Health Effects of Aerosols - Aerosols: Measurement, Dosimetry, and Health Effects - page 606

mortality from daily changes in air pollution [221]. A recent study that evaluated deaths in New
York City from 1985 to 1994 found that risk of death per interquartile range of PM 10 was greater for
persons whose death certiicates indicated an underlying respiratory disease [221]. This was par-
ticularly true for those aged 75 years and older, in whom a similar increased risk was also observed
for cancer deaths. Numerous other studies have supported these indings. Table 23.14 summarizes
representative examples of studies based on investigations of single cities or areas (a more detailed
discussion of the multicity studies in the United States and Europe follows). Studies from around
the world (based on time series and case crossover) have generally reported associations between
increased daily PM aerosol (or surrogates such as SO 2 and NO 2 ) and daily nonaccidental mortal-
ity, both from all causes and particularly from cardiovascular and respiratory causes. Although
Schwartz and colleagues, in time series studies, reported that out-of-hospital deaths had a stronger
association than in-hospital deaths, a case-crossover study from the Seattle area did not ind any
association between out-of-hospital deaths due to cardiac arrest and daily changes in any measure
of PM aerosol [222] (Table 23.14). Two of the studies in Table 23.14 are of particular note. Clancy
et al. [223] reported 10% and 15% decreases in daily cardiovascular and respiratory deaths in rela-
tion to daily changes in air pollution in the years that followed the 1990 ban on the sale of soft coal in
Dublin, Ireland. These decreases in daily mortality were associated with decreases in average black
smoke levels in 1984-1990 of 50-15 μg/m 3 in the years 1990-1996 (36% reduction) [223]. Vidal and
colleagues studied daily changes in mortality in relation to ambient pollutants in an environment
with low average PM 10 levels (see footnote in Table 23.14). These investigators found an association
between respiratory mortality and NO 2 and SO 2 ; PM 10 was only associated with all-cause mortality
at a 2-day lag. The explanation for why pollutants strongly associated with the PM aerosol seemed to
have more speciic effects on mortality was not explained by the author. A study from Seoul, Korea,
points to the dificulties of ascribing effects to PM aerosol [224] (Table 23.14). In this study, the PM 10
association was a 1.5% increase/interquartile interval; however, when O 3 was included with PM in
the regression model, the PM effect was observed (2.7% increase) when ambient O 3 concentrations
were above 13 ppb (Table 3 of Ref. [224], Table 23.14). This suggests that the overall chemical milieu
is relevant to the interpretation of the magnitude of PM-mortality associations. Finally, a study in
Mexico City found that over all lags and averages, the RR of death per 10 μg/m 3 of PM 10 was greater
for persons outside of hospitals at the time of death compared to deaths for persons in hospitals at
the time of death [225]. Since hospitals usually have controlled environments, this supports a role
for the ambient environment.
Subsequent to the publication of the time series studies noted in Table 23.16 and many other time
series not cited in the table or this chapter, a problem was noted with the software that had been
used by many of the studies to control the confounding effects of long-term trends in ambient PM
aerosol, meteorological, and other time-dependent confounding factors [226] (see footnote in Table
23.15 for details). A number of these studies were reanalyzed, and although some of the effect esti-
mates were reduced, the overall conclusion of an association between daily changes in PM aerosol
and daily mortality was not altered (the same was observed for associations with hospital admission,
which are not covered in this chapter) [227].
Given the heterogeneity of risk estimates that have been derived across studies carried out in
individual cities, two large studies were undertaken in the United States and in Europe (Table 23.15),
which attempted to study multiple cities with a broad range of PM aerosol environments using uni-
form statistical methodological approaches and to explore sources of heterogeneity between the
estimates derived from individual cities. Since both studies were reanalyzed to account for the soft-
ware problem noted earlier, only the reanalyzed data are summarized. The basic structure of each
study is summarized in Table 23.16.
NMMAPS estimated that for each 10 μg/m 3 increase in daily PM 10 , daily mortality increased
0.7% (posterior SE 0.06), 0.21% (0.06), and 0.10% (0.06) for lags of 0, 1, and 2 days, respectively
(Figure 23.24, top panel). At lag 1, the effects were largest for cardiorespiratory diseases (Figure
23.24, middle panel), and the lag 1 PM 10 effect was found to be robust to the inclusion of other