The proposed standards should not be approved. The research relied upon to justify the proposed standards does not show epidemiologically significant relative risks from existing ambient particulate levels. In fact, the consistent finding of insignificant relative risk effectively demonstrates that there is no causal link. And significant research papers call into question the linkages used to justify the proposed standards.
Air Pollution Trends and Policy
Toxic air pollution existed in the past, and still may occasionally occur in some places on the planet as a local phenomenon, as particulate and other noxious air pollution in industrial areas, from various sources. Certainly air in big cities has been fouled and even when not toxic, smelly and visible. But trends in air pollution in the past 30 years as reported and confirmed by the EPA have all been positive, attributable to changes in industrial processes, regulatory efforts and cleaner petroleum and coal consumption.
- The United States has made tremendous progress in reducing air pollution during the last forty years. Air pollution has declined dramatically since the 1960s and 1970s, and virtually the entire nation now meets federal health standards for carbon monoxide (CO), sulfur dioxide (SO2), and nitrogen dioxide (NO2). Many areas of the country still exceed health standards for ground-level ozone (“smog”) and airborne particulate matter (PM), but both of these pollutants continue to decline as well. Half of the nation’s ozone-monitoring locations exceeded the federal one-hour ozone standard in the early 1980s, but only 13 percent exceeded the standard by the end of 2002. PM measurement methods have changed a number of times during the last forty years, but all trend data show PM levels dropping. Average levels of PM2.5—the form of PM now of greatest regulatory concern—have declined by a third during the last twenty years. (Schwartz, 2003, 1)
Rather than focus on making rational marginal improvements on this progress, all too often some air pollution researchers, activist groups, and even the EPA focus on crisis and pushing the envelope of defining health threats so that the public and the media are largely unaware of the progress being made.
A good example is in the September 9, 2004 issue of New England Journal of Medicine, in which C. Arden Pope describes killer air in Belgium in 1930, Pennsylvania in 1948, and London in 1952 and uses those incidents as examples of why he thinks there is good reason to pay attention to a study in that same issue that claims to show a causal relationship between non-toxic air pollution and children's pulmonary functions. But the study, like so much of the epidemiological research on air pollution health effects, shows no epidemiologic proof, just "associations," which are nothing more than statistical blips in population studies subject to bias, confounders etc., and at insignificant causation ranges.
People do not go out into the streets of America, choke and die. The days of the people of London and Pittsburgh wearing dark clothes to mask the effect of soot and smoke are gone. In America ambient air pollution did not kill anyone, last week, last year, or in the last ten years. The crisis is long past, and the real health effects that remain require sober, scientific approaches, not hype.
Air Pollution and Death
One of the authors, Dr. Dunn, is an emergency physician much more familiar with what kills people than economists and public health officials who don't know the business end of a ventilator and live in the world of death certificates and mortality data. People die for many reasons and under many circumstances in America, but air pollution doesn't kill them, even the worst levels of outdoor air pollution one might imagine in America don't create a toxic level.
One doesn't die from an exposure to air pollution; one dies from failed medical therapy, arrythmias caused by long-term coronary disease, stroke, pulmonary embolism, which are not caused by air pollution. The death and illness rates during smog and air pollution catastrophe periods in the past were affected by less effective medical management and heavier cigarette smoking but also significantly higher air pollution than exists anywhere in the United States today.
Deaths from acute respiratory failure in the past were more common and less preventable, but that is an independent factor related to medical advances and not due to air pollution itself. Airway diseases, the main effect of any air pollution, were less treatable before the 1970s. Pulmonary medicine has changed dramatically for the better since 1970. Many airway diseases were more dangerous in the past and medical therapies frequently failed to control disease and death. Medical expertise in respiratory illness and cardiovascular disease has changed.
Even the asthma problem is an increasing problem not related to air pollution, since the rate of asthma is increasing with decreasing air pollution. The research clearly shows that deaths from asthma have far more to do with socioeconomic phenomenon than with air pollution. In fact, recent studies show that the well-forested state of Maine leads the nation in per capita asthma cases.
It is time to retire the air pollution health effects studies of crude death tallies and instead to look at real relative risks.
