Originally Posted by kathymuggle
That is incorrect. You need to read the study.
They were found to be insignificant because they did not meet all 3 of the criteria for significance as designed by the study: an odds ratio of more than 3, having 20 or more case, or being consistent in both countries:
"We regarded three criteria as signal strengthening: analysis based on 20 or more vaccine exposed cases (reliability of analysis); a rate ratio of 3.0 or more (strength of association); and significantly increased rate ratios in both countries when analysed separately (consistency)."
I am interested in odds ratios of 2, or even less than 2, so the "3" is high, to me.
The number 20 makes sense in some ways - but will fail to capture rare(ish) reactions. The authors acknowledge rare reactions might be missed. If you keep in mind 300 000 were given the vaccine, 19 could have had a reaction above and beyond the expected rate in the population, and it would not be considered significant. I am going to round for ease of math, but 20/300 000 means 1/15 000. They could dismiss something with an adverse rate of 1/15 000. Depending on the type of reaction or issue we are talking about, that could be pretty significant.
I do understand why they would want consistency between the countries - although if there is not consistency, they need to look at why.
The rate of diabetes is 1/5,000.
Odds ratios shouldn’t be taken out of context of their statistical power. With so few cases, the error is huge. Type 1 diabetes, for example, was barely above unity for odds ratio, which means there is no risk. At 1/15,000 you’re at the level of background noise.The argument that just because they observe something it must be causal, requires a lot more information than that. Again, the background levels of diabetes is about 1/5000, so you’re asking an epidemiological study to find something that is so rare that it would be incapable of showing causality.
The authors eliminated the three effects for numerous reasons, one of which was the background rate (NOT the rate compared to the non-vaccinated group). But they mentioned others like random spread of time after vaccination to disease onset. Diabetes may have actually started well before the vaccination, or the day before, or naturally a week later. Random spread implies non-causality.
Odds ratios of 2 or 3 are not that significant because of the low numbers. It’s a logical fallacy to believe that because we didn’t see an adverse effect, it might exist. No, that’s not how science works. Until there’s evidence, there is no claim. Especially if you’re starting to look for things that are causal, because the numbers are so small that you’re trying to invent a causation when there’s barely a correlation.
It’s ironic, but for a bunch of adverse effects, they showed LOWERED rates of said effects in the vaccinated group compared to the unvaccinated group. Notice that neither I nor the authors made a big deal of that, because the numbers were so small. So, if you are going to ignore logic and statistics for results that you cherry pick to prove you point, then by logic, you need to accept the data that says the adverse effects were actually lowered by vaccination. It’s only fair. :D