Let me run through a few miscellaneous news items and updates from the COVID19 Crisis.
*** One of the minor (or perhaps not so minor) mysteries of the COVID Crisis Era is just what the story is, who is behind the Worldometers.info website. This has been a concern and question of mine since the very beginning of the epidemic. I wrote about my effort to get to the bottom of it here, all the way back on March 16th. The gist is that the site has a vast amount of Coronavirus data. But it provides close to no information on who is behind the site or owns it or anything more than the most generic information on how or by what criteria it collects its data. Ordinarily that lack of transparency would be totally disqualifying. But it does directly link to sources in all cases and those sources are invariably government or legitimate media sources. (When I say link I mean directly link to the same data on the government website.) I was eventually persuaded to add it to our list of Coronavirus links when I saw that Statista, a well-known data subscriptions service cited their numbers. I later saw many other news organizations doing the same.
Now CNN has this lengthy investigation based on the fact that Prime Minister of Spain has gotten into a major controversy over citing them and it turns out that Johns Hopkins, which runs the canonical global COVID19 data hub, is using them as one of their data sources! Everyone seems to have it was Worldometers citing Johns Hopkins!
The upshot remains a mystery. As I said above, Worldometer is pretty consistently accurate and in the relative few cases of inaccuracy it doesn’t appear to be faked information or un-cited information. It’s questionable decisions about interpretation or categorization, about what tests to count as “tests” – is it just PCR or infection tests or do antibodies tests count too? These are basically methodological questions which more training and best practices might lead you to avoid. They’re not even ‘errors’ precisely.
The other side of this is they collect an almost unbelievably large amount of information – again, all linked – in a very timely fashion. The owners of the site won’t even talk to the press. Reading between the lines it seems to be a boiler room type operation run with the intent of making ad revenue. If it’s not “run by an international team of developers, researchers, and volunteers” as the site claims, how on earth do they so effectively and rapidly pull together all the data? It’s like the ‘uncanny valley’ of data.
*** A new study (pre-peer review) from Hong Kong suggests that wearing masks is highly effective as slowing the spread of COVID. The study used hamsters – not as comical as it sounds – and the numbers were relatively small. But the study found that masking reduced the transmission by respiratory droplets by as much as 75%. In case you’re wondering, no this wasn’t like some cute animal meme site where they created tiny hamster masks. If I’m understanding the study description, they basically placed masks over the cages. And the cages were next to each other. No masks, 66.7% of the healthy hamsters got COVID after a week. That percentage dropped to 16.7% when masks were placed on the sick and healthy cage. It was 33% when the mask was only on the healthy cage. Peer review will be important on a study like this since this is obviously not precisely like a human wearing a facial mask. But it does suggest real efficacy for the basic barrier concept. And there are genetic reasons why hamsters are similar to humans in terms of the respiratory spread of COVID.
*** A new, very large (over 70,000) and apparently well-designed antibodies (serology) study from Spain suggests that about 5% of Spaniards have been infected with COVID. As in the United States, prevalence varies greatly by region, with a high of 14.2% in Soria province and 11.3% in the Madrid region all the way to under 2% in Asturias, the Canaries and the North Africa enclaves of Ceuta and Melilla. Two key takeaways. This study suggests that roughly 90% of cases go undetected in official government tallies and that the true infection mortality rate is 1.1%. We’ll return to this topic later. But the 90% number is notable since it is broadly in line with other serology studies which show a ‘true’ prevalence of infection usually in the 10x to 12x range compared to official tallies.
*** On this line of emerging serology studies another study was done in Indiana. This study was a mix of PCR tests (for active infections) and antibodies (serology) tests. They came up with a total of 2.8% of Indiana residents having a current or past COVID infection. That translated to official infection tests and official tallies missing 10 out of 11 cases in the population. Again, note that this is quite close to the result from the Spanish study. See a discussion of the results here from epidemiologist Natalie E. Dean.
*** Students began returning to schools today in South Korea.