This snapshot taken on 16/07/2006, shows web content selected for preservation by The National Archives. External links, forms and search boxes may not work in archived websites.
Epidemiological Evidence for Health Effects of Particles
Summary
Few epidemiological studies have addressed directly the independent health effects of the fine as compared with the coarse fraction of particles. The limited evidence that is available comes from time-series studies that have related short-term fluctuations in mortality, hospital admissions and other measures of morbidity to temporal changes in the concentrations of air pollutants; and from cross-sectional and longitudinal investigations that have compared rates of disease or death in populations according to indices of their long-term exposure to pollutants.
The ability of these studies to discriminate between the effects of different particulate fractions is limited by the generally high correlation between different measures of particles, and especially between PM10 and PM2.5. Also, in the studies of long-term exposure, full adjustment for potential confounders has not always been possible. Overall, however, the variation in risk across the range of ambient levels (e.g. 10th to 90th percentile) has been similar for PM10 and PM2.5. PM10-2.5 has shown less consistent associations with health effects and appears less robust than PM2.5 in multi-pollutant models.
Sulphates, which form part of PM2.5, have also been associated with health effects, but probably do not account for all of the effects of PM2.5. In studies of summer acid haze in eastern North America, the effects of sulphates have been difficult to separate from those of acid aerosol and ozone. Black Smoke tends to show similar effects to those of PM10 and to be robust in multi-pollutant models. This suggests that material monitored by the Black Smoke method is responsible for some of the toxicity of PM2.5. The limited evidence that is so far available suggests that associations with PM2.5 are similar to those with counts of particle numbers in the sub-micrometre range.
Because fine particles are more uniformly distributed over a given area than coarse particles, it is likely that a central monitor will provide a less accurate reflection of population exposures to PM10 than to PM2.5. Such misclassification of exposures would normally be expected to bias estimates of risk towards the null. Thus, even where morbidity and mortality appear to be more strongly associated with PM2.5 than with coarser fractions of particles, it does not necessarily follow that the former is more toxic.
Current epidemiological evidence suggests that PM2.5 and PM10 are equally effective for detecting and monitoring adverse health effects from particulate pollution in the general population. There is some indication that the toxicity of PM10 resides largely in the PM2.5 fraction, but because of methodological limitations, this is far from conclusive.
Introduction
88. Epidemiological evidence may be conveniently divided into that which concerns short-term effects and that which concerns long-term or chronic effects. These are based on very different methodologies, each having its particular strengths and weaknesses. The purpose of this chapter is to examine the evidence that PM2.5 or any of its components is associated with health effects and if so to what extent PM2.5 accounts for the effects of PM10, of which it is a part. Epidemiological studies measure this fraction either directly as PM2.5 or indirectly (and partially) by measuring specific components such as sulphate (SO4), nitrate (NO3), acid aerosol (H+) or Black Smoke (BS). The often strong intercorrelations between these components in the varying mixtures of particles occurring in the ambient situation makes it difficult to distinguish the individual effects of particle species.
89. There is a general lack of epidemiological evidence concerning fine particles as measured by PM2.5 because it is not routinely monitored. In North America, modern time-series studies using data from the late 1980s now use PM10 rather than the older TSP measure, which included particles above 10 µm in diameter or the Coefficient of Haze, a reflectance method which requires site-specific calibration with PM10. Some studies have used sulphate or acid aerosol, which may be considered to be components of PM2.5. Until recently, in Europe, the main measures of particles were TSP and Black Smoke but some studies using PM10 are now appearing. Published reports of the health effects of PM2.5 are restricted to a few panel studies. Black Smoke should be considered in any assessment of the effects of alternative metrics of particles because it comprises particles <4.5 µm, and is a good indicator of primary emissions from combustion sources.
90. Direct comparison of coefficients per unit increase of pollutant gives information about the relative toxicity of two particle measures, provided that the units of measurement are the same. However, more important in public health terms are the relative effects of the measures across the ambient ranges. This indicates the relative contribution of each pollutant to variations in morbidity and mortality in a community. Risk estimates are therefore often expressed in relation to an interpercentile range (e.g. 10-90th or interquartile) of a pollutant. Where one fraction is a component of another (e.g. PM2.5 is a part of PM10, or sulphate is a part of PM2.5), comparisons are difficult because the measures are not independent. The best way of comparing measurements is to compare separate entities (e.g. PM2.5 and PM10-2.5, or sulphate and Black Smoke). However, these may still be correlated and it is therefore necessary to examine their independent effects in multi-pollutant models. This presents difficulties because the effects are often small and multi-pollutant models may be inconclusive.
