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

*Zoinks* (Or some other sound of shock)

https://www.theguardian.com/world/2...ses-at-record-level-with-78610-new-infections

Highest number of recorded cases from any day of the pandemic, 15% higher than the peak of the January wave. And Omicron is only just getting started. Now dominant in London:

 

Look at that Omicron trajectory in London:

 

Just to note that the 78,000 figure is for confirmed *New* infections. If you've previously been infected in previous waves, your positive PCR doesn't count in the UK figures. This was perhaps understandable when numbers were relatively low, but with Omicron, it means that the reported rates are going to be a large understatement of positive tests!

 

we're spiking with Omicron here as well. Reading some stats that it's anywhere between 48% to 80% of our new covid cases now. (Ontario, Canada).

Fly safe

 

There's been suggestions that combining Paxlovid with Molnupiravir may raise efficacy. That was before the lower efficacy numbers came out for Molnupiravir.

 

BBC interviewed some immunologists who say that not only do boosters increase antibodies, they produce "better" antibodies and B and T cells.

https://apple.news/A9ALynJXhSLO9rGA2hMme1w

But we will mostly hear about boosters raising antibody titers by x times, not that there are "better" antibodies and better response from B and T cells. Hard to quantify the effects of the latter.


There were earlier studies also indicating that about 6 months after vaccination, B and T cells would produce better response and possibly sustain durable immunity. However, Israel said it saw rising hospitalizations of fully-vaccinated people, which is why they rolled out boosters quickly and broadly.

Today, there are studies showing protection against Omicron infection is well under 50% for 2-dose vaccination and around 70% or less for protection vs. hospitalization for older people.

So if B or T cells are suppose to become better over time, it doesn't seem to have been born out by first Delta and then Omicron. At least not yet.


As far as boosters producing both quantitative and qualitative improvements in immune response, maybe by the spring they will be advocating another round of boosters, if Omicron-specific boosters aren't ready or don't show vastly-improved efficacy.

 

Hong Kong researchers find that Omicron replicates 70x faster than Delta in the bronchus but 10x slower in the lungs.




Slower replication in the lung does not necessarily correlate to lower severity however:

Maybe Omicron is much more efficient at infecting cells in the upper respiratory tract but not the lung or maybe other organs as well?

It goes strong to establish a beach head but then slows down and the immune system can catch up.

May also indicate why vaccines are so much less effective, as antibodies in the nasopharyngeal linings are thought to be low and not long-lasting. The better efficacy of boosters may be short-lived, even if there are still high titers in the blood overall but maybe not where you're getting infected.

 

I don't know if this has been already shared, but here it goes (I can't embed the video):
https://rumble.com/vqtk6q-pldoras-p...-dr.robert-malone-a-los-padres-y-abuelos.html

He was one of the first mRNA researchers and he's not an antivaxer. I found this video by chance and I tried looking it up somewhere else (YT), to be able to embed it here, but I couldn't. It seems many of his videos are being removed on main video platforms for spreading misinformation.

I found other videos, though, like this one:


I'm vaccinated and I still think it was the best choice, but I have a few questions regarding possible permanent side-effects that we haven't had the time to confirm/dismiss, etc.

 

Why does everyone persist in using the raw count. As stated so many times, if people are testing more, you'll catch more positive cases. These values need to be normalised against testing if you want a more realistic picture of what's actually happening. Cases have increased by 19% in the past seven days, but then testing has increase 15%! So without any change in the disease spread, you'd still be looking at a 15% increase in number of cases reported from the change in testing!

Considering the amount of complaining that's happened about sources and numbers and unrepresentative data, I can't understand why people still perpetuate raw counts. It's as wrong as the UKHSA demographics in the vaccinated infection rates. If you aren't happy with data 'showing' infections are higher in vaccinated people because it's inaccurate, you shouldn't be happy using raw case counts without factoring in testing rates. It's inconsistent and unscientific.

