> that certain very specific substances are bad for (for the general population) if you consume them too often

There are people who smoke their entire lives, die at 90, and tobacco had nothing to do with their individual deaths or even really any tangible negative health outcomes. There are people who drink or smoke pot every day and it has nothing to do with their deaths or quality of life. There are people who have steak and eggs cooked in butter every morning with no cholesterol or cardiac problems.

As a whole, people who do these things have a statistically higher probability of having negative outcomes. On an individual basis, there is a lot of variation as to what the actually effect might be.

Poison moves this probability towards 100%.