This, and also hospitalizations are a less exploitable metric. Self-selection bias isn't much of a problem, and the number of tests being done doesn't influence

This, and also hospitalizations are a less exploitable metric. Self-selection bias isn't much of a problem, and the number of tests being done doesn't influence the results. I would look at either hospitalizations or deaths once vaccinations reach a large percentage of the population.