A detailed review of the evidence for health
effects of PM is beyond the scope of this entry.
Suffice it to say, abundant data are available on the
health effects of PM, characterized using different
PM indicators, from epidemiological and laboratory animal studies using a range of indicators of
morbidity and mortality [65, 68, 84]. In setting the
PM standards, the greatest weight, appropriately,
has been given to the epidemiological evidence
and, specifically, to those studies which can provide quantitative exposure-response information.
Two large studies have yielded especially valuable information that has played a key role in the
setting of the PM NAAQS using the PM 10 and
PM 2.5 indicators. The first of these is the Harvard
Six Cities study designed by Ferris and colleagues
[20, 23] to specifically evaluate the health effects
of criteria air pollutants. In my opinion, this is one
of the best studies ever conducted of the influence
of air quality on health. It involves a 14–16-year
follow-up of 8,000 subjects living in six communities in the United States selected to provide a
gradient in several indices of air quality. Extensive
efforts were made to characterize air quality.
A second air quality study made opportunistic
use of data from a population of 500,000 individuals who self-enrolled in 1982–1989 in an American Cancer Society Study [54]. A subgroup of
240,000 of these individuals lived in 50 metropolitan areas in which PM2.5 measurements were
made for regulatory compliance purposes.
The vast amount of information available from
the Six Cities and ACS studies has been reviewed
by the EPA [68, 70, 84]. Selected data will be
presented here to illustrate some of the key results
that have influenced the setting of the PM
NAAQS. Figure 7 summarizes an impressive
array of information from the ACS study on allcause mortality, cardiopulmonary mortality, lung
cancer mortality, and all other cause mortality
[54]. With control for smoking, education, and
national status, the controlled forward stepwise
inclusion of additional covariants had little influence on the estimated associations between fine
particulate (PM 2 . 5 indicator) air pollution on cardiopulmonary and lung cancer mortality. As may
be noted, the inclusion of cigarette smoking attenuated the estimated relative risk of PM 2 . 5 , which is
not surprising. When this is done it is necessary to
normalize the smoking history to pack-years, for
example, 1 pack/day for 25 years or 2 packs/ day
for 12.5 years equals 25 pack-years. The relative
risks for an average current smoker (men and
women combined, 22 cigarettes per day for
33.5 years with smoking initiated before age
18 years) (equivalent to 37 pack-years) were
equal to 2.58, 2.89, and 14.80 for all-cause, cardiopulmonary, and lung cancer mortality, respectively. The magnitude of these relative risks is
substantially greater than that of PM 2.5 and
emphasizes the critical importance of considering
cigarette smoking in any study of air pollution.
Moreover, it is important that investigators follow
the lead of the ACS investigators and report the
results of the analyses done for cigarette smoking
at the same time they report relative risks for
various air pollutants. These results on cigarette
smoking provide valuable perspective for considering the results of analyses for the various air
pollution indicators. A comparison of the relative
risk estimates they obtain for smokers with those
of other investigators serves as a “reality check”
on their analyses.
The results from the Six Cities study and ACSbased studies are generally similar [28, 53]. Summary information from the two studies is given in
Table 4 with a comparison made of the risk ratios
for the most polluted versus least polluted cities.
For comparison, results are also shown for typical
smokers. Note that in this case the typical smoker
has been defined as about a 25 pack-year smoker,
about two thirds that of the current smoker
discussed earlier for the ACS study. Interestingly,
the relative risk, for example, for lung cancer
mortality is about two thirds of the value cited
earlier, 9.73 versus 14.80. The range of the confidence intervals for the association between PM 2.5
and cardiopulmonary mortality, even for these
studies with thousands of subjects, emphasizes
the challenges faced in teasing out small effects
of air pollution in developed countries.
Laden et al. [29] extended the observations for
the Six Cities Study Cohort by including deaths
for 1990–1998 and compared those observations
to the earlier time period (1974–1984) and for the
entire period. The estimated city-specific average
28
Air Quality Guidelines and Standards
effects of PM is beyond the scope of this entry.
