This blog discusses effects of second-hand smoke. The information regarding the effects comes from two different sources: Hirayama’s report, “Non-smoking wives of heavy smokers have a higher risk of lung cancer: a study from Japan” and Chapter 6 in Schneider’s book, Introduction to Public Health.
Hirayama’s study followed 91,540 wives of smokers from 1966-1979. These women were 40 and above and did not smoke themselves. The study found that men who smoke heavily (>20 a day) had a higher risk of lung cancer but not stomach cancer, cervical cancer or ischemic heart disease. The advantage of how this study was designed it incorporated of women of different ages. People are not going to develop lung cancer overnight; therefore looking at women who have been with smokers for different lengths of time was beneficial for accurate results.
I was surprised at the results stating that wives of smokers in agricultural settings are more at risk than the women in urban areas. By definition women in urban areas are in more contact with others (and particulate matter) than rural women. The proportion of smokers in urban areas is reasonably the same as it is in rural areas (or at least what I have observed). So if the proportion of smokers are reasonably the same in both rural and urban areas (who are more at risk), why would more rural ladies have lung cancer? Yifan Ding (from the Institute of World Development and Research Center) seems to have an answer for that question – economic reasons.
Ding published the paper “Social and Economic Disparities”. (http://www.eolss.net/Sample-Chapters/C13/E4-25-04.pdf) She found that there is an economic disparity between people living in urban and rural areas. Rural people having a lower income level than those living in urban settings. Now how does this relate to Hirayama’s study? Well, most studies pay the participants in studies, as Christine Grady (PhD in bioethics and is in charge of human subjects research at the National Institutes of Health) states that “not only [is it]... ethical to pay people, it might be unethical not to pay them”. (http://stanmed.stanford.edu/2008summer/just_another_lab_rat.html) While this study did not pump participants with untested drugs, it is unlikely that the participants were not paid. If someone is short on money - mostly rural women, (as Ding showed in her paper) they are willing to jump at chance to make money. The study would have wanted similar numbers of non-smoker women with non-smoker husbands, non-smoker women with smoker husbands and women with smoking habits (their husbands not specified). Poorer women are more likely to tell those running the study whatever necessary to ensure their ability to participate. Thus one cannot guarantee that all the female participants were not smokers or did not have other conditions that would increase their chances of cancer. Therefore, the study should have considered the different economic statuses as an error possibility.
Schneider also discusses the problems that come with human studies, such as lifestyle and other conditions that would affect the human subjects and thus messing with the correlation results. Schneider’s comments do not necessarily support or reject Hirayama’s results. Her comments simply remind us that there are other factors, not just the specific ones that a study considers and that should be taken into account.
Hi Kat - thank you for your blog! You made some really interesting points, and I love that you thought about the possible selection bias with rural vs urban participants. I agree that if money is offered, it can attract some more than others, and it makes sense that rural populations are poorer than urban ones and so might be tempted to participate when they otherwise wouldn't. Just be careful in your assumptions - studies have to disclose when they're paying their participants in the methods section so that the reader can draw their own conclusions about biases. But nice job thinking about this and looking up more background information!
ReplyDeleteAlso be careful with percentages versus total numbers. In your third paragraph, you mention that more rural women have lung cancer, which is puzzling to you because the percentage of women who smoke is the same between urban and rural settings. The study actually does not look at the overall proportion of women who smoke or who don't smoke or break it down by seting in which they live (rural vs urban) - the authors select participants based on their smoking status or lack of it. What the study shows is that among non-smoking women who have smoking husbands, lung cancer rates are higher than among non-smoking women who have non-smoking husbands. So in other words, it's looking at a proportion and not at the overall number of women in any particular setting.
I liked your inclusion of causality criteria. This study is very well-done, with a very large sample size, long enough follow-up period to allow outcomes to occur, covering a broad geographic area so as to decrease the likelihood that region exposures are causing any found associations, and measures multiple outcomes (emphysema, CAD, cervical cancer) as well as several possible confounders. It is a convincing study and meets many of the causality criteria such as large study, large relative risk, biologic basis, and dose-response findings.
Good job on this blog - I look forward to reading more!