I'm curious how health risks (mortality, lifetime probability of cancer, etc) change with cigarette consumption. Specifically, treatring cigarette consumption like a continuous variable, rather than a binary for "smokers" and "nonsmokers". I've got access to academic literature, but I'm not a doctor or a PhD health researcher. This article states in the abstract that the dose-response relationship between cigarettes and incidence of lung cancer is linear. This website (CDC) states that "smokers" have a 25-fold increase in risk of lung cancer.
How to reconcile the two? The latter reports odds ratios, which means that they did a logistic regression, which means that the raw logistic regression coefficient was 3.2. If the average smoker smokes 20 cigarettes a day, and the dose-response relationship is linear, then the relative risk for someone who smokes one per day should be exp(0.32) = 1.37. So someone who smokes one cigarette per day would be 1.37 times more likely to get lung cancer than someone who never smoked (on average).
Is this a reasonable estimate? I can certainly understand why public health types wouldn't promote this sort of conclusion, as smoking one cigarette can be a slipper slope to smoking a lot more cigarettes.