[BostonCard]

Also, to be fair to him, he cites the following (news) article in Science, which makes the same mistake about serology data and mortality data:

https://www.sciencemag.org/news/2020/08/...xplain-why

Scientists who surveyed about 10,000 people in the northeastern cities of Nampula and Pemba in Mozambique found antibodies to SARS-CoV-2 in 3% to 10% of participants, depending on their occupation; market vendors had the highest rates, followed by health workers. Yet in Nampula, a city of approximately 750,000, a mere 300 infections had been confirmed at the time. Mozambique only has 16 confirmed COVID-19 deaths. Yap Boum, a microbiologist and epidemiologist with Epicentre Africa, the research and training arm of Doctors Without Borders, says he found a high prevalence of SARS-CoV-2 antibodies in people from Cameroon as well, a result that remains unpublished.

So what explains the brobdingnagian gap between antibody data on the one hand and the official case and death counts on the other? Part of the reason may be that Africa misses many more cases than other parts of the world because it has far less testing capacity. Kenya tests about one in every 10,000 inhabitants daily for active SARS-CoV-2 infections, one-tenth of the rate in Spain or Canada. Nigeria, the continent’s most populous nation, tests one out of every 50,000 people per day. Even many people who die from COVID-19 may not get a proper diagnosis.

But in that case, you would still expect an overall rise in mortality, which Kenya has not seen, says pathologist Anne Barasa of the University of Nairobi who did not participate in the country’s coronavirus antibody study. (In South Africa, by contrast, the number of excess natural deaths reported between 6 May and 28 July exceeded its official COVID-19 death toll by a factor of four to one.) Uyoga cautions that the pandemic has hamstrung Kenya’s mortality surveillance system, however, as fieldworkers have been unable to move around.

But again, that caveat ("the pandemic has hamstrung Kenya’s mortality surveillance system, however, as fieldworkers have been unable to move around.") is pretty important.

By the way, I didn't mean to be harsh on ecological analyses earlier.  Correlations from ecological analyses are very important to generate hypotheses.  Many (in my experience) most have innocent explanations and don't pan out, but every once in a while an important insight will come from a hypothesis that arose from an ecological correlation.  That being said, a skeptical eye is warranted, especially before invoking "just so stories" as explanations.

BC

[BostonCard]

Also, to be fair to him, he cites the following (news) article in Science, which makes the same mistake about serology data and mortality data:

https://www.sciencemag.org/news/2020/08/...xplain-why

Scientists who surveyed about 10,000 people in the northeastern cities of Nampula and Pemba in Mozambique found antibodies to SARS-CoV-2 in 3% to 10% of participants, depending on their occupation; market vendors had the highest rates, followed by health workers. Yet in Nampula, a city of approximately 750,000, a mere 300 infections had been confirmed at the time. Mozambique only has 16 confirmed COVID-19 deaths. Yap Boum, a microbiologist and epidemiologist with Epicentre Africa, the research and training arm of Doctors Without Borders, says he found a high prevalence of SARS-CoV-2 antibodies in people from Cameroon as well, a result that remains unpublished.

So what explains the brobdingnagian gap between antibody data on the one hand and the official case and death counts on the other? Part of the reason may be that Africa misses many more cases than other parts of the world because it has far less testing capacity. Kenya tests about one in every 10,000 inhabitants daily for active SARS-CoV-2 infections, one-tenth of the rate in Spain or Canada. Nigeria, the continent’s most populous nation, tests one out of every 50,000 people per day. Even many people who die from COVID-19 may not get a proper diagnosis.

But in that case, you would still expect an overall rise in mortality, which Kenya has not seen, says pathologist Anne Barasa of the University of Nairobi who did not participate in the country’s coronavirus antibody study. (In South Africa, by contrast, the number of excess natural deaths reported between 6 May and 28 July exceeded its official COVID-19 death toll by a factor of four to one.) Uyoga cautions that the pandemic has hamstrung Kenya’s mortality surveillance system, however, as fieldworkers have been unable to move around.

But again, that caveat ("the pandemic has hamstrung Kenya’s mortality surveillance system, however, as fieldworkers have been unable to move around.") is pretty important.

By the way, I didn't mean to be harsh on ecological analyses earlier.  Correlations from ecological analyses are very important to generate hypotheses.  Many (in my experience) most have innocent explanations and don't pan out, but every once in a while an important insight will come from a hypothesis that arose from an ecological correlation.  That being said, a skeptical eye is warranted, especially before invoking "just so stories" as explanations.

BC