This problem exists with or without finding an attenuated strain. A better way to think about this is let’s say we had two strains of the virus isolated - one that killed 10% of people infected and the other that killed less than 0.0001%. Which one should we use in a challenge trial?