Few environmental issues have generated as much controversy in recent years as have airborne particles. Large numbers of studies have found associations between airborne particles and measures of mortality and morbidity. Most studies have linked temporal, day-to-day variations in outdoor particle concentrations to day-to-day variations in total and cause-specific mortality, and to variations in emergency and unscheduled admissions to hospitals. Some studies have linked geographic differences in long-term exposure to particles to differences in death rates or survival, and to differences in the prevalence of chronic respiratory symptoms or lung function indices. Many questions still surround the causality of the observed associations and the mechanisms that could be responsible for them. Studies that have examined cause-specific mortality have consistently shown that in addition to respiratory deaths, cardiovascular deaths are increased with increased particle exposure. We still do not know very well how the cardiovascular system is affected by inhalation of fine particles. The few studies published so far of long-term effects of fine particles have suggested that even relatively low concentrations, when inhaled over a lifetime, can be associated with reduced survival. In this issue of the American Journal of Respiratory and Critical Care Medicine, two studies, both from the western United States, are reported that provide new insights into the effects of particulates, and add a few more questions to the many that already exist.

Pope and coworkers (1) tested the hypothesis that exposure to particulate matter acutely reduces blood oxygenation, which could explain some of the cardiovascular effects seen in the population studies briefly summarized earlier. No effect on blood oxygenation was found, but the investigators did find that increased exposure to airborne particulate matter slightly increased heart rate (HR). As indicated in their article, the biologic significance of these changes remains unclear, since there is little if any information available to help determine the prognostic value of small, transient changes in HR. Perhaps more important than the small (0.8 bpm) increase in the population mean HR was the observation that the odds of having one's HR increase by 5 or 10 bpm for each increase in PM₁₀ of 100 μg/m³ were also increased, showing once more that it is more important to document what goes on in the tail of the distribution than in the middle of it. Cardiovascular studies have tended to focus on single measurements of HR and HR variability as risk factors for subsequent cardiac events in patient populations, and to a more limited extent in random population samples. Such studies indicate that low HR variability and a high resting HR are important risk factors for subsequent cardiovascular events (2-4), but it is far from clear whether this is then also true for small, temporary changes in HR and HR variability that might be caused by changes in air pollution exposure.

Other investigators have also started to look at short-term changes in cardiovascular endpoints in relation to short-term changes in air pollution exposure. Preliminary results of these studies indicate that increases in exposure to respirable particles (PM_2.5) may lead to transient reductions in HR variability (5, 6). An increase in blood pressure with PM_2.5 was found in one of the two studies just cited (5) but not in the other, and the one study that looked at oxygen saturation and HR could find no effect of PM_2.5 on either of these endpoints (6). The study by Pope and coworkers (1) reported in this issue of the Journal was conducted in an area with low concentrations of NO₂, SO₂, and (wintertime) ozone. CO concentrations were also considered, but were generally not found to be associated with the cardiovascular endpoints studied. It seems reasonable to assume from this study that the associations found were primarily related to particulate matter, as in earlier studies from the Utah Valley.

Other hypotheses concerning effects of particulate matter on the cardiovascular system have centered on ultrafine (< 100 nm) particles being able to provoke alveolar inflammation, with release of mediators capable of increasing blood coagulability (7). A study retrospectively looking at blood coagulability data obtained during an air pollution episode in January 1985 found evidence that plasma viscosity was indeed increased during the episode (8). Concentrations of many pollutants were increased during the episode, however, including SO₂, and no data on particulate matter other than a rather crude measure of total suspended particles (TSP) was available for analysis. The extent to which the findings of this particular study were related to particles, ultrafine particles, or some other compound of the air pollution mixture therefore remains unclear.

Clearly, we are seeing only the beginning of more mechanistically oriented studies on acute effects of air pollution on cardiovascular endpoints. New studies, such as the one conducted by Pope and colleagues, provide the first building blocks of a set of hypotheses that are in need of further testing.

