The age range choices for the CDC datasets have confounded many analyses I've attempted to conduct, mostly revolving around age-adjusted CFR's & IFR's.
It's doubly frustrating because the CDC obviously HAS the data, and chooses not to provide it to the public in a form that we can verify their work with
In some cases, it's good enough to estimate. There have been times I've laid out a curve and interpolated. In this case, envelope math was clearly good enough.
But yes...there is no reason not to have open source public health data, except to allow for a handful of public health officials to have control over what gets published. I believe that if we ever get a good look behind the curtain, we're going to find that every aspect of the pandemic and medical data was rife with publication bias. Of course, you know that very specifically in your corner of the research.
We will never see that data, it has long been destroyed, purposely not collected or done in a way that it will never give clear answers. Imagine that in the first 12 months the US did not collect any data on who already had the virus nor did they test to confirm it properly.
When i got Alpha, I tested, then tested with a different lab, then did a government test as well. 3 tests said I had it. Then antibody test 3 months later and 8 months later. Same for my whole family. No one cares to know this yet we all know it provides some sort of immunity.
Small point - is it possible that the lockdowns impacted the all cause mortality in the age you noted? (Less car accidents, less homicide etc because of the decrease in driving and interaction)
Friends of mine live on a usually busy main road in our small city. During the lockdowns the traffic on their street was nearly zero. The result, however, was that the few drivers left on the road turned it into their own personal Indy 500. My friends said the noise (and the sirens) were unprecedented. They were relieved when normal traffic returned.
Very little impact on excess mortality, tho. Rates for the general population for suicide, murder, and accidents were within normal range comparing 2019-2021.
I will look at this site later, but I have doubts. Remember that 2021 totals are not yet nearly complete. There are larger numbers of R-codes deaths with causes that will not settle until later.
"Still, there are some people who refuse to believe in even the need for due diligence with respect to some authorities. This article is primarily for them. I hope some will read it, and at least see the need to dig a little deeper."
With deep respect and appreciation for the mathematical/ statistical analysis it is only faith in your skill and honesty that leads someone like me to trust those figures lead to the conclusion. While it is glaringly obvious to you and others who assimilate and manipulate numbers with ease many like me are overwhelmed by the matrix of figures. Offered in the spirit of the clueless user included in Beta test applications.
What we need are simple representations in bar charts, pie charts or graphs to translate the relationships and really grasp what the numbers clearly demonstrate. Most folks don't like to admit to being confused or unable to understand, which only helps the faction promoting talking points like 98% effective that we can feel confident we understand. If the goal is broad understanding more pictures less math is the target audience who struggle to be informed but cannot. <3
Quarterly reports are just raw numbers. Annual reports get into rates and "adjusted" comparisons to suicide in general population. Annual reports also cover data quality, procedures, programs, etc.
I would not trust these numbers yet for multiple reasons.
1. Ambulatory reports of suicide attempts spiked across the military in 2021.
2. There is currently focus on the military, making the hiding of deaths very possible. The R-code deaths won't be fully sorted for months, and are higher than usual, so all 2021 mortality numbers should carry an asterisk, with assumption that tallies are partial.
For active duty suicide (I have no idea about reserve), I am under the impression that the Pentagon's "Defense Suicide Prevention Office" gets its PRIMARY data from the dead service member's chain of command - - not the DoD medical establishment. There is an administrative form to be filled out by chain of command (not doctors). In the Marine Corps, if chain of command fails to forward the form within the prescribed 72 hours, the chain of command will be in trouble. I suspect other services are, like the Marine Corps, diligent about forwarding the form on time.
My point here is that the Defense Suicide Prevention Office reports of completed suicide are primarily driven by data from chain of command - - not doctors entering diagnostic codes. However, suicide ATTEMPT data in Defense Suicide Prevention Office ANNUAL reports probably come from diagnostic codes coming out of clinics or hospitals.
Also, military suicide has the attention of Congress. If DoD is faking suicide data, Congress will be displeased.
As regards sieves, the Bardenheier study showed the problem with using the same people for the treatment arm and for control. Survivorship bias using a nursing home population. The all cause mortality rr for treatment was 0.34.
Thanks for writing this. The two jab regimen has inflated effectiveness by design. It’s why the Jensen jab lost in uptake and why boosters almost always look almost completely worthless. In which there are less vaccinated infected who count as unvaccinated cases due to only a 14 day window vs. 35 day window.
