In
this study by Takeshi Hirayama, the researchers hoped to prove that smoking not
only affects those directly in contact with the smoke (the smokers themselves),
but also those surrounding the smoker (the wives of the smokers). To do this they performed a prospective
population survey of 265,118 adults in Japan on health and things such as
alcohol, drinking, and occupation. After
this study, they performed a follow up and found 346 women had died from lung
cancer. The factor that had the
strongest correlation with those women that contracted and even died from lung
cancer was if the woman was the wife of a heavy smoker. No other factors, such as the husband’s
drinking habits, seemed to have a statistically significant correlation.
For what the researchers were trying
to accomplish, the survey was certainly the best choice of experimental
design. It allows them to reach out to
and obtain information from thousands of people. If they had done any other type of
experiment, they would have only reached out to a few hundred. With so few people being tested in this
scenario, they would not have produced convincing results. To make their results even more convincing,
the researchers addressed many confounding variables. Their ability to discredit or provide
evidence against confounding variables is probably what makes the argument so
strong.
One problem, however, arises. The survey really acts as a correlational
study. In other words, the results do
not prove that smoking causes lung cancer in the wives of smokers. It only proves that the incidence of women
contracting lung cancer is higher in women that are married to a heavy
smoker. What we are left with is a
strong case for the causative role of smoking on the development of lung
cancer, but not a true factual argument.
This is probably the biggest weakness of the article, but is nonetheless
unavoidable. To really prove causation,
this study should be paired with an experimental study. Unfortunately, it is ethically wrong to force
someone to start smoking and then see if his or her significant other contracts
lung cancer. Therefore, the study at
hand will have to do.
I found this study to be compelling
for many reasons. Instead of proving
that smoking causes cancer, this study proves that second-hand smoke also
causes cancer. Essentially, the argument
against smoking gains the upper-hand because now it is known to affect people
not even smoking, making it more deadly than previously thought. Additionally, I found it interesting that the
highest effects were found in agricultural families. I personally would have thought that city
dwelling and the cramped life therein would have proven to have stronger
effects. Therefore, I think it would be
interesting to do a follow up study to track down the reason why the
relationship was strongest in rural areas.
Tracy -
ReplyDeleteI completely agree with everything you've said. Obviously we all read the same article, and so the overall idea that we all received from that article was probably the same. Though I had not written it in my post myself, the experiment was well made with that one exception. But like you mentioned towards the end, it's not as if you can force others to smoke in order to do a "test run." I guess you could always try using lab rats, having a control and an experimental, but I wonder as to how affective it would be in providing evidence for the argument.
I also, like you, found it interesting that the highest effects were found in agricultural families. Why do you think this is? It really would make more sense for people to have the highest effects by living in the city...but maybe it's because of the more exposure that those in the city have that they are not as easily as affected because they're body is more "used" to it? I'm not too sure if that makes sense but something that I had been thinking about while reading that paragraph you wrote.
Hi Wren,
ReplyDeleteThank you for your post. You make many good points about the article - that it is a large study with long follow-up that allows enough time for the outcomes of interest (lung cancer, other cancers, emphysema, etc) to be realized. Often, studies that are too small or too short find negative results not because they're not there, but because they are under-powered (not large enough) or not long enough to find statistically significant results.
Since you seem to have such a good grasp of the article, I want to challenge you on a few of the points you made. First, is this really the only study design that could be so large? Or so long? Second, I agree with your point that this is an association and not confirming causation, though, as you point out, the authors did control for many known confounders like the husbands' alcohol intake. Is it possible that there are other factors that they have not controlled for? (For example, cooking with open fires inside the home, which occurs in rural settings and is known to be associated with emphysema, or childhood exposure to second-hand smoke?). Be careful in the last paragraph when you say it proves that second-hand smoke causes lung cancer, when you've just had a full paragraph saying that it doesn't. That being said, this is really a powerful study which, for the study of second-hand smoking, is a convincing and appropriate design. study design in humans that makes the strongest argument for causation is a randomized control trial, which as you say you cannot do in this scenario because it's unethical.
Finally, the article does discuss the possibility that rural couples have been together longer, since people who live in rural settings tend to get married younger than urban couples. This could be one reason why lung cancer has a stronger association among this population.
You did a great job with this article - I can see that you thought about the important aspects of it and put real thought into it - nice job!
I think in order to make sure that people are happy about participating in the study is that participants were likely compensated. However the ethical questions of these sorts of studies remain.
ReplyDeleteI am also surprised that agricultural areas are more affected. I wonder if there was an even proportion of rural and urban areas. Is there also a chance that some people are predisposed to cancer and in the study they got people who were related?
Kat