On March 6, the Journal of the American Medical Association published a new study of the relationship between air pollution and mortality.
“Lung Cancer, Cardiopulmonary Mortality, and Long-term Exposure to Fine Particulate Air Pollution,” by C. Arden Pope and coauthors, contends people living in areas with higher levels of fine, airborne particles are likely to die earlier than people living in areas with less fine-particle pollution.
Specifically, the Pope study contends living in an area with fine particulate levels 70 percent higher than average, results in a 6 percent increase in the risk of death over a 16-year period.
Although the authors claim to have demonstrated a substantial risk from air pollution, they may have mistakenly attributed to air pollution health risks that are actually caused by other factors omitted from their analysis. Furthermore, even taking the results at face value, the study found a relatively small risk from particulates when compared with other risks people face.
Confounding breeds confusion
The Pope study is an “ecological study”: a study in which researchers observe people’s health, behaviors, and risk factors in the real world, rather than in the tightly controlled conditions of the laboratory. Ecological studies are inherently limited because researchers cannot obtain detailed ongoing information about the relationships between people’s health and the potential causes of health risks.
For example, the behavior of the subjects cannot be prescribed, and their diet, exercise, and other health habits cannot be effectively monitored. Pollution exposure can be estimated only roughly, based on a few regional measurements rather than actual individual exposure. Researchers don’t know how much time people spend outdoors, or whether they continue to live and work in the same place after their initial entry into the study.
It is thus difficult to tell in an ecologic study whether observed health outcomes are the result of pollution exposure … or other differences between people who live in high- and low-pollution areas. For example, if it turned out that people in high-pollution areas are more likely to drink or smoke, there’s a danger of inadvertently confusing an effect of alcohol consumption or smoking with an effect of pollution. This problem is known as confounding. Other confounders include diet, exercise frequency, income, marital status, “body-mass index” (BMI, a measure of obesity), and educational attainment.
The Pope study researchers accounted for most of these confounders in their analysis. But the factors were assessed only when people entered the study in 1982 and not afterward. If any of these factors changed after 1982, and if the changes were correlated with pollution levels, then the study results would suffer from uncontrolled confounding.
For example, if people in areas with higher pollution were also likely to get fatter between 1982 and 2000 when compared with people in lower pollution areas, researchers could mistake an effect of body weight for an effect of air pollution.
Similar concerns apply to other confounders, such as diet and smoking. For example, if the prevalence of smoking decreased more slowly in higher pollution areas during the last 20 years, then smoking might have actually been responsible for effects the Pope study attributes to air pollution.
Because the risks of smoking and obesity are so much larger than the risks the Pope study estimated for fine particulates, even a small difference in smoking and obesity trends between areas with differing pollution levels could swamp the claimed effect of differences in air pollution.
For example, the Pope study found that a 70 percent increase in the concentration of fine, airborne particle levels increases risk of dying prematurely by 6 percent. But for a six foot, 200-pound, non-smoking man, gaining just 15 pounds increases the risk of an early death by 17 percent.
Two other findings in the Pope study suggest the authors’ efforts to control for confounding were incomplete. First, the study found particulate exposure increases the risk of lung cancer for men, but not for women. Second, the association of air pollution and either cancer or cardiopulmonary mortality held only for people with a high school education or less.
There are also other potentially confounding factors the Pope study did not assess at all, including income and wealth, and physical activity levels. These factors also have a strong relationship to health and could have changed over time in ways that could cause misattribution of health effects to air pollution when they were actually due to other factors.
Assumptions aren’t always accurate
While it is very important to determine whether low-level exposure to airborne particles poses a risk to human health, such research must be based on sound assumptions in order to provide valid information about health risks. Nevertheless, the Pope study authors made a number of questionable assumptions that should have steered them away from claiming to have generated the “strongest evidence to date” regarding the relationship between airborne particles and human health.
Regarding exposure, the study’s authors assumed that everyone within a zip code was exposed to the same level of particulate pollution, though evidence shows such exposure can vary widely. Not only does exposure vary from place to place in concentration, but it also varies in the chemical composition of the particles.
The authors also assumed people told the truth on the initial questionnaires regarding how much they smoked and drank. But survey researchers have observed people tend to under-report those behaviors. Thus, if the authors used survey responses to account for smoking and drinking, they could be underestimating that risk in their study population.
The authors also assumed health-related behaviors did not change after the entrance survey in 1982. For example, the authors assumed that persons who were nonsmokers when they entered the study never took up smoking. Similarly, they assumed that people who filled out surveys in 1982, and later died in the same area where they originally lived, had stayed there throughout. This problem also applies to other health-related factors, such as change in weight and diet after entering the study.
Context offers clarity
The Pope study authors contend the study provides “the strongest evidence to date that long-term exposure to fine-particulate air pollution common to many metropolitan areas is an important risk factor for cardiopulmonary mortality.” Media reports on the study were uncritical in repeating the study’s findings, giving short shrift to its many limitations.
But even if we take the results at face value, the study greatly exaggerates actual risks, and does a poor job of placing the risk of particulate air pollution into the context of other risks people face.
The study’s authors may well have mistakenly attributed to air pollution health risks that are actually caused by other factors. But even ignoring that concern, the study found a relatively small risk from particulates. For example, the study found reducing particulate levels by 60 percent would reduce the risk of dying during a 16-year period by about 6 percent. A six foot, 215 pound, non-smoking man can achieve the same risk reduction by losing about five pounds; he can get three times the risk reduction by losing about 20 pounds.
The Pope study also found most of the health benefits from reducing airborne particulate levels accrue from reducing particulates down to a concentration in air of about 18 micrograms per cubic meter (mcg/m3). Reductions below this level provided little or no additional health benefit.
But according to the Pope study’s pollution measures, all but 2 of 51 metropolitan areas were already below 18 mcg/m3 as of 1999-2000. National fine-particulate monitoring data also show few areas of the country now have particulate levels above 18 mcg/m3. Thus, even if the small additional health risk reported by the Pope study is real, few people are exposed to it.
Ongoing reductions in particulate levels also mean future particulate levels will be even lower, further reducing risk. For example, total particulate emissions dropped 75 percent between 1940 and 1997, while per-capita emissions dropped more than 85 percent. Total particulate levels in air declined about 50 percent between 1960 and 1990, while coarse plus fine particulates declined 19 percent between 1991 and 2000. The health hazards discussed in the Pope study occur only after many years of exposure to elevated particulate levels. Continuing declines in particulate pollution suggest that remaining risks will not persist for long enough to damage health in the future.
Bang for the buck
Everyone deserves air that’s safe to breathe, and most Americans agree on the importance of improving air quality in places where air pollution is high enough to threaten health. But most Americans also want to know public health resources are going where they’ll get the biggest bang for the buck—that is, toward larger, more certain risks, rather than toward small risks that might not even be real.
Exaggerating the public’s risk from air pollution is no better than ignoring real air quality problems. If society misspends scarce resources based on inaccurate information, more people will suffer, not fewer.
Kenneth Green is chief scientist, and Joel Schwartz is senior scientist, of the Reason Foundation’s Environmental Program.
For more information …
The Pope study, “Lung Cancer, Cardiopulmonary Mortality, and Long-term Exposure to Fine Particulate Air Pollution,” was published in the March 6, 2002 issue of the Journal of the American Medical Association. For more information, visit the JAMA Web site at http://jama.ama-assn.org.
More information on improving air quality in the United States is available in Clearing the Air, Indur Goklany, available through Amazon.com at http://www.amazon.com/exec/obidos/ASIN/1882577833/theheartlandinst.