A widely shared study on ivermectin might be wrong

Is the anti-parasitic drug ivermectin effective for preventing or treating COVID-19, as some commentators claim? I’m afraid I’ve haven’t been following the debate closely at all. So the short answer is: I don’t know. (The FDA, incidentally, says you shouldn’t take it for COVID.) However, I came across one study on ivermectin – which has been widely shared on social media – that I didn’t find very convincing. Here I’d like to explain why.

The study, by Hisaya Tanioka and colleagues, is titled ‘Why COVID-19 is not so spread in Africa: How does Ivermectin affect it?’ (I came across the study via this tweet, which as of my writing, has more than 16,000 likes.) What did the authors find? They begin by noting that Sub-Saharan Africa has been less affected by COVID-19 than other world regions. And they hypothesise that widespread administration of ivermectin (to treat the parasite disease onchocerciasis) might provide an explanation.

According to the authors, 99% of onchocerciasis infections have occurred in 31 African countries – basically those located in the central, equatorial region of the continent (there’s a map in the viral tweet above). For their analysis, the authors simply divided African countries into two groups: the 31 with “community directed treatment with ivermectin”, and the other 22 non-endemic countries. They find that the average COVID-19 death rate (taken from the WHO’s coronavirus disease dashboard in January) is significantly lower in the ivermectin group.

Why am I not convinced? One reason is that, due to lack of testing and healthcare infrastructure, we can’t trust the official COVID-19 death numbers for most countries in Sub-Saharan Africa. But shouldn’t this problem have affected both of the two groups equally? No.

To begin with, all the countries in North Africa are in the non-ivermectin group, and these countries have higher levels of economic development. As a result, they may have reported more accurate (and hence higher) COVID-19 death rates. Looking at the authors’ tables, Tunisia and Morocco both reported quite high numbers (as did Libya; though that country isn’t very economically developed these days). On the other hand, Algeria and Egypt reported quite low numbers, most likely due to delays or other kinds of misreporting.

What’s more, South Africa – generally considered the most economically advanced country in Sub-Saharan Africa – is also in the non-ivermectin group. And it reported the highest COVID-19 death rate of any country in the sample.

According to the authors’ tables, the average death rate in the ivermectin group is 14.4 per million (compared to 121.9 per million in the non-ivermectin group). If we look at the other countries in the non-ivermectin group (i.e., not South Africa and not the ones in North Africa), we find that many of them reported numbers just as low as those in the ivermectin group. For example: Botswana’s figure is 30.9; Eritrea’s figure is 1.6; Madagascar’s figure is 9.6; Somalia’s figure is 8.2; and Zambia’s figure is 29.2. It seems plausible that underreporting explains both these low death rates and the low death rates in the ivermectin group.

Can underreporting really explain a disparity as large as 14.4 versus 121.9? In principle, yes. For example, the official death toll in Nicaragua, as of this writing, is only 199. Yet the number of excess deaths in that country is at least 7,000. Even South Africa’s numbers are afflicted by underreporting. For example, the official death toll in that country is 81,595. Yet there have been at least 150,000 excess deaths.

A second reason why I’m not convinced is that the authors didn’t control for average age. (Though they do mention the relative youth of Africa’s population in their Discussion.) Even in the absence of underreporting, you’d expect more COVID-19 deaths in the non-ivermectin group because those countries are slightly older. In fact, there seems to be strong inverse correlation between average age and prevalence of onchocerciasis in Africa. The continent’s oldest countries are: Tunisia, Morocco, Libya, Algeria and South Africa. And all of these are unaffected by onchocerciasis.

I don’t know whether ivermectin is effective for preventing or treating COVID-19. But the study by Tanioka and colleagues doesn’t provide strong evidence that it is. (Note: the observation that COVID-19 death rates are lower in African countries that administer ivermectin was even mentioned at a press conference by the chairman of the Tokyo Medical Association.) While Tanioka and colleagues’ findings are consistent with their hypothesis, it seems more likely that they’re explained by some combination of underreporting and differences in average age.

Image: Joachim Huber, Okavango Delta, Botswana, 2007

The Daily Sceptic

I’ve written three more posts since last time. The first examines whether population density explains variation in COVID-19 death rates across Europe. The second summarises a recent study finding that natural immunity protects better against infection than the Pfizer vaccine. The third summarises a new study finding that the switch to remote learning caused a large decline in US school enrolment.

Thanks for reading. If you found this newsletter useful, please share it with your friends. And please consider subscribing if you haven’t done so already.