Not long ago Steve called me and said that had some real unaltered vaccination records and wanted me to see what they revealed about vaccine safety. Here’s my take on it. There’s so much more to dig in here, but I leave that as an exercise to the reader. Steve has made the data public here.
Turns out the vaccine increases death rates in the first 6 months after the shot which persists for about a year before going back to baseline. And it’s the last dose that kills you – pulling forward death into that first year post shot.
Let’s start with death rate. Here we see the number of deaths per 100,000 person days after the last shot. More on that metric later, but it’s the least biased way we can look at the data we have. The blue line is from the vaccine records, the red is from some synthetic data I created that was explicitly designed to NOT have any vaccine effects (more on that below, too).
Solgar - Chelated Zinc...
Buy New $16.49
(as of 07:30 UTC - Details)
I don’t know about you, but I clearly see a peak in death rate about 6-months out from the last shot and that stays elevated above baseline (the red) for about a year.
Another way to look at the data: let’s just look at the people who died and see how long they survived after their last shot. I tend think in probability density, so that’s what I have here (I labelled it “Fraction of People”, but for those in-the-know, it’s probability density, the area under the curve is 1).
If the vaccine didn’t cause death, then I’d expect to see what the red data shows, a constant death rate (with Poisson noise) after the last shot. But that’s not what we see, the deaths show up with a 6-month-ish peak and stay elevated for about a year.
What about age stratification? What about vaccine batch? What about dose count and timing? Yeah, that’s all in the data, slice and dice and see what you find. I wanted to get the big picture and see if there’s signal. If that’s not a clear signal, then I don’t know what to tell you. Have at the data, it’s here.
Ok, nerd stuff follows below.
What is the data?
We have “record-level data”. Which is an entry in the data for every vaccination a person received with the date and dose number they received it and death date if that happened. I think this is the first time we have true unaggregated records that allow us to do real correlations.
Garden of Life Zinc Vi...
Buy New $9.79
(as of 06:35 UTC - Details)
It’s not perfect, though. We only have vaccinated people in this data, so we can’t get a true unvaccinated death rate over the same time frame. From what I gather, people end up in one of three data sets and we only have one of them here, so we don’t have everyone’s visits – practically that means it often looks like someone skipped a dose because that dose may have been recorded elsewhere. So we have to be creative to analyze the data because it can be very easy to find a survivorship bias or other anomalies since we are only looking at a subset of people (those who were vaccinated and died in the time frame we have and also at a facility that was reporting to this data set).
What’s the deal with the synthetic data in these plots?
I get squeamish when statisticians talk about all the weird controls and things they do with data, so I wanted to try another way to make sense of this data. I’m a computational modeler, what I try to do is create models that describe the natural world and use that to gain understanding (of course, all models are wrong, some are useful). So rather than go through all kinds of gyrations to account for any of the reporting bias that may show up in the data, I decided to create synthetic (fancy word for made-up) data that 1) is guaranteed to not have any vaccine effects in it and 2) would have the same reporting bias as the real data. The only difference between the real data and the synthetic data is the effect of vaccination.
Let’s look at the constraints in the real data: a person only shows up in the records after their first (recorded) vaccination. That means even a constant death rate (just people randomly dying) will appear in the data as an increasing death rate because more people are entering the records as they get their first vaccination. We also only have people who died after a vaccine – they wouldn’t be in the data set if they died before getting a vaccine. Since the data set stops at some time, we also don’t have people who might have died after the data set was sent to us. And as people die near the end of the reporting period the death counts go to zero since we’re running out of people.
