The Year of Ozone

Ozone was rivaled only by climate change as the key environmental issue of 2015 in the United States. On October 1, 2015, after a flurry of lobbying and predictions of doom, the U.S. Environmental Protection Agency (EPA), ending (perhaps) a process begun under the Bush Administration, issued a final rule setting the National Ambient Air Quality Standard for ozone at 70 ppb for an 8-hour average, down from 75 ppb. This was met not merely by the usual lawsuits but by a legislative assault leading to riders that overturned the rule, which were only dropped in hard negotiations, against the background of shutting down the government. They will be back as independent bills in 2016. Why all the fuss? And why did this process take so long?

The last decade has witnessed a collision between developing science on the health effects of ozone and a wrenching change in industrial technology in the United States. Historically, EPA estimates of health benefits of ozone reduction were focused on elegant exposure chamber studies showing reversible reductions in lung function and increases in symptoms after exposure compared with filtered air. Although these results were clear, chamber studies necessarily cannot capture more severe effects of both short-term and long-term exposure to ozone. It became harder to justify tighter and more expensive standards based on such short-term, modest effects. However, even in the 1980s, there were signs that more serious effects occurred, such as increased hospital admission for respiratory disease (1). More recently, studies of long-term exposures of monkeys to ozone demonstrated permanent changes in lung morphology (2).

For effects of short-term exposure, these epidemiologic results soon became a flood. Studies from all over the world reported associations of ozone exposure with hospital admissions for respiratory disease (3–9) and with mortality (10–17). Moreover, a large multicity study found no evidence of a threshold down to about 20 ppb (18). Other studies reported associations with biomarkers that make those findings plausible (19–23). A chamber study this year reported that 100 ppb of ozone for 4 hours resulted in an increase in C-reactive protein during the next 24 hours and a shift to more sympathetic tone in heart rate variability, indicating that cardiovascular as well as respiratory effects are plausible (24). So if ozone at current concentrations is killing people and putting them in the hospital, why has it taken so long to tighten the standards?

The fuss comes from the nature of ozone formation. In most of the United States, that formation depends on emissions of nitrogen oxides, which in turn come from combustion sources, primarily motor vehicles and large industrial and utility coal boilers. Motor vehicles have been subject to increasingly stringent controls of nitrogen oxides since 1980. Further tightening will occur with the Tier III standards scheduled to begin in 2017. Given the half-life of cars, these reductions will phase in quickly. For coal boilers, many of which are more than 50 years old, turnover is slow, and the “grandfather” policy of not requiring retrofits has resulted in old plants without controls. At the same time, inexpensive natural gas has become a cost-effective competitor with coal for these processes, and the “grandfathered” status a regulatory subsidy that keeps them from being replaced by cleaner facilities. The tougher ozone standard, therefore, threatens the existence of coal as a cost-competitive fuel for industrial and utility boilers. Internal combustion engines for nonautomotive use may also require additional controls. Hence, the controversy and the crucial role that the identification of more serious health effects has played in justifying these standards. As the controversy has grown, the need for such understanding of ozone’s health effects has likewise grown.

What EPA did not rely on in setting or defending its ozone standards is the developing evidence on the effects of long-term exposure. In this issue of the Journal, Turner and colleagues (pp. 1134–1142) indicate that the evidence for chronic effects is growing (25). Jerrett and coworkers (26), using the Cancer Prevention Study II cohort, reported a significant effect of long-term ozone exposure on respiratory mortality rates, but no significant findings for cardiovascular mortality rates. Turner and colleagues followed the same cohort, but with additional years of follow-up and twice as many deaths. In addition, they used greatly improved exposure estimates based on combinations of land use regression and chemical transport models. They report an association (95% confidence interval) of annual ozone and all-cause mortality with a hazard ratio of 1.02 (1.01–1.04), and with cardiovascular mortality of 1.03 (1.01–1.05) in models controlling for particulate matter with a diameter smaller than 2.5 μm and NO₂. Moreover, this comes on the heels of another study, the CanCHEC study (27), looking at a cohort of 2.5 million Canadians, which reported essentially identical associations with all-cause mortality and somewhat larger effects for cardiovascular mortality.

Turner and colleagues’ article also includes extensive sensitivity analyses showing no evidence that the ozone effects are confounded by individual and area-based measures of socioeconomic status or of pollution modeling strategy. This greatly strengthens the argument for causality. Also supporting causality is the article by Zanobetti and Schwartz, who used a difference in differences approach to causal modeling and reported associations of annual ozone exposure with mortality in a cohort of sensitive individuals (28). To eliminate confounding by spatially varying factors, they only looked within a city and showed that year-to-year variations of summer ozone concentrations around the ozone trend line for each city were associated with year-to-year variations in mortality rates around their trend lines.

A key finding in Turner and colleagues’ study is that ozone was associated with cardiovascular mortality rates. Because the baseline mortality rates for cardiovascular death are so much higher than for respiratory deaths, even modest increases in risk of those deaths implies a large increase in attributable deaths. The Centers for Disease Control and Prevention estimate that 610,000 people die each year from heart disease (29). The hazard ratio from Turner and colleagues’ study, if causal, indicates that a 10-ppb reduction in average ozone concentrations in the United States would prevent 18,000 early deaths a year. Compared with this, the risk estimates from the time series studies the EPA relied on are small. Other evidence this year supports an association with heart disease. For example, another cohort study reported an association between long-term ozone exposure and factor VII coagulant activity (30), a chamber study reported that ozone affected fibrinolytic activity (31), and a toxicology study reported that after ozone exposure, isolated coronary vessels exhibited greater basal tone, enhanced susceptibility to serotonin stimulation, and impaired response to acetylcholine (32).

None of this is to say that the issue of effects of ozone on cardiovascular disease is settled. An English cohort found no association, for example (33). But during the last decade, the evidence for more serious chronic health effects of ozone has been growing. But so has the political conflict. The Obama administration, after the 2008 partial remand of the 75-ppb standard, asked the Clean Air Scientific Advisory Committee whether new evidence weakened their recommendation of a standard between 60 and 70 ppb. The committee responded that it had strengthened. Even so, moving the standard to 70 ppb took a long time and a court order for a proenvironmental administration. The backlash has been considerable. A bill to change the composition of the committee, removing academic scientists and replacing them with industry scientists, passed the House of Representatives and will be back next year. We have much to learn about the cardiovascular effects of ozone, and little time to do so.