The literature used to justify these proposed regulations—analyzed in detail in the staff report and Federal Register article for this docket—suffers dramatically from findings of relative risk far below acceptable thresholds for drawing conclusions. The studies by Dockery (1993), Pope (1995), and Samet (2000) provide great examples. In each case their results showed that large studies with adequate power could not demonstrate relative risk of any significance. They showed effects of less than 10 percent, when the epidemiological standard for showing causation in death effects is 200 to 300 percent. Some epidemiologists require relative risk of 4, i.e. a 400% effect, when evaluating poorly controlled cohort studies.
In the Federal Register article for this docket (January 17, 2006, Part II EPA) Figures 1 and 2 show the staff analysis of the research on the relative risks of mortality and morbidity associated with elevated ambient particulate levels. The overwhelming majority of the studies show relative risks of less than 1.5, with only a few coming even close to the minimal epidemiological standard of 2 for reaching conclusions of causation.
The problem is that many researchers and EPA staff have pushed aside epidemiological research standards for drawing conclusions of causation, and have instead embraced the much lower standard of statistical significance. A result that is statistically significant is one where we know the statistical tools work in showing a relationship—in this case, that higher ambient particulate levels are in some way related to higher mortality and morbidity rates, but are not necessarily causal in nature. But statistical significance does not mean the effects of the relationship are significant. Hence, the very small relative risks found in almost all the research cited in the staff report for this proposed rule may be statistically significant, but they are epidemiologically insignificant.
The EPA has a serious public health responsibility and a federal mandate to find toxins with legitimate science, promulgate appropriate solutions for the public benefit and then assess the effectiveness of what it has done. Many existing rules have been effective; many have not. The improved air quality in all major cities in the country is proof enough that some have worked in concert with technological changes that reduce emissions. Effectively achieving current standards is where we are likely to see the most positive health effects. But the EPA spends little time on the effectiveness of current rules, and works tirelessly at promulgating new ones.
The EPA also has a mandate to act only on the basis of acceptable scientific evidence of health effects. Under Federal Rules of Evidence 702, the EPA would struggle to show that the research used to justify this proposed rule meets the standard. The Bradford Hill (BH) criteria for toxicology are elementary and establish biological plausibility for toxin effects. They require the toxicologist to establish plausibility, dose effect, reproducibility, time relationship, and a pattern of predictable and observable effects.
The research used to justify the proposed standards does not meet these tests. A raft of "associations" and relative risks that do not meet epidemiological standards of proof is not the proper basis for scientific decision-making. Research that assumes away most real work effects and barriers to establishing causality has to be used to guide investigation, not set policy. Challenges these studies fail to overcome include:
- mobile populations;
- unreliable, non-continuous and fixed monitor information;
- no monitor information on some pollutants all the time (2.5 micron particles for example) or part of the time (10 micron and others);
- an attempt to assess long-term chronic health effects of air pollution by death studies, an acute phenomenon;
- death certificates and raw death data used without autopsies;
- inside air quality ignored for populations living indoors, particularly during old age, advanced medical illness, and terminal illness; and
- most of all, no biological plausibility because the deaths are in the setting of non-toxic levels of air pollution, an insupportable straight-line effect toxicology.
Applying minute and insignificant relative risks to the whole U.S. population to come up with large number effects of death and sickness is fabrication. The precautionary principle approach embodied by the proposed rules should be abandoned and replaced by focusing effort and resources on the real health threats that exist.
Research Calling Into Question Effects Used to Justify New Rules
Enstrom (2005) studied deaths in elderly Californians in 25 counties 1973-2002. He found that the relative risk was extremely small and insignificant—1.04 in the first part of the study (1973-1982), then relative risk of death from air pollution disappeared altogether in the second part of the study (1983-2002). For the entire period the relative risk was 1.01, which is in fact no risk at all. No increased death effects of any kind were shown in the counties with higher levels of air pollution, eliminating any dose-response effect, and in fact some of the higher pollution counties had lower relative risks.
He also reviewed the cohort studies on health effects of fine particulates and mortality by Pope et al (1995), Pope et al (2002), Dockery et al (1993), McDonnell et al (2000), and Lipfert et al (2000) (all studies used in the staff research report supporting this proposed rule), and found that their results were fairly similar to his, with the weakest health effects being present during the most recent years. Only the Dockery study published in 1993 in a small cohort shows a relative risk above 1.1 at 1.15. All the other studies show relative risk similar to Enstrom, in the range of 1.07 or less.
