This may be a repeat reply, something weird seems to be going on with interwebs..... To your point, I was assuming the population is not on board for 3x per year boosters, but I've been surprised by plenty of other things in the last 2 years, so who knows 🤷♂️
Thank you! I have no background in either statistics, science, or medicine yet the data collection and presentation upon which public policy and opinion seem to have been formed has seemed not to have been based on the right questions to frame it in any meaningful way, leading me to wonder if I'm just going mad, or living in an abjectly corrupt world. Even though I might have to revisit this multiple times to fully understand your work you restore some of my hope.
I'm looking forward to reading your update tonight or tomorrow. Already grateful that you are not only using your skills to bother delving under the hood of misrepresentation or stupidity but also sharing publicly what you see, I am now even more grateful you are going the extra mile of making this more accessible for the ill-equipped. Thank you.
I still struggle to believe that our freedoms have been squashed, and we're not even *allowed* to see underlying data that caused our freedoms to be squashed.
Yes. But even that which is revealed can be so easily shown to be based on deliberate distortions or outright lies that skew the whole picture, always in the direction to fit the agenda. A gross, "obvious" example is the definition of "vaccinated" and "unvaccinated" that quietly shifted without notice. We don't get to see the coding that shunts data into its various buckets which I bet would reveal a lot of cushioning and culling in the "desired" direction, or to explore those "buckets" themselves that categorize in ways so crude the distortions again seem to serve an agenda rather than truth. Even I, a person with no training, can see that the failure to account for the timelines that account for waning VE and the shifts of vaccine roll-out by age group creates big distortions. These are just the tip of the iceberg. Without access to raw data, it may be impossible to see the accrued flaws that lead to its even more flawed presentation, to put it kindly. My short stint creating databases was enough to show me the importance of HOW data is collected is vital to the picture it creates. I imagine all these variables can be a huge challenge for any statistician but it seems to me that the driving force behind it was never impartial.
A slightly positive VE should be normalized to the adverse effects to determine actual risk. The problem is unreliability of VAERS, but we probably have enough insight to draw risk conclusions. Ultimately everyone needs to make a go-nogo decision about the next shot. Individual risk calculation is highly variable, but on average seems to re moment against vaccination.
A more important omission is causes of individual risk. Overall stats show most people are unaffected or mildly affected, and we have a lot of indications of what conditions increase risk -- obesity, vitamin D, diabetes -- conditions that degrade immune health. The bureaucracy wants to sell vaccines, so have suppressed this information, but can probably be reasonably extrapolated from the data we do have.
Great analysis - Thanks for working through that! Do you think that part of the leveling off in death % levels since Week 31 or so could be the booster campaign (since I think you're right to recognize some short-term gains in reducing severe outcomes)?
Are these graphs total all-cause deaths, or covid deaths? We should look at total deaths, because cause of death might be misattributed. For example, Bartram notes that vaccinated people may take longer to die following positive PCR, and if they take more than 28 days they are not counted as covid. https://bartram.substack.com/p/are-deaths-in-the-vaccinated-delayed
These are COVID deaths. Was really focusing this one on why VE statistics may be misleading in short term. Will have to think about how adding all-cause could add another layer to this picture.
Another factor supporting use of all-cause is that covid vax could elevate non-covid deaths. So even if reduced risk from covid, net risk might be elevated over time. We are seeing dramatic heart-related deaths in professional athletes, presumably all vaxxed have this risk, perhaps not to the same degree.
There was a TV series 50 or so years ago about a doc with a terminal illness decided to go out with a lot of risky adventures. If someone with terminal illness falls off a mountain, was death due to the mountain or the illness? Most covid deaths were past their shelf life, and were probably goners anyway. Allocation of cause of death is really a question of perspective, not of fact.
When you reach a significant level of vaccination, the people who aren't getting vaccinated will contain a significant proportion of people who weren't vaccinated because it was medically inadvisable -- i.e. they are so old and frail we thought that even a mild reaction to the vaccine would likely kill them, or they are on strong immuno suppressant drugs, and cannot mount a challenge to the vaccine agent no matter how many times you jab them, so it's a waste of time and money to do so, or they are in the middle of treatment for some other disease and getting vaccinated with anything is considered a bad idea at this time. These people aren't all that numerous, but they are also not going away. And when they catch covid, they unsurprisingly die at rates much greater than rest of the population. But it is very disingenuous to use their deaths as an argument that we need to get that tiny percentage of unvaccinated people vaccinated.
The UK has been great in releasing its data, but there still are times where you wonder 'why aren't you tracking <this thing>'? What percentage of Brits have been told it is not medically advisable for to take the vaccine should be something that we can get a rough estimate of, but if that figure is somewhere, I haven't been able to find it.