 

I think your right on the too many variables. Maybe the air was more polluted. Maybe it was time of day, maybe it is so specific that it depends what part of the tissue it enters. Maybe it depends on muscle use after. It is crazy how little we know at the same time we know so much. Like the quote about the island of knowledge expanding the shoreline of ignorance. Still area is radius squared and perimeter isn't so that quote is funny because I'm sure the speaker thought about that and hence is really saying the ratio of known to known unknowns increases.

 

When they make a booster semi-mandatory I do hope they still have an adenovirus carrier (or subunit) vaccine. I love the science for mRNA vaccines, but I love it more applied to other people

PS. I wouldn't refuse a mRNA vaccine ... except self-amplifying, no need to make the general population guinea pigs for that, that can wait.

 

It's still a useful measure, though what's being calculated is the derivative that matters, not the actual numbers. So if you're getting an exponential graph and the derivative over time is sloping upwards, you're getting exponential growth. More testing won't cause this, since that's like saying more testing increases acceleration, which it doesn't. The positive % by sample of tests has to be increasing for an exponential growth to occur.

Or to put simply. If your positive rate is 5% and moves 19% more over 7 days.
100 tests = 5 positive.
15% increase in testing = 115 tests
= 5.75 people should be getting sick.

You test 115 (15% more) but you get back 23 people are sick, you're now in 20% positivity rate. The case rate is growing.

 

But it's not! It's static. We are presented numbers like 60,000 cases one day followed by 70,000 the next to show us the disease is increasing at an astronomical rate. But most of this increase is derived from an increase in testing; we actually have an exponential increase in testing at the moment.

​

I'm not saying the virus isn't spreading, but the numbers are nonsense and no-one's caring to use the right ones, confusing the genuine understanding.

On the converse, if you divide the number of cases by the number of tests, you get the rate.

Instead, we get, as per your example:

100 tests = 5 positive.
15% increase in testing = 115 tests
= 6 people tested positive.

News reports there were 6 new cases today versus yesterday's 5, a 20% (6/5) growth of cases. Where the growth is actually nil, zero, zip, nada, static, none whatsoever.

The positive case rate has been between 4 and 6% in the UK since the summer, when LFT came in around March I think. We had 1.3 million tests a day from mid March (selective PCR and generalised LFTs), the highest daily test rate being 1.9 million on 21st March. That dropped to ~750k a day in August, upped and downed to 900k a day in the Autumn. There were 670k performed on 30th October, and then the rate has risen to 1,630k on the 15th of this month.

This is the case rate in the UK, factoring in the number of samples taken, case rate / test rate to December 14th...

​

All the increasing case rates of the fortnight 1st - 14th December have been due to more testing, not more virus. We had Trump laughably dismissing the increasing viral presence in the US in the beginning of the pandemic by saying, "if you test more people, you'll catch more virus." What's happening now is the opposite dumb - now we have the testing capacity to track the virus at the population level, it's being ignored to just pick up an ever increasing number that isn't necessarily driven by virus activity.

How can anyone make sane, informed decisions without using sane, informed data? Misinformation and wonky stats have really punctuated the mess of this pandemic.

 

hmm.

Do you have a data source I can work with that you'd be okay to also use as being the base set of metrics? Sort of curious if I can solve this and dump the results out into a public notebook that you can use from this point forward.

Yes, I do agree that re-writing numbers to change the narrative can totally mess people up. I don't live in UK, so it's hard for me to really understand what's happening over there, I can only look at raw numbers if you have them.


Is it okay if we use this one?
https://coronavirus.data.gov.uk/

and to confirm the reported news in UK is specifying this particular number?



Or a different one? I'll have to look at the positive case rate vs tested on this site.

I just need to be clear, that if the virus is dropping off, increased testing would plummet the positivity %. So increased testing is not directly correlated to increased positivity rates if the samples are large enough, which I can see are fairly large here.

edit: I'm just going to do some graphing, but positivity graph doesn't really seem anomalous from what I can see.
edit 2: oh, you're spiking over there

 

Cases up 31.4% over the 7 days to 16th December. Number of tests up by 18.5% in the 7 days up to 15th December:

https://coronavirus.data.gov.uk/

Given that the results take 24 hours or more to come through, that looks as increase over an above testing levels to me. Not forgetting that the figure doesn't include reinfections. I presume the tests taken by those with reinfections still count in the testing statistics? If so, that would skew the numbers even further.

