Japan: Second-hand smoke indicative of lung cancer

This week we read the paper by Takeshi Hirayama titled "Non-smoking wives of heavy smokers have a higher risk of lung cancer: a study from Japan."  The study was a cohort study, measuring the difference in risk of disease over time.  The study followed a sample of 540 non-smoking wives aged 40 and above for 14 years in 29 health centre districts in order to assess the incidence of lung cancer in relation to the smoking habits of their husbands; it also addressed the dose-response relation.  The results indicated that non-smoking wives of husbands who smoke heavily were indeed at greater risk for developing lung cancer.  Interestingly, the risk was greater in agricultural families.  The experimenter determined that the  husbands' smoking habits did not increase their wives risk for other disease such as stomach cancer, cervical cancer, or heart disease.


There are two major things I found particularly interesting in this study.  The first is the finding that the risk was greater in agricultural families most likely due to the longer period of mutual contact between couples in rural families.  The study pointed out that in urban families, some couples meet only for a short period in the day.  I found this ironic because it suggests a wife is hurting herself by spending more time with her husband.  In addition, I thought the statistic that 73% of Japanese men were smokers, while only 15% of women was interesting.  To me, this was an important piece of information to readers as it suggests the problem of second-hand smoke is exaggerated between spouses in Japan verses, for instance, the United States.  This difference must be inherently a social construct, but I wish the experimenter had included a sentence explaining why there is such a difference.  I'm not sure what this statistic would be in the United States, but I would guess that the numbers are much closer.  Also, the p value 0.0001 describes the close relation between the amount husbands smoked and the mortality of their non-smoking wives from lung cancer.  As we learned in class, the p value measures the probability that something actually happened by chance alone and a small p value indicates a statistically significant relation.


As mentioned, the experimenter performed a cohort study. Every design has its own advantages and disadvantages, and in lecture we learned about the strengths and weaknesses of this design.  Because of its design, the experiment was able to assess the study of relatively common exposures and calculate incidence while examining--and ruling out--other possible outcomes linked to exposure.  The study required a large sample size, which was met, but although expenses were not mentioned, the fourteen-year long follow up was probably costly.    


Although it is never directly written, the p value and strong correlations between smoking husbands and non-smoking wives incidence of lung cancer suggests a causal relationship.  The mechanism by which second-hand smoke amounts to lung cancer is not stated, however, so this cause cannot be proven concretely.  Schneider explained how a large number of subjects is more likely to yield valid results compared to small numbers, and he also emphasizes that the stronger the association measured between exposure and disease makes for a cause-and-effect relationship; Hirayama meets both of these standards.  Hirayama also assesses the dose-response relationship between exposure and risk of disease, which Schneider states as additional evidentiary support for a causal relationship.  Thus, although cause and effect may not be proven, there is strong statistical evidence in support.