Many of us can do this analysis. You and others have done a great job. But somehow, the public health officials cannot. They are clearly both incompetent and corrupt. Only a corrupt person would try to do such a thing and only an incompetent would think that they wouldn’t be caught.
A very balanced piece. Remember that the "unvaccinated" are a different age distribution from the "vaccinated" in this data set and there is almost certainly miscategorisation bias built in. So please people be wary of using this dataset "as is".
Not at all, it was a generic comment. If you are comparing age bands like for like you should see a pattern, however you still need to account for exposure time. The "unvaccinated" include people who are end-of-life so you should probably remove the first 2 months of data (if you see that there is a higher mortality in Jan-Mar 2021 in this group). After March 2021 you should see comparative cohorts but to calculate exposure time you would need to count from 1st March 2021 in the unvaccinated against the date of vaccination in the vaccinated (whichever group). For the between group (vaccinated) analysis it's important to check the start date and expected SMR for the age group included, and count the expected mortality rate from that time point for that age group. The reason is that it is possible that there is a confounder driving the differential between moderna and pfizer, which could merely be exposure time or age distribution.
Yes, indeed. I marked 1st March 2021 in my worksheet as the date at which it appears they stopped "prioritising" the most vulnerable. Before that the vaccinated mortality rate is higher, as expected. Although, I would be confident in stating that it is still underestimated due to miscategorisation. Have you looked at the data to confirm the possibility of time bias?
I've looked at it in brief along with @openvaet who I think fortunately has a bit more time available to spend on it. My first impression is that the groups are so disparate by age (even Pfizer vs moderna) that a direct comparison is not possible without bootstrapping matched samples. We did something similar with the New Zealand data and proved that it showed nothing. I have a similar feeling that this data has been released to tie us up in knots, but will revert when I know more.
Hi Joel. How have you determined the death rate among the unvaccinated? For starters, what was the unvaccinated population in this age group in order to have an accurate denominator. Does the raw data specify the date of last vaccination relative to death, so that misclasdification due to the use of an inoculation window can be elucidated? As Steve pointed out, we don't know the health status of the unvaccinated, which can be a confounder in some countries where the jab was administered more judiciously.
As I understand, it contains the entire population, so if there is no date of first vaccination, I assumed they were unvaccinated (notwithstanding the obvious misclassification I have elucidated).
In the data of most countries I have seen , unvaccinated applies to any person given ANY shot 1 2 or boosters within the 14 day period after their jab. It was an easy way to hide deaths that happened as an AE immediately following jab
This is record-level data so it hasn't been corrupted by the public health and statistics agencies. However, it is evidently still replete with errors.
Many of us can do this analysis. You and others have done a great job. But somehow, the public health officials cannot. They are clearly both incompetent and corrupt. Only a corrupt person would try to do such a thing and only an incompetent would think that they wouldn’t be caught.
Can't put lipstick on this Pfizer pig of death.
A very balanced piece. Remember that the "unvaccinated" are a different age distribution from the "vaccinated" in this data set and there is almost certainly miscategorisation bias built in. So please people be wary of using this dataset "as is".
Even within the narrow band of 70 to 79 year olds that I have used?
Not at all, it was a generic comment. If you are comparing age bands like for like you should see a pattern, however you still need to account for exposure time. The "unvaccinated" include people who are end-of-life so you should probably remove the first 2 months of data (if you see that there is a higher mortality in Jan-Mar 2021 in this group). After March 2021 you should see comparative cohorts but to calculate exposure time you would need to count from 1st March 2021 in the unvaccinated against the date of vaccination in the vaccinated (whichever group). For the between group (vaccinated) analysis it's important to check the start date and expected SMR for the age group included, and count the expected mortality rate from that time point for that age group. The reason is that it is possible that there is a confounder driving the differential between moderna and pfizer, which could merely be exposure time or age distribution.
Yes, indeed. I marked 1st March 2021 in my worksheet as the date at which it appears they stopped "prioritising" the most vulnerable. Before that the vaccinated mortality rate is higher, as expected. Although, I would be confident in stating that it is still underestimated due to miscategorisation. Have you looked at the data to confirm the possibility of time bias?
I've looked at it in brief along with @openvaet who I think fortunately has a bit more time available to spend on it. My first impression is that the groups are so disparate by age (even Pfizer vs moderna) that a direct comparison is not possible without bootstrapping matched samples. We did something similar with the New Zealand data and proved that it showed nothing. I have a similar feeling that this data has been released to tie us up in knots, but will revert when I know more.
Hi Joel. How have you determined the death rate among the unvaccinated? For starters, what was the unvaccinated population in this age group in order to have an accurate denominator. Does the raw data specify the date of last vaccination relative to death, so that misclasdification due to the use of an inoculation window can be elucidated? As Steve pointed out, we don't know the health status of the unvaccinated, which can be a confounder in some countries where the jab was administered more judiciously.
As I understand, it contains the entire population, so if there is no date of first vaccination, I assumed they were unvaccinated (notwithstanding the obvious misclassification I have elucidated).
In the data of most countries I have seen , unvaccinated applies to any person given ANY shot 1 2 or boosters within the 14 day period after their jab. It was an easy way to hide deaths that happened as an AE immediately following jab
This is record-level data so it hasn't been corrupted by the public health and statistics agencies. However, it is evidently still replete with errors.
Forgive me, but is there a reason to believe that record-level data isn't subject to corruption? TY