Posts

Comments

Comment by EngineerDude on Iceland's COVID-19 random sampling results: C19 similar to Influenza · 2020-03-31T19:48:17.739Z · LW · GW

Good analysis, close to my calculations of 0.04% - 0.17%(from Iceland data) averaging around 0.1% which is almost perfectly in line with seasonal flu. I actually suspect that coronavirus will in the long run(after a vaccine and treatments/medication) actually be a lot closer to swine flu(0.03%), implying that seasonal flu might be quite a bit worse than Cov2 NOT the other way round as reported thus far by the media. The only mitigating factor I can think of is the shorter incubation period for swine flu which might mean that Cov2 has a little more time to get around the population asymptomatically thus increasing it's transmissibility or R_0.

I also did a crude but interesting analysis based on the symptomatic testing conducted in the States. It turns out that about 16% of all symptomatic people tested have the virus in the US. If we consider the widely accepted figure of ~3 flu like illnesses for every adult per year(more for kids) we can estimate that in the US alone some 990 million flu like infections occur every year. Flu season and similarly "cold" season lasts for about 16 weeks from mid December to mid April where thereafter the regular summer/late spring/early fall rate of the flu virus drops to less than half(0.4 to be exact). Using this information we can work out that for a period of two weeks during flu season some roughly 52 million Americans have some sort of flu-like virus(a lot of whom show no symptoms). Knowing that Cov2 is about 19% as common now as would be other viruses any other year this leaves us with ~9.925 million infections. As of today there have been 3,402 deaths and 3,893 that are critical. Given that the mortality rate of all critical/serious cases is around 27% we can assume that the total death tally will be around 4453 assuming no more spreading occurred as of about a week ago( a big assumption I know, but there is also a lag in testing percentages so I think this is a safe assumption and works for our example). You might ask why I am using a two week interval and well I am assuming that most cases have occurred within that 2 week interval(I did say this was a crude analysis). In any case, if we crunch the numbers this works out to about 0.045% IFR, which is actually surprisingly similar to Iceland and within it's range of 0.04% - 0.17%.


Something interesting to note is that the deCODE tests are done on asymptomatic cases only(according to Kári Stefánsson). This wildly skews the results since as much as 50% of cases are asymptomatic or so mildly symptomatic as to go mostly unnoticed. Which means we could safely double the rate of infection his data reveals...?