This study is unscientific, but the criticisms lodged by many on the thread are without merit. Yes, everyone has a 100% risk of death, and smoking does not increase this probability. What the study claims is, however, made clear in the following quote: "Smoking status data was available for all household members aged 15 and over, and
death rates were monitored for three years after the two censuses."
The death rate increases by 15% for those who live with smokers, compared with those who do not, over the three-year period following the census. If one's risk of death is 1 death per 100,000 people per year living with non-smokers, then according to this study, one's risk of death would be 1.15 per year per 100,000 people. That's all.
However, the study fails in many ways.
- By not including the cause of death, the connection to smoking is shaky. Car accidents?
- Error bars are not included - what if one's probability of death was 1.0 +/- .6 per year per 100,000 people in non-smoking households, and 1.15 +/- .6 per year per 100,000 people in smoking households - presto! the effect is smaller than the error bars, and might be statistically insignificant. This is, in fact, the case with the WHO study that found a similar effect for the health detriment of secondhand smoke (SHS) - the effect was well below statistical significance.
- The study takes many things into effect, but not everything. "Even after taking into account age differences, ethnicity, marital status and socio-economic position," still leaves diet and exercise differences, for example. This is the problem of the uncontrolled variable.
- The study also fails to distinguish between correlation and causation. It might be that people who smoke have a genetic tendency towards addiction, and this tendency also is responsible for a very slight increase in death rate, and non-smoking family members share the genetic trait. This may be far fetched, but it can't be excluded based on this article.
- Finally, the study is intended to make you conclude smoking is bad, even if you don't smoke yourself. And this might, in fact, be true. But what it will be used for (undoubtedly!!) is to push legislation that bans smoking in some or all public places, or in some or all private places. As bad as SHS might be, this study does not make the case that legislation that bans smoking is beneficial. (For an analogy, guns might indeed be used for murders, but laws banning guns affirmatively does not reduce murder rates.) It does not even attempt to make this case. Laws that prohibit common activities anger people, cost money to enforce and create black markets, among other things.
- angry people increase murder rates (for example, in NYC a bouncer was murdered over a smoking ban),
- enforcement costs money that could have been better used elsewhere, for health care or quitting programs, for example, or simply not taken from people in the first place, making them richer - and poverty increases death rates too.
- black markets are criminal enterprises that can lead to violent crime and hence higher death rates.
This study certainly fails to even attempt to account for these effects, and they will become issues if, in fact, any prohibitions result from people citing this study.
- Even if we were to grant that living with smokers all the time does increase one's risk of death some small amount, the case is not made that exposure to SHS at the hour-a-day or hour-a-week levels are proportionally as harmful, or even harmful at all. It might even be beneficial, look up "hormesis." (For example, alcoholism certainly increases death rates, but this does not say that the best amount of alcohol is none - a glass of red wine a day increases life expectancy both by its beneficial antioxidant content, and by keeping one's liver in shape without killing it.) A study that compared death rates of non-drinkers who lived with alcoholics vs. those that didn't would certainly find a hazard to living with alcoholics, and probably far more so than than non-smokers living with smokers, but if one banned alcohol based on this, one would miss the red wine effect altogether, to say nothing of the gang violence, the ascendency of the Kennedy family into politics via corruption, and the expansion of federal power that did result from alcohol prohibition - namely, the BATFE and the first federal gun control law, NFA'34.
By the way, I'm a non-smoker and don't like cigarette smoke, but I really can't stand junk science, and especially people using junk science to expand the nanny state.
I'm bumping the War On Drugs list for reference of the proto-war-on-tobacco, coming soon to a country near you, and because the same arguments apply to other drugs that were made here for tobacco.
As a followup to this post, the very next day the following article appeared:
Study: Lesbian & Bisexual Girls at Heightened Risk for Tobacco Use . If gay women are slightly more inclined to smoke, then non-smokering gay women might be slightly more inclined to room with smokers, in which case the detrimental health effects of being lesbian come out as appearing to be the result of SHS. The article does not allow one to exclude this hypothesis as the explanation for the perceived effect.