4 ms·

Here's a source for the reinfection rate. I've seen a few studies that all came to the < 1% conclusion: https://www.medrxiv.org/content/10.1101/2021.01.15.2124

by Exmoor 6y ago

Here's a source for the reinfection rate. I've seen a few studies that all came to the < 1% conclusion: https://www.medrxiv.org/content/10.1101/2021.01.15.21249731v1 https://www.medrxiv.org/content/10.1101/2021.01.15.21249731v... The Manaus data is interesting. The basis for assuming there's reinfection happening seems to be a seroprevalence survey done over the summer which estimated 76% of the population had antibodies to the virus. The accuracy of that survey has been called into question for a variety of reasons. The survey was done through blood donors and apparently they incentivized blood donation by offering people the results of their antibodies test as part of donation. Since testing capacity during peak infection was limited, this would give people who thought they might have had the virus (and likely did) incentive to donate blood and made the sample not nearly as random as it was presented to be.

timr 6y ago
> The basis for assuming there's reinfection happening seems to be a seroprevalence survey done over the summer which estimated 76% of the population had antibodies to the virus. The accuracy of that survey has been called into question for a variety of reasons. This is exactly correct. Extrapolating from the Manaus seroprevalence paper is scientifically tenuous, at best. They say so in the Lancet commentary cited in the CIDRAP piece cited by OP: "In the Lancet report, experts said four factors may be at play, some possibly related. First, scientists might have overestimated the attack rate for the first surge, and infections might have been below the herd immunity threshold" > The survey was done through blood donors and apparently they incentivized blood donation by offering people the results of their antibodies test as part of donation. Since testing capacity during peak infection was limited, this would give people who thought they might have had the virus (and likely did) incentive to donate blood and made the sample not nearly as random as it was presented to be. Not just that, but if you read the paper carefully, you'll see that they've applied a number of fairly arbitrary "corrections" to the data, which pull the seroprevalence from an observed 20% range up to the cited "76%" number (Figure 2A in this link): https://science.sciencemag.org/content/371/6526/288 https://science.sciencemag.org/content/371/6526/288 While any of these adjustments might be individually merited, when you see a >300% aggregate "adjustment" to the raw data, it should make you very skeptical about any claims made with that data. Particularly when the raw data is below the likely herd-immunity threshold, and has been "corrected" to be above it. More simply: the most likely conclusion is that the original paper was wrong. This is perfect example of bad science being propagated by journal title and headline. Whomever wrote that (IMHO, terrible, editorialized) CIDRAP piece didn't bother to read the original data, and just took the first paper's claim at face value. Journalists have been (and continue to be) insufficiently skeptical when it comes to Covid science -- except when it suits the narrative they're trying to create. If a paper supports their desired narrative, any half-baked, highly implausible claim is treated as worth serious consideration. If a paper conflicts with their claim, no counterargument is too implausible to be considered.

by Exmoor 6y ago

timr 6y ago

> The basis for assuming there's reinfection happening seems to be a seroprevalence survey done over the summer which estimated 76% of the population had antibodies to the virus. The accuracy of that survey has been called into question for a variety of reasons. This is exactly correct. Extrapolating from the Manaus seroprevalence paper is scientifically tenuous, at best. They say so in the Lancet commentary cited in the CIDRAP piece cited by OP: "In the Lancet report, experts said four factors may be at play, some possibly related. First, scientists might have overestimated the attack rate for the first surge, and infections might have been below the herd immunity threshold" > The survey was done through blood donors and apparently they incentivized blood donation by offering people the results of their antibodies test as part of donation. Since testing capacity during peak infection was limited, this would give people who thought they might have had the virus (and likely did) incentive to donate blood and made the sample not nearly as random as it was presented to be. Not just that, but if you read the paper carefully, you'll see that they've applied a number of fairly arbitrary "corrections" to the data, which pull the seroprevalence from an observed 20% range up to the cited "76%" number (Figure 2A in this link): https://science.sciencemag.org/content/371/6526/288 https://science.sciencemag.org/content/371/6526/288 While any of these adjustments might be individually merited, when you see a >300% aggregate "adjustment" to the raw data, it should make you very skeptical about any claims made with that data. Particularly when the raw data is below the likely herd-immunity threshold, and has been "corrected" to be above it. More simply: the most likely conclusion is that the original paper was wrong. This is perfect example of bad science being propagated by journal title and headline. Whomever wrote that (IMHO, terrible, editorialized) CIDRAP piece didn't bother to read the original data, and just took the first paper's claim at face value. Journalists have been (and continue to be) insufficiently skeptical when it comes to Covid science -- except when it suits the narrative they're trying to create. If a paper supports their desired narrative, any half-baked, highly implausible claim is treated as worth serious consideration. If a paper conflicts with their claim, no counterargument is too implausible to be considered.