The other study of particulate matter published in this issue is the paper by Abbey and coworkers (9), reporting an association between long-term inhalable particles and mortality in nonsmokers. For obvious reasons (expense and duration), long-term cohort studies examining effects of air pollution on mortality are scarce. The evidence we have so far on long-term effects of particulate matter on survival comes largely from two studies conducted in the United States (10, 11). The study by Abbey and coworkers now supplements these two studies with new data on effects of long-term exposure to various air pollution components, including PM₁₀ and suspended sulfate as an indicator of fine particulate matter. The authors found that all-natural–cause mortality, nonmalignant respiratory mortality, and lung cancer mortality were related to PM₁₀ in males but not in females. Sulfate concentrations were not related to mortality, but SO₂ was related to lung cancer mortality in both men and women, ozone was related to lung cancer mortality in men, and NO₂ was related to lung cancer mortality in women. To some extent these results are in line with those reported earlier by Dockery and associates (10) and Pope and colleagues (11), but there are also several differences. The gender difference for the associations with particles was much less pronounced in the two earlier studies than in the one by Abbey and coworkers. The authors argue that this could have been due to males spending more time outdoors than females, which would increase their exposure to outdoor PM₁₀ and reduce exposure misclassification. However, this argument seems weak on several counts. First, ambient and personal exposure to PM₁₀ have been shown to be reasonably highly correlated (12). Second, differences in time spent outdoors between population groups tend to be gradual rather than substantial, having been expressed in categories of < 4 h, 4 to 16 h, and > 16 h per week—or, expressed in another way, < 90%, 90 to 98%, > 98% of time not spent outdoors. The differences in relative risks (RRs) reported, from 1.07 via 1.18 to 1.31 for the association between CRC mortality and PM₁₀, are nonetheless impressive; they must almost certainly be related to other personal characteristics associated with time spent outdoors, as no reasonable assumptions about breathing patterns and indoor/outdoor concentrations of PM₁₀ can account for more than a fourfold difference in RR being associated with what in effect are marginal differences in time not spent outdoors. The third reason why this argument is not convincing is that several studies have suggested that fine respirable particles are more hazardous than coarse particles, which contribute to PM₁₀ concentrations. As the authors note, fine particles readily penetrate indoors, and if it is the fine particles that matter, small differences in time spent outdoors cannot matter all that much.

Another striking difference between the study by Abbey and colleagues and the two previously reported cohort studies is the lack of association between air pollution and cardiopulmonary mortality in the new study. Cardiopulmonary deaths were most consistently associated with fine particles in the two previous studies, and the absence of an association for this, by far the largest group of causes of death, is puzzling and unexplained. The study may be plagued here by an absence of data on fine particles, and it is to be hoped that with the accumulation of data on fine particle exposures over time, we will learn whether the discrepancy with the other two studies remains.

The particle data used in the study by Abbey and colleagues, spanning a long period, came from measurements of TSP for the period prior to 1987. TSP was correlated with PM₁₀ for a 2-yr period of colocated measurements, and TSP measurements were converted into estimated PM₁₀ values. Because concentrations and contrasts between concentrations were higher in the first part of the study, this means that the exposure estimates for particulate matter were in fact dominated by TSP measurements. Since the authors themselves argue that some of the discrepancies may disappear when more PM_2.5 data becomes available, it is puzzling that at the same time they claim that using TSP to estimate PM₁₀ results in “minimal loss of precision.” It is true that PM₁₀ concentrations, estimated from site- and season-specific regression equations, were cumulated over a 2-yr period. However, TSP levels may have changed differently with time than may levels of PM₁₀ or PM_2.5 (as in the Six Cities study [10]), so that the actual loss of precision may have not been quite so minimal. Also, PM_2.5 usually correlates more highly with PM₁₀ than does TSP with PM₁₀. If it is worth waiting for more PM_2.5 data to become available (which I think is certainly the case), having an exposure metric that is dominated by TSP measurements makes this study less useful as a source of conclusions about the role of fine particles than were the previous two cohort studies, which used actual measurements of fine particles. Another issue related to measurement error is that the AHSMOG cohort came predominantly from three air basins (San Diego, Los Angeles, and San Francisco). Although the estimated particulate exposures showed a good spread between study subjects, they were based largely on TSP concentrations estimated for home and work addresses only, and it is possible that within air basins, a significant amount of time was spent outside of these two locations in commuting, shopping, leisure activities, and for other reasons. Fine particle concentrations tend to show more spatial homogeneity than do coarse particle concentrations, which are more influenced by factors such as windblown dust. Therefore, the reported exposure variation in the AHSMOG study may actually overestimate to some extent the within-area contrast in fine particle exposure. The effect of this may lead to further exposure misclassificaiton, especially when study subjects are from a few air basins only and are therefore likely to move into each other's exposure categories. A difference between the AHSMOG study and the other two cohort studies is that in those studies, exposure contrasts were essentially determined by contrasts between rather than within communities. Because subjects are much less likely to move between than within communities, the net effect could be that true exposure contrasts were better captured in the Six Cities and American Cancer Society (ACS) cohorts (although these relied on few monitoring sites per location) than in the AHSMOG study with its elaborate exposure classification scheme. This could also be a reason why in the AHSMOG study a relationship was found between ozone and lung cancer mortality: the exposure distribution for ozone in the Los Angeles area has very little overlap with the exposure distribution in San Francisco (13), so that for this pollutant, the true intersubject variation in exposure may have been sufficient to show an association. Ozone was not evaluated in the ACS study, and exposure to it showed little contrast between the locations studied in the Six Cities study, and the finding that ozone was related to mortality in the AHSMOG study is intriguing.

