All deductions would include deaths avoided by the vaccine. If there are ten poisoned people, and you give them all an antidote, then one dies of poison and on

All deductions would include deaths avoided by the vaccine. If there are ten poisoned people, and you give them all an antidote, then one dies of poison and one dies from a stab wound; the net effect was eight lives saved. If the antidote killed one person you might argue it was net seven lives saved. Google "Net breakthrough deaths". You are not describing a real metric. You would be better off describing this as "total deaths from covid among vaccinated individuals". > If anyone is willing to think, the increasing rate is problematic, because, naturally, there is a lag between when one is vaccinated and infected. Another lag between infection and hospitalization. Another lag between hospitalization and death. Because we can actively compare current deaths of vaccinated individuals vs unvaccinated individuals, and compare them to our bayesian priors as you tried to imply you were comfortable with. If the vaccine is a flat 50% reduced chance of death, and you vaccinate the highest risk population, you might still see the vaccinated population having a higher death rate even if vaccine is working perfectly. Your math is pointless.