Completely agree with that line of thinking and a previous post discussed evidence that the vaccinated are a healthier group overall. My general philosophy with my posts is to be as conservative as possible in my hypotheses and assertions. Since what I write can quickly get me labeled all sorts of things by the "mainstream", I don't want to give any ammo to critics that I might be exaggerating a point. Since I don't know how to properly account for the point you raise, I excluded it... would only make the possible implication in this post even stronger.
Nice work. One of the challenges with this type of analysis is the confounding factor of natural protection. For example, in any age cohort there will be vaxxed individuals who would have survived perfectly well without having been vaxxed.
What do you think of the situation in Britain? The fully vax rate is roughly 70% but the vaxed constitute roughly 80% of all deaths from Covid. It would seem that being vaxed make you somewhat more susceptible to dying of Covid.
This article seems to completely miss the point of what is in the data. There is no mention of whether these are Covid- or non-Covid-deaths. Non-Covid-deaths dominate the counts largely, and should not depend at all on the vaccination status (except for those people that are killed by the vaccin). The vaccination curve and death curve are shifted simply because there is a (differential) delay in reporting, and the population numbers are very rapidly changing.
Based on the context (discussing VE, etc) as well as saying "COVID deaths" at least half a dozen times, I thought I was clear. Nevertheless, I went back and added "COVID" before any mention of death where I thought it might be unclear. I have no way of knowing what the delay is in reporting, I'm working with the best data I know of.
Thanks for the clarification. Then, if you are speaking of COVID-deaths, it is quite clear where the delays are coming from: it takes a while between infection and death on average, so any COVID-death when reported shall be compared to the vaccinated numbers earlier on -- take your best bet for the delay, I'd use 2 weeks, but regardless you will see the trends to get much closer and implications will change. Another simple exercise would be looking at the non-cumulative numbers, weekly numbers of vaccinations and deaths. You will see two bell-shaped curves quite self-similar, again showing that all that is going-on is just a business of delays.
My friend, the whole point of the post was looking at the delay between vaccinations and covid deaths. My hypothesis is that the proper way to compare the curves is to off-set them by a period of time where the vaccines are expected to be pretty effective (probably ~4 months) plus a few weeks time between infection and death. So when we compare the % of folks dying today who are vaccinated, we should compare it to the % of folks who were vaccinated x weeks ago, not % of folks vaccinated as of today.
Well, too bad for your hypothesis then that the offset between the curves is clearly <1month. Nothing to learn about VE here. Just check the differential plots, which will also get rid of those arbitrary rescaling you have to apply to compare the cumulative curves.
I'm really not sure what post you are reading. The horizontal distance between the curves changes over time, but I very clearly show the distance is 18-22 weeks when looking at COVID deaths in week 38. Re-scaling? No idea what you mean. The horizontal axis is weeks and the vertical axes are synchronized.
Just plot the differential distributions and check the spacing between the peaks. Taking offsets from cumulative distributions makes no sense, it's entirely dependent on tail effects.
Imagine Netflix releases a new episode of the Crown, and further imagine the graph of it's viewing adoption over time as a percentage of the population is the same as the vaccinated line. Then, in any age group graph, replace the vaccinated line with the Crown line. What does the graph show? It shows a correlation between watching the Crown and dying of Covid, but no causation.
What is needed is a randomized double blind placebo controlled study of test groups with and without the vaccine. If those two lines cross you have determined how long vaccine effectiveness lasts. (Perhaps one reason for the rush to vaccinate everyone is to make it more difficult to find unvaxxed candidates for such a study.)
But what I am saying is something different. You are saying correlation does not prove causation, 100% agree. The argument here is more of a "without correlation, there is not causation". This is not necessarily true (the correlation could be hidden by other confounders), but in this case, I think that's pretty strong. My point above is that a possible time shift between vax & covid death might remove nearly all (negative) correlation, which would imply in the long term there is no causation of vaccines preventing COVID deaths.
Seems to support the outcome of that large Swedish study. Given that I would predict VE to continue to worsen into the negative realm by 8 months. Will be interesting to see if that happens on your graphs.
Nice work. One of the challenges with this type of analysis is the confounding factor of natural protection. For example, in any age cohort there will be vaxxed individuals who would have survived perfectly well without having been vaxxed.
Steady-state? If the boosters are coming every 6 months, then 3, then omicron vax, then booster for omicron… What steady-state????
This may be a repeat reply, something weird seems to be going on with interwebs..... To your point, I was assuming the population is not on board for 3x per year boosters, but I've been surprised by plenty of other things in the last 2 years, so who knows 🤷♂️
Oh. I am pretty sure they will be on board for whatever.
Odd months - boosters for the current vaccine.
