[N5Gooner (OP)]

How accurate are the comparisons given by different Countries on their infections and deaths ? I see we changed our death recording to only those who died within 28 days of being tested. Who is actually telling the truth?

[alancalverd]

Nobody knows the truth. As a police officer once told me "If the head is not detached from the body, cause of death is an opinion, not a fact".

Deaths of those infected with COVID  range over a number of common causes such as pneumonia or cardiovascular failure, which are fairly easy to diagnose from clinical records without a postmortem, so in busy periods (and who wants to conduct a postmortem on a known infectious body?) one of these is more likely to be recorded as cause of death., maybe with COVID as an afterthought. 

The only valid statistic is excess deaths compared with the recent 5-year average. This seems to correlate fairly well with population density D multiplied by government incompetence G. Hence the USA and Brazil, with very low D but astronomcal G, are in the same league as the UK (high D, moderate if confused G) and way ahead of e.g. Singapore and South Korea with UK-level population density  but near-competent governments.   

[Bored chemist]

"If the head is not detached from the body, cause of death is an opinion, not a fact".

I think he was poisoned before they cut his head off.

[alancalverd]

I think this is an example of an undistributed middle. Literally.

[set fair]

Who is actually telling the truth?

Most likely South Korea, Belgium, Vietnam, Singapore.

[alancalverd]

And how do they know? Have they really carried out postmortems on everyone? Even in non-pandemic times, "cause of death" is often an educated guess in "nonsuspicious" circumstances.
The only certainty is that all deaths are registered, so "excess deaths" is precisely calculable.

[Bored chemist]

The only certainty is that all deaths are registered, so "excess deaths" is precisely calculable.

How many of them were caused by a cancer that wasn't treated promptly because the patient was too scared of covid to go to hospital?

[alancalverd]

That would, in my book, count as an excess death, and in the absence of any other endemic cause, I'd assign the statistic as "due to COVID".

As it happens I'm in the midst of  refereeing a research project on the impact of COVID on radiotherapy. Some treatments have been delayed, some accelerated, and some have modified the adjuvant chemotherapy protocol. It is possible that the overall effect on quality of life may have been positive (through hypofractionation and abandoning chemo) negative or neutral. Watch this space!

[chris]

As Belgian health minister Maggie De Block observed "across the EU, no one counts Covid the same..."

What she is highlighting is the difference in case definitions, inclusion criteria on death certificates, testing volumes and regimes etc.

Looking at Russia, you can see something very strange appears to be going on - huge numbers of "cases" but very few deaths. It turns out that the Russians are using "a very specific definition of a Covid death" which may well account for the potentially suspiciously low death rate. Conversely, perhaps the rest of the world are using too loose a definition and over-calling covid deaths as a result...

[Petrochemicals]

I believe South Korea and japan's reporting, their cultures have long had face masks as a social norm and they are fastidious about presentation cleaning etc. Germany I do not believe, their mortality reporting is very different, if you have emphysema in Germany and catch corona and pass away, your cause of death is listed as emphysema, unlike UK USA Italy etc where it is mentioned.

[Bored chemist]

Imagine, just as a thought experiment, that I could provide you with perfectly accurate statistics for deaths from covid by date, age group, sex and location- say the postcode/ zip code or "what three words" address.

What would you do with the data?

[alancalverd]

Ignore it.

Very few people die from COVID. Many suffer from it, and quite a few die from secondary infections because their immune system has over-responded to COVID, so the prevalence of COVID is far more important than the statistics of "COVID-only"  deaths .

Since I can't alter the age, sex or location of anyone, that data is of little clinical value though it will doubtless generate lots of academic papers and political argument. Suppose there was a huge peak of deaths among females aged 49 in Lincoln: so what? Postmortem statistical finesse is unlikely to lead to prevention or cure.

It would be of interest to know what every COVID-positive person was doing in the 2 weeks before testing positive, but even there, we know the answer before asking the question: avoid the company of anyone who is not proven clean, and institute a national quarantine to get rid of the disease.

[Bored chemist]

so the prevalence of COVID is far more important than the statistics of "COVID-only"  deaths .

Well, since it's a flight of fantasy, imagine I can provide that data too.
And, if you like, a "%due to covid" figure for every single death.

What use is the data?

[alancalverd]

The prevalence is important.

If, as our politicians seem to want, COVID becomes endemic, prevalence is a determinant of perceived risk and thus determines public behavior and the budget assigned to prevention and cure.

In a more rational society, the change of prevalence is an indication of the success or otherwise of your preventive measures.

"Probability of causation" (% of a single death due to X)  is a useful parameter in determining compensation for early death in industry where the symptoms are indistinguishable from natural incidence. I know it has been used (incorrectly) in the nuclear power industry and (more correctly) where chemical carcinogens have been involved, but I don't see a lot of use in this case, except where deliberate or negligent infection is alleged - say where infectious patients have been discharged into care homes.