The infuriating part is that to produce the paper someone prepped the data of first prick to death or no prick to death. By cleverly splitting the deaths into covid and non-covid they never report the most important metric, all cause mortality by shot status (from first jab date). See above for more discussion.
We live in the POST-FACT Information economy; raw data, facts, probabilities, these are all ‘injected’ with biased opinion likely layered in logical fallacies. Truth discovery in todays world requires a time, talent, and willingness to dig, uncover, sort, and reveal.
Isn't this a simple cohort mismatch issue? We know that the choice to get vaccinated or not is not a random choice across society. We know certain political leanings are more likely to get vaccinated. We know that blacks are less likely to get vaccinated, as well as education, income, etc. We also know that all these factors matter in other aspects in life, like overall health, risky jobs, living in safe vs less to plain non-safe neighborhoods, etc.
I actually came across this yesterday, when looking at a J&J vs mRNA article from the CDC, that was based on the same data. You here noticed the same thing ('mRNA rulez' according to the CDC), but the CDC actually has a separate article pointing this out. When I looked at it, I already suspected a cohort mismatch, as two separate studies done elsewhere actually showed a more subtle effect, and we know the mRNA shots less effective than these numbers due to known data and definition issues with the unvaccinated. But this study actually proves my suspicions, that indeed we are looking at a cohort mismatch.
It's not that I think CDC numbers are honest, but I do believe there's a plausible explanation for lower overdose and suicide deaths that you haven't considered. It could be a selection effect. The kinds of young people who are inclined to suicide and drug overdoses and drunk driving are not the kinds of young people who line up to be vaccinated.
Vaccination status might be a predictor of social status and (therefore) general health. However, the race/ethnicity part of the table that you cut off provides a similar picture.
Reporting the numbers of deaths is quite pointless. They should be weighted by length of time period spent in that vaccination status. Unvaccinated people are unvaccinated the whole time whereas time in dose-1 limbo should be around two months (less for most that get a 2nd jab, but some people remain in limbo). Multiply the dose-1 figures by 7/2, and you are in the ballpark of the unvaccinated figures.
The middle education range has the highest vaccination rate. The tails have the lowest, with those who have PhDs being the least likely to get vaccinated.
Among my wealthy friends, few are vaccinated.
I don't think status is going to result in 60+ percent relative efficacy. Besides, if you run the numbers, these stats make it look like the vaccinated suddenly have a 400+ year life expectancy. Nothing on any actuarial table I ever saw looks like that.
"Reporting the numbers of deaths is quite pointless. They should be weighted by length of time period spent in that vaccination status."
The normalization is in person-years. The units in the VSD study are fine. The problem has to be elsewhere. It's in the algorithm.
I see. The report also hints at that in the table captions but I have no idea when exactly the adjusting is done. For example, J&J 45-64: Table 1 lists headcounts of 158,157 (J&J) and 624,106 (comp.). Table 2 has 130 vs 497 deaths. That's 0.082% (J&) vs 0.080%. Is this an inappropriate computation, and if so, why? How can I get from headcount to person-time?
I didn't bother hunting all the calculations. I'm dyslexic to a point that my concentrated reading time is limited--particularly after 27 months of pandemic research at a rate of 80 hours a week with very little breakage. At this point, if I don't need to read it to make the point, I'm moving on to the next higher priority task.
At the end of the day, one major problem is that all the public health data being used isn't simply open source. In that case, we would have simple answers, quickly, and without even the potential for corruption.
Either we change that going forward, or we give up to a Pharmauthoritarian dystopia (or the war all that stupidity brings).
It seems much more likely that someone would get a vaccine out-of-network, than they would be to get medical treatment out-of-network. So in the case of a harmful vaccine, that would mean a lot of injuries would show up in VSD as an injury with no associated vaccination. So I'm wondering how much this may skew the numbers.
First let me state that I am blond, and did not read that slowly. :-) But, wouldn't we need to add up ALL of the deaths, from the mRNA jabs combined, and then compare that number to the unjabbed? In that case, if I am reading the chart correctly, the jabbed are dying more from all cause that is not COVID than the unjabbed. (Apologies, but I cannot call it "vaccinated" or "unvaccinated", since it is not truly a "vaccine".) By splitting out the mRNA injections by brand, it automatically makes the number of deaths in each column smaller. Then by further breaking it down by number of jabs, etc., the same thing happens. I think splitting everything out is another slight of hand, for this data.
