and Latin America, strongyloides can cause, among other things, diarrhoea, fatigue and weight loss. However, they only pose a graver threat if their numbers grow out of control. Such “hyper-infection”, which is often fatal, becomes far more likely if a patient is receiving corticosteroids, which both suppress the immune system and appear to make female worms more fertile. And dexamethasone, a corticosteroid, is now a standard treatment for severe covid-19, because it prevents the immune system from going into overdrive and attacking the body’s own cells.
Building on observations by David Boulware, a professor of medicine at the University of Minnesota, Dr Bitterman concluded that strongyloides may account for the conflicting results of studies about the effectiveness of ivermectin as a treatment for covid-19. In trials conducted in countries where the parasites are common, many people could have both covid-19 and strongyloides infections. Covid-19 might already have weakened their bodies’ defences against the worms; treating the coronavirus with corticosteroids would let the parasites run wild.
In the groups who received ivermectin during trials, the drug would keep strongyloides in check. But patients in control groups would be left at the worms’ mercy. This would make it look as if ivermectin were preventing deaths caused solely by covid-19, when in fact it was preventing those caused by the parasites or by a combination of the two infections. This mechanism would explain why most studies conducted in places where strongyloides are rare showed no benefit from taking ivermectin. “Ivermectin doesn’t treat covid,” Dr Bitterman wrote. “It treats parasites (shocker) that kill people when they get steroids that treat covid.” He concluded that “taking strongyloides endemic populations, putting them into a control group with corticosteroids is a death sentence”.
In July 2020 a group of doctors argued in the Journal of the American Medical Association that it was “reasonable to consider presumptive treatment with ivermectin for moderate- to high-risk patients not previously tested or treated for strongyloides”, and said that the risk of infection by the worms in covid-19 patients should be “based on factors such as country of origin and long-term residence”. The World Health Organisation also recommends ivermectin in this context. However, most people in rich Western countries like America—where demand for ivermectin, driven by advocates on social media, is so high that some people have resorted to taking the equine version of the drug—do not fit this description. At least when treating patients who have never been to countries with widespread strongyloides, the evidence suggests that mainstream doctors in such places are right to avoid prescribing ivermectin.
None of this, of course, suggests that vaccinations are less efficacious than ivermectin in preventing death from the virus alone, much less, as Bret Weinstein and Heather Heying suggest, you’re better off not getting vaccinated than getting vaccinated. Just get tested for worms if you get covid and live in a roundworm-infested part of the world!
*Note that Fisher has been canceled.
Chet Brinestein and Weather Weying are new bloggers on the scene, who are challenging the norms of prescription of ivermectin for treatment of parasitic worms.
Their solution : coronavirus vaccination.
[ 100% satire above! Not real! I made it up for s laugh … hopefully!…]
Excellent! File under Conclusions that were jumped to, and Correlation is not necessarily causation. A future textbook example.
Interesting, and in retrospect not at all surprising.
Shamefully, Fisher has even been cancelled by the Royal Statistical Society.
I now make sure to mention him whenever I teach Experimental Design – I give a very brief introduction to it on about 10 different courses now and whenever I ask for a show of hands find that everybody has heard of control groups but very few people have heard of blocking or factorial designs, let alone Fisher.
This really shouldn’t surprise anyone, frankly. It’s a bit like saying that diabetic patients (even previously undiagnosed ones) with Covid-19 who get treated “aggressively” for their diabetes while on steroids (which worsen blood sugar levels) for Covid-19 do better than those who receive no diabetes treatment. But diabetes medications don’t in and of themselves treat Covid-19 (as far as anyone knows).
And people who have touched hot stoves have a higher incidence of burns than people who have not come into contact with anything hot. And people who are in moving vehicles are more likely to be involved in vehicular accidents than people who are sitting on a sofa at home.
Don’t get me wrong, it is interesting and important that this article was written, published, and should be spread. But the fact that it needed to be is disheartening.
The elegance in pointing it out is that strongyloides is unknown to Westerners unless they have studied parasitology. Because of it’s peculiar life cycle it is the only roundworm that can quickly kill you if your immune system is suppressed. Even though I saw a case of strongyloides hyper-infection when I was training, in a patient being treated for kidney transplant rejection —he died—I didn’t make the connection until I read about it recently. (I don’t think ivermectin was known back then to be useful for strongyloides, just for river blindness which it has almost wiped out.). The other point is that ivermectin doesn’t look very impressive anyway, so retired people like me don’t put a lot of thought into trying to figure out why it might work in some people.
In respect of what another commenter said about correlation not causation, I’d say, “Not so fast.” Yes it’s true that a regression of two variables picked at random that yields a correlation at an alpha rejection of 0.05 will be falsely positive 1 time in 20 even if there is in truth no correlation at all. But if a correlation is found when testing an a priori hypothesis and is statistically robust, e.g., from being replicated in a new dataset, then very likely there is in truth some kind of causal link driving or contributing to the correlation. The rub is that the cause isn’t always the first thing that pops into your mind, or even the one you had carefully considered before you started analyzing the data in the first place, or even in the direction you were hoping to find. The last is a particular problem in vicious circles and feedback loops. Now of course to test if A indeed does cause B you need to do a carefully designed controlled experiment.
The ivermectin business even illustrates how lack of correlation can hide true causation. Lumping all the non-fraudulent studies together shows no correlation (probably.). Yet if you control for background prevalence of any worms > 8.1% a weak correlation emerges, which becomes plausibly causal. Notice how you only think to control for worms if that causal hypothesis occurs to you. Imagine if you knew for every individual patient whether he had strongyloides specifically, not just that he lived in a country where any old worms are common, before starting treatment. If you then found very strong, statistically highly significant correlation between ivermectin and outcome in patients with strongyloides given steroids, then you would probably just go ahead and give it. No RCT needed because ivermectin is already known to be good for worms.
It appears that Strongyloides infection reduces your immune system’s efficacy in general, so even without steroids Ivermectin may benefit a worm infested patient. Vaccines may also be less effective in these patients.
Yes, and with hindsight it is obvious.
Definitely some “amusing” aspects to Covid and ivermectin. Even if some of that may qualify as gallows humour.
But some here might be interested in the Wikipedia article on ivermectin and the Antiviral Research article it links to.
The former notes that “ivermectin has antiviral effects against several distinct positive-sense single-strand RNA viruses, including SARS-CoV-2”, although the “half maximal inhibitory concentration” (IC50) is some 9 to 21 times the level that was used in a computer model.
But the model itself is based on a level – 600 μg/kg – that is 3 times “the highest regulatory approved dose of ivermectin”:
“Even with most generous assumptions for clinical translation, the in vitro IC50 is > 9-fold and >21-fold higher than the day 3 plasma and lung tissue simulated Cmax respectively, following a high dose ivermectin regimen of 600 μg/kg dose daily for 3 days. (Smit et al., 2019) This dose scenario, which ignores consistent exposure, exceeds the highest regulatory approved dose of ivermectin, being a 200 μg/kg single dose for the treatment of Strongyloidiasis.”
https://en.wikipedia.org/wiki/Ivermectin#COVID-19
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7172803/
May well be some benefit to taking ivermectin – apart from potential deworming – but it seems unlikely to be a very significant one.