>You’re right about calculating risk, but when’s the last time a vaccine in normal, longer term stage three trials resulted in a higher fatality rate than COVID (for any age group)? The link for the SARS vaccine candidate was a failure that was caught in a mouse model, which unsurprisingly they also did with the new vaccines before the human trials started. To echo the parent comment, these were immediate side effects on challenge (which would likely been caught in stage 2 trials even if they only happened in humans and not in animal models).
It's basically like saying: "the unit tests have all passed now, the regression tests all pass, so let's roll out the fix straight to prod, because it's urgent and we don't have enough time to do the normal amount of staging environment testing". Sure, probably it's fine. But mistakes happen. In any other context, a 1/100,000 chance of error would be considered incredibly low, a great achievement. But in this case, a 1/100,000 chance of seeing after 2-3 years a failure like happened early on in those previous trials would be an incredible tragedy, if the vaccine was taken by people with less than 1/100,000 chance of being killed by covid.
https://onlinelibrary.wiley.com/doi/10.1111/ijcp.13795 makes a better case than I can that "a finite, non-theoretical risk is evident in the medical literature that vaccine candidates composed of the SARS-CoV-2 viral spike and eliciting anti-SARS-CoV-2 antibodies, be they neutralising or not, place vaccinees at higher risk for more severe COVID-19 disease when they encounter circulating viruses"
>If we want to go with unusual reactions that only show up over time, what about the chance that whatever long term side effect you’re imagining from the vaccines instead happens for people who have been infected with COVID 5 years from now? Once you decide to make decisions based on rare and novel events with unquantifiable risks, you’ll find they show up absolutely everywhere if you’re being intellectually honest.
The difference is that covid does not have a big qualitative difference from other respiratory viruses, and no circulating respiratory viruses are known to cause serious long-term symptoms in people who were short-term asymptomatic. Whereas the MRNA treatment is quite a different mechanism from previous vaccines.
Separately, ethically it's generally considered worse if a death results from deliberate human action than from an "act of nature". E.g. nowhere is it morally acceptable to murder somebody just because their organs could be used to save ten people. In this case that means it would be considered morally worse if people were coerced into taking a vaccine that ended up killing them than if, absent the vaccine, they died from natural causes of equivalent risk.
>That data’s N is a little low, but let’s take it seriously for a moment. The vast majority of vaccinated people in that dataset did not go on to infect others, and none of them were epicenters for super-spreader events. Eyeballing it, it’s consistent with a sterilizing immunity in excess of 80%. If the vaccines turn out to be that effective at preventing transmission, that’s an excellent outcome (it is higher than most vaccines).
There's one super-spreader there, a prison cook. I pulled the data from the backing rest API (if you want I could upload the notebook and show you):
pd.Series(vax_links).describe()
count 17.000000
mean 2.705882
std 3.670230
min 1.000000
25% 1.000000
50% 1.000000
75% 2.000000
max 14.000000
dtype: float64
pd.Series(unvax_links).describe()
count 127.000000
mean 2.181102
std 1.965647
min 1.000000
25% 1.000000
50% 2.000000
75% 2.000000
max 15.000000
dtype: float64
While there are way fewer vax cases than unvax cases, looking at the average number of infectees it doesn't appear that vaccinated people infect fewer on average than the unvaccinated do, at least given the limited data.
It's basically like saying: "the unit tests have all passed now, the regression tests all pass, so let's roll out the fix straight to prod, because it's urgent and we don't have enough time to do the normal amount of staging environment testing". Sure, probably it's fine. But mistakes happen. In any other context, a 1/100,000 chance of error would be considered incredibly low, a great achievement. But in this case, a 1/100,000 chance of seeing after 2-3 years a failure like happened early on in those previous trials would be an incredible tragedy, if the vaccine was taken by people with less than 1/100,000 chance of being killed by covid.
https://onlinelibrary.wiley.com/doi/10.1111/ijcp.13795 makes a better case than I can that "a finite, non-theoretical risk is evident in the medical literature that vaccine candidates composed of the SARS-CoV-2 viral spike and eliciting anti-SARS-CoV-2 antibodies, be they neutralising or not, place vaccinees at higher risk for more severe COVID-19 disease when they encounter circulating viruses"
>If we want to go with unusual reactions that only show up over time, what about the chance that whatever long term side effect you’re imagining from the vaccines instead happens for people who have been infected with COVID 5 years from now? Once you decide to make decisions based on rare and novel events with unquantifiable risks, you’ll find they show up absolutely everywhere if you’re being intellectually honest.
The difference is that covid does not have a big qualitative difference from other respiratory viruses, and no circulating respiratory viruses are known to cause serious long-term symptoms in people who were short-term asymptomatic. Whereas the MRNA treatment is quite a different mechanism from previous vaccines.
Separately, ethically it's generally considered worse if a death results from deliberate human action than from an "act of nature". E.g. nowhere is it morally acceptable to murder somebody just because their organs could be used to save ten people. In this case that means it would be considered morally worse if people were coerced into taking a vaccine that ended up killing them than if, absent the vaccine, they died from natural causes of equivalent risk.
>That data’s N is a little low, but let’s take it seriously for a moment. The vast majority of vaccinated people in that dataset did not go on to infect others, and none of them were epicenters for super-spreader events. Eyeballing it, it’s consistent with a sterilizing immunity in excess of 80%. If the vaccines turn out to be that effective at preventing transmission, that’s an excellent outcome (it is higher than most vaccines).
There's one super-spreader there, a prison cook. I pulled the data from the backing rest API (if you want I could upload the notebook and show you):
While there are way fewer vax cases than unvax cases, looking at the average number of infectees it doesn't appear that vaccinated people infect fewer on average than the unvaccinated do, at least given the limited data.