Look at the charts. The gradient for positive tests seems to be rather steeper than the increase in testing. This would indicate positivity rates are increasing.

Denmark, currently undergoing an Omicron surge, has good surveillance for reinfections, which is pretty telling:



The WHO, UKHSA, ECDC, SSI (in Denmark) and many other similar organisations think Omicron is an enormous threat. I'm interested to know why you appear to think differently, Shifty?

Ultimately, it is all academic with the way the data is pointing. It will undoubtedly be clear before Christmas as doubling times appear to be 1.5 to 2 days at present. Most of the infections become symptomatic would have occurred a couple of doublings ago. Say, 3 more doublings to become apparent by this time next week (assuming people start to behave more cautiously now rates are rocketing). We'll be on hundreds of thousands of cases a week by then though we won't have the testing capacity to count them all - and probably won't count them all if we don't start to count the reinfections.

 

If you use OurWorldInData, they already tap the same source and do the various comparisons for us so we don't need to crunch the numbers ourselves. Although I don't think they use specimen date cases. In fact I'm not quite sure what their numbers are as they values don't seem to align with any ONS source. But as a quick reference without too much error, this is definitely convenient:

https://ourworldindata.org/covid-cases

Yes, but it's not quite as simple as that. Firstly, testing data lags somewhat. Secondly, cases are mentioned on the day they are reported, but some are from previous days. You have to wait a few days to get Cases By Specimen date consolidated numbers. The news might report 54037 cases on December 11th, for example, but the real number of +ve cases sampled that day was 45814, with the daily numbers factoring results from previous days' sampling being published on the 11th. There's +/- 20% on Reported Date values versus Specimen Date values.

The accurate 'cases by specimen date' numbers are lagging considerably behind the current Omicron spike so we won't know quite how that's going for real. However, my point has nothing to do with Omicron and is about correct stats!

To be clear, I'm not drawing any conclusions. I'm just asking people use the right data to make arguments. Whether cases are going up or not, the data needs to be correct, and for the past fortnight it's not been. Whatever changes happen with Omicron need to be tracked accurately. The point about the true flat rate in the UK from 1st to 14th December is just to illustrate how misleading the wrong numbers are in the hopes it'll get people to consider the importance of correct data science and not just take whatever numbers are thrown around at face value.

What are you using for the 'cases up' number? And why are you counting increase to the 16th December without a valid test count for that day?? For all you know, testing doubled on the 16th. I don't understand how someone with the smarts you have and critical eye for 'dodgy data' that I've referenced in the past can end up using such misaligned datapoints. Surely if our limiting number for comparing cases with tests is the 15th December, we should be comparing cases %age increase to the week up to the 15th? Why have you not done this?

Why look at a curve when the data isn't properly aligned? Why not look at the actual plots of case rates using appropriate datapoints? And again, this isn't about what's happening right now, but your choice to quote case counts, and indeed Chris Whitty's choice to use case counts.

I'm not saying it's not an enormous threat. Where have I said it's not a problem? I'm asking you (and everyone else) why you are using case rates without considering the impact of increased testing? If Omicron is spreading quickly, it'll be apparent in the data. But no-one should be making claims it is and then using unrepresentative data, like "78,610 new cases,15% higher than the peak of the January wave," especially when they know how to use data correctly and know that these numbers aren't at all comparable because the testing is completely different between now and January.

The REAL comparison is that the highest number last winter by specimen date, instead of randomly spread out over several days, was 81,477 on 29th December. The number of tests performed that day was 344,775. That's a case positivity rate of 23.6%. Testing was of course far more targeted then on suspected case. The highest case count we have now is 76,946 for 13th December but we're still counting and that'll be higher in a couple days when we'll have a final figure. Tests performed that day were 1,307,252. Case positivity rate is 5.8%. Even factoring in the most egregious caveats one can think of, there's no way these figures are comparable, so why are they presented as useful data, either by you or Chris Whitty? Why isn't real, accurate, meaningful data used to the support the argument instead?

