> Treating any modeling we have of the impact of Covid 19 as reliable is an exercise in insanity.
In detail, yes. In general, you can still make some good back-of-an-envelope calculations.
Conservative estimates are that the coronavirus has a CFR of about 1% given adequate medical care (South Korea's statistics, where testing has been comprehensive enough that we should have identified any wide pool of asymptomatic cases), and another low estimate of its R0 factor is 2. Left unsuppressed, this implies that the disease would spread to infect about half of the population, and it would kill 1% of those infected.
For the United States, that implies that a "flattened curve" -- where mitigation prevents medical resources from being overwhelmed but does not fully suppress the disease's spread -- will kill about 1.6 million people.
Beyond that, we know that the disease requires intensive care at some multiple of the death rate (say 2x) and hospitalization at another multiple (say a further 3x). These estimates are reasonably consistent with New York's numbers (https://nymag.com/intelligencer/article/new-york-coronavirus...). Testing shortfalls could make these multiples worse, if there are hospitalized or ICU patients positive for the virus but not included in these totals.
Given overwhelmed medical services, we can presume that a large fraction (say half?) of ICU patients would die for want of care, and a smaller but still significant fraction of hospitalized patients would do the same (1/8?).
This implies that an un-flattened curve would have roughly triple the death rate, with the excess caused by inadequate care. With a 3% inadequate-care CFR, if left to run its course the disease would then kill about 4.8 million Americans.
> It's just to say that they are conservative and acting on knowingly incorrect information.
If policymakers are acting on "knowingly incorrect information," it's because their assumptions are too benign rather than too severe. I believe that my estimates above should be uncontroversial, and to the extent they err I've tried to err on the less-deadly, less-contagious side.
Various models and data sets put asymptomatic cases at between 20% and 50%. Those are fully asymptomatic - i.e. total end-to-end progression with either very mild symptoms indistinguishable from a minor cold, or no symptoms at all.
It's almost impossible to estimate expected population mortality with limited and noisy data, but I've seen estimates from 1.5% to 0.05%.
The only thing that can be said with certainty is that social distancing, testing, and tracking all do a lot to prevent initial infection, and good access to ICU hugely improves chances of survival after infection.
The rest is guesswork at this point. Having said that - my current hand-wavy estimate of deaths in the UK is high five, low six figures. Multiply by five or so for the US.
> Various models and data sets put asymptomatic cases at between 20% and 50%. Those are fully asymptomatic - i.e. total end-to-end progression with either very mild symptoms indistinguishable from a minor cold, or no symptoms at all.
That's why I use South Korea as a model. They've tested enough that they should have found the majority of asymptomatic cases, and they still have CFR above 1% (1.7% as of this writing).
That also puts a bound on reasonable levels of occult spread. We can support maybe 50% of cases being totally asymptomatic and undetected, but if that is significantly greater then we'd see contact-tracing (again, SK-style) entirely fail as a control measure.
So it seems like an absolute best-case CFR is 0.5%, if it would be 1% among symptomatic cases and there again that many that never notice / are diagnosed with the disease. For the UK, that would give an optimistic projection of (66e6 * 50% * 0.5% =) 165k deaths in a "herd immunity" outcome with a "flat curve", so this is consistent with the range of your "hand-wavy estimate."
We will not know how effective testing has been until we have serological tests. There are a few assumptions being made regarding the efficacy of testing, and contact tracing will certainly miss pockets of asymptomatic people.
In detail, yes. In general, you can still make some good back-of-an-envelope calculations.
Conservative estimates are that the coronavirus has a CFR of about 1% given adequate medical care (South Korea's statistics, where testing has been comprehensive enough that we should have identified any wide pool of asymptomatic cases), and another low estimate of its R0 factor is 2. Left unsuppressed, this implies that the disease would spread to infect about half of the population, and it would kill 1% of those infected.
For the United States, that implies that a "flattened curve" -- where mitigation prevents medical resources from being overwhelmed but does not fully suppress the disease's spread -- will kill about 1.6 million people.
Beyond that, we know that the disease requires intensive care at some multiple of the death rate (say 2x) and hospitalization at another multiple (say a further 3x). These estimates are reasonably consistent with New York's numbers (https://nymag.com/intelligencer/article/new-york-coronavirus...). Testing shortfalls could make these multiples worse, if there are hospitalized or ICU patients positive for the virus but not included in these totals.
Given overwhelmed medical services, we can presume that a large fraction (say half?) of ICU patients would die for want of care, and a smaller but still significant fraction of hospitalized patients would do the same (1/8?).
This implies that an un-flattened curve would have roughly triple the death rate, with the excess caused by inadequate care. With a 3% inadequate-care CFR, if left to run its course the disease would then kill about 4.8 million Americans.
> It's just to say that they are conservative and acting on knowingly incorrect information.
If policymakers are acting on "knowingly incorrect information," it's because their assumptions are too benign rather than too severe. I believe that my estimates above should be uncontroversial, and to the extent they err I've tried to err on the less-deadly, less-contagious side.