After updating my dashboard with the most recent two UK Vaccine Surveillance Reports (covering weeks 2-5 of 2022), I played around to look at the data in a new way.
Amazing analysis to so clearly represent all that data in one graph. It would be interesting to reduce each age/vax cohort to a probability. A review of my age group seems to indicate increasing risk with more jabs. I wonder if that's consistent across all age groups.
This really helpful and many thanks! Is there any statement you can make about the 29% vax status for the <70 or working group where the USA insurance companies are seeing such a dramatic rise in mortality? We cannot trust USA numbers or classification. Thank you again!
I can only speak in general terms... it can't be the case that the data from the UK, Israel, and the US can all be taken at face value, they are just too different and human biology is not. Personally, I put far more faith in the UK & Israel data with I believe good reason. This is why I spend so much time with the UK data despite being an American. I don't spend much with Israel data even though I believe its of general higher quality because I don't know Hebrew.
I spent a lot of time this week trying to wrap my head around raw incident rates in dynamic populations and why no one is calling out basic incidence/population * 100k stats and comparisons as being unreliable this far in to a pandemic.
You have movement between vaccinated/unvaccinated groups over time and worse; an unknown distribution of previously infected that are likely protected against severity from omicron (and infection pre-omicron). For the first one I discovered using person-time but it doesn't solve the second problem.
This wasn't a big deal initially but now that the UK for instance has 27% (18m people) of the population in previous known infections an uneven distribution of those people has the power to severely skew raw population adjusted rates.
Even without uneven distribution (and with no population movement), you would expect to see smoothed incidence rates decrease because you are still dividing by the entire population cohort even though a larger and larger portion of them have/had some kind of immunity.
As an extreme for the last UK vax surveillance report. If you subtracted all ~18m prev infections from the vaccinated population and do the raw rate for total vaccinated deaths, it goes from 5 per 100k to 10 per 100k.
If you do the same for the unvaccinated side, you are only left with ~1.2 million people and the death rate goes from ~5 per 100k to 55 per 100k.
I'm not really a numbers guy so maybe I am totally wrong here but it seems to me that at best comparisons of crude rates probably stopped being a reliable metric a long time ago. The "correct" denominator is unknowable.
I never realized you made a UK dashboard, it's super.
If Omicron is mild, severe / deaths efficacy should go out of the window. Just like you wouldn't see different rates of burglary for gated and ungated houses in a city where the horde of violent murderous burglars all suddenly took up jobs as cleaners.
Of course, we've been measuring hospitalization / deaths from burglaries by "hospitalizations or deaths after presence of burglar genes in house." So the stats will still rack up as people who hire cleaners (now ex-burglars) occasionally go to the hospital / die for normal reasons.
If basically all the gated and ungated houses regularly hire cleaners (now ex-burglars), the gated houses will look like the burglars "broke the protection of the gates" and the ungated houses will look like they are having just as many burglaries as before, because they *were* being beaten up and killed by the burglars back when they had no gates.
So we should expect the rise in "with not from" deaths in the unvaccinated to be disguised by the cessation of "from" deaths. That seems to account for why there's no Omicron deaths wave for the unvaxxed in the raw counts.
The only wrinkle here is that the boosters might still be providing a bit of infection efficacy vs Omicron. But probably not at all for the 70+ groups, since they are ~14 weeks post 3rd dose in the week 2-5 snapshot.
There was a pretty significant increase in the raw number of deaths in the last 4 reports. Without any good reporting of from/with, I don't know how we determine if there are many people dying from COVID or if the numbers simply must increase as a by-product of the widespread "cases" of Omicron.
There's a raw death increase in the 60-69 sets but it doesn't scale with the case increases (the per-100k plots can safely be used for this group to measure longitudinal changes), which is not what you would see if the deaths were primarily "with" false positives before Omicron - the "with" deaths should go up with cases. Even more so for the 40-59 sets, where absolute counts are pretty flat, though here things are a bit shiftier since there's actually been a recent reduction in unvaxxed rates in these ages.
