Something to keep in mind: The deaths/total-known-cases ratio does
settle down to roughly 1% once the dust has settled and data is more
complete (look at South Korea, China outside of the initial
out-of-control Wuhan/Hubei outbreak, and the Diamond Princess) _under
certain circumstances_.
Those circumstances being, availability of first-world quality intensive
hospital care for ALL of the (very roughly) 5% fraction of cases bad
enough to require mechanical ventilation. Given that, roughly 4 out of
5 of those severe cases will survive, and the remaining 1 out of 5 (out
of a hundred cases overall) die, producing that ~1% overall death rate.
BUT. If the spread of infections is NOT effectively contained early,
the number of severe cases can surpass the locally available number of
intensive-respiratory-care hospital slots. At which point, essentially
ALL of those needs-mechanical-ventilation cases that don't get it _will die.
_So the real deaths/cases rate can go up to nearer 5% than 1% in places
that let their hospital capacity be overwhelmed. Wuhan/Hubei in the
original outbreak, and now Iran, Italy, and likely other places soon.
RE undetected cases skewing the deaths/cases rate higher, that's also a
factor, but less of one than many (including me) hoped early on. The
Diamond Princess data (where everyone on board was tested) shows
something under 50% of total infections are asymptomatic. (45%, last I
looked, and dropping slowly as more infectees develop delayed symptoms.)
Henry
__
On 3/20/2020 2:19 PM, David Summers wrote:
Take your current death rate from the virus, divide by the approximately 1% death rate. Once your positive test reach the level, you are testing enough. (But all numbers need to be adjusted for exponential growth, of course)
On Fri, Mar 20, 2020, 11:07 AM Uwe Klein <uwe@xxxxxxxxxxxxxxxxxxx <mailto:uwe@xxxxxxxxxxxxxxxxxxx>> wrote:
Am 20.03.2020 um 21:20 schrieb David Summers:
> While the infection rate may be unknown, the death rate is easier to
> know. Barring a coverup, the numbers seem right.
>
Thing I haven't really tried to get a grip on:
You do a number of tests ( with cause for testing assumed
i.e. not shotgun style testing. )
from this you get a certain number of positives : infected.
for the US from official numbers 50+k tests with ~16k infected
found or
~30% positive.
for Germany I could only find 35k in the last week for the last weeks
delta of ~9k positives. 25% ? ~ same domain.
( haven't looked anywhere else.)
When is a point of overtesting reached?
What positives rate should be aimed for to get good coverage?
( obviously testing everybody and his/her hamster is out of the
question
at the moment // for ever?
I'd go for shotgun antibody testing later on to get expanded
information
including (silent, unnoticed) spread )
Uwe