Saturday, February 9, 2013

The Smoker's Wife

In Non-smoking wives of heavy smokers have a higher risk of lung cancer: a study from Japan, a paper published in the British Medical Journal in 1981, Dr. Hirayama set up an observational study to see if second hand smoking is related to lung cancer. From 1966 to 1979 (14 years), Dr. Hirayama tracked the prevalence of lung cancer in 91,540 non smoking wives aged 40 and above. He categories the smokers (and non smokers) into groups of different amounts of cigarettes and compares the prevalence of lung cancer in the wives of these smokers between these different categories of cigarette usage. With a p value of 0.00097, Dr. Hirayama concludes that the wives of smokers have a statistically significant higher risk of lung cancer than the wives of non-smokers do; sequentially, he reports that, although not as much as first hand smoking, second hand smoking causes lung cancer.

In the way and manner this paper is written, finding out that second hand smoking can cause harm seems to be shocking and somewhat cutting edge. This study probably reached headlines and causes many a smoker to quit on behalf of the health of their loved ones; this, of course, is a good thing because it most likely reduced the number of smokers in the world, decreasing the potential years of life lost globally. On top of that, companies can produce and develop alternative methods for nicotine addict to get their fill of nicotine without lighting up and causing second hand smoke (which I think could have stimulated the economy).

Of course in order to include so many subjects in the study, Dr. Hirayama had to do an observational study. He couldn't, with relative ease, create a true experimental study. Doing an observational study would have been the best choice economically and logistically.

But the main problem of doing an observational study is that there are many more variable which are not controlled which causes the association to not immediately mean causation.If we look at Hill Criteria for causation, all the criteria checks out (strength of association, consistency of association, high number of subjects). Just like the study that concluded that coffee caused pancreatic cancer with a significantly low p value (when it was really caused by tobacco), the association found in this study may be cause by a confounding factor(s). For example, husbands who smoke may be more depressed, and this depression is shared with the wives, which in turn causes a pathway that leads to lung cancer; or perhaps, smoking husbands have less disposable income, so the wives are more stress in the accounting department, which causes lung cancer. I do believe that that second hand smoke causing lung cancer seems to have a logical biological pathway, but there could still be a middle man factor that is the actually cause of lung cancer.

In any case, it may be better for wives, kids, husbands to reduce cigarette consumption for their own and the health of other. If needed be, those who really needs the buzz or effects from nicotine have many alternatives to cigarette to chose from nowadays; There are electronic cigarettes, nicotine patches, and gums. Of course they are relatively new and the long term effects of using them are not known yet, but, without a doubt, they don't produce second hand smoke that causes harm to people other than the user.


  1. I especially liked reading your fourth paragraph of your blog post. It was helpful to include concrete details and terms such as p value and Hill Criteria to thoroughly explain the results of and problems with the study. I also liked how you included various other options a smoker could do to have the same feelings of smoking but lessen the affects it has on others. Overall, nice job on this week's blog assignment. I really enjoyed reading it.

  2. Thanks for the post!

    For decades, public health advocates have attempted to link cigarette smoking and lung cancer. The attempt to make a causative link between smoking and lung cancer has been stymied by the long timespan prior to the development of lung cancer, the relative rarity of the disease and the numerous confounding factors that are concurrent in individuals who smoke. This landmark prospective cohort study sought to show an association between second-hand (aka side-stream) smoke exposure, lung cancer incidence, overall mortality and lung cancer attributable mortality. For more information about criteria of causation please see the Bradford Hill criteria:

    The study designers utilized a cohort design in order to analyze the risk associated with second hand smoke exposure (risk ratio = relative risk). Cohort studies are often difficult to accomplish due to the length of follow-up time, the overall cost of the study and the possibility of a large number of study participants to be lost to follow-up. In contrast, case-control studies identify cases (those with disease) and controls (those similar but without disease) and attempt to distinguish different rates of past exposure. This study sought to prove that smoke exposure was specifically related to an increase in lung cancer incidence, environment of residence is not a confounder of this relationship and that smoke exposure and lung cancer incidence has a dose response relationship (more smoke exposure causes more cancer). The authors further point out that considering their results, future comparisons of the health status between smokers and nonsmokers should control for the secondhand smoke exposure of nonsmokers.

    No study is perfect and you rightly discuss some other possible confounding factors that could be at play in the association of smoking, lung cancer and lung cancer related mortality. Other alternatives to a randomized control study (in this case would be unethical) would be to see if the lung cancer related mortality improves after a husband stops smoking. This study attempts to show a dose-response relationship between smoking, lung cancer and mortality, further supporting the lung cancer/smoking association. While this study followed conventional understanding, there were strong economic interests on behalf of tobacco companies to fight tooth and nail any association of their product with adverse outcomes. Here is some more on the topic: