This is clear up confusion from last night (and add some post publication peer review) to this paper in @lancetmicrobe thelancet.com/journals/lanmi… 1/n
The headline finding of ‘far from herd immunity’ (also emphasized by this piece) stands up. But once you get into the discussion, things turn weird. thelancet.com/journals/lanmi… 2/n
First “Among 452 returnees evacuated from Hubei province in March, 2020, the seroprevalence was 4%, with the majority (88%) being confirmed by microneutralisation assay.” 3/n
But earlier in the results “665 serum samples were collected from 452 returnees from Hubei province (of which 364 [80·5%] were from Wuhan)”
So it’s not a random sample of Hubei (or likely even Wuhan but let’s not get distracted) 4/n
So it’s not a random sample of Hubei (or likely even Wuhan but let’s not get distracted) 4/n
Then later, “Of the 17 seropositive returnees…16 individuals who were seropositive had been staying in Wuhan”
So again it’s mostly Wuhan, not Hubei. And we know that there were few deaths outside the city. 5/n
So again it’s mostly Wuhan, not Hubei. And we know that there were few deaths outside the city. 5/n
This matters, because later in the discussion these data end up being used to estimate the total number of cases and infection fatality rate for Hubei! It comes in at 0.16% - much much lower than estimated elsewhere. @nicolamlow spotted this 6/n
Then as pointed out by @viciykevin and @chenchenzh. If we limit ourselves only to the returnees from Wuhan (with caveat that they’re not necessarily a representative sample) 16/364 are seropositive, or 4.4%. 4.4% of Wuhan (popn 11 million) is nearly 500,000. 7/n
Now there were about 4000 deaths in Wuhan (NYT). So what would that imply for the IFR in Wuhan? It would be about 0.8% That is a number that *is* consistent with the data from elsewhere (HT @unsorsodicorda for the link) 8/n
Yet again, a story that seems to indicate that #COVID19 is actually mild turns out to say nothing of the kind once you dig into it (can a note to that effect be added @onisillos?) Thanks to @nicolamlow @chenchenzh @viciykevin and @sudha_lakshmi for making me come back to it 9/end
