Coronavirus Pandemic (COVID-19) (SARS-CoV-2) [2020]

Well...that graph is very different to others! What are the dates for the near 0 (should that be 1?) rates before the spike?

This is the graph I get plotting the numbers in Excel, downloaded from the site you linked to:

upload_2021-12-17_16-26-41.png

Here's the Excel table:

upload_2021-12-17_16-27-36.png

This is up to the 12th as that's the latest final figure for specimen date cases. The fluctuations are likely caused by people testing differently on different days, so oscillating in 7 day cycles with Monday being the day of most tests.

That's very different to yours! A large spike is incoming on the 13th and 14th because testing didn't increase much but cases did. At the moment though, we haven't got a methodology for tracking case rate. I wonder what data you are getting from the API? Possibly only one of the testing Pillars, so perhaps comparing case rates to only PCR tests or LFTs?
 
Last edited:
Well...that graph is very different to others! What are the dates for the near 0 (should that be 1?) rates before the spike?
Regrettably, I just realized the dates are in reverse order on the graph, most recent is the very left!! This is why you need to be able to read them!
LOL.

You are AOK!

Updated graph above, I removed the old one to discard any possible misinformation.
 
Last edited:
Just had my booster at my local pharmacy. Pfizer this time round - I had double AZ first time.

It was rather busy, the guy said they'd been doing about 200 per day. On Tuesday the queues were out of the door he said.
 
I wonder what data you are getting from the API? Possibly only one of the testing Pillars, so perhaps comparing case rates to only PCR tests or LFTs?
I'm only using
cumCasesByPublishDate'
cumTestsByPublishDate'

I form the delta by subtracting the last day. I've updated the graph above.
 
Last edited:
I see you've updated the graph. It now seems right, aligning with the given data. As we can see the number of cases largely corresponds with +ve rate but the last couple of months have been anomalous and the case rate not illustrative. The early valley of the spring also has a ridiculously low case rate because testing was way, way, way over the typical!

And as if it's bad enough trying to understand what's going on in one country, different countries counting different data in different ways makes it nigh impossible to apply data science, kinda defeating the point of Big Data. We really need an international standard for data collation and reporting.
 
I see you've updated the graph. It now seems right, aligning with the given data. As we can see the number of cases largely corresponds with +ve rate but the last couple of months have been anomalous and the case rate not illustrative. The early valley of the spring also has a ridiculously low case rate because testing was way, way, way over the typical!

And as if it's bad enough trying to understand what's going on in one country, different countries counting different data in different ways makes it nigh impossible to apply data science, kinda defeating the point of Big Data. We really need an international standard for data collation and reporting.
its about 80% of the job sadly. Getting data to align because everyone does it differently. There are no standards and there is a significant amount of cleaning required. But that's also why AI works, the AI converges towards a value, but it never says it's 100% sure. And that's also how we should be viewing things. Most people are probably correct, but there is always an off-chance you could be wrong, despite how small. But it's okay to make decisions on the probable bet.
 
London's declared a Major Incident over Omicron. Ignoring the confusing issue of case rates, which also are irrelevant if no-one's actually getting sick, we're seeing hospitalisation rates increasing with considerable increase in London.

upload_2021-12-18_22-31-37.png

Reports say the majority are unvaccinated, and London's ethnic populations have not been as willing to accept the vaccines. Sadly no real numbers - hospitalisation with breakdown by vaccinated/not isn't a maintained number AFAICS.
 
Holland going into lockdown, so we will be able to compare and contrast how different mitigations (or the lack of them, in the case of the UK) are likely to play out.

Here's something unusual and surprising (though not bad!) in the numbers out of South Africa:


A very rapid decline in cases and perhaps hospitalisations after the initial rapid peak, which indicates there is something else we don't know about this variant. Is it perhaps something like those infected with Beta (which we know could evade existing immune response from past infection or vaccination) have good protection against Omicon? Might explain why the surge there seems to be subsiding quickly if a lot of the population have innate immunity/protection from their past infection with a particular variant. If so, we wouldn't expect to see the same in most countries where Beta never took off or was suppressed. As I seem to be saying in every post these days, we'll likely know in a couple of weeks...
 
The other one is hospitalisations, which finally got a most recent update for SA and shows a large increase by 11th December.

upload_2021-12-19_10-16-23.png

Yet deaths (per million population), dated to the 18th so including a week of these hospitalisations, remains very low with only a marginal increase.

upload_2021-12-19_10-17-36.png

Edit: Apparently 90% of serious Omicron hospitalisations are from the unvaccinated. Additionally, 50% of hospitalisations are incidental, so not because of C19. This kinda screws with the hospitalisations numbers. I believe this is the same case for SA, that where 'hospitalisations' are up, these are often incidental and not cases where C19 has caused a need for hospital treatment. We really need an 'excess hospitalisations' figure to show the impact on the health services.
 