Suffice it to say, abundant data are available on the
health effects of PM, characterized using different
PM indicators, from epidemiological and laboratory animal studies using a range of indicators of
morbidity and mortality [65, 68, 84]. In setting the
PM standards, the greatest weight, appropriately,
has been given to the epidemiological evidence
and, specifically, to those studies which can provide quantitative exposure-response information.
Two large studies have yielded especially valuable information that has played a key role in the
setting of the PM NAAQS using the PM 10 and
PM 2.5 indicators. The first of these is the Harvard
Six Cities study designed by Ferris and colleagues
[20, 23] to specifically evaluate the health effects
of criteria air pollutants. In my opinion, this is one
of the best studies ever conducted of the influence
of air quality on health. It involves a 14–16-year
follow-up of 8,000 subjects living in six communities in the United States selected to provide a
gradient in several indices of air quality. Extensive
efforts were made to characterize air quality.
A second air quality study made opportunistic
use of data from a population of 500,000 individuals who self-enrolled in 1982–1989 in an American Cancer Society Study [54]. A subgroup of
240,000 of these individuals lived in 50 metropolitan areas in which PM2.5 measurements were
made for regulatory compliance purposes.
The vast amount of information available from
the Six Cities and ACS studies has been reviewed
by the EPA [68, 70, 84]. Selected data will be
presented here to illustrate some of the key results
that have influenced the setting of the PM
NAAQS. Figure 7 summarizes an impressive
array of information from the ACS study on allcause mortality, cardiopulmonary mortality, lung
cancer mortality, and all other cause mortality
[54]. With control for smoking, education, and
national status, the controlled forward stepwise
inclusion of additional covariants had little influence on the estimated associations between fine
particulate (PM 2 . 5 indicator) air pollution on cardiopulmonary and lung cancer mortality. As may
be noted, the inclusion of cigarette smoking attenuated the estimated relative risk of PM 2 . 5 , which is
not surprising. When this is done it is necessary to
normalize the smoking history to pack-years, for
example, 1 pack/day for 25 years or 2 packs/ day
for 12.5 years equals 25 pack-years. The relative
risks for an average current smoker (men and
women combined, 22 cigarettes per day for
33.5 years with smoking initiated before age
18 years) (equivalent to 37 pack-years) were
equal to 2.58, 2.89, and 14.80 for all-cause, cardiopulmonary, and lung cancer mortality, respectively. The magnitude of these relative risks is
substantially greater than that of PM 2.5 and
emphasizes the critical importance of considering
cigarette smoking in any study of air pollution.
Moreover, it is important that investigators follow
the lead of the ACS investigators and report the
results of the analyses done for cigarette smoking
at the same time they report relative risks for
various air pollutants. These results on cigarette
smoking provide valuable perspective for considering the results of analyses for the various air
pollution indicators. A comparison of the relative
risk estimates they obtain for smokers with those
of other investigators serves as a “reality check”
on their analyses.
The results from the Six Cities study and ACSbased studies are generally similar [28, 53]. Summary information from the two studies is given in
Table 4 with a comparison made of the risk ratios
for the most polluted versus least polluted cities.
For comparison, results are also shown for typical
smokers. Note that in this case the typical smoker
has been defined as about a 25 pack-year smoker,
about two thirds that of the current smoker
discussed earlier for the ACS study. Interestingly,
the relative risk, for example, for lung cancer
mortality is about two thirds of the value cited
earlier, 9.73 versus 14.80. The range of the confidence intervals for the association between PM 2.5
and cardiopulmonary mortality, even for these
studies with thousands of subjects, emphasizes
the challenges faced in teasing out small effects
of air pollution in developed countries.
Laden et al. [29] extended the observations for
the Six Cities Study Cohort by including deaths
for 1990–1998 and compared those observations
to the earlier time period (1974–1984) and for the
entire period. The estimated city-specific average
28
Air Quality Guidelines and Standards