I read that Germany is not testing the dead, so if someone wasn’t tested while alive then their death is attributed elsewhere. Italy is testing the dead to get a more accurate picture. I find it baffling that there isn’t an international standard for testing in a situation like this so that we can get a handle on what is actually going on. If we don’t have data, how can we possibly draw inferences and learn anything?

There are international standards and the standard is that flu is not normally considered a cause of death. It's not realistic to expect autopsies to be performed and coroners reports to be written on every person who dies of pneumonia during an epidemic.

And yet I understand this is what is happening in Italy where the suspected cause of death is Covid19. If we don’t make it realistic to expect this (during a pandemic), then how can we trust the numbers?

Right, for example what if there is a second strain going around that is obscured by the large inconsistencies in testing and reporting?

A huge factor here is time. 3 weeks ago I saw a bunch of people saying "Look, South Korea tested a lot of people, finding the majority of the mild cases, and thus only has a death rate of 0.5%". Since then, however, deaths/cases in SK has risen to 1.59%, and increasing daily. They don't have many new cases, what's happening is the cases from 3-4 weeks ago are dying.

Germany has tested a lot and found a lot of cases earlier in the disease progression. I'm sure their death rate will rise like S.K.'s did. Not all the way to the 8% - 10% we see in Spain and Italy, but I'd guess up to 2%.

Wildly different case fatality rates might be caused by the same infection fatality rate if milder infections are systematically less likely to turn into medical cases in some countries.

Compare the age distribution of infections in Iceland versus the Netherlands for instance.

https://twitter.com/alexandreafonso/status/12435570137597009...

We have every reason to think that just as large a fraction of the 10-19 year olds in the Netherlands have gotten this in the Netherlands as in Iceland but when it stays as a cough and light fever many medical regimes are never going to test those cases.

And countries have different demographic profiles, though not enough to explain a 10x difference. And medical systems get overwhelmed in some places which could possibly be up to a 10x difference in a really cases.

You can't compare those numbers without an in-depth understanding and comparison of who is being tested in each country.

Which means the numbers are not consistent.

I'd just say the numbers are different.

Another case in which the numbers are oddly (one might say "suspiciously") different are the UK's number of critical patients (163) compared to their known number (19522), which is very low at 0.8% compared to France's 11.5%. That's a massive difference in the number of seriously critical patients, given that the UK has a sufficiently large sample size now.

> critical patients

What does that even mean? The UK isn't testing unless you need hospital care (or you're special, e.g. Prince Charles, Boris Johnson, Idris Elba), aren't all of those patients somewhat critical? Or is "critical" just "hooked up to a ventilator" in which case it could be limited by the number of ventilators instead?

I like to thing of it as a unit problem. The numbers are presented as x "items", but the definition of "items" for each of the numbers is not consistent (or, even, actually known).

There are multiple variables at play. For example in Italy there is a high percentage of people living with their parents in the age 20-40s bracket, which should lead to more elderly contamination. Also the health services between Germany and Italy should not be even comparable in terms of equipment and capacity.

> health services between Germany and Italy should not be even comparable in terms of equipment and capacity.

Italy's health care is among the best in the world, and Lombardy's is several points above the Italian average.

Germany has apparently 2.6 times the amount of IC beds available in proportion to the population- in the Covid epidemic this should offset the reach of maximum capacity by a week.

Do you think that's enough to account for a 146 to 6 difference in deaths per million between Spain and Germany?

After you run out of ventilators the sky is the limit when it comes to deaths.

When you still have ventilators, only a few very elderly/unlucky die...

So a country with 10,000 cases and 1,000 ventilators can have 20 deaths, and another with 10,000 cases and 500 ventilators might have 400 -- if of the 10,000 cases in both 520 need ventilator... (and some die, whether with ventilator or not, but many wouldn't if they had ICU/ventilator treatment).

Mortality from ARDS is 50% even with ventilators. Source: wife is a pulmonologist and I am a radiologist.

Are there other reasons why C19 patients would need ventilators? Or is ARDS the main reason?

Is that mortality rate consistent across different underlying causes?

ARDS is the main reason. 40% mortality from just ARDS. Mortality goes up with covid19, age, sepsis, and renal failure.

You'll need to back up the claim that "only a few very elderly/unlucky die"

What % of people actually survive after being on a ventilator? Because I've seen various studies and anecdotes of this for C19 specifically, with results ranging from "basically no-one" to "80%, at least temporarily, but hardly anyone is actually out of hospital yet so maybe they're going to die later".

Beyond C19, this study suggests roughly something like 50% long-term survival IIUC: https://www.ncbi.nlm.nih.gov/pubmed/26003390

Ventilators are invasive things that themselves can do lung damage. Fairly common for people to choose not to go on them in end-of-life directives.

Medical system capacity and utilization explain the vast majority of the variation in mortality among different countries, after accounting for stage of the epidemic.

When things are not above capacity, mortality is 0.5% or less. After reaching capacity, mortality jumps 10x or more.

Age distribution can only account for variations within, at most, a factor of two.

So Spain's (and France's to a lesser degree) medical system got overwhelmed that much faster than Germany's or the UK's?

They might have been very late with testing. I know this was barely on the radar in France even 2-3 weeks ago. MRSA has also been suggested as a component.

In Italy the hospitals were overwhelmed so you should expect the mortality to increase very fast when you can't even give basic treatment to patients in need. The same thing happened in Wuhan and that led to massive death toll. But as I said, many variables at play and we will surely have a better understand of what mattered the most to prevent mortality.