Scientific Evidence Portal
Assessment of risk from active and passive smoking, with particular reference to the epidemiological evidence | P.N. Lee
Article Published: 1991
Type: Articles and Dissertations
Further Information "Many claims of possible health effects of active and passive smoking are made in the scientific literature and in the media. This document is intended to assist the lay reader in gaining a better understanding of the strengths and weaknesses of such claims."
It is divided into 8 main sections;
2. Types of epidemiological study
3. Quantification of risk
4. Problems of causal inference
5. Types of bias and error
6 . Some major sources of epidemiological data on smoking and health
7. Risks of smoking
8. Risks of ETS exposure.
9. Effects in pregnancy, infancy and childhood"
This educational paper explores risk based on epidemiological evidence, and explores why, at best, epidemiology allows some educated guesses, but it never can establish causality – ever. Therefore, when we read in the mass-media that an epidemiological study has “demonstrated”, “established”, or “proven” that A "causes" B, the layperson immediately thinks of sage men in white coats with microscopes observing a phenomenon right under their eyes. Then, if they “believe” that “smoking kills” because they don’t like their own habit or hate smoking, that becomes “conclusive proof” – especially if pronounced by “authority” – as it confirms their beliefs. Conversely, if it is contrary to their belief, alternative explanations are immediately searched out to minimize the disappointment and cling to the belief (example: "are they paid by the tobacco industry?", or "are their credentials good enough"?).
In reality, that epidemiological study has scientifically demonstrated nothing. It only reports the persuasion, opinion, beliefs and biases of those who are interpreting data that is usually gathered with questionnaires, the answers to which are most of the times unverifiable.
Furthermore and foremost, epidemiological studies concern themselves with populations and never with individuals, and this is something that the layperson has trouble understanding. That trouble is dishonestly exploited by antismoking propaganda on an emotional basis. The fact, for example, that smokers are at greater statistical risk of lung cancer does not and cannot — ever — prove or demonstrate that your uncle Tom died of lung cancer because he smoked. The lung cancer could have been caused by any of the other suspected factors, in part or in total. You cannot know — and neither can your doctor in spite of whatever his persuasion may be.
As it is stated at page 33 of this document (emphasis added):
“Having considered all the possibilities noted above, it
may be possible to conclude with reasonable certainty that an epidemiologically observed association is due to a cause and effect relationship. It should be noted that this does not imply that one can conclude whether any specific individual died from or contracted the disease of interest as a result of exposure.”