9 Comments

BrilliantđŸ’•

Expand full comment

These non-Covid deaths were misclassified because they'd had one jab at time of death or two jabs but less than 14 days at time of death? Is that correct?

Expand full comment
author

The bulk of the erroneous data appears to be after dose 1, probably 14 days after.

Expand full comment

Great work, thanks. The early January peak in mortality rates for the vaccinated in your corrected curves does not match with the peak in vaccination rates per weak, probably hence not real. Most likely meaning your 1week shift needs further adjustement, and you might end up spreading the vaccinated excess mortality more evenly. Presumably it's the weekly vaccination rate, convolved with a pretty wide right-tailed kernel.

Expand full comment
author

The peaks shouldn't align if the vaccinated deaths are counted as unvaccinated for the first 14 days and then the data is aggregated weekly? Bottom line is the data has been manipulated and there is insufficient granularity to infer the exact extent.

Expand full comment

Uhm, actually the number I am speaking of would simply follow from integrating the curves in your last plot. Seems to be 1 vaccine death every 1000 vaccinations in this age category. Kind of huge.

Expand full comment
author

Oh, I get you. Yeah, that wouldn't surprise me. Nursing home residents were dying within hours of being injected. Yes, some were on the brink of death anyway but such a high fatality rate would not surprise me at all.

Expand full comment

Yes, I agree with your conclusion, the raw data is misleading here, but you have done already a very good job at reconstructing the trend, and adding back deaths to the vaccinated category where they belong, including presumably those in the first 14 days. If you look at your last plot though, you see it's not just a shift, but also the red curve is broadened. With some further simple hypothesis you might be able to reconstruct the kernel which converts vaccination rates to mortality from vaccine in a self consistent way. That would be a great number to have, integrated over a few months. Also, would be really useful to have this for the other age categories.

Expand full comment
author

I have only shown dose 1 in that chart. I should have included dose 2 as well. I think the right side of the kernel is influenced by dose 2, 14-day "adjustment".

Expand full comment