Last edited:
UK is on the rise, but it still looks okay for now at least with respect to mechanical ventilators, hospital admissions.
Cases and admissions are going up, but death rate still seems stable for now.
I'll get around to putting in tolerances for variance that way you can use that or a trend line instead of seeing the reporting seasonality.

wICKVtx.png

z5bLgrZ.png

IGtJNww.png
 
One of the issues with admissions is that they are regionalised which won't show in national stats. You could have counts going down nationally but exceed capacity in a region like London. But then admissions data doesn't show whether it's a problem with C19 or not as admissions for other causes that happen to test +ve for C19 are included in the C19 counts - for that you need ICU data I guess. I don't know if the C19 death registration is going to make the distinction between 'caused by C19' or stick with the absolute 'they had C19 28 days ago and have died since from something unrelated but we'll count that as a C19 mortality'. That's be revealed in the 'causes of death' data, with lower respiratory death ratio to C19 cases.

The other issue is medical staff being off sick or having to isolate for multiple days, leaving the hospitals short-staffed and that accumulating a backlog of problems.
 
Low death counts is not enough good news if you are worried about hospital overcrowding.
And even a variant with low fatality can become a lot more lethal if hospitals become so strained that they can't provide proper treatment for those that need it...
 
The issue with the UK numbers at present is that previously-confirmed cases don't count towards the stats shown on the dashboard. This means that reinfections (of which there will be many with Omicron) don't appear in the figures shown. You'd hope that reinfections will lead to mild disease in the vast majority of cases, but that doesn't help with an understanding of the general spread of Omicron and the impact will change as it finds its way into older age groups. All eyes will be on London where the S-gene dropout numbers would tend to indicate that 90% of cases are now Omicron. Hopefully, the hospitalisation data over the next week will show definitively that the vaccines can greatly reduce risk.

I note that the Danish authorities have started to ignore positive case rates (due to overwhelmed testing capacity), but are instead focusing on hospitalisations. I expect it's the way all countries will be heading in due course.
 
The issue with the UK numbers at present is that previously-confirmed cases don't count towards the stats shown on the dashboard.
Where's that clarified. I've not seen that explained in the Dashboard explanations. It states cases are counted when someone submits a positive test, either LTF or PCR, and I see no mention that they won't include a positive case when it's someone who's previously been infected. There's an ambiguous phrase saying repeat positives won't be counted, but I take that to mean over the short term, as obviously someone testing every day for the same infection to see when they are well again shouldn't be counted as a new case every one of those days. Admittedly, it could include all positives ever, but that would be a daft oversight.

Edit: You're right - I asked and here's the reply.

Currently the dashboard is not showing reinfections. People are only counted as cases once, on the date of their first positive test, regardless of the length of the interval between positive tests.

We are actively working on an episode-based definition, whereby if a certain amount of time has elapsed between positive tests a person can be counted more than once.

We don’t yet have a confirmed timescale for this as it is complex to implement and requires agreement across all nations of the UK. In the meantime we will look into clarifying the current metadata as you suggest.

For England we do publish data on reinfections in the UKHSA surveillance report: https://www.gov.uk/government/stati...d-19-surveillance-reports-2021-to-2022-season
 
Last edited:
Our data always included hospitalizations, ICUs etc as well as positive tests in absolute numbers and relative to the total number of tests, and the data is always also available in detail for regions.

Incidentally I thought it was very interesting that the average age of an unvaccinated person on the ICU was 20 years below that of a vaccinated person. That says a lot already by itself. Will be interesting to see if Omikron affects that.

I’ve had covid in April and got extremely tired for a few days but after a week I was already mostly recovered, and vaccine 3 months later. Under normal circumstances a booster wouldn’t feel necessary yet, but if it continues to be necessary for the greater good then so be it.

We have no margins here, our hospitals were already full and postponing other care again due to optimistic policies and so we already have a lockdown again, which we need partly because we just cannot afford omikron causing more hospitalizations.
 
I’ve had covid in April and got extremely tired for a few days but after a week I was already mostly recovered, and vaccine 3 months later. Under normal circumstances a booster wouldn’t feel necessary yet...
It possibly isn't.
https://www.bmj.com/content/375/bmj.n3047

The pandemic has revealed a form of “super-immune” response among some who have been both infected with the virus and vaccinated against it.

“However, we found that if someone were to contract SARS-CoV-2 first, and then receive a single dose of the vaccine, this mounted a hybrid immune response that is 20-30 times stronger than [in] those who have been vaccinated, even with two doses (without contracting the virus).”
Certainly in terms of protecting yourself and resisting the disease, your body is likely well trained and doesn't need any additional support. But if the matter is not spreading Omicron, boosters may still be valuable *if* they stop infection for the period after the shot thanks to antibodies.
 
I'd imagine that they would have the numbers somewhere. The problem is that there are going to be so many Omicron infections (and reinfections) that it is only going to be possible to sequence a tiny fraction of cases. I'd imagine they have their data models worked out to try and calculate the probable rate of reinfections as the Omicron wave progresses. The S-gene dropout will give a reasonably accurate indication of prevalence as most of the PCR tests which show this feature are going to be Omicron.
 
Back
Top