This isn't a discussion about the disease or course of action, but about correct process, which is absolutely fundamental to the selection of best courses of action. As someone who's be very thorough in identifying rogue data, why are you not similarly thorough about how case numbers are presented, or even selecting appropriate, comparable datapoints in your own example here?

 

I just dashed my reply out earlier and so took the 7 day averages from the government site. Lots of confounding factors, but no doubt in the overall signal. This should become clearer in just a few days. Thing is, I don't really see the need to delve down too deeply because it seems pretty clear that the figures are going to become ludicrously high, very quickly. It's then just a case of how that translates to hospitalisations and deaths. It has been moving so fast that it is easy to forget it is less than 3 weeks since the WHO gave this variant the Omicron tag. In those 3 weeks it has spread around the world and is heading to be the dominant variant in many countries very soon indeed. Less than a week, in fact. The fact it has moved so fast means that we still don't know just how high the risk of serious disease is in our population. The claims Omicron is 'mild' coming from some commentors about South Africa do tend to ignore the fact that they've had not far off twice as many deaths as the UK in previous waves from a smaller and younger population (using excess death data). If a lot of the most vulnerable are no longer there to become infected, it may skew the numbers somewhat alongside the prior immunity that exists. All but impossible to compare with different demographics and vaccination rates, I suspect, though somebody will be able to retrospectively calculate a nice mathematical model in a few months, by which time it won't be much use.

You'll soon be presented with lots of other 'dodgy' data to look at. This epidemiologist at the UKHSA is saying that testing capacity will be maxxed out by the middle of next week so absolute daily case numbers won't have much relevance to the real spread:



A good job we have reasonable seroprevalence surveys to fall back on, though I wonder if they are doing to have large enough samples to provide an accurate estimate of actual case rate!

 

They aren't date aligned though. Testing average stops at 9th December, meaning a four day's difference with 'Date Reported' numbers. Obviously if you are in a point of exponential growth, a week between sampling dates is going to give a massive delta, which is what your "Cases up 31.4% over the 7 days to 16th December. Number of tests up by 18.5% in the 7 days up to 15th December:" numbers are comparing. Comparing date aligned testing and cases, I already provided the data showing that all 'growth' for the first two weeks of December is just from testing. Everything after that is decidedly unclear.

Hence the need to be particular about number choice.

Just tried it manually, I see the rolling average for cases by specimen date to the '13th' actually seems to include all the data up to the 16th, so that up-tick you see in the curve of cases is artificially early.

 

It boggles my mind that some contries are relying SOLELY on mRNA vaccines. Even if the risk of an unknown long term adverse effect is low, do you really wanna put MOST of your population through it? All at once? But, hey, saying that is "irresponsible" because Vaccines=Good and they will "end this thing" because someone in a lab coat said that is "the science" on TV...

I understand the emergency use in a time of crysis, but it would at least be wise to have a good portion of your population use the atenuated-virus vaccines instead of rna, for the sake of long term populational robustness.

 

Yea thanks for pointing this out, I'll change my columns from published date to specimen date to see a better picture.
If I did that, it would likely remove the reporting seasonality from the results in this graph. But this is a raw new cases/tested graph, no other calculations were done except to convert it to a percentage. This was done to normalize the increased testing that was occurs.

This is UK positivity rates per day for 2021 up to the data that is available. Sadly, I failed to get Seaborn to make a proper x-axis tick, so I'm still working on that. But UK-COVID api is straight forward and really easy to use. I can release the notebook if you guys are interested in tracking covid yourself as opposed to relying on sites.

You have an interesting set of data about LFD tests and PCR tests, and LFD tests that were then reconfirmed by PCR tests. Lots of good stuff here. A lot of work was done here.

edit: goodness, you guys have it good. At least there is an invokable API. For where I live, I have to do a manual search and download the results manually before I can access them. So much faster to have an API.

edit: as per my error below, I caught a bug in the way the graphing tool was interpreting date (as a string as opposed to a date). Resolved, this is the updated graph.