Table 10 in Enstrom's paper pulls together the findings of the reviewed studies and his own data and adjusting for age, sex and homogenaity finds relative risks all very close to 1.00, reflecting insignificant risks. In fact, the Pope et al (2002) study, a 16 year follow up to the earlier Pope et al (1995) study shows a declining cumulative risk from 1.07 to 1.04, first half to second. Hence the relative risk in the second decade must have been well below 1.04. So, these vanishingly small and insignificant risks are declining in the data.
Suresh Moolgavkar has an impressive body of work on epidemiologic and methodology issues of air pollution—several cited in the staff research for this docket. In Moolgavkar (2005) he asks "Can contemporary epidemiological and statistical tools reliably detect miniscule risks, particularly with strong risk factors as potential confounders?" Moolgavkar objects to the methodology of proportional hazards modeling because "it is highly unlikely that proportionality of hazards would hold over the entire period of time covered by these studies." (The long-term air pollution health effects studies). He asserts that it can be argued that "the SO2 effect wipes out the PM signal in joint pollutant models."
He goes on to point out that "evidence fell far short of supporting a causal association between particle mass concentration and human health." He goes on "the results of observational epidemiology studies can be seriously biased, particularly when estimated risks are small, as is the case with studies of air pollution. The Agency (EPA) has largely ignored these issues. . . . I conclude that a particle mass standard is not defensible on the basis of a causal association between ambient particle mass and adverse effects on human health."
The research used to justify the proposed rule does not demonstrate results that meet the epidemiological threshold for concluding that ambient particulate levels cause increased morbidity or mortality. In fact the consistent finding of insignificant relative risk effectively demonstrates that there is no causal link. Moreover, important research studies have shown that low relative risk results and pervasive confounders make it very unlikely that the proposed rules will have measurable health effects. The EPA should abandon this precautionary-principle driven effort and focus instead on achieving success with current rules that have been shown to have meaningful health benefits.
Samet JM, Dominici F, Curriero FC, et.al. Fine particulate air pollution and mortality in 20 U.S. cities, 1987-1994. NEJM 2000; 343:1742-9.
Wong JD, Shapiro MF, Boscardin WJ, et. al. Contribution of major diseases to disparities in mortality. N Engl J Med 2002;347:1585-92.
Fitzpatrick R. Ed. Social status and mortality. Ann Intern Med 2001 134;10:1001-2.
Lantz 1998 Lantz PM, Lepkowski JM et. al. Low income was an independent risk factor for premature death after controlling for health behaviors. JAMA 1998; 279:1703-8.
Dockery DW, Pope CA 3d, Xu X, et. al. An association between air pollution and mortality in six U.S. cities. N Engl J Med 1993;329:1753-9.
Pope CA, Thun MJ, Namboodiri MM, et. al. Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am J Respir Crit Care Med 1995;151:669-74.
Pope CA, Burnett RT, Thun MJ et al. Lung cancer, cardiopulmonary mortality a, and long-term exposure to fine particulate air pollution. JAMA 2002; 287:1132-41.
McDonnell WF, Nishino-Ishikawa N, Petersen FF, et.al. Relationship of mortality with the fine and coarse fractions of long-term ambient PM10 concentrations in non-smokers. J Expos Environ Epidemiol 2000;10:427-436.
Lipfert FW, Perry HM, Miller JP, et.al. The Washington University—EPRI veteran's cohort mortality study: preliminary results. Inhal. Toxicol. 2000, 12 S4:41-73.
Pope CA. Ed. Air pollution and health -- good news and bad. N Engl J Med 2004 351;1132-1134.
McFadden ER jr., Warren EL. Observations on asthma mortality. Ann Intern Med 1997;127:142-7.
Enstrom J. Fine particulate air pollution and total mortality among elderly Californians, 1973-2002. Inhalation Toxicology 2005; 17:803-16.
Moolgavkar S. Let. Fine particles and mortality. Inhalation Toxicology 2006;18:93-4.
Moolgavkar S. A review and critique of the EPA’s rationale for a fine particle standard. Reg Tox Pharm 2005; 42:123-44.
Schwartz, J. No Way Back: Why Air Pollution Will Continue to Decline, (Washington: American Enterprise Institute, 2003).
Author Contact Information
For questions please contact Dr. Adrian Moore, firstname.lastname@example.org, 661/477-3107.