Short-Term Health Effects: Epidemiological Evidence from Time-Series Studies
91. Short-term effects of ambient particles have been extensively investigated using time-series studies in which estimates of daily exposure are analysed in relation to health outcomes occurring on the same day or lagged up to several days. There are two types: panel studies, which obtain data at an individual level and employ outcomes such as lung function, symptoms or use of medications, and ecological time-series studies, which use aggregated data comprising counts of deaths or health care utilisation usually obtained from routinely recorded data sources. Analyses of time-series use statistical techniques to minimise potential confounding due to trends, seasonal, meteorological and other influences, and to allow for serial correlation often found in such data. In attempting to identify and interpret an association between a pollutant and health there is also the problem of confounding by and interaction with other pollutants. This is difficult and sometimes impossible to resolve because of the complex interplay between pollutants and the small associations involved.
Ecological Time-Series Studies of PM2.5
Mortality
92. Dockery et al (1992) analysed daily all-cause mortality for one year (1985/6) in St Louis and in eastern Tennessee surrounding Kingston/Harriman (Table 5). The only statistically significant relationship was with PM10 in St Louis; however, the PM10 coefficient was of similar size in Kingston/Harriman. Coefficients for PM2.5 and sulphate were also increased but not significantly. The effects of each pollutant are best compared using a method that adjusts for the range of pollutant values and in this case the range from zero to the mean was used. The effects of PM2.5 and of sulphate appeared to be similar to those of PM10. In St Louis, coarse particles (10-2.5 µm) were also positively associated with mortality but neither fine nor coarse particles showed a stronger association than the other when considered simultaneously. Analysis of the coarse fraction was not mentioned for Kingston/Harriman. In St Louis, elemental concentrations of aluminium, calcium, chromium, iron and silica were all correlated with PM10 at r = 0.5 or higher and were all positively, though not significantly, associated with mortality. Similar findings relating to elemental concentrations applied to Kingston/Harriman. The effects of acid particles were small and non-significant. Relative lack of statistical power due to the short time-period limited the depth of analysis required to examine the independent effects of the different fractions of particles. The results indicate that PM2.5 was an important component of PM10 and that sulphate was more important than acid aerosol.Table 5. Total daily mortality and air pollution in St Louis and Kingston/Harriman. Comparison of PM10 and PM2.5 (Dockery et al 1992).
PM10 (µg/m3) PM2.5 (µg/m3) Sulphate (µg/m3) Mean St Louis 27.6 17.7 8.0 Kingston/Harriman 30.0 21.0 8.7 Relative Risk
(95% CI)St Louis 1.042 1.031 1.050 (1.003, 1.083) (0.996, 1.066) (0.957, 1.152) Kingston/Harriman 1.049 1.049 1.072 (0.959, 1.147) (0.970, 1.134) (0.969, 1.158) 93. The most comprehensive ecological time-series designed specifically to compare the effects of PM2.5 on mortality with those of other particle metrics was reported by the Harvard group (Schwartz et al 1996). This took advantage of a particle monitoring system set up as part of the Six Cities air pollution epidemiology programme. The modest correlation between the fine and coarse particle fractions provided an opportunity to distinguish the effects of these two measures. Daily mortality was analysed for each of the six cities over an eight year period between 1976 and 1987 and presented both individually and as a combined analysis. The summary estimates are shown in Table 6. The effect of PM2.5 was more significant than that of PM10 but the relative risks for the 5-95th interpercentile range of pollutant were very similar. The effects of the coarse fraction were small and not significant. They were reduced even more when analysed with PM2.5 in the model. The association with sulphates was also significant but the relative risk was less than that for PM2.5. It was concluded that the effects of PM10 were largely located in the fine fraction. The effect of sulphate did not account for all the effects of PM2.5, suggesting that particles resulting from primary combustion sources were also important. The paper addressed the question of whether the results could have been influenced by differences in measurement error between the fine and coarse fraction, but concluded that this was unlikely.
Table 6. Total daily mortality and air pollution in six eastern US cities (Schwartz et al 1996).