Now, looking at the chart the crazy way I am, J & J would still show fewer all cause mortality, yet it is the one that the use has been halted in, in several countries, due to the number of concerning side effects, including death. Again, something isn't adding up.
I'm still trying to finish my coffee after a crazy working weekend, so if my thoughts aren't making sense, feel free to ignore them, and go back and read my first sentence.
People who get vaccines are often better integrated into society, prone to more security than risk seeking etc.
It's classic selection bias and has always been that way. Vaccinated people will never get into traffic accidents at the same rate as unvaccinated, you need to clean/match the data meticulously which is pretty much impossible.
That is precisely why trials need to be double blindly placebo controlled.
Alas, the governments are now preparing the narrative that vaccine!=medication, so they don't need controlled trials. (I read a lot of lobbyist literature, it's all over the place if you just look for it.)
Looking at various pro-vaccine studies, it was obvious to me that the raw data "pea" was being hidden under the adjusted data "shell."
Raw data isn't being reported.
Bias in the adjusted data is what would be called "systematic error" in chemistry and physics.
Systematic error should be caught and prevented with proper procedure. Systematic error should never make it into the reported data.
The age range choices for the CDC datasets have confounded many analyses I've attempted to conduct, mostly revolving around age-adjusted CFR's & IFR's.
It's doubly frustrating because the CDC obviously HAS the data, and chooses not to provide it to the public in a form that we can verify their work with
In some cases, it's good enough to estimate. There have been times I've laid out a curve and interpolated. In this case, envelope math was clearly good enough.
But yes...there is no reason not to have open source public health data, except to allow for a handful of public health officials to have control over what gets published. I believe that if we ever get a good look behind the curtain, we're going to find that every aspect of the pandemic and medical data was rife with publication bias. Of course, you know that very specifically in your corner of the research.
We will never see that data, it has long been destroyed, purposely not collected or done in a way that it will never give clear answers. Imagine that in the first 12 months the US did not collect any data on who already had the virus nor did they test to confirm it properly.
When i got Alpha, I tested, then tested with a different lab, then did a government test as well. 3 tests said I had it. Then antibody test 3 months later and 8 months later. Same for my whole family. No one cares to know this yet we all know it provides some sort of immunity.
Small point - is it possible that the lockdowns impacted the all cause mortality in the age you noted? (Less car accidents, less homicide etc because of the decrease in driving and interaction)
Traffic mortality increased during lockdown periods, and homicides shot up.
Thank you - that is interesting! Really appreciate your posts and work Mathew.
Thank you for a good question. The answers are not likely intuitive to everyone, so I made an addendum to the post.
Friends of mine live on a usually busy main road in our small city. During the lockdowns the traffic on their street was nearly zero. The result, however, was that the few drivers left on the road turned it into their own personal Indy 500. My friends said the noise (and the sirens) were unprecedented. They were relieved when normal traffic returned.
Very little impact on excess mortality, tho. Rates for the general population for suicide, murder, and accidents were within normal range comparing 2019-2021.
I'm not sure what you mean. The changes in PDT were unprecedented in my lifetime.
Suicides:
2019...47511
2020...47653
2021...47978
Murders:
2019...19141
2020...19196
2021...19344
Poisoning
2019...75795
2020...76029
2021...76539
https://deadorkicking.com/death-statistics/us/2019/
I will look at this site later, but I have doubts. Remember that 2021 totals are not yet nearly complete. There are larger numbers of R-codes deaths with causes that will not settle until later.
The site states that its numbers about the previous year are preliminary for several months.
If you can find a breakdown by age, sex, and cause of death, you might see suicides up in younger men.
Mathew, great article!
That statistical MMWR report was so ridiculous.
I wrote about it in October also. Let's compare notes.
https://igorchudov.substack.com/p/boring-is-there-an-error-in-cdc-mmwr
I commented that CDC used incorrect person-year denominators and the CDC article is either a giant mistake, or a giant lie.
I will take a deeper look after VRBPAC tomorrow. Assuming I don't take a few days off, which is very possible at this moment.
"Still, there are some people who refuse to believe in even the need for due diligence with respect to some authorities. This article is primarily for them. I hope some will read it, and at least see the need to dig a little deeper."
