4 ms·

Unfortunately in this case the original post is mostly correct. It claims that this vaccine study did not establish that the vaccines were protective against de

by emtel 5y ago

Unfortunately in this case the original post is mostly correct. It claims that this vaccine study did not establish that the vaccines were protective against death. This is both true at a glance (the number of deaths are clearly too small to establish statistical significance, i.e. the trial was underpowered to detect efficacy against death), and corroborated by the study you link to which (on page 68) lists the 95% CI of efficacy against death as NE-100%. NE = non-estimable). Now, we do know that vaccines are protective against death from covid-19, but this study did not establish that fact. Why do I bother writing this? I believe the only way to defeat "misinformation" is to rebut it accurately. An inaccurate rebuttal is likely to be interpreted as "my opponents don't even understand the data, therefore I believe my original position even more strongly". I think a better rebuttal to this argument is "yes, the original trials did not prove efficacy against death - because they were underpowered, not because the effect did not exist. Subsequent observations of death rates among vaccinated and non-vaccinated populations are basically impossible to explain unless the vaccines protect against death, even though those observations weren't taken during an RCT."

vannevar 5y ago
Your point is well taken, but I doubt the casual reader would make the distinction that you do. And in regard to Covid deaths in the study, the people who ran it were well aware that the small number of deaths were not, by themselves, enough to establish efficacy. But in combination with the numbers for severe illness (2 in the vaccinated vs 106 for the placebo group) and everything we understood about how the disease kills, I would argue that the study does in fact overwhelmingly support the conclusion that the vaccine prevents death, just as we have since confirmed.
- emtel 5y ago
  I 100% agree that if you have proof of efficacy against severe disease, its reasonable to assume efficacy against death. In fact I think it would be absurd not to. But, the usual rules of the game (and certainly the rules the FDA plays by as I understand it) are that you either reject the null hypothesis with a sufficient degree of statistical significance or you don't, and if you don't, you haven't "proved" anything. So if someone claims that studies proved something according to those rules, its fair to call them out if there was in fact no proof. To your first point: Yes, 99% of the population does not understand statistics, null-hypothesis testing, what "under-powered" means, etc. But, I think we under-estimate the ability of the public to understand good explanations of these topics. Maybe I'm too optimistic, but I believe the median reader of a page like the linked article can grasp explanations of these topics even if they are initially somewhat ignorant of them. And, while I can't prove it, I believe that the most epistemically rigorous rebuttals of what we feel are false claims are ultimately going to be the most durable.
Asooka 5y ago
> Now, we do know that vaccines are protective against death from covid-19, but this study did not establish that fact. Can you point to a source not associated with Pfizer that proves that claim? The only thing I've noticed among those I've seen with covid is that the outcome depends on age, overall health before infection, and how quickly treatment was administered, not with vaccination status.

by emtel 5y ago

vannevar 5y ago

Your point is well taken, but I doubt the casual reader would make the distinction that you do. And in regard to Covid deaths in the study, the people who ran it were well aware that the small number of deaths were not, by themselves, enough to establish efficacy. But in combination with the numbers for severe illness (2 in the vaccinated vs 106 for the placebo group) and everything we understood about how the disease kills, I would argue that the study does in fact overwhelmingly support the conclusion that the vaccine prevents death, just as we have since confirmed.

emtel 5y ago
I 100% agree that if you have proof of efficacy against severe disease, its reasonable to assume efficacy against death. In fact I think it would be absurd not to. But, the usual rules of the game (and certainly the rules the FDA plays by as I understand it) are that you either reject the null hypothesis with a sufficient degree of statistical significance or you don't, and if you don't, you haven't "proved" anything. So if someone claims that studies proved something according to those rules, its fair to call them out if there was in fact no proof. To your first point: Yes, 99% of the population does not understand statistics, null-hypothesis testing, what "under-powered" means, etc. But, I think we under-estimate the ability of the public to understand good explanations of these topics. Maybe I'm too optimistic, but I believe the median reader of a page like the linked article can grasp explanations of these topics even if they are initially somewhat ignorant of them. And, while I can't prove it, I believe that the most epistemically rigorous rebuttals of what we feel are false claims are ultimately going to be the most durable.

Asooka 5y ago

> Now, we do know that vaccines are protective against death from covid-19, but this study did not establish that fact. Can you point to a source not associated with Pfizer that proves that claim? The only thing I've noticed among those I've seen with covid is that the outcome depends on age, overall health before infection, and how quickly treatment was administered, not with vaccination status.