Given that I’ve dedicated my life to treating cancer and researching the biology of cancer, the ultimate goal being to use that knowledge of cancer biology to develop ever more effective treatments directed at the specific molecular derangements that lead to cancer, it’s not surprising that I’m very much anti-tobacco. After all, arguably there is no known behavior that causes more cancer-related deaths than smoking tobacco. It’s not just lung cancer, either. Smoking causes a wide variety of cancers, chronic obstructive pulmonary disease (COPD), and cardiovascular disease, all of which result in huge numbers of deaths among smokers each and every year. Clearly, smoking tobacco is so harmful due to a variety of health problems that it can cause that it is rare to find anyone who tries to argue otherwise. Unfortunately, there are estimated to be over 1 billion smokers in the world, and the toll of their addiction includes cancer, lung disease, and heart disease, and it’s estimated that during the 20th century over 100 million people worldwide died due to tobacco-related diseases.
While it is clear that smoking tobacco is deadly to those who actually become addicted to it, what has also becoming increasingly clear is that secondhand smoke is also a health hazard. It turns out that I’ve dealth with this topic before, mostly in the context of dealing with those who would argue against the contention that secondhand cigarette smoke increases the risk of cardiovascular disease (and cardiac events in those with preexisting cardiac disease), chronic lung problems, and lung cancer. For instance, denialists like to try to argue that epidemiology can never demonstrate a relative risk less than two, a contention that no reputable epidemiology organization accepts and a contention that, if accepted, would force us to deny a whole range of health risks, including the relative risk of myocardial infarction due to, for example, Avandia. Indeed, six years ago Tim Lambert ably described the origin of the claim that epidemiology can’t confirm a relative risk of a magnitude less than two. Not surprisingly, this claim originated with Philip Morris as a strategy to downplay the risk due to secondhand smoke.
I may have left my blog (mostly) inactive during the recent Thanksgiving holiday weekend, but there was a story describing a study that appeared on the Friday after Thanksgiving and tries to come up with an estimate of the death toll worldwide that can be attributed to secondhand tobacco smoke:
Secondhand smoke kills more than 600,000 people worldwide every year, according to a new study.
In the first look at the global impact of secondhand smoking, researchers analyzed data from 2004 for 192 countries. They found 40 percent of children and more than 30 percent of non-smoking men and women regularly breathe in secondhand smoke.
Scientists then estimated that passive smoking causes about 379,000 deaths from heart disease, 165,000 deaths from lower respiratory disease, 36,900 deaths from asthma and 21,400 deaths from lung cancer a year.
Altogether, those account for about 1 percent of the world’s deaths. The study was paid for by the Swedish National Board of Health and Welfare and Bloomberg Philanthropies. It was published Friday in the British medical journal Lancet.
“This helps us understand the real toll of tobacco,” said Armando Peruga, a program manager at the World Health Organization’s Tobacco-Free Initiative, who led the study. He said the approximately 603,000 deaths from secondhand smoking should be added to the 5.1 million deaths that smoking itself causes every year.
Upon reading the story, I had to wonder: Does this study actually show that secondhand smoke results in such a large number of deaths worldwide? So, as usual, over the weekend I decided to go to the source, namely the World Health Organization study that was published online in The Lancet entitled Worldwide burden of disease from exposure to second-hand smoke: a retrospective analysis of data from 192 countries. The study, supported by the Swedish National Board of Health and Bloomberg Philanthropies, sought to estimate the global burden of disease due to secondhand smoke.
Before we look at risks due to secondhand smoke, it should be pointed out that this study found that large numbers of people worldwide are exposed to secondhand smoke. For instance, it was estimated that, worldwide, approximately 40% of children (aged 0-14 years for purposes of this study), 33% of male non-smokers, and 35% of female non-smokers were regularly exposed to second-hand smoke. In this case “regularly” was defined differently for children as for adults. For children the definition of secondhand smoke exposure included having one or both parents who smoked indoors or having another person living with them who smoked. These estimates were obtained mainly from the global youth tobacco survey (GYTS), which is a school-based survey of children aged between 13 years and 15 years that covers more than 120 countries. For adults, the definitions of exposure were based on either having a spouse who smokes or regular exposure to tobacco smoke at work and could be further characterized by the number of cigarettes smoked by the spouse, the duration of exposure, or the frequency of exposure in number of days per week. Overall, the highest proportions of the population exposed to secondhand smoke were found in Europe, the western Pacific, and parts of southeast Asia, with more than 50% of some population groups exposed.