But my point is just as much about how, despite milder Omicron, the hospitalization / death rates don't *drop* for this group as why they don't surge - there's an increase in "with" and a decrease in "from" at the same time.
70+ unvaxxed seem to show more movement but now we're in the territory where regular flu can kill, so the mildness discount is not as great. You get "withs" and "froms" here. Even for the vaxxed. If any severe efficacy shows in this group vs Omicron it's the infection efficacy wrinkle and healthy user bias, just like with the flu vaccines.
It's really the middle aged where the severe efficacy was showing the most plausible effect in 2021. I think this will totally disappear absent a comeback of the "Wuhan" lineage.
My difficulty in reading your charts is red/green color blindness. Anything that is red or green with the same color value takes me a long time to read, sometimes I just give up because it is too difficult to tell color markers apart. I can read the last chart better than the original. My suggestions, for example, if you changed the the outline color of the unvaxxed symbols to black it would be most helpful.
Thank you for demonstrating the effect of the vaccines on Covid-19 mortality. You have demonstrated so clearly that the booster has a significant effect on preventing death
If I understand your analysis, in the +80 age groups 3-vaccianted are 4.3% of the population and 38% of all deaths, while 2-dosed are only 0.2% of the population, but 15.5% of the deaths, that would calculate to 3-dosed have an 8-times higher chance of dying that 2-dosed! How can 4.4% of the over-80 population be 29% of the deaths?
You have also shown this with the 70-79 age group: Boosted are 7.7% of the population, and 11.6% of deaths, while vaccinated are 0.3% of the population and 6.3% of deaths. So the booster reduces deaths by >90%! How can 3.75% of the 70-79 population be 35% of the deaths?
Thank you for your analysis. You have shown boosting all +70 year-olds, almost 20% of the deaths could have been prevented.
Let me go ahead and complete your analysis. In the 80+ group, 0.2% of the population is not vaccinated and are 6.3% of all deaths, so the double-dosed are >2x higher chance of dying than the unvaccinated! For the 70-79, they are 0.4% of population and 3.8% of deaths -> the double-dosed are again >2x more likely to die than the unvaccinated!
So, I guess it would be a tough call for people over 70.... do you get vaccinated and double your risk of dying until your 3rd dose?
Perhaps there is more to the story with regard to interpreting rates when people are moving from one cohort to another and we are using arbitrary definitions (e.g. still considered double-dosed for 2 weeks past receiving a 3rd dose): https://boriquagato.substack.com/p/new-data-on-vaccine-efficacy-from?r=tv61s
If I understand now, it's 0.2%/6.3% Un-vax, 0.2%15.5% vax and 4.3/38% boosted, for the +80? Unvaccinated have a 3.5-times the chance of death as boosted, and vaccinated have 8.8 times the death rate.
Whose data are you using - NHS or ONS? As I'm sure you have seen (and was explained in this report), there is some question regarding the reliability of the UK population size, and that this can influence the number of unvaccinated people. The NHS know the total number of people vaccinated, so they have to calculate the unvaccinated. As the number of unvaccinated is small, any small change can influence the denominator, and can influence the calculations (especially when using the NHS data, which overestimates the total population, and therefore the number of unvaccinated).
The source of the data is the UKHSA Weekly Surveillance reports (as linked to in the post). There is a reason my post does not focus on rates (that was your introduction)... because they are hard to present in a fair way due to imprecise or conflicting population figures, small denominators, strange definitions (e.g. 2 week window post jabs), and differences between the vaxxed & non-vaxxed populations (non-vaxxed tend less healthy overall, poorer & likely more frail at older ages).
This post was primarily a neutral display of the UK statistics. My only editorial is to say that it seems we continue to act as though vaccinating more people (and children) will change the course of things. It clearly will not... virtually all old people are already vaxxed, nevertheless those are the people who are dying. Forcing vaccinations on more young people through mandates will quite obviously have no effect on the big picture.
