4 ms·

Unfortunately it turns out that the way vaccination effectiveness is measured could create this apparent effect even in the case that the vaccination was saline

Unfortunately it turns out that the way vaccination effectiveness is measured could create this apparent effect even in the case that the vaccination was saline. This sounds absurd and wrong, but, as is sometimes the case with statistics, it's possible. Prof Norman Fenton has described the issue and also demonstrated it visually in Excel: https://twitter.com/profnfenton/status/1460339552397275149 https://twitter.com/profnfenton/status/1460339552397275149 https://probabilityandlaw.blogspot.com/2021/11/is-vaccine-efficacy-statistical-illusion.html https://probabilityandlaw.blogspot.com/2021/11/is-vaccine-ef... and there's another worked example here: https://boriquagato.substack.com/p/bayesian-datacrime-defining-vaccine https://boriquagato.substack.com/p/bayesian-datacrime-defini... The problem is obscure but already known since before 2021. It's called Immortal Time Bias: https://catalogofbias.org/biases/immortal-time-bias/ https://catalogofbias.org/biases/immortal-time-bias/ The cause is the definitional game playing public health engages in, whereby people are labelled as unvaccinated for a period of time even after they've already taken a vaccine. This effectively makes the vaccinated "immortal" for two weeks, because if they'd died within that two week period they'd have been counted as unvaccinated. Thus even if a vaccine did nothing at all it would be given positive effectiveness by the methodology public health researchers are using. The problem with COVID vaccines is especially serious because there's increasing amounts of evidence that they may have some sort of immunosuppressive effect, and that people are much more likely to catch COVID in the period immediately after taking the vaccine than they were normally. Up until a few days ago this effect could only be seen by observing that many countries experienced a case surge coincident with the start of their vaccination programme. Then it was discovered that somebody in a Canadian public health agency who didn't get the memo had accidentally published proof this was happening: https://web.archive.org/web/20220108064918/https://www.alberta.ca/stats/covid-19-alberta-statistics.htm#vaccine-outcomes https://web.archive.org/web/20220108064918/https://www.alber... (scroll down to the graph labelled "number of days between first immunization and COVID diagnosis"). I have to link to the Wayback Machine because the moment this graph was noticed, it was deleted from the website. Hence why I say someone didn't get the memo. It's extremely likely that other public health agencies have the same data but they consistently refuse to publish data in which "vaccinated" is defined correctly. Public health will not allow the population to see un-distorted data about VE.

majewsky 5y ago
> number of days between first immunization and COVID diagnosis Survivorship bias. People end up in this graph because they got a COVID diagnosis after the first immunization, but before getting the second immunization. That's only a very short time window on most vaccination schedules (3-6 weeks), so the only surprise here is that there is a long trail of people who got the first shot and then held off on the second shot for a long time. I can understand why they took down that graph. It's very misleading when taken out of context.
- native_samples 5y ago
  People not getting second shots is not a surprise, anyone who has a bad reaction to the first will do that. The lack of any proper tracking of such reactions means it has to be inferred from the data, but the explanation there is fairly obvious. W.R.T. taking it down, if public health statistics is open to misinterpretation they should explain why and how to correctly interpret it. Doing that sort of work is the reason we have statistical agencies in the first place. W.R.T. survivorship bias, you seem to be implying that the shape of the graph is expected for any such plot for any vaccine? I don't quite understand what argument you're getting at, sorry. Could you be more explicit? If we zoom in then there's a clear wave shape in which after immunization cases quickly peak at 1000/day about a week after the first shot, then declines again. This is very different to the same data plotted from date of second immunization, where we see a small wave that subsides (same pattern) but then the much bigger one as immunity wears off after about 180 days.
  - majewsky 5y ago
    > People not getting second shots is not a surprise, anyone who has a bad reaction to the first will do that. I don't know any such person, but I know lots of people who had immune reactions to the first shot (like a localized inflammation) and who got the second shot anyway because that's nothing out of the ordinary. Including myself. In fact, most doctors will tell you outright that this is probably going to happen when you get the shot. After all, the vaccine is supposed to elicit an immune reaction. I would personally be more worried if I receive a vaccine and there is no immune reaction at all. > you seem to be implying that the shape of the graph is expected for any such plot for any vaccine For any vaccine with a two-shot schedule, yes. Most people will be getting the second shot after N weeks, so if you plot the incidence of anything across all people who have only received the first shot, you will have the bulk of the distribution between 0 and N weeks because receiving the second shot removes people from the statistical population.

majewsky 5y ago

> number of days between first immunization and COVID diagnosis Survivorship bias. People end up in this graph because they got a COVID diagnosis after the first immunization, but before getting the second immunization. That's only a very short time window on most vaccination schedules (3-6 weeks), so the only surprise here is that there is a long trail of people who got the first shot and then held off on the second shot for a long time. I can understand why they took down that graph. It's very misleading when taken out of context.

native_samples 5y ago
People not getting second shots is not a surprise, anyone who has a bad reaction to the first will do that. The lack of any proper tracking of such reactions means it has to be inferred from the data, but the explanation there is fairly obvious. W.R.T. taking it down, if public health statistics is open to misinterpretation they should explain why and how to correctly interpret it. Doing that sort of work is the reason we have statistical agencies in the first place. W.R.T. survivorship bias, you seem to be implying that the shape of the graph is expected for any such plot for any vaccine? I don't quite understand what argument you're getting at, sorry. Could you be more explicit? If we zoom in then there's a clear wave shape in which after immunization cases quickly peak at 1000/day about a week after the first shot, then declines again. This is very different to the same data plotted from date of second immunization, where we see a small wave that subsides (same pattern) but then the much bigger one as immunity wears off after about 180 days.
- majewsky 5y ago
  > People not getting second shots is not a surprise, anyone who has a bad reaction to the first will do that. I don't know any such person, but I know lots of people who had immune reactions to the first shot (like a localized inflammation) and who got the second shot anyway because that's nothing out of the ordinary. Including myself. In fact, most doctors will tell you outright that this is probably going to happen when you get the shot. After all, the vaccine is supposed to elicit an immune reaction. I would personally be more worried if I receive a vaccine and there is no immune reaction at all. > you seem to be implying that the shape of the graph is expected for any such plot for any vaccine For any vaccine with a two-shot schedule, yes. Most people will be getting the second shot after N weeks, so if you plot the incidence of anything across all people who have only received the first shot, you will have the bulk of the distribution between 0 and N weeks because receiving the second shot removes people from the statistical population.