The FT trumpet new research, which, according to the FT, shows that the new #coronavirus may already have infected half the UK population. Which would be great news, suggesting very low death rates.
But probably not. (THREAD) 1/
But probably not. (THREAD) 1/
Via reddit -- reddit.com/r/CoronavirusU… -- here's the paper in question (by Lourenço et al): dropbox.com/s/oxmu2rwsnhi9… 2/
The paper presents results of a standard SIR model of the spread of infectuous disease. The model has two types of input: Empirical data on deaths and assumptions about other parameters. 3/
Two parameters are crucial: (i) the proportion of the population at risk of severe disease; (ii) the proportion of the population that dies if severely ill.
For the latter they choose a distribution with a mean of .14 and back this up with some references. 4/
For the latter they choose a distribution with a mean of .14 and back this up with some references. 4/
For the former -- the proportion of the population at risk of severe disease -- they *assume* values of .01 or .001. That is, in various versions of the model, they assume that only one percent or only one tenth of a percent are at risk of becoming severely ill if infected. 5/
Using those values, they run the model. The basic insight is rather simple:
a. There is a considerable number of deaths in the UK
b. If we assume that only a small fraction of those who are infected die,
then it follows that
c. many in the UK must already have been infected. 6/
a. There is a considerable number of deaths in the UK
b. If we assume that only a small fraction of those who are infected die,
then it follows that
c. many in the UK must already have been infected. 6/
Indeed, that's logical.
.
The problem is that their assumptions about the proportion of the population that will become severely ill are backed up by nothing. Quite literally: Their table 1 (p. 5) shows "support" for their various assumptions. The cell in question is empty. 7/
.
The problem is that their assumptions about the proportion of the population that will become severely ill are backed up by nothing. Quite literally: Their table 1 (p. 5) shows "support" for their various assumptions. The cell in question is empty. 7/
Their optimistic assumptions about the proportion of the population that will become seriously ill deviate from what everybody else thinks. In fact, we already know that their more optimistic assumption is wrong: The population of Lombardy is ~10 Million; >4,000 have died. 8/
That's impossible under the most optimistic assumption that .001 of the population become severely ill if infected. That's the assumption the dramatic "half of the population" claim is baed on.
So I think you can forget about the apparent implication of very low death rates. 9/
So I think you can forget about the apparent implication of very low death rates. 9/
If you find an error in my reasoning, or in my maths, let me know.
Source for # of deaths in Lombardy: thestar.com.my/news/world/202…
10/10
Source for # of deaths in Lombardy: thestar.com.my/news/world/202…
10/10