Well, the most recent 4 weeks had ~6K total COVID+ deaths. 7 of those were in the unvaccinated under 18 group. Have to remember that "cases" in the under 18 are through the roof. 567K new cases among that group during that time, so it's almost certainly the case that most or all of those 7 deaths were "with" not "from" COVID.
Is this since the beginning of the pandemic, or this year? If it's all time, I'm confused about unvaxxed deaths, because all the initial deaths were unvaxxed.
Either way... the oldest people continue to be the ones most likely to die, regardless of vaccination status? Assuming that most people 70 and up were at least double vaxxed (is that a valid assumption for the UK?) does this mean the shots did nothing?
"Tallying up the above, the double & triple-dosed 70+ population is 12.5% of the population & makes up 71.3% of all COVID related deaths."
This is for the 4 most recently reported weeks, Weeks 2-5 of 2022. 95% of the 70+ population has at least 2 doses. Given all the statistical hijinks, it's hard to say how much (if any) effect vaccinations are having on deaths. What is clear, however, is that the overwhelming majority of deaths remain among the vaccinated elderly. Trying to force vaccinations on younger people (or in the US, getting the vaccine approved for 2-5 year olds) will have nearly zero impact on the overall picture.
Amazing analysis to so clearly represent all that data in one graph. It would be interesting to reduce each age/vax cohort to a probability. A review of my age group seems to indicate increasing risk with more jabs. I wonder if that's consistent across all age groups.
This really helpful and many thanks! Is there any statement you can make about the 29% vax status for the <70 or working group where the USA insurance companies are seeing such a dramatic rise in mortality? We cannot trust USA numbers or classification. Thank you again!
I can only speak in general terms... it can't be the case that the data from the UK, Israel, and the US can all be taken at face value, they are just too different and human biology is not. Personally, I put far more faith in the UK & Israel data with I believe good reason. This is why I spend so much time with the UK data despite being an American. I don't spend much with Israel data even though I believe its of general higher quality because I don't know Hebrew.
Huh, I thought the jab was supposed to stop death by covid. Of course, it would have been so much worse...
They would have died more painfully, rather than alone and intubated in … uh … 🤔
Never mind.
I spent a lot of time this week trying to wrap my head around raw incident rates in dynamic populations and why no one is calling out basic incidence/population * 100k stats and comparisons as being unreliable this far in to a pandemic.
You have movement between vaccinated/unvaccinated groups over time and worse; an unknown distribution of previously infected that are likely protected against severity from omicron (and infection pre-omicron). For the first one I discovered using person-time but it doesn't solve the second problem.
This wasn't a big deal initially but now that the UK for instance has 27% (18m people) of the population in previous known infections an uneven distribution of those people has the power to severely skew raw population adjusted rates.
Even without uneven distribution (and with no population movement), you would expect to see smoothed incidence rates decrease because you are still dividing by the entire population cohort even though a larger and larger portion of them have/had some kind of immunity.
As an extreme for the last UK vax surveillance report. If you subtracted all ~18m prev infections from the vaccinated population and do the raw rate for total vaccinated deaths, it goes from 5 per 100k to 10 per 100k.
If you do the same for the unvaccinated side, you are only left with ~1.2 million people and the death rate goes from ~5 per 100k to 55 per 100k.
I'm not really a numbers guy so maybe I am totally wrong here but it seems to me that at best comparisons of crude rates probably stopped being a reliable metric a long time ago. The "correct" denominator is unknowable.
Yes. Part of the issue discussed here: https://boriquagato.substack.com/p/new-data-on-vaccine-efficacy-from?r=tv61s
I never realized you made a UK dashboard, it's super.
If Omicron is mild, severe / deaths efficacy should go out of the window. Just like you wouldn't see different rates of burglary for gated and ungated houses in a city where the horde of violent murderous burglars all suddenly took up jobs as cleaners.
