The study is an example of a cohort study, since it was done with a large number of subjects and over an extended period of time. It dealt with some confounding variables (i.e rural vs. urban environments, age groups, drinking habits of the husbands) and it also considered the dose-response relationship between exposure and risk of disease.
In my opinion, asserting that second hand smoking is a "cause of lung cancer" rather than a "factor of risk" is a very strong accusation, especially because of unavoidable bias in the control group that might have lead to associating variables by chance.
In this particular case, bias might not been a problem in reporting, because the population was interviewed and followed by establishing a record linkage system between the risk-factor records and a residence list obtained by special yearly census, with death certificates. However, I do see bias as a big problem regarding the controlled variables in the group of non-smoker women. The variables were set upon what their husbands were doing, but what about what the women were doing? Were they drinking? Were they doing exercise? What were their diets like? Did other people who they were around continuously smoked? All of these confounding variables are not mentioned in the study, and that makes me question the certainty of the results because of a bias selection of control groups.
Also, I find incredible how passive smoking is associated with a disease as complex as cancer and not with the obvious asthma that comes with breathing contaminated air. The weak association between the exposure and disease and the lack of a biological explanation that addresses this association also takes away from this study credibility and reliability.
When we are analyzing a study of such a complex nature and large dimension it is important to recognize the limits of epidemiological research. For the most part, this epidemiologic study no matter how well designed it was to avoid error, it cannot prove cause and effect because of its susceptibility to many sources of error. As it was proven in this study, confounding factors may influence results, suggesting an association that cannot be entirely proved; and bias might have also played its part in the selection of control cases. Asserting the wrong conclusions could have enormous consequences, so one must always be skeptical on reading on these studies and very careful and perfectionist when conducting one.
With a large population of randomly selected women, the habits of the women will approximate the population. Because the only factor being studied is then the one being selected for: nonsmoking with husbands who are smokers. The risk-taking behavior of wives of smokers has not been shown to be higher than those of wives of nonsmokers. Also the study mentions that the differences between rural and urban may mostly be due to differences in habits, but both groups show increased risk of lung cancer. I believe your view of this article does not take into account that the researchers were careful in limiting the effect of confounding variables.
ReplyDeleteThe study itself was limited to the relationship between nonsmoking wives with heavy smoking husbands. I can see why you wonder about the other possible factors that can influence the study, but i believe the study took care of many important confounding variables. The difference between agricultural and urban couples helps us determine that the risk of lung cancer is increasing because of second hand smoking rather than other factors such diet, drinking,etc. My only concern with this study is that it was only performed in one country, making it hard to generalize these statistics.
ReplyDeleteAlthough I still think that this study is a very well conducted study, I do see many of your points. I agree that there are many confounding variables regarding the behaviors of the wives and who they were around. Though I think that most women in Japan, especially middle age women, are housewives who hang around with other housewives, so I don't think that knowing that would change things too much. I mean, this is from what I know about Japanese culture. I do still see your point that there is no "perfect" study, but since there is there is no perfect study, this is very close to it.
ReplyDeleteI like that you are thinking about cause and effect, and that one cohort study like this does link but doesn't prove causation. However, we do have multiple studies that do point to causation for smoke and passive smoking - it is important to note that there are some things we can't study using randomized controlled trials. But when there are many cohort or case control trials showing the same thing, we can accept that as evidence. Passive smoking and elevated cancer risk is one of them.
ReplyDelete