Okay a few starting notes as has been my want recently.
Between the Russian and Ukrainian Embargos on wheat shipments and Russia and Belarus now also refusing to ship Potash (they are the top two providers in the world), making sure you have backup food supplies just went from important to literally life or death. You thought the Toilet Paper crisis and Gas Crisis in the US were bad (you know when the pipeline was Ransomware’d and people were putting gasoline in grocery bags to take home?), wait until the thing you can’t get at the store is food. People will be killed in those riots. At a bare minimum, you need to be able to stay home and away from stores for a few weeks at a time.
Tracking COVID data is still vital to ensure we can crush any attempt to return to mandates and lockdowns.
I’m having conversations with the UK HSA about Immortal Time Bias in their data. No idea if it’s going to lead anywhere, but they have been very polite and fairly responsive so far. It’s generally a 2 or 3 day lag between responses, but I’ll let you know if anything substantive comes out of it.
I’ve been playing with visualization and helping another charter on Twitter who is mired in an entirely manual process. It has both distracted me a little but also helped me to really laser in on some good charts to visualize what is going on.
So something is off with either my current 40-49 calculations or my old ones. I’m not certain. I know I updated my vaccination population chart, but things don’t line up with previous weeks. I’m still looking into this, but will post the data as I have it.
All Data From:
National Flu and COVID Report (Vaccine Uptake Table)
UKHSA Vaccine Surveillance Reports (I have data going back to report 36, the first with a table of rates)
Okay, so I’m changing how the 2 Dose/3 Dose plots look. 2 new charts. The first is 3 weeks of JUST the delta of 3 Dose Case Rate - 2 Dose Case Rate. This is where I spotted the change in 40-49. It was clearly increasing before, but now it shows decreasing for the last three weeks. In part of this investigation I realized that they are using the vaccination populations from 2 weeks back (which is kind of a poor mans average of the last 4 weeks of data) and I’ve been using the current week. I fixed that for these charts.
So aside from 40-49 declining which is an interesting signal given it’s the highest delta, you can see that 70-79 actually flipped over 2 weeks ago, not this week. One could argue that the more recent 3 Dose rate would be more accurate given the 2 week lag period, but since those cases are being counted as 2 Dose cases, it actually gives a benefit to the boosted category. There is something off about the 40-49 rate but I’m not sure if it’s something I fixed this week or something that I broke. The rates seem plausible and my numbers match the UK data, but this declining delta is not what I expected.
Here it is easy to see that the 80+ rate delta is moving up fairly quickly. I’m still not certain that it will cross over, there is a fair way to go.
Second chart is my attempt to show how crazy this swing has been. I can plot from week 2 to week 10 because I snagged that early mistaken release and saved off the original data (which turned out to be a good match for the 2 Dose ONLY column. The Mortality seems a little suspect and low, but cases is right in line with the following weeks.
So I’m bringing back the Case Rate Delta plots of old. But instead of Unvaccinated - 2 Doses, this is Boosted - Unvaccinated. The swing in case rates here is truly remarkable. The shift in Under 18s is basically non-existent, because Under 18s aren’t actually boosted very much at all. But it’s interesting to see that 80+ got worse then better right around weeks 3 and 4. On this chart, negative numbers show boosters working better than 2 Dose, but positive numbers show the opposite.
But is that what they show? Because notice that there is a VERY tightly grouped set of deltas for 18-49 with 50-59 closer to that grouping than the 60+ grouping (one could argue it’s really a 70+ grouping with 60-69 close to it) . We can ignore Under 18s here pretty much.
Okay so why does that matter? Because of this.
Those peaks are in absolute numbers of shots given for each week. Notice how tightly coupled 18-49 peaks are in time (even as 18-29 is lower) and then we see those same three 2 Dose rates so much higher than the boosted rates and tightly coupled. (The very negative rate difference of 12,000!) Those same three ages are very tightly coupled in time on the rate delta plot.
Then the 50-59 (the next most recent peak), followed by 60-69 and their rates are commensurate in the data.
It would appear that 70-79 and 80+ are no longer showing similar effects in this data. I do not know if it is possible to determine whether the cause is that there is enough time since they were dosed, or whether age related immunological factors are the primary drivers here. It is possible that 80+ is simply so immunologically inferior to even 70-79 year-olds that they do not respond in the same way.
