Read on Twitter
much has been made recently about the rise in reported covid deaths in the last few weeks.
the question is: is it real?
i normalized 3 series to april peak to make them easy to compare.
CDC covid deaths and all cause deaths align well and show about a 20% drop from april.
the question is: is it real?
i normalized 3 series to april peak to make them easy to compare.
CDC covid deaths and all cause deaths align well and show about a 20% drop from april.
because these CDC series are real day of death, we can get a true picture of what happened.
the covid tracking project is different. they report deaths on the day they are filed, not the day they happened.
so you can get temporal mismatch.
this is illustrated below.
the covid tracking project is different. they report deaths on the day they are filed, not the day they happened.
so you can get temporal mismatch.
this is illustrated below.
this is a key chart from @justin_hart .
it shows that most of the deaths reported in december in the US actually occurred in november.
so the CTP is really shooting behind the pigeon in terms of what they report.
it shows that most of the deaths reported in december in the US actually occurred in november.
so the CTP is really shooting behind the pigeon in terms of what they report.
he has a good thread on this here.
you need real day (or week) of death data to draw real epi curves. https://twitter.com/justin_hart/status/1343651223933599745?s=20
you need real day (or week) of death data to draw real epi curves. https://twitter.com/justin_hart/status/1343651223933599745?s=20
the data from CTP cannot be used this way. it's not actually linked to disease curves and can mistake reporting boluses for death spikes.
we can see covid spike 3 times here as seasonal waves pass.
both seem to peak about 20% lower than april.
BUT: this data lags.
we can see covid spike 3 times here as seasonal waves pass.
both seem to peak about 20% lower than april.
BUT: this data lags.
it takes time to fill. the 3 weeks at the end of each series in red may be materially incomplete and as we see from justin's work above, backfill takes time.
even the last 5-8 weeks may still get increased.
this makes it difficult to say anything definitive yet.
even the last 5-8 weeks may still get increased.
this makes it difficult to say anything definitive yet.
the CTP series and the CDC series diverge meaningfully after the week ending 12/5. but as we see from the long term chart, CTP is generally late, not early in seeing a top.
it has also consistently overstated deaths in downslopes.
it has also consistently overstated deaths in downslopes.
and so this leaves us without a really good truth standard.
given the mad crush at hospitals and in states right now to find covid death and get CARES act money for year end, it's hard to trust the CTP day of report data.
OTOH, we know the CDC data lags several weeks.
given the mad crush at hospitals and in states right now to find covid death and get CARES act money for year end, it's hard to trust the CTP day of report data.
OTOH, we know the CDC data lags several weeks.
trying to guess which will converge on which is challenging. my guess is the peak on CDC data rises and moves back a couple weeks. whether it will match or exceed the april peak is not clear. https://twitter.com/boriquagato/status/1343301300306362368?s=20
people have tried to use cases to guess, but that generally fails
reported cases are such an artifact of testing levels as to render them epidemiologically irrelevant.
84% of the value is predicted by sample rate.
reported cases are such an artifact of testing levels as to render them epidemiologically irrelevant.
84% of the value is predicted by sample rate.
you can see it linearly here:
but what we can say is this:
covid surges and deaths have been highly seasonal and true second surges in places that had a big first surge look very uncommon. https://twitter.com/boriquagato/status/1336294267367649281?s=20
covid surges and deaths have been highly seasonal and true second surges in places that had a big first surge look very uncommon. https://twitter.com/boriquagato/status/1336294267367649281?s=20
despite what you hear, places like NY are not getting second surges. they are just testing a ton.
adjust "cases" for testing level and it's ~88% below april.
adjust "cases" for testing level and it's ~88% below april.
this is confirmed by deaths.
they look like the adjusted chart above and nothing like the raw figure shown in orange that is being used in such misleading fashion.
they look like the adjusted chart above and nothing like the raw figure shown in orange that is being used in such misleading fashion.
this refusal of deaths and all cause deaths to confirm the wild cases figures being thrown around is the signature of a "casedemic"
it's testing and overly inclusive definitions (like sweden counting all deaths within 30 days of a + PCR test as a covid death no matter the cause)
it's testing and overly inclusive definitions (like sweden counting all deaths within 30 days of a + PCR test as a covid death no matter the cause)
the data on covid is of terrible quality. tests are different, testing levels, definitions of case and death, all of it.
so you really need to take something of a mosaic approach.
when all cause deaths fail to confirm, you should get suspicious. https://twitter.com/FatEmperor/status/1337470384929730569?s=20
so you really need to take something of a mosaic approach.
when all cause deaths fail to confirm, you should get suspicious. https://twitter.com/FatEmperor/status/1337470384929730569?s=20
PCR testing is overstating clinical covid, covid deaths, and covid hospitalization by a significant margin.
the question is by how much, not if it is doing so, especially in light of the bizarrely inclusive "death with covid" definition in wide use. https://twitter.com/boriquagato/status/1330277330128936960?s=20
the question is by how much, not if it is doing so, especially in light of the bizarrely inclusive "death with covid" definition in wide use. https://twitter.com/boriquagato/status/1330277330128936960?s=20
no other pathogen is counted that way. it's inflating covid figures and making comps to flu and pneumonia very difficult if not outright meaningless.
but the one thing we can say is that all the mitigation efforts that have imposed such dire costs have not shown any efficacy whatsoever.
the virus will be the virus.
the policy has been just damage, not cure. https://twitter.com/boriquagato/status/1343965548619517952?s=20
the virus will be the virus.
the policy has been just damage, not cure. https://twitter.com/boriquagato/status/1343965548619517952?s=20
