Sizing Up Air Pollution Research

The focus of epidemiologic research on the health effects of air pollution is undergoing a significant shift. Evidence that outdoor particulate air pollution is associated with excess mortality and morbidity has accumulated through a series of studies conducted on four continents over the last decade. 1–9 Most of these studies considered either exposure to total suspended particles (TSP) or exposure to total respirable particles (particles less than 10 μm in diameter, or PM10). Both of these measures of pollution are associated with remarkably consistent increases in mortality and morbidity, and the excess seems to be largely independent of other pollutants. 1–4 When the authors of two large cohort studies 10,11 examined mortality in relation to fine particles less than 2.5 μm in diameter (PM2.5), however, it appeared that much of the effect of particles might be due to this fine fraction. Based largely on this evidence, the U.S. Environmental Protection Agency moved in 1997 to establish a health-based air quality standard for fine particles. Since then, the epidemiologic lens has been focused downward, to examine smaller particles, and inward, to elucidate potential mechanisms linking ambient particle levels with adverse health effects. Two articles in the current issue of Epidemiology are indicative of this trend. 12,13 Schwartz and Neas 12 re-analyzed data from two previous studies of respiratory health among children to examine the evidence for differential effects of fine and coarse airborne particles, while Peters and her colleagues 13 examined the relation of cardiac arrhythmia with air pollution episodes in a small, but elegant follow-up study. Schwartz and Neas 12 conclude that fine particles have “much stronger acute respiratory effects than coarse particles” at between 2.5 and 10 μm in diameter. They report odds ratios of 1.05 (95% CI = 0.90–1.23) and 1.29 (95% CI = 1.06–1.57) for coarse and fine particles, respectively, from analyses of lower respiratory symptoms recorded by diary in the Six Cities Study. From combined analyses of data from follow-up studies in two Pennsylvania towns, they report a small, positive change in peak expiratory flow with a one interquartile-range increment in coarse particles, and larger, negative changes with equivalent increments in both fine particles and fine sulfate. In broad terms, these results support previous findings of associations between fine particles and other health endpoints. 10,11,14 Respiratory complaints and reduced lung function are transient responses, but they are associated with more serious conditions, like asthma, that may result in hospitalization and even death. With respect to the particles themselves, fine particles may be more biologically effective than coarse particles: they penetrate more deeply into the lung, and recent research in the laboratory suggests that they may also have greater acute toxicity. 15,16 Nevertheless, readers who study the details of Schwartz and Neas’s data 12 may not agree with their conclusion that fine particles are the principal cause of acute respiratory symptoms. In the abstract of their paper, Schwartz and Neas emphasize a limited set of findings, citing multiple-pollutant models for combined lower respiratory symptoms in the Six Cities Study and only one of the two Pennsylvania cities where they studied lung-function. Other aspects of the data tell a less consistent story. In single variable models for the Six Cities Study, for example, the association of lower respiratory symptoms with fine particles was only modestly stronger than that with coarse particles (odds ratio 1.14), and every 95% confidence interval contained the odds ratio for all four particle indicators. In another analysis of the Six Cities data, which evaluated cough without other respiratory symptoms, the odds ratios were around 1.2 for all classes of particles, except when fine and coarse particles were evaluated in the same model; in that situation, coarse particles had the stronger effect. Schwarz and Neas downplay their results for cough, but another observer looking at their data from the Six Cities Study might conclude that all respirable particles are associated with symptoms, with coarse particles primarily provoking coughing, and fine particles principally inducing other symptoms. Schwartz and Neas also argue that fine sulfates play an important and possibly predominant role in producing the effect of fine particles. This conclusion must be based entirely on the results of the two lung function studies. In those studies, the regression coefficient for fine sulfate was 1.67 times the coefficient for PM2.1 in a combined analysis, but in the Six Cities study, the odds ratios for sulfate and PM2.5 were roughly equal. The independence of the contribution of sulfate particles is difficult to assess, because, in both studies, sulfates were evaluated only in single-pollutant models. These results certainly support the proposition that airborne particles have adverse health effects, but they provide only limited evidence that the fine sulfate component is the agent primarily responsible for the effect. Peters and colleagues 13 followed 100 patients with implanted cardioverter defibrillators to examine the effects of air pollution on the occurrence of heart-rate abnormalities. Electronic records downloaded from the defibrillator indicated when it intervened to stabilize cardiac rhythm. This design provides a convenient solution to the ordinarily difficult problem of detecting transient cardiovascular events. The investigators hypothesized that episodes of arrhythmia, indicated by defibrillator activation, would be associated with excursions in the level of PM2.5. This hypothesis is supported by epidemiologic observations of increased cardiovascular morbidity and mortality on high air pollution days 17,18 and by experimental findings of heart rate abnormalities in animals exposed to particles. 19 Peters and her co-workers nevertheless found a stronger association with nitrogen dioxide (NO2) than with fine particles among most participants. Among individuals with repeated defibrillator discharges, a potentially susceptible subgroup, arrhythmic events were associated with a wider array of air pollutants, including fine particles, coarse particles, elemental carbon, and carbon monoxide, in addition to nitrogen dioxide. When considered along with other recent research, these two new studies add to the emerging body of evidence that airborne fine particles are associated with some of the most serious health effects of air pollution. 1 Some results from these same two studies also suggest, however, that neither the coarse fraction of respirable particles nor gaseous pollutants should be neglected in the enthusiasm to identify a single agent that could account for the principal health effects of air pollution. This point was underscored by findings communicated recently at the Third Colloquium on Particulate Air Pollution and Human Health, in Durham, North Carolina. My colleagues and I presented the results of a re-analysis of data on mortality and air pollution from Mexico City. We previously reported associations between fine particles and mortality from all causes, respiratory diseases, and cardiovascular diseases, 20 and hypothesized that coarse particles would be less strongly associated with these outcomes. Instead, we found that coarse particles were associated with all three categories of mortality more strongly than were fine particles, and that when both types of particles were analyzed in the same model, the effect of coarse particles was dominant. 21 At the same conference, van den Eden and colleagues reported that hospital admissions for cardiovascular diseases in Los Angeles were also associated more strongly with coarse particles than with other air pollutants (S.K. van den Eden, personal communication, 1999). These new findings are a reminder of the importance of perspective in epidemiologic research. Multiple studies are almost always needed before strong causal inferences can be made, because human populations and natural systems are so variable. Epidemiologic research on air pollution exemplifies the problem: air pollution is a complex mixture whose level and composition varies over time and space according to a large number of factors ranging in scale from personal microenvironments to regional climate. Such differences in the nature of air pollution are among the possible explanations for the divergent findings about coarse particles in Los Angeles and Mexico City, which have warmer, drier climates than the eastern United States, where most previous studies 11–14 have been conducted. Because populations differ markedly with respect to age, underlying health, and concomitant exposure to other insults, it is also reasonable to anticipate that different populations exposed to air pollution in different places may show a range of distinct responses. While investigation of the health effects of fine particles should continue, the results of some of the latest research 12,13,21 suggest it would be unwise to focus too narrowly, to the exclusion of other components of air pollution. The preoccupation with the smallest particles may be due in part to the tendency to look where the light is: although most researchers live in North America and western Europe and most previous studies on the health effects of air pollution have been conducted in those areas, many of the world’s largest cities and much of the population currently exposed to high levels of air pollution are located in developing countries. For public health purposes, it is clearly important to examine the health effects of air pollution in these areas. Research in these settings may also yield far-reaching scientific benefits by allowing the effects of air pollution to be understood in a context more representative of the world’s climate and health.