PM10 (µg/m3) PM10-2.5 (µg/m3) PM10-2.5 (µg/m3) H+ (µg/m3) Sulphate (µg/m3) 5-95th %ile 59.8 38.8 29.1 53.3 17.0 Relative Risk (95% CI) 1.05 1.058 1.02 1.012 1.038 (1.032,1.069) (1.042, 1.074) (0.992, 1.049) (0.981, 1.044) (1.022,1.055) 94. Ostro (1995) used an index based on visibility, which had been calibrated against PM2.5 to investigate the association of fine particles with daily mortality in the Los Angeles metropolitan area from 1980 to 1986. In the summer, there were small and significant effects on all-cause and respiratory mortality, but no significant associations were found when the all-year series was analysed. The results were robust to sensitivity analyses using different assumptions about the relationship between the visibility index and PM2.5. Loomis et al (1999) reported an association between fine particles and infant mortality in Mexico City. The studies of Ostro and Loomis, while consistent with an effect of fine particles on mortality, tell us nothing about any effects of the coarse fraction.
95. In view of the difficulties encountered in separating the effects of fine from coarse particles, Schwartz et al (1999) investigated the effects of dust storms, which are associated with coarse particles of crustal origin, on mortality in the city of Spokane, WA. The mean PM10 level in 17 dust storms was 263 µg/m3, but no effect on mortality could be demonstrated. In contrast, Ostro and colleagues (1999) reported associations between PM10 and daily mortality in the Coachella Valley, a desert area of California where coarse particles comprise the larger part of PM10. The relevance of either of these findings to particulate pollution in the UK is unclear.
96. Two recent studies provide a further comparison of the effects of the coarse and fine fractions on daily mortality. In Phoenix, Arizona, significant associations with daily total mortality in the elderly were observed for the coarse but not the fine fraction (Mar et al 2000). Because data are not presented for the fine fraction, it is not possible to evaluate the extent of the different effects. For cardiovascular mortality, associations with both the coarse and fine fractions were significant and estimates of effects across the interquartile range were larger for the fine fraction. In Mexico City, both the coarse and fine fractions were significantly associated with daily mortality (Castillejos et al 2000). In multi-pollutant models, the effect of the coarse fraction was more robust than that of the fine fraction. These studies are consistent with the idea that the coarse fraction should not be excluded from consideration of health effects.
97. Levy and associates (2000) reported a meta analysis of 29 estimates of particulate matter associations with mortality, 19 of which related to US cities. The overall effect was estimated at 0.7% per 10 µg/m3 increase in particulate matter but the finding of relevance to the present discussion is that the particulate matter effect estimate was greater in cities where the proportion of particulate matter attributable to PM2.5 was greater. These results provide some evidence that it is PM2.5 which is the more important fraction.
98. To inform the thinking of EPAQS, an analysis of PM2.5, PM10, PM10-2.5, sulphate and Black Smoke and daily mortality in the West Midlands of the UK was commissioned by the Panel. This was the only area in the UK with measurements of PM2.5 over an extended period. The results are summarised in Appendix 1. In all-year analyses, no measure of particles showed a significant effect, but in the warm season there were significant associations between all- cause mortality and PM10, PM2.5, Black Smoke and sulphate, but not PM10-2.5. Overall, the findings did not help in distinguishing the effects of PM2.5, PM10, sulphate or Black Smoke but the effects of PM10-2.5 were somewhat less and in some cases negative.
Hospital Admissions
99. Three studies from Canada enable the effects of different particle fractions on the use of hospitals for cardio-respiratory disease to be compared. All limited their analyses to the summer months. The study of Thurston et al (1994) used a variety of measures of particles and analysed their associations with daily hospital admissions for respiratory diseases to hospitals in Toronto during July and August 1986-1988. TSP, PM10, PM10-2.5, PM2.5 and sulphate were all strongly intercorrelated. In univariate models, all these measures, except for PM10-2.5, were significantly associated with increased respiratory admissions, with PM2.5 showing the most significant association, followed by sulphate. In models including ozone, all associations with particles become non-significant with the rankings being H+ > sulphate > PM2.5 > PM10 > TSP in descending order. This led the authors to conclude that the finer fractions were more important. Overall, however, the dominant pollutant was ozone with relative risks of over 10 times those for the most important particulate metric (H+ in this instance).100. The relevant results from a study by Delfino et al (1997) of emergency room admissions for respiratory diseases are shown in Table 7. PM10, PM2.5 and sulphate were all positively and significantly associated with admissions. The effects per unit of mass were greater for PM2.5 than for PM10, but the effect of an increase from zero to the ambient mean was greater for PM10, reflecting its greater mass. For sulphate, an increase to the mean was associated with the least percentage increase in admissions. In multiple pollutant models, ozone was more robust than any of the particle indicators.