Of course, we've been measuring hospitalization / deaths from burglaries by "hospitalizations or deaths after presence of burglar genes in house." So the stats will still rack up as people who hire cleaners (now ex-burglars) occasionally go to the hospital / die for normal reasons.
If basically all the gated and ungated houses regularly hire cleaners (now ex-burglars), the gated houses will look like the burglars "broke the protection of the gates" and the ungated houses will look like they are having just as many burglaries as before, because they *were* being beaten up and killed by the burglars back when they had no gates.
So we should expect the rise in "with not from" deaths in the unvaccinated to be disguised by the cessation of "from" deaths. That seems to account for why there's no Omicron deaths wave for the unvaxxed in the raw counts.
The only wrinkle here is that the boosters might still be providing a bit of infection efficacy vs Omicron. But probably not at all for the 70+ groups, since they are ~14 weeks post 3rd dose in the week 2-5 snapshot.
There was a pretty significant increase in the raw number of deaths in the last 4 reports. Without any good reporting of from/with, I don't know how we determine if there are many people dying from COVID or if the numbers simply must increase as a by-product of the widespread "cases" of Omicron.
There's a raw death increase in the 60-69 sets but it doesn't scale with the case increases (the per-100k plots can safely be used for this group to measure longitudinal changes), which is not what you would see if the deaths were primarily "with" false positives before Omicron - the "with" deaths should go up with cases. Even more so for the 40-59 sets, where absolute counts are pretty flat, though here things are a bit shiftier since there's actually been a recent reduction in unvaxxed rates in these ages.
But my point is just as much about how, despite milder Omicron, the hospitalization / death rates don't *drop* for this group as why they don't surge - there's an increase in "with" and a decrease in "from" at the same time.
70+ unvaxxed seem to show more movement but now we're in the territory where regular flu can kill, so the mildness discount is not as great. You get "withs" and "froms" here. Even for the vaxxed. If any severe efficacy shows in this group vs Omicron it's the infection efficacy wrinkle and healthy user bias, just like with the flu vaccines.
It's really the middle aged where the severe efficacy was showing the most plausible effect in 2021. I think this will totally disappear absent a comeback of the "Wuhan" lineage.
My difficulty in reading your charts is red/green color blindness. Anything that is red or green with the same color value takes me a long time to read, sometimes I just give up because it is too difficult to tell color markers apart. I can read the last chart better than the original. My suggestions, for example, if you changed the the outline color of the unvaxxed symbols to black it would be most helpful.
Sorry about that George. I have updated the post by adding at the end, the chart split into 3, one for each vaccination cohort. Hope this helps.
Thank you for demonstrating the effect of the vaccines on Covid-19 mortality. You have demonstrated so clearly that the booster has a significant effect on preventing death
If I understand your analysis, in the +80 age groups 3-vaccianted are 4.3% of the population and 38% of all deaths, while 2-dosed are only 0.2% of the population, but 15.5% of the deaths, that would calculate to 3-dosed have an 8-times higher chance of dying that 2-dosed! How can 4.4% of the over-80 population be 29% of the deaths?
You have also shown this with the 70-79 age group: Boosted are 7.7% of the population, and 11.6% of deaths, while vaccinated are 0.3% of the population and 6.3% of deaths. So the booster reduces deaths by >90%! How can 3.75% of the 70-79 population be 35% of the deaths?
Thank you for your analysis. You have shown boosting all +70 year-olds, almost 20% of the deaths could have been prevented.
You're most welcome.
Let me go ahead and complete your analysis. In the 80+ group, 0.2% of the population is not vaccinated and are 6.3% of all deaths, so the double-dosed are >2x higher chance of dying than the unvaccinated! For the 70-79, they are 0.4% of population and 3.8% of deaths -> the double-dosed are again >2x more likely to die than the unvaccinated!