So far, United States-based time series studies of ozone have suggested acute effects on hospital admissions but not on mortality; however, the European experience is different, showing that in several areas, ozone was related to daily mortality (14). This could be related to the absence of air conditioning in most homes in Europe, which induces opening of windows and doors during warm weather with high ozone concentrations, effectively increasing exposure to outdoor air when ozone concentrations are high. One could imagine that the frail subjects at greatest risk of dying acutely are not likely to be outdoors much, and that the day-to-day variation in their personal ozone exposures is therefore not very well captured by ambient monitoring when air conditioning is present. However, gradual deterioration of health related to long-term exposure, possibly leading to the reduced survival detected in cohort studies, takes place over a lifetime, and substantial differences in long-term, outdoor concentrations of ozone are related to true differences in personal exposure, as was shown by Künzli and associates (13). In view of the intriguing findings for ozone reported by Abbey and colleagues, it would be useful to reanalyze the ACS study, with its many locations spread over the United States, for associations between mortality and ozone in addition to the analyses already reported for fine particulate matter.

One more intriguing observation in the AHSMOG study received only one small paragraph in the paper, so that it might easily be missed. This was the finding that the relative risk of CRC mortality in relation to PM₁₀ was much larger among subjects with low antioxidant vitamin consumption at baseline (RR = 1.26) than for those with high antioxidant vitamin consumption (RR = 1.08). Modulation of air pollution effects by nutritional factors is an area not very well investigated until now, and it seems that identification of individual determinants of susceptibility to air pollution (including nutrition) is worth further pursuit. A recent publication in the Journal showed that acute effect of air pollution on the lung function of shoe shiners in Mexico City were modulated by antioxidant supplementation (15).

Information from other parts of the world about the effects of long-term exposure to air pollutants on survival is scarce. There is a general lack of adequate measurements of inhalable or respirable particles covering long periods. Consequently, European studies have focused more on gaseous components and black smoke, which prohibits direct comparison with the results of the United States-based studies. Nevertheless, preliminary results of a long-term French study have suggested that survival was decreased in areas that had high SO₂ concentrations in the period from 1974 to 1976 (16). The cautious interpretation of such findings would be that expanding the current, still narrow data base by clever utilization of ongoing cohort studies may, in a few years' time, provide significantly more knowledge about effects of long-term exposure to air pollution on survival than is now available. Obviously, a more ideal solution would be to start new cohort studies specifically designed to document the effects of current, long-term exposures to both fine particles and other air pollution components on survival. The stakes are high, and it seems that in addition to squeezing all possible information out of ongoing studies, the time is right for another investment in a few, properly designed, long-term studies of the effects of fine particles on survival.

References

Section:

ChooseTop of pageReferences <<CITING ARTICLES