Toxicity of fine particles.

Abstract

To the Editor: Higher concentrations of fine particles [particulate matter less than 10 or 2.5 μm in aerodynamic diameter (PM10 or PM2.5)] in the air are often associated with higher death rates from all causes or from cardiovascular and respiratory causes, and many authors have concluded that the associations may be causal. 1–3 Others have questioned the idea as being biologically implausible, or because they consider the statistical evidence to be inadequate to demonstrate a causal relationship. 4 From day-to-day variations in concentrations of PM10 and the associated variations in the number of deaths on the following day in 20 U.S. cities, Samet et al1 calculated that increases of 10 μg of PM10 per cubic meter of air resulted in an average increase of 0.51% in deaths from all causes, and an increase of 0.68% in deaths from cardiovascular and respiratory causes. From the consistency of these and other results, they and Ware 2 concluded that increases in concentration of PM10 were the causal factor in the relationship. Effects on all-cause mortality varied from about −1.0% in Atlanta to about +2.0% in Oakland. As the data from all of the cities were treated uniformly, these variations did not result from different methods of analysis. These real differences may have resulted from the people being more susceptible, the environment being more dangerous, or the PM10 itself being more toxic in some cities than in others. The differences did not depend on average concentrations of PM10 or ozone, or on city-specific demographic characteristics. The differences showed that increases in PM10 were not consistently harmful. From the stated consistency of the effects, Samet et al1 concluded that PM10, the particles themselves, were responsible for the associations. PM10 comes from many sources, and so it may have varied in composition from city to city. Between-city concentrations of PM10 and other pollutants were positively correlated. It was concluded that this “correlation structure generally reflects the common sources of the primary combustion-related gases,”1(p 1744) implying that this showed that the PM10 came from similar sources in the 20 cities, and would be likely to have similar composition and effects. However, proportions of PM10, sulfur dioxide (SO2), and carbon monoxide varied widely from city to city. This suggests that the sources of PM10, and its composition and possible toxicity, may have varied quantitatively between cities. The effects of daily variations in the concentrations of ozone were also calculated, and overall these were said to be concentrated near zero. In fact, each increase of 10 ppb in ozone concentration was associated with a small mean reduction of about 0.35% (±0.36; 95% confidence limits = −0.01, 0.71) in deaths. This association varied between winter and summer. From the data in Tables 1 and 2 of Samet et al, 1 we have calculated the correlations between death rates from respiratory and cardiovascular causes and mean concentrations of listed pollutants. The relationships with PM10 and SO2 are graphed in Figures 1 and 2.FIGURE 1: Mean daily death rates from cardiovascular and respiratory causes in 20 U.S. cities plotted against mean PM10 concentrations. Data from Tables 1 and 2 of Samet et al. 1FIGURE 2: Mean daily death rates from cardiovascular and respiratory causes in 15 U.S. cities plotted against mean SO2 concentrations. Data from Tables 1 and 2 of Samet et al. 1Between-city variations in death rates were weakly and nonsignificantly correlated with variations in mean concentrations of PM10 (r = +0.26;Figure 1), as well as carbon monoxide (r = +0.25), nitrogen dioxide (r = +0.27), and ozone (r = −0.09) (data not shown in figures). They were, however, highly correlated with variations in concentrations of SO2 (r = +0.88, P < 0.01;Figure 2). Day-to-day variations in concentrations of SO2 were not associated with corresponding daily variations in deaths. These apparently anomalous and inconsistent results were not discussed by Samet et al. 1 They could perhaps be explained if PM10 varied in toxicity from city to city, and if the higher death rates after high PM10 pollution days were merely hastening by a few days the deaths of those already near death. Samet et al, 1 and other authors whom they cited, have considered this latter explanation but have generally dismissed it. In our opinion, the statistical data do not show that reducing concentrations of PM10 will prolong life. We thank Grant McKenzie, Animal and Food Sciences Division, Lincoln University, for composing the graphs. Pat Palmer David Saville