So, I guess it would be a tough call for people over 70.... do you get vaccinated and double your risk of dying until your 3rd dose?
Perhaps there is more to the story with regard to interpreting rates when people are moving from one cohort to another and we are using arbitrary definitions (e.g. still considered double-dosed for 2 weeks past receiving a 3rd dose): https://boriquagato.substack.com/p/new-data-on-vaccine-efficacy-from?r=tv61s
Now I'm confused: the 0.2% relates to double-vaccinated or unvaccinated? Do you have a breakdown of deaths by age/vaccination status?
For the 80+ group, 0.2% of group is double-dosed & 0.2% is unvaccinated. Everything can be seen here: https://public.tableau.com/views/UKReportsRiskRatios/UKCharts?:language=en-US&publish=yes&:display_count=n&:origin=viz_share_link
Meaning 0.2% of UK population, not of that age group.
If I understand now, it's 0.2%/6.3% Un-vax, 0.2%15.5% vax and 4.3/38% boosted, for the +80? Unvaccinated have a 3.5-times the chance of death as boosted, and vaccinated have 8.8 times the death rate.
Whose data are you using - NHS or ONS? As I'm sure you have seen (and was explained in this report), there is some question regarding the reliability of the UK population size, and that this can influence the number of unvaccinated people. The NHS know the total number of people vaccinated, so they have to calculate the unvaccinated. As the number of unvaccinated is small, any small change can influence the denominator, and can influence the calculations (especially when using the NHS data, which overestimates the total population, and therefore the number of unvaccinated).
The source of the data is the UKHSA Weekly Surveillance reports (as linked to in the post). There is a reason my post does not focus on rates (that was your introduction)... because they are hard to present in a fair way due to imprecise or conflicting population figures, small denominators, strange definitions (e.g. 2 week window post jabs), and differences between the vaxxed & non-vaxxed populations (non-vaxxed tend less healthy overall, poorer & likely more frail at older ages).
This post was primarily a neutral display of the UK statistics. My only editorial is to say that it seems we continue to act as though vaccinating more people (and children) will change the course of things. It clearly will not... virtually all old people are already vaxxed, nevertheless those are the people who are dying. Forcing vaccinations on more young people through mandates will quite obviously have no effect on the big picture.
More saturated colors are easier to see, yes.
The double-jabbed are at much higher risk, obviously. I wonder if/when the tripple-jabbed will catch up.
I'm surprised that under-18 unvaccinated are even 0.1%. That's one in 1000 of the deaths. I'd have thought that it was even less than that.
Well, the most recent 4 weeks had ~6K total COVID+ deaths. 7 of those were in the unvaccinated under 18 group. Have to remember that "cases" in the under 18 are through the roof. 567K new cases among that group during that time, so it's almost certainly the case that most or all of those 7 deaths were "with" not "from" COVID.
The "white on white" makes it hard to see but very good info.
Thanks
Added a version with some different colors. Is that better?
Yes that is much more readable. Thanks
Is this since the beginning of the pandemic, or this year? If it's all time, I'm confused about unvaxxed deaths, because all the initial deaths were unvaxxed.
Either way... the oldest people continue to be the ones most likely to die, regardless of vaccination status? Assuming that most people 70 and up were at least double vaxxed (is that a valid assumption for the UK?) does this mean the shots did nothing?
"Tallying up the above, the double & triple-dosed 70+ population is 12.5% of the population & makes up 71.3% of all COVID related deaths."
This is for the 4 most recently reported weeks, Weeks 2-5 of 2022. 95% of the 70+ population has at least 2 doses. Given all the statistical hijinks, it's hard to say how much (if any) effect vaccinations are having on deaths. What is clear, however, is that the overwhelming majority of deaths remain among the vaccinated elderly. Trying to force vaccinations on younger people (or in the US, getting the vaccine approved for 2-5 year olds) will have nearly zero impact on the overall picture.
Thanks! That's very helpful.