Next, the investigators attempted to estimate the burden of disease from secondhand smoke as deaths and disability-adjusted life-years (DALYs) for children and adult non-smokers. The basis of these calculations were relative risk estimates for specific diseases and area-specific estimates of the proportion of people exposed to secondhand smoke, using comparative risk assessments using data from 192 countries. Specific diseases and conditions were chosen based on recent reviews that judged the evidence to be sufficient to infer a causal association with secondhand smoke and if national incidence statistic were available. Ultimately, the health outcomes chosen for children were:
- Lower respiratory infections (relative risk 1.55)
- Asthma onset (relative risk 1.32)
- Acute otitis media (relative risk 1.38)
For adults, the health outcomes studied were:
- Adult-onset asthma (relative risk 1.97)
- Lung cancer (relative risk 1.55)
- Heart disease (relative risk 1.21)
There are a number of other adverse health outcomes that have been linked with secondhand smoke, but investigators either didn’t have adequate data to study these outcomes or deemed the existing evidence not to be strong enough to support a causal relationship.
Once these outcomes were decided upon, the attributable burden of disease, in deaths or DALYs, was estimated for every outcome by multiplication of the population attributable fraction by the total burden attributable to that disease. To put it more simply, the attributable risk is the difference in the incidence of a condition between an exposed population and an unexposed population. Similarly, the attributable burden of disease is the incidence of disease that can be–as you might guess–attributed to the exposure to secondhand smoke. When it was all put together, the authors estimated:
This exposure was estimated to have caused 379,000 deaths from ischaemic heart disease, 165 000 from lower respiratory infections, 36 900 from asthma, and 21,400 from lung cancer. 603 000 deaths were attributable to second-hand smoke in 2004, which was about 1·0% of worldwide mortality. 47% of deaths from second-hand smoke occurred in women, 28% in children, and 26% in men. DALYs lost because of exposure to second-hand smoke amounted to 10·9 million, which was about 0·7% of total worldwide burden of diseases in DALYs in 2004. 61% of DALYs were in children. The largest disease burdens were from lower respiratory infections in children younger than 5 years (5,939,000), ischaemic heart disease in adults (2,836,000), and asthma in adults (1,246,000) and children (651,000).
One aspect of the results of this study that were particularly disturbing is that deaths due to secondhand smoke were skewed toward poor and middle-income countries, where children tended to die of lower respiratory infections associated with secondhand smoke. In Europe’s high-income countries, only 71 child deaths were recorded for this study, while 35,388 deaths were in adults. In contrast, in Africa, there were an estimated 43,375 deaths potentially attributable to secondhand smoking in children compared with 9,514 in adults. This is very much like the case with infectious diseases, where the death toll in poor African countries is much higher than in wealthier, more developed countries.
By any stretch of the imagination, these are huge numbers. However, although there’s little doubt that secondhand smoke is an indoor pollutant that has the potential to result in health problems in nonsmokers, just how good are these estimates? Part of the problem with this study is that these estimates rely on a number of assumptions. These include the accuracy of the meta-analyses used to estimate the relative risk for each disease due to secondhand smoke and assuming that the fatality rate for each condition associated with secondhand smoke was the same as the mean case fatality rate for the disease in general. This latter assumption I found particularly troubling, not because it isn’t a relatively reasonable assumption in most conditions but because making such an assumption over 192 nations may not be valid. One reason is that fatality rates for each disease might be very different depending upon the nation, given the large differences that exist in access to health care and the quality of that health care in all these countries.
Other uncertainties in epidemiological studies of this type also include problems with the underlying health data available for analysis; inaccuracies in the exposure data used; the choice of the population under study; the effect sizes and whether they are transferable to other populations and exposure conditions; the burden of active smoking (deduced from the total burden before estimation of the burden from secondhand smoke); and the susceptibility of ex-smokers. The investigators performed various sensitivity analyses, in which they tested the effects of changing different variables on the the final estimates for the total number of deaths attributable to secondhand smoke. For example, making changes in these assumptions could lower the death toll estimate by as much as 28% or increase it by as much as 30%. One aspect of this study that is somewhat reassuring is that the estimates don’t vary by that much depending upon changes in major assumptions and that the estimates of attributable risk due to secondhand smoke in this study are in line with similar estimates from other studies.
With over 1 billion people smoking, the global burden of disease due to tobacco is staggering, even if secondhand smoke were perfectly safe. However, there has been a body of evidence growing over the last 30 years implicating secondhand tobacco smoke as a significant health risk. Because it’s primarily epidemiological in nature, it’s full of flaws and potential biases such that no single study can demonstrate definitively the relative risk of death due to secondhand smoke. That’s why one has to assess at the totality of evidence in coming to conclusions, and that totality of evidence clearly points in the direction of secondhand smoke being a major health risk. The question that follows from data such as this study is: What should we do about it?