With deep respect and appreciation for the mathematical/ statistical analysis it is only faith in your skill and honesty that leads someone like me to trust those figures lead to the conclusion. While it is glaringly obvious to you and others who assimilate and manipulate numbers with ease many like me are overwhelmed by the matrix of figures. Offered in the spirit of the clueless user included in Beta test applications.
What we need are simple representations in bar charts, pie charts or graphs to translate the relationships and really grasp what the numbers clearly demonstrate. Most folks don't like to admit to being confused or unable to understand, which only helps the faction promoting talking points like 98% effective that we can feel confident we understand. If the goal is broad understanding more pictures less math is the target audience who struggle to be informed but cannot. <3
I can sometimes be every step in the process of synthesis-to-simplicity, but there are not enough hours in the day. I've over 200 working articles.
Trivia: The Department of Defense has an office dedicated to tracking active duty and reserve suicide. The office issues quarterly and annual reports. FYI, I link to Q4 report for CY 2021. Report depicts numbers by active duty vs reserve, Army, Navy, USMC, Air Force, CY quarter, year going back to 2015. https://www.dspo.mil/Portals/113/Documents/2021QSRs/TAB%20A_20220317_OFR_Rpt_Q4%20CY21%20QSR.pdf?ver=PcN7dgtBKM3RG2uP1rAZ-A%3d%3d
Quarterly reports are just raw numbers. Annual reports get into rates and "adjusted" comparisons to suicide in general population. Annual reports also cover data quality, procedures, programs, etc.
I would not trust these numbers yet for multiple reasons.
1. Ambulatory reports of suicide attempts spiked across the military in 2021.
2. There is currently focus on the military, making the hiding of deaths very possible. The R-code deaths won't be fully sorted for months, and are higher than usual, so all 2021 mortality numbers should carry an asterisk, with assumption that tallies are partial.
For active duty suicide (I have no idea about reserve), I am under the impression that the Pentagon's "Defense Suicide Prevention Office" gets its PRIMARY data from the dead service member's chain of command - - not the DoD medical establishment. There is an administrative form to be filled out by chain of command (not doctors). In the Marine Corps, if chain of command fails to forward the form within the prescribed 72 hours, the chain of command will be in trouble. I suspect other services are, like the Marine Corps, diligent about forwarding the form on time.
My point here is that the Defense Suicide Prevention Office reports of completed suicide are primarily driven by data from chain of command - - not doctors entering diagnostic codes. However, suicide ATTEMPT data in Defense Suicide Prevention Office ANNUAL reports probably come from diagnostic codes coming out of clinics or hospitals.
Also, military suicide has the attention of Congress. If DoD is faking suicide data, Congress will be displeased.
As regards sieves, the Bardenheier study showed the problem with using the same people for the treatment arm and for control. Survivorship bias using a nursing home population. The all cause mortality rr for treatment was 0.34.
That study was totally worthless.
Can you post the study? I don't know the author.
Thanks for writing this. The two jab regimen has inflated effectiveness by design. It’s why the Jensen jab lost in uptake and why boosters almost always look almost completely worthless. In which there are less vaccinated infected who count as unvaccinated cases due to only a 14 day window vs. 35 day window.
Every new window of time segmentation results in a copy of whatever bias is in play. That's necessary to keep the illusion going.
I wrote about this last fall, see here: https://paulkraken.substack.com/p/cdc-vaccine-protective-effect-expected?s=w
The infuriating part is that to produce the paper someone prepped the data of first prick to death or no prick to death. By cleverly splitting the deaths into covid and non-covid they never report the most important metric, all cause mortality by shot status (from first jab date). See above for more discussion.
We live in the POST-FACT Information economy; raw data, facts, probabilities, these are all ‘injected’ with biased opinion likely layered in logical fallacies. Truth discovery in todays world requires a time, talent, and willingness to dig, uncover, sort, and reveal.
(Thanks for doing what you do!)
Isn't this a simple cohort mismatch issue? We know that the choice to get vaccinated or not is not a random choice across society. We know certain political leanings are more likely to get vaccinated. We know that blacks are less likely to get vaccinated, as well as education, income, etc. We also know that all these factors matter in other aspects in life, like overall health, risky jobs, living in safe vs less to plain non-safe neighborhoods, etc.
I actually came across this yesterday, when looking at a J&J vs mRNA article from the CDC, that was based on the same data. You here noticed the same thing ('mRNA rulez' according to the CDC), but the CDC actually has a separate article pointing this out. When I looked at it, I already suspected a cohort mismatch, as two separate studies done elsewhere actually showed a more subtle effect, and we know the mRNA shots less effective than these numbers due to known data and definition issues with the unvaccinated. But this study actually proves my suspicions, that indeed we are looking at a cohort mismatch.
