A similar argument can be made about cigarettes and lung cancer. One cannot point at a tumour, even in a heavy smoker, and say with certainty "well, that's fro

A similar argument can be made about cigarettes and lung cancer. One cannot point at a tumour, even in a heavy smoker, and say with certainty "well, that's from the smoking", as opposed to air pollution, or just bad genetic luck. (Or quite possibly a synergy of all three.) When you look at it statistically, it's like what you say, smoking lowers your life expectancy on average. Yet if you took two large and matched cohorts, one smoking and one non-smoking, after 50 years the smoker population would be much smaller than the non-smoking. On the population level, I think it's reasonable to attribute those missing people to having died from smoking, even though individually we can't say much.