When that happens, which is pretty normal (ie: Einstein's theory of relativity is also not all proven and still a theory, and it doesn't mean we ignore it outright) we use risk analysis instead, which have nothing to do with a coin toss.
There's really two possible outcomes here:
1) a person with natural immunity and vaccine is more likely to reduce infection and transmission, saving lives.
2) a person with natural immunity and vaccine is no more likely to reduce infection and transmission than a person with natural immunity and no vaccine, saving no additional lives.
One option has a potential to save lives, and in the worst absolute case, it'll match the second option. Thus, it's a no brainer.