Table 7. Fine particles and emergency room visits during June to September in Montreal 1992/93. (Delfino et al 1997)
PM10 (µg/m3) PM2.5 (µg/m3) SO4 (µg/m3) Mean 21.7 12.2 3.34 Relative Risk 1.16 1.12 1.06 (95% CI) (1.04, 1.28) (1.02, 1.21) (0.99, 1.11) Correlation with PM2.5 (r) 0.96 1.0 0.93 101. The effects of different particle fractions on hospital admissions for cardio-respiratory disease in Toronto during the summers of 1992-4 were analysed by Burnett et al (1997) (Table 8). PM10, PM2.5, PM10-2.5, sulphate and Coefficient of Haze were all significantly associated with respiratory admissions and had similar effects for an increase in pollutant from the 25-75th percentile, though the effect of PM2.5 was somewhat greater (RR 1.037) than that of PM10-2.5 (RR 1.023). Cardiac admissions were significantly associated with PM10 and PM10-2.5 and were similar in size of effect to that of PM2.5, though the latter was not significant. The ranking of effects across the interquartile range were, in decreasing order, PM10-2.5 > PM10 > PM2.5 > sulphate. This ranking does not correspond to that which might have been expected had PM2.5 best reflected the active components of the aerosol.
102. The only other published study of admissions to analyse PM2.5 is from Seattle where a major source of particles is smoke from open wood fires (Sheppard et al 1999). Significant associations between PM10, PM2.5 and coarse fraction and daily admissions for asthma under the age of 65 years were found but the effects of the coarse and fine fractions could not be distinguished.
Table 8. Effect of different particle fractions on cardio-respiratory admissions in Toronto during the summers of 1992-4 (Burnett et al 1997).
Total Particles PM10 (µg/m3) Fine Particles (PM2.5) (µg/m3) Coarse Particles (PM10-2.5) (µg/m3) Coefficient of Haze (103 in ft) Sulphate (µg/m3) H+ (nmol/m3) Interquartile range 14.3 11.0 4.75 0.25 3.84 5.25 RESPIRATORY ADMISSIONS Relative risk for interquartile range 1.030 1.037 1.023 1.037 1.029 1.024 (1.048, 1.013) (1.015, 1.059) (1.010, 1.036) (1.023, 1.051) (1.015, 1.043) (1.010, 1.038) CARDIAC ADMISSIONS Relative risk for interquartile range 1.033 1.031 1.036 1.062 1.017 1.024 (1.004, 1.063) (0.997, 1.066) (1.015, 1.057) (1.040, 1.084) (0.996,1.039) (1.004, 1.045) Correlation with fine particulates 0.89 1.00 0.72 0.45 0.79 0.76 103. Hospital admissions were also examined as part of the commissioned study of the West Midlands described in Appendix 1. In an all-year analysis of respiratory admissions, there were no significant associations with any measure of particles but in seasonal analyses there were associations with sulphate. In some of the age-diagnosis specific series there were significant associations with PM10, PM2.5 and Black Smoke, but none with PM10-2.5. Generally, none of the particle measures showed strong and consistent association with respiratory or cardiac admissions. Examination of the lag patterns showed PM10, PM2.5 and Black Smoke shared similarities whilst the lag patterns for sulphate and PM10-2.5 in particular were somewhat different and less consistent.
104. A number of other studies from North America have examined the effects of fine particles on mortality or admission for respiratory and/or cardiac admissions during the summer months, using sulphate and acid aerosol as the particle indicator but generally without any measurement of PM10 or PM2.5 (Thurston et al 1992; Burnett et al 1994; Delfino et al 1994; Burnett et al 1995; Gwynn et al 2000). Nevertheless, these studies contribute in a general way to the evidence that fine particles may have adverse effects. All of these studies found associations with sulphate or acid aerosols or both. These were usually independent of ozone levels, which also tended to be associated with increased admissions. The findings indicate, at the least, that the "acid summer haze", which is characteristic of the north eastern USA and adjacent Canada has effects on daily admissions but it is often difficult to clearly separate the effect of sulphates from those of the acid aerosol or ozone component. The relevance of these findings to the UK, where acid levels are much lower, is therefore unclear.
