We found significant associations between air pollution and hospital admissions for MI and pneumonia. Similar associations have been reported before. These results differ in that there was a clear difference by season, with the MI associations only in the summer, and the pneumonia associations only in the winter. In addition, the use of PM2.5, PM non‐traffic, and BC allowed us to better focus on sources of pollution. For pneumonia admissions, we found the largest estimates for BC, a surrogate for traffic particles. PM2.5 showed a weaker effect, and CO, also a traffic pollutant, a significant, but weaker association. This suggests a more specific association with traffic particles than with fine particles or other markers of traffic pollution.
In contrast, for MI admissions in the warm season, the CO association showed the largest estimated effect, but the NO2 and PM2.5 associations were essentially similar in magnitude. The association with BC was noticeably weaker, and there was a significant association with non‐traffic particles. Taken together this presents a more complex picture, with both traffic pollutants and non‐traffic pollutants triggering MI. The lack of association with ozone combined with the significant association with non‐traffic particles, which in Boston in the warm season are dominated by regional secondary particles, primarily sulphate particles, suggests sulphates and other secondary particles are the principal non‐traffic pollutant of concern.
This is partially consistent with the results of Peters and coworkers, who found stronger associations with NO2 and BC than with PM2.5.
We previously found an association between pneumonia and PM10
in a multicity time series analysis.16
In Boston we similarly found an association with pneumonia admissions. The BC and CO effects were the largest. Fusco and coauthors,24
found a strong effect of CO and NO2
on respiratory admission in Rome, where pollution (including particles) is dominated by traffic. They did not have BC or PM2.5
measurements in that study. Given the lack of plausible mechanism for CO to influence pneumonia, we believe the CO findings should be interpreted as probably reflecting some other traffic related pollutant.
Atkinson for the APHEA 2 group1
found an association with asthma and all respiratory admissions, but not chronic obstructive pulmonary disease admissions and PM10
, and a smaller estimate for black smoke.
There are several recent studies of intermediate markers of health in Boston showing associations between PM2.5 and/or components of fine particles.
De Meo and coworkers41
found a significant effect of ambient particulate air pollution on decreased oxygen saturation at rest in a cohort of older people studied during the summer of 1999 in Boston. This suggests that pollution may results in low levels of hypoxaemia, which may in turn affect the cardiovascular system in ways that influence cardiopulmonary events.
Another study in Boston42
suggests that changes in PM2.5
lead to within‐person increases in resting and exercise blood pressure among vulnerable patients with cardiovascular disease. In a Los Angeles panel study of patients with chronic obstructive pulmonary disease43
and in a large cross sectional German study of older adults,44
higher levels of air pollution were also associated with higher blood pressure.
exposure to summertime air pollution and heart rate variability was examined in a panel study of 28 elderly subjects and we found stronger associations with BC, an indicator of traffic particles, than with PM2.5
; CO had similar patterns of association to black carbon.
Another study in Boston46
used implanted cardioverter defibrillator records of ventricular tachyarrhythmias to assess the role of air pollution as a trigger of these potentially life threatening events. The authors found that associations of ventricular tachyarrhythmias with fine particle mass, CO, NO2
, and BC suggest a link with motor vehicle pollutants.
Reducing levels of pollution from diesel engines and coal burning power plants, even in cities with comparatively low concentrations, will prevent serious health consequences.
What is already known on this topic
Significant associations between PM10 and hospital admissions for MI and respiratory causes have been reported before.
Hypotheses to explain the potential mechanisms for these particle effects might entail systemic inflammation, changes in autonomic function, or oxidative stress capable of influencing both cardiovascular and pulmonary physiology.
Zelikoff has reported that animals infected with strep pneumonia, and subsequently exposed to concentrated air particles, had double the bacterial burden in the lungs 48 hours later, compared with animals exposed to filtered air.47
Other human and animal studies48
support the role of particles in increasing systemic inflammation, with pollution related increases in plasma viscosity,49
C reactive protein,50,51,52
white blood cell counts,54,55
and blood pressure.43,44
Our findings in this Boston study provide support for another link in the ongoing research into the biological mechanism of air pollution and health.
The paper of Laden and colleagues analysed the elemental composition of PM2.5 to identify source related fractions of fine particles. The data used were for the years 1979–1988 and in that study mobile sources in Boston accounted for 29% of PM2.5, while power plant based sulphate particles accounted for 50% of fine mass. In our analysis traffic accounted for a somewhat larger fraction of the PM2.5 than in the earlier study, but this is consistent with the increased traffic over the period and the modest reduction of particles from coal burning power plants (included in our “PM2.5 non‐traffic”) because of the acid rain control provisions of the 1990 clean air act. Laden and coworkers report power plant particles were more associated with respiratory deaths and traffic particles with cardiovascular deaths.
One limitation of the study is that we could not analyse other components of the particles such as organic carbon, metals, and sulphates, because these data were not available for the time period when we had hospitalisation data.
Another limitation of this study is the use of a central site for monitoring air pollution. Recent studies comparing personal with ambient exposure have reported good correlations between day to day changes in central station PM2.5
and personal exposure.56,57
The correlation for BC is weaker, and this added exposure error may have weakened the relative effects of BC. Of particular interest, Sarnat and coworkers have shown this to be the case in Boston,58
and that day to day variations in ambient CO and NO2
are better predictors of personal exposure to BC than of personal exposure to themselves. Hence the use of ambient monitors for particle exposure seems justifiable, while the associations with gases from traffic sources may be more general surrogates for traffic exposure, or specifically traffic particles.
What the paper adds
This manuscript is one of the first to examine PM2.5 and its components, as well as gaseous air pollutants in their association with specific causes of hospital admissions (heart attacks and pneumonia) in an elderly popluation.
In future studies the use of GIS based exposure measures in large population studies could be useful to focus on the health effects of more individualised measures of exposure to air pollution.
This manuscript is important to the field of public health as it is one of the first to examine PM2.5 and its components, as well as gaseous air pollutants in their association with specific causes of hospital admissions (heart attacks and pneumonia) in an elderly population. We report the pattern of associations varied by season, and that both traffic pollutants and non‐traffic particles were associated with heart attacks.