Does Smoking Cause Lung Cancer?
Of course it does. But what was the evidence in favor of the hypothesis, what was the evidence against it and how was a judgment about what lawyers call general causation finally reached? Fifty years ago a remarkable paper was published that demonstrates how a causal inference is rationally reached.
The paper itself is important to anyone trying to understand the rise of epidemiology, its methods and the profound respect it earned; and its reasoning is important to anyone who wants see how a causal hypothesis is properly dipped in the acid bath of skepticism; and what it looks like if it survives.
In the case of cigarette smoking the effect, lung cancer, was nine fold higher in smokers than in non-smokers. In retrospect it was an easy case. The question today of course is whether epidemiology can shed any light on subtler risks attributed to suspected carcinogens which are themselves just one part of a staggeringly complex causal web involving genes, epigenes, pathogens and other as-yet undiscovered causes. More about that in a future post.
