It's definitely not 0. Taking an aspirin will thin your blood just slightly, and irritate your stomach lining just a little bit. There's a chance (though of course only a small chance) that one of those things will have consequences that kill you.
There's also a chance that it'll have consequences that save your life -- e.g., making congestive heart failure a little bit less likely.
It seems uncontroversial that there may be nonlinear effects at large doses -- after all, if some dose is large enough that it almost certainly kills you then it's simply not possible for twice the dose to be twice as likely to kill you. But surely the estimates you're concerned about are not obtained by linear extrapolation from such large doses; they're the result of extrapolating from known statistics for doses small enough to be unlikely to kill, but not so small that the risk can't be measured because it's overwhelmed by background noise.
Those extrapolations could still be wrong, of course. But comparing against something that's obviously wrong like going from 100 aspirins to 1 aspirin is not a fair comparison, and scores much better on rhetorical effect than on actual evidential force.
There's also a chance that it'll have consequences that save your life -- e.g., making congestive heart failure a little bit less likely.
It seems uncontroversial that there may be nonlinear effects at large doses -- after all, if some dose is large enough that it almost certainly kills you then it's simply not possible for twice the dose to be twice as likely to kill you. But surely the estimates you're concerned about are not obtained by linear extrapolation from such large doses; they're the result of extrapolating from known statistics for doses small enough to be unlikely to kill, but not so small that the risk can't be measured because it's overwhelmed by background noise.
Those extrapolations could still be wrong, of course. But comparing against something that's obviously wrong like going from 100 aspirins to 1 aspirin is not a fair comparison, and scores much better on rhetorical effect than on actual evidential force.