It's not that I think CDC numbers are honest, but I do believe there's a plausible explanation for lower overdose and suicide deaths that you haven't considered. It could be a selection effect. The kinds of young people who are inclined to suicide and drug overdoses and drunk driving are not the kinds of young people who line up to be vaccinated.
Selection effects are never this brutal.
Lay it out in a spreadsheet with the ten largest causes of death, and try to make the numbers not look comical. You'll give up quickly, I promise.
Vaccination status might be a predictor of social status and (therefore) general health. However, the race/ethnicity part of the table that you cut off provides a similar picture.
Reporting the numbers of deaths is quite pointless. They should be weighted by length of time period spent in that vaccination status. Unvaccinated people are unvaccinated the whole time whereas time in dose-1 limbo should be around two months (less for most that get a 2nd jab, but some people remain in limbo). Multiply the dose-1 figures by 7/2, and you are in the ballpark of the unvaccinated figures.
The middle education range has the highest vaccination rate. The tails have the lowest, with those who have PhDs being the least likely to get vaccinated.
Among my wealthy friends, few are vaccinated.
I don't think status is going to result in 60+ percent relative efficacy. Besides, if you run the numbers, these stats make it look like the vaccinated suddenly have a 400+ year life expectancy. Nothing on any actuarial table I ever saw looks like that.
"Reporting the numbers of deaths is quite pointless. They should be weighted by length of time period spent in that vaccination status."
The normalization is in person-years. The units in the VSD study are fine. The problem has to be elsewhere. It's in the algorithm.
I see. The report also hints at that in the table captions but I have no idea when exactly the adjusting is done. For example, J&J 45-64: Table 1 lists headcounts of 158,157 (J&J) and 624,106 (comp.). Table 2 has 130 vs 497 deaths. That's 0.082% (J&) vs 0.080%. Is this an inappropriate computation, and if so, why? How can I get from headcount to person-time?
Yeah, those tables are a bit confusing.
I didn't bother hunting all the calculations. I'm dyslexic to a point that my concentrated reading time is limited--particularly after 27 months of pandemic research at a rate of 80 hours a week with very little breakage. At this point, if I don't need to read it to make the point, I'm moving on to the next higher priority task.
At the end of the day, one major problem is that all the public health data being used isn't simply open source. In that case, we would have simple answers, quickly, and without even the potential for corruption.
Either we change that going forward, or we give up to a Pharmauthoritarian dystopia (or the war all that stupidity brings).
It seems much more likely that someone would get a vaccine out-of-network, than they would be to get medical treatment out-of-network. So in the case of a harmful vaccine, that would mean a lot of injuries would show up in VSD as an injury with no associated vaccination. So I'm wondering how much this may skew the numbers.
First let me state that I am blond, and did not read that slowly. :-) But, wouldn't we need to add up ALL of the deaths, from the mRNA jabs combined, and then compare that number to the unjabbed? In that case, if I am reading the chart correctly, the jabbed are dying more from all cause that is not COVID than the unjabbed. (Apologies, but I cannot call it "vaccinated" or "unvaccinated", since it is not truly a "vaccine".) By splitting out the mRNA injections by brand, it automatically makes the number of deaths in each column smaller. Then by further breaking it down by number of jabs, etc., the same thing happens. I think splitting everything out is another slight of hand, for this data.
Now, looking at the chart the crazy way I am, J & J would still show fewer all cause mortality, yet it is the one that the use has been halted in, in several countries, due to the number of concerning side effects, including death. Again, something isn't adding up.
I'm still trying to finish my coffee after a crazy working weekend, so if my thoughts aren't making sense, feel free to ignore them, and go back and read my first sentence.
There's an intrinsic vaccine bias.
People who get vaccines are often better integrated into society, prone to more security than risk seeking etc.
It's classic selection bias and has always been that way. Vaccinated people will never get into traffic accidents at the same rate as unvaccinated, you need to clean/match the data meticulously which is pretty much impossible.
That is precisely why trials need to be double blindly placebo controlled.
Alas, the governments are now preparing the narrative that vaccine!=medication, so they don't need controlled trials. (I read a lot of lobbyist literature, it's all over the place if you just look for it.)