This to me a a very strong indication that boosting is correlated with elevated 2 Dose case rates, which it should not be unless it increases the risk of contracting COVID-19 given enough prevalence in the community (there wasn’t much when the 70+ cohorts were being boosted) and much of that risk is in the first 2 weeks after boosting. I might write in a bit more detail about this later because I believe there is more to it than just raw booster counts.
But I still think this is very strong evidence due to the unbelievably rapid swing in rates. The first thing that came to my mind was the Denominator Scaling effect. With a significantly smaller denominator, a single case change has a larger effect and so over time a drop in cases changes the rates by much more, but the biggest swings are in the groups with the largest populations still in the 2 Dose cohort. Meanwhile, the case rate spikes immediately follow the administration of very large numbers of booster doses.
I played around with some different ways to normalize my standard charts and have settled on normalizing by boosted rates. The major benefit here is we can see what the multipliers are. How many case/admissions/deaths in the Unvaccinated vs the Boosted for instance. I know the US likes to push very large numbers. What does the UK Data show?
I mean we already knew this though right. Of course we really only knew the 2 Dose part from above. But yes, very low rates continue in the “Unvaccinated” as well. Non-normalized the rates look like this
This is very much a continuation of the prior trend. All rates falling, but Unvaccinated and 2 Dose rates falling faster.
Admissions is where it gets really interesting. For the first time ever the Unvaccinated admissions rate in a young age group has dipped below the boosted rate. The 18-29 year-olds are just below boosted. More importantly, the MAX “Unvaccinated” ratio to boosted sits at about 2.3X this week. That’s quite different than the CDC tells it.
It’s too early to tell if this trend is continuing, but 30-39 both Unvaccinated and 2 Dose vaccinated look like they might be on the way as does the 18-29 2 Dose cohort. This is definitely a thing to keep our eyes on.
Interestingly, this chart is NOT normalized, but the boosted rates nearly look normalized. This might be a first sign of the spring mini-peak in the seasonal trend of Covid-19.
Here we continue to see the highest relative rate is the 2 Dose 70-79 cohort. Interestingly, that group doesn’t have the highest Unvaccinated multiplier, 40-59 is where the peak is there with 40-49 mortality rates increasing into this week quite significantly. It’s counter trend to everyone else and I’m very confused by it, but it does match the data in the report. It simply appears there was an unusually low number of 40-49 year-old booster deaths. This is a good thing, but it is kind of a data point that is out of line with the rest of the data. We shall see if it corrects.
Honestly, it isn’t even a blip on the chart when scaled to the 80+ rates,
The crude CFR charts still tell the same story of efficacy against bad outcomes if you do happen to get sick.
This brings me to the pie charts I played with this week. I don’t normally play in raw numbers, but if you never play in raw numbers, you miss out on seeing the forest for the trees a little bit.
Notice that cases this month are pretty similar from 0-49. Admissions are kind of consistent as well though Under 18 has a higher share of admissions than expected. If I were forced to guess, it would be more caution especially with very young children. It isn’t translating into bad outcomes though because you can see the percentage of mortality is very low and grows only slowly until 50+ where it expands rapidly. In fact 64% of all mortality is coming from the 80+ population and just 3.4% of cases. Hospitalizations really take off in the 70s which is again, likely due to much more caution (this time very warranted by outcomes.
I really like this view as a sanity check on what really matters here. I’m not looking at vax status or anything like that. I did work with someone looking over those statistics. While he very reasonably wanted to make sure every chart he put in was one he’d touched himself to be sure the data was right, I helped him run through a bunch of data to see whether it told the story we thought it did. (It took me long enough to write this that he got his Substack post out before I finished)
Take a look. He has some really good twitter work on the NYC Vaccine Record Matching (VRM) and how it might be affecting the CDC numbers.
Hope this has been helpful and if you got to here, let me know what you think about the pie chart dashboard!
So. Food shortages are likely incoming..... It's almost as though they knew, or have manufactured the crisis to suit some sort of agenda.
I think when I fully grasp the circle style charts, I will love them. Still a change mentally now. Love your work so far, glad I stopped by.