You seem to be missing another option: that treatments get better so that hospitalization leads to fewer deaths than it did earlier. Which seems obviously true.

You seem to be missing another option: that treatments get better so that hospitalization leads to fewer deaths than it did earlier. Which seems obviously true. Why would deaths be the best proxy for hospitalization? Why don't you have actual hospitalization numbers?