Read on Twitter
Results from the first high-quality antibody study for Covid-19 are out, and confirm an IFR estimate of at least 1.1% (confidence interval: 0.5% to 2.5%):
https://www.hug-ge.ch/medias/communique-presse/seroprevalence-covid-19-premiere-estimation
Sadly, this really does show that any optimism that this disease was milder than thought is wrong
https://www.hug-ge.ch/medias/communique-presse/seroprevalence-covid-19-premiere-estimation
Sadly, this really does show that any optimism that this disease was milder than thought is wrong
3.5% of the population had had it by April 6th-10th and 5.5% by April 14th-17th. This compares to 190 deaths in a population of 499k to April 20th.
Given time from symptoms to death is 2-6 weeks, the first number is likely to be more relevant, meaning IFR is 1.1%
Given time from symptoms to death is 2-6 weeks, the first number is likely to be more relevant, meaning IFR is 1.1%
In reality, if the Korean experience is similar, this should rise over coming weeks to closer to the 1.6% implied by the following age-related fatality rates:
<49: 0.1% or lower
50-59: 0.5%
60-69: 1.6%
70-79: 5.2%
80+: 18.6%
<49: 0.1% or lower
50-59: 0.5%
60-69: 1.6%
70-79: 5.2%
80+: 18.6%
This is higher than the estimates used by @neil_ferguson and team, but within their confidence interval, because mortality in the South Korean population was higher than that in China, likely due to better data collection
This would reflects a 1.2% to 1.6% infection fatality rate in the UK population, a 1.0-1.4% IFR in the US population, and means that data from Italy coming out over the weekend is likely to be terrible, with seroprevalence at 10-20% and an IFR tragically higher than 2%
For avoidance of doubt: there’s no possible way that you can now sustain the narrative this is only as bad as the flu
If you want to know how you could have already been sure about this over a month ago, I wrote it up here: https://medium.com/@BenGardiner/covid-19-calculating-the-death-rate-isnt-actually-that-hard-7a61ed5715e8
