Paradoxes in the reporting of Covid19 vaccine effectiveness: Why current studies (for or against vaccination) cannot be trusted and what we can do about it
Given the limitations of the randomized controlled trials (RCTs) for Covid19 vaccines, we must increasingly rely on data from observational studies to determine vaccine effectiveness. But over-simplistic reporting of such data can lead to obviously flawed conclusions due to statistical paradoxes. For example, if we just compare the total number of Covid19 deaths among the vaccinated and unvaccinated then we are likely to reach a different conclusion about vaccine effectiveness than if we make the same comparison in each age category. But age is just one of many factors that can confound the overall results in observational studies. Differences in the way we classify whether a person is vaccinated or is a Covid19 case can also result in very different conclusions. There are many critical interacting causal factors that can impact the overall results presented in studies of vaccine effectiveness. Causal models and Bayesian inference can in principle be used to both explain observed data and simulate the effect of controlling for confounding variables. However, this still requires data about relevant factors and much of these data are missing from the observational studies (and the RCTs). Hence their results may be unreliable. In the absence of such data, we believe the simplest and most conclusive evidence of vaccine evidence is to compare all-cause deaths for each age category between those who were unvaccinated and those who had previously had at least one vaccine dose.