I’m guessing there is a term for this in epidemiology and public health I just don’t know what it is or what all goes into the calculation.
But if 800 people died today, the virus takes ~3 weeks to kill you and has maybe a 2% death rate for infected people, does that mean 21 days ago about 40,000 new infections occurred in one day?
I mean it doesn’t take exactly 3 weeks and the actual IFR isn’t known for sure.
It is a better indicator than anything else we’ve got but actually calculating the number infected roughly 3 weeks ago requires information we are completely lacking.
My WAG is based on current estimates for fatality rates of those sick enough to be diagnosed (even if they do not get one) of maybe 1% (or less), and guess-estimates that there are 8 or 9 asymptomatic or very mildly symptomatic infected individuals to every one of them. Roughing it that means if 800 died today it would mean roughly 800,000 new infections in one day about 3 weeks ago. (Most of whom never had any identifiable disease.)
I think that addition of 8 or 9 non-symptomatic case per sick case is too optimistic, and fails the San Marino Test. Because it means that the actual case fatality rate is 0.1 %, so the 26 deaths in San Marino would correspond to 26,000 infections (out of a population of 33,000), as of three weeks before the last death there (yesterday). Everyone would have already got it, and we shouldn’t be seeing any more cases in that country within the last week - but that’s not what seems to have happened.
1% seems far more reasonable - rising to 2% or 5% in an overwhelmed health system. That seems consistent with South Korea’s 0.6% so far
I think the term is back-casting, or at least that is what is used in historical demography. One trickiness would be the differential between a ‘natural’ death [person gets sick, gets worse, dies] versus a treated death [person gets sick, gets worse, goes into hospital, gets into ICU and is kept alive while the hosipital system is still capable, hangs on in a desperate battle for a week or more, then finally goes]. there is a nursing home here where a number of elderly have died over the course of several weeks. I’m not sure if its a lingering and staggered contraction of the disease from one person to another, or some just were kept in treatment for a long time before succumbing].
San Marino doesn’t get any sort of ‘The Biggest …’ awards very often, although I suspect we will see Trump reasonably soon contextualising US death rates as being ‘still less than San Marino’s, so sad, I love their sunny beaches and sombreros’.
Was there a general shutdown 3 weeks ago? I mean a general, global shutdown that we have now. Surely that would affect estimation.
My recollection is that mid march was when some states in the US started limiting gatherings to 250 people and closing restaurants for anything other than carry out or delivery.
As to San Marino - small n pool can give goofy numbers. They currently have a rate of deaths per million from COVID over 4 times Italy’s current.
Using a small n sample as a test would be ill advised.