Black Smoke and PM10
105. Black Smoke measures particles that are both dark in colour and less than about 4.5 µm in diameter. It is therefore relevant to compare the effects of Black Smoke with those of PM10 to see how much of the PM10 toxicity resides in this component. In Amsterdam, Black Smoke was more strongly and more significantly related to daily mortality than PM10 when considered in unit terms (Verhoeff et al 1996). However, the effect of an increase in pollutant from the 10-90th percentile was similar (PM10 +3%, Black Smoke +3.3%; Table 9). In multi-pollutant models PM10 was less robust than Black Smoke.Table 9. Comparison of effects of Black Smoke and PM10 on daily mortality in Amsterdam (Verhoeff et al 1996). The most significant lag of those examined (same day) is shown.
PM10 (µg/m3) Black Smoke (µg/m3) 10-90th %ile of pollutant 50 19 Relative Risk 1.030 1.033 (95% CI) (0.993, 1.070) (1.004, 1.039) 106. An analysis of daily mortality in London from 1992-4 also allows the effects of PM10 to be compared with those of Black Smoke (Bremner et al 1999) (Table 10). The effect of Black Smoke on all-cause mortality was stronger than that of PM10 but both associations were significant. The associations with respiratory mortality were both significant and similar in scale. In two-pollutant models, the effect of PM10 was reduced when Black Smoke was included, whereas the effect of Black Smoke was not sensitive to the inclusion of PM10. This is consistent with the idea that Black Smoke includes an important toxic component of PM10.
Table 10. Comparison of effects of Black Smoke and PM10 on daily mortality in London 1992-4 (Bremner et al 1999).
PM10 (µg/m3) BS (µg/m3) PM10 adj* BS BS adj**PM10 (10-90th %ile) 30.7 16.1 Relative Risk 1.040 1.031 1.013 1.033 (95% CI) (1.009, 1.073) (1.004, 1.060) (0.97, 1.057) (0.994, 1.072) Note: See paragraph 106 for explanation
* adjusted for the effects of Black Smoke
** adjusted for the effects of PM10107. A number of panel studies have measured both PM10 and PM2.5 (Schwartz et al 1994; Neas et al 1995; 1996; Romieu et al 1996; Peters et al 1997a). Two papers from Germany present data on PM10 and sulphate (Peters et al 1996; 1997b) and one, data on sulphate alone (Peters et al 1997c). Two papers enable comparison of mass measures of fine particles with numbers of particles in the fine and ultrafine fractions (Peters et al 1997a; Pekkanen et al 1997).
108. Schwartz et al reported a large daily morbidity study in six eastern US cities among 1844 school children (about 300 in each city) over one year (Schwartz et al 1994). Particles measured were PM10, PM10-2.5, PM2.5, sulphur and H+. The correlation between PM10 and PM2.5 was 0.92. Both PM10 and PM2.5 were significantly associated with an increased incidence of cough. The authors could find no evidence that any other measure of particles including PM2.5 was better that PM10 in predicting lower respiratory symptoms.
109. Neas et al studied 62 symptomatic and 46 asymptomatic children with diaries and peak expiratory flows during the summer of 1991 (Neas et al 1996). State College, Pennsylvania was chosen because of relatively high levels of acidity. The focus of the study was the effect of mould spores but results for PM2.1, black soot, sulphate and particle strong acidity were also reported. None of the particle measures was significantly associated with reductions in PEFR. However, for symptoms of cough, wheeze and colds, PM2.1 and H+ stood out as having the most consistent positive associations. PM10 was measured but no results were reported.
110. Another report by Neas et al described a panel of 83 children in Uniontown, Pennsylvania, followed using daily diaries during the summer of 1990 (Neas et al 1995). PM2.5 was associated with a significant increase in evening cough episodes and the association strengthened when weighted for time spent outdoors. Whereas particle strong acidity, PM10 and sulphate were associated with significant reductions in peak flow rate, the effect of PM2.5 was smaller and non significant.
111. Schwartz and Neas (2000) have reanalysed three panel studies of children in the eastern US to compare the effects of fine and coarse particles. In the largest of these, based on the Harvard Six City Diary Study, it was found that PM2.5 had the larger and more stable effect on all lower respiratory symptoms combined. In the analysis of cough without other symptoms, there was a significant association with the coarse fraction. The results for the other panels, all based in Pennsylvania, while pointing towards associations between lung function and PM2.5, do not show clear differences between PM2.5 and PM10-2.5.