Even months - boosters for omicron.
And in between, whatever, pills against myocarditis.
and turning away calls from life insurance salesmen
Thank you! I have no background in either statistics, science, or medicine yet the data collection and presentation upon which public policy and opinion seem to have been formed has seemed not to have been based on the right questions to frame it in any meaningful way, leading me to wonder if I'm just going mad, or living in an abjectly corrupt world. Even though I might have to revisit this multiple times to fully understand your work you restore some of my hope.
That's super kind of you to say. You also inspired me to try to make the idea for this post a bit more clear. See update 👆
I'm looking forward to reading your update tonight or tomorrow. Already grateful that you are not only using your skills to bother delving under the hood of misrepresentation or stupidity but also sharing publicly what you see, I am now even more grateful you are going the extra mile of making this more accessible for the ill-equipped. Thank you.
I still struggle to believe that our freedoms have been squashed, and we're not even *allowed* to see underlying data that caused our freedoms to be squashed.
Yes. But even that which is revealed can be so easily shown to be based on deliberate distortions or outright lies that skew the whole picture, always in the direction to fit the agenda. A gross, "obvious" example is the definition of "vaccinated" and "unvaccinated" that quietly shifted without notice. We don't get to see the coding that shunts data into its various buckets which I bet would reveal a lot of cushioning and culling in the "desired" direction, or to explore those "buckets" themselves that categorize in ways so crude the distortions again seem to serve an agenda rather than truth. Even I, a person with no training, can see that the failure to account for the timelines that account for waning VE and the shifts of vaccine roll-out by age group creates big distortions. These are just the tip of the iceberg. Without access to raw data, it may be impossible to see the accrued flaws that lead to its even more flawed presentation, to put it kindly. My short stint creating databases was enough to show me the importance of HOW data is collected is vital to the picture it creates. I imagine all these variables can be a huge challenge for any statistician but it seems to me that the driving force behind it was never impartial.
A slightly positive VE should be normalized to the adverse effects to determine actual risk. The problem is unreliability of VAERS, but we probably have enough insight to draw risk conclusions. Ultimately everyone needs to make a go-nogo decision about the next shot. Individual risk calculation is highly variable, but on average seems to re moment against vaccination.
A more important omission is causes of individual risk. Overall stats show most people are unaffected or mildly affected, and we have a lot of indications of what conditions increase risk -- obesity, vitamin D, diabetes -- conditions that degrade immune health. The bureaucracy wants to sell vaccines, so have suppressed this information, but can probably be reasonably extrapolated from the data we do have.
Great analysis - Thanks for working through that! Do you think that part of the leveling off in death % levels since Week 31 or so could be the booster campaign (since I think you're right to recognize some short-term gains in reducing severe outcomes)?
Definitely possible. Not sure I have the data (or idea) on how to analyze the impact.
Are these graphs total all-cause deaths, or covid deaths? We should look at total deaths, because cause of death might be misattributed. For example, Bartram notes that vaccinated people may take longer to die following positive PCR, and if they take more than 28 days they are not counted as covid. https://bartram.substack.com/p/are-deaths-in-the-vaccinated-delayed
These are COVID deaths. Was really focusing this one on why VE statistics may be misleading in short term. Will have to think about how adding all-cause could add another layer to this picture.
Another factor supporting use of all-cause is that covid vax could elevate non-covid deaths. So even if reduced risk from covid, net risk might be elevated over time. We are seeing dramatic heart-related deaths in professional athletes, presumably all vaxxed have this risk, perhaps not to the same degree.
There was a TV series 50 or so years ago about a doc with a terminal illness decided to go out with a lot of risky adventures. If someone with terminal illness falls off a mountain, was death due to the mountain or the illness? Most covid deaths were past their shelf life, and were probably goners anyway. Allocation of cause of death is really a question of perspective, not of fact.
Also, I don't want to be negative -- I very much appreciate your analyses. Your idea of looking at the vax/death curves is excellent!
When you reach a significant level of vaccination, the people who aren't getting vaccinated will contain a significant proportion of people who weren't vaccinated because it was medically inadvisable -- i.e. they are so old and frail we thought that even a mild reaction to the vaccine would likely kill them, or they are on strong immuno suppressant drugs, and cannot mount a challenge to the vaccine agent no matter how many times you jab them, so it's a waste of time and money to do so, or they are in the middle of treatment for some other disease and getting vaccinated with anything is considered a bad idea at this time. These people aren't all that numerous, but they are also not going away. And when they catch covid, they unsurprisingly die at rates much greater than rest of the population. But it is very disingenuous to use their deaths as an argument that we need to get that tiny percentage of unvaccinated people vaccinated.
