Wednesday, March 11, 2020

Covid-19 update


A very quick update on my last post.  The above shows the same graph as before but with more data.  Korea continues to get their infection under control, Italy is really struggling to do the same (and reports of conditions in the health care system there are getting really gruesome).  Hopefully the Italian curve will start to bend now that the country is in lockdown.

My slapdash extrapolation of the US case curve (the dashed line) has performed just about flawlessly - the US is on track to have a higher number of cases (relative to population) than China within a few days (well, it's not clear how meaningful the comparison is, since it's not clear that either country's numbers are very accurate).  There is no sign of any meaningful containment in the US so far (bending of the curve), and I do not expect any in the next week as the US is still in the process of solving its testing problems, and has not instituted mandatory social distancing at any scale.

The UK has a similar trajectory to the US but a few days ahead (no doubt due to greater proximity to Italy).  Japan is doing much better than most places - possibly because it did things like closing all schools fairly early, but still has not fully contained its epidemic.

This second graph shows the crude death rate - total deaths as a fraction of total known cases.  Of course, neither number might be exact - some cases never got diagnosed, or are not diagnosed yet, some people died misdiagnosed as some other kind of pneumonia.  These are just the best numbers we have.


Korea continues to have a very low death rate, reflecting the benefit of a very well organized and extensive testing regime.  The US death rate was for a while amongst the worst (representing the particularly poor testing regime in the US).  However, Italy now has the worst ratio - presumably reflecting the fact that the health care system is overwhelmed and many patients are not receiving good care.

3 comments:

  1. Rather than crude death rate, I have been plotting death rate among resolved cases: deaths/(deaths + recovered). It shows a pretty consistent pattern by nation, with an initial peak of high mortality tailing off to an asymptote that globally is around 6% right now. Italy looks...bad; they've started a new peak rather than tailing off. South Korea looks effective.

    ReplyDelete
  2. The number of death per infected person seems very high in Italy
    But once you get the data in a table sorted by age, the death rate is in line with what is seen in other countries.


    Based on the Repubblica Article:https://www.repubblica.it/cronaca/2020/03/10/news/coronavirus_l_epidemia_in_italia_non_e_arrivata_dalla_cina-250882478/?refresh_ce

    •No death below 40 years old.
    •98.9% of deaths above 60
    •All death ratio below the Chinese Numbers
    •The Article also says that 90% of patient in intensive care are above 50 (46% 50 to 70 and 46% 70 and above)



    ReplyDelete
  3. Hi Stuart,
    You might want to check out how this guy estimates mortality rates with imperfect data
    https://medium.com/@tomaspueyo/coronavirus-act-today-or-people-will-die-f4d3d9cd99ca

    ReplyDelete