It's extremely hard to extract meaningful insight from population data, because it's massively confounded. Recently some locations locations have reported omni

It's extremely hard to extract meaningful insight from population data, because it's massively confounded. Recently some locations locations have reported omnicron case _rates_ being higher in the vaccinated population than unvaccinated. Your logic would have us conclude that the vaccination increases the risk of becoming a covid case with this strain. ... in reality, the vaccinated population is likely more at risk and the vaccine protection isn't enough to offset the difference. But it's extremely hard to tell, which is why RCTs are so important. Those one in a million events also don't get found outside of trials: They go unreported because outside of trials surveillance is very low, or if noticed they'll get attributed to something other than vaccination because without a control group you can't establish a baseline rate narrowly enough to attribute it to the treatment with any confidence.