The UK has been great in releasing its data, but there still are times where you wonder 'why aren't you tracking <this thing>'? What percentage of Brits have been told it is not medically advisable for to take the vaccine should be something that we can get a rough estimate of, but if that figure is somewhere, I haven't been able to find it.
Completely agree with that line of thinking and a previous post discussed evidence that the vaccinated are a healthier group overall. My general philosophy with my posts is to be as conservative as possible in my hypotheses and assertions. Since what I write can quickly get me labeled all sorts of things by the "mainstream", I don't want to give any ammo to critics that I might be exaggerating a point. Since I don't know how to properly account for the point you raise, I excluded it... would only make the possible implication in this post even stronger.
Nice work. One of the challenges with this type of analysis is the confounding factor of natural protection. For example, in any age cohort there will be vaxxed individuals who would have survived perfectly well without having been vaxxed.
Agree. Do you know of any data pointing the prevalence of previous natural infection among the vax vs. non-vax?
What do you think of the situation in Britain? The fully vax rate is roughly 70% but the vaxed constitute roughly 80% of all deaths from Covid. It would seem that being vaxed make you somewhat more susceptible to dying of Covid.
This article seems to completely miss the point of what is in the data. There is no mention of whether these are Covid- or non-Covid-deaths. Non-Covid-deaths dominate the counts largely, and should not depend at all on the vaccination status (except for those people that are killed by the vaccin). The vaccination curve and death curve are shifted simply because there is a (differential) delay in reporting, and the population numbers are very rapidly changing.
Based on the context (discussing VE, etc) as well as saying "COVID deaths" at least half a dozen times, I thought I was clear. Nevertheless, I went back and added "COVID" before any mention of death where I thought it might be unclear. I have no way of knowing what the delay is in reporting, I'm working with the best data I know of.
Thanks for the clarification. Then, if you are speaking of COVID-deaths, it is quite clear where the delays are coming from: it takes a while between infection and death on average, so any COVID-death when reported shall be compared to the vaccinated numbers earlier on -- take your best bet for the delay, I'd use 2 weeks, but regardless you will see the trends to get much closer and implications will change. Another simple exercise would be looking at the non-cumulative numbers, weekly numbers of vaccinations and deaths. You will see two bell-shaped curves quite self-similar, again showing that all that is going-on is just a business of delays.
My friend, the whole point of the post was looking at the delay between vaccinations and covid deaths. My hypothesis is that the proper way to compare the curves is to off-set them by a period of time where the vaccines are expected to be pretty effective (probably ~4 months) plus a few weeks time between infection and death. So when we compare the % of folks dying today who are vaccinated, we should compare it to the % of folks who were vaccinated x weeks ago, not % of folks vaccinated as of today.
Well, too bad for your hypothesis then that the offset between the curves is clearly <1month. Nothing to learn about VE here. Just check the differential plots, which will also get rid of those arbitrary rescaling you have to apply to compare the cumulative curves.
I'm really not sure what post you are reading. The horizontal distance between the curves changes over time, but I very clearly show the distance is 18-22 weeks when looking at COVID deaths in week 38. Re-scaling? No idea what you mean. The horizontal axis is weeks and the vertical axes are synchronized.
Just plot the differential distributions and check the spacing between the peaks. Taking offsets from cumulative distributions makes no sense, it's entirely dependent on tail effects.
Imagine Netflix releases a new episode of the Crown, and further imagine the graph of it's viewing adoption over time as a percentage of the population is the same as the vaccinated line. Then, in any age group graph, replace the vaccinated line with the Crown line. What does the graph show? It shows a correlation between watching the Crown and dying of Covid, but no causation.
What is needed is a randomized double blind placebo controlled study of test groups with and without the vaccine. If those two lines cross you have determined how long vaccine effectiveness lasts. (Perhaps one reason for the rush to vaccinate everyone is to make it more difficult to find unvaxxed candidates for such a study.)
But what I am saying is something different. You are saying correlation does not prove causation, 100% agree. The argument here is more of a "without correlation, there is not causation". This is not necessarily true (the correlation could be hidden by other confounders), but in this case, I think that's pretty strong. My point above is that a possible time shift between vax & covid death might remove nearly all (negative) correlation, which would imply in the long term there is no causation of vaccines preventing COVID deaths.
Thanks. Is it accurate to rephrase your conclusion that after some number of months vaccine effectiveness might be dropping to zero?
Seems to support the outcome of that large Swedish study. Given that I would predict VE to continue to worsen into the negative realm by 8 months. Will be interesting to see if that happens on your graphs.
Nice work. One of the challenges with this type of analysis is the confounding factor of natural protection. For example, in any age cohort there will be vaxxed individuals who would have survived perfectly well without having been vaxxed.