[riccia]

I think along the lines you're thinking, you would actually want to come up with an expected cost of an infection of an individual ("cost" being in a misery-illness-utility sense, not financial). This would presumably include the average expected illness severity of the person infected but also the average illness burden in others indirectly attributable to that person being infected.

So you'd want to give it first to people who are very likely to become severely ill themselves, and whose infection would cause others to become severely ill, directly or indirectly (eg through disease spread or through lost resources, as if an ER doc in an understaffed area became severely ill).