112. The paper by Peters et al (1997a) is important because it allows different mass fractions of particulate matter to be compared along with number and size distributions. Data were obtained on 27 non-smoking adult asthmatics in Erfurt (former East Germany) during the winter of 1991-2. Significant associations with symptoms and lung function were observed for the ultrafine as well as various other mass fractions. Overall, the authors concluded that the effects of the number of particles were stronger than those of their mass but their respective effects were difficult to separate. In comparing the effects of PM10 with smaller size/mass fractions, there seemed to be little difference in respect of reductions in lung function and greater effects on symptoms, but the data presented do not provide separate estimates for the coarse fraction (a sample of the findings is shown in Table 11).
113. A panel study of asthmatic children in Mexico City, reported by Romieu et al (1996), included measures of both PM2.5 and PM10. Compared with the UK, mean levels were high (85.7 and 166.8 µg/m3, respectively); also, the mean one hour concentration of ozone was 190 ppb. Both measures of particles were associated with significant reductions in peak flow rate and increases in symptoms. The effect of PM2.5 remained after allowing for that of ozone.
Table 11. Associations between particulate fractions and respiratory symptoms and lung function in a panel of non-smoking asthmatic adults (extracted from Peters et al 1997a). Effect sizes refer to an interquartile range of particle concentration.
Particle size (mass concentration) Five day mean Feeling ill during the day Morning Peak Expiratory
Flow (l/min)Odds ratio 95% CI Effect 95% CI PM0.01-2.5 1.21 1.06, 1.38 -1.42 -2.57, -0.28 PM10 1.47 1.16, 1.86 -1.51 -3.20, 0.19 114. Pekkanen et al (1997) compared the effects on peak flow rate of PM10, Black Smoke and particle number concentrations in a panel of 39 asthmatic children in Kuopio, Finland, where particle levels are quite low. Daily variations in Black Smoke and in particle number concentrations between 0.032 and 0.32 µm and between 1.0 and 10 µm were highly correlated (around 0.9), but correlations between PM10 and particle number concentrations were somewhat less (<0.7). All these measures were associated with reductions in peak flow rate but only PM10 and Black Smoke were significant. It was concluded that the number concentration of ultrafine particles was no more strongly associated with variations in peak flow than with PM10 or Black Smoke. PM2.5 was not examined specifically. In a further panel study from the same group in the same city (Tiittanen et al 1999), 49 children with chronic respiratory symptoms were followed for six weeks in the spring along with various measures of particles, including fine and coarse fractions. The correlations between PM10, PM10-2.5 and PM2.5 were very high (all above 0.9) and levels were generally low. PM10 was as strongly associated with reductions in lung function as the number of ultrafine particles. Cough symptom was significantly associated with PM10, PM10-2.5 and PM2.5 at a 2 day lag. Inconsistent associations at a variety of lags were observed with all of the various fractions, including coarse fraction and particle number concentrations.
115. Several other panel studies in a variety of environments have shown associations between PM2.5 and symptoms or physiological outcomes. Koenig et al (1993), in Seattle, where wood smoke is an important source of particles, found negative associations between fine particles (estimated using a light- scattering device) and lung function in a panel of children. In three summer camp studies, Thurston et al (1997) found associations between ozone and reduced lung function but considered that sulphates and acid aerosols, which are part of the acid summer haze, might also have an independent effect. Gold et al (1999) in Mexico City, where mean levels of particles and ozone are much higher than in the UK, found that the effects of PM10 and PM2.5 on symptoms and lung function in children were similar but that PM10-2.5 had little effect on lung function. Finally, Liao et al (1999), studied the effects of PM2.5 on heart rate variability in elderly residents of a retirement home in Baltimore. Measurements of PM2.5 from inside and immediately outside the residence were associated with reduced heart rate variability. The mechanism may be related to an effect on autonomic control of the heart. The significance of this finding is that reduced heart rate variability is a predictor of clinical cardiac events. This study did not, however, provide comparative data on PM10 or coarse particles.
116. If fine particles are more uniformly distributed over a given area than coarse particles, it is likely that a central monitor will provide a more accurate reflection of the population exposure to PM2.5 than to PM10. Assuming similar toxicity, such misclassification of exposure from the use of a central monitor is generally thought to bias estimates of the health risks towards the null. From this it has been argued by Lipfert and Wyzga (1997) that differences between fine and coarse particles in their associations with mortality and morbidity, may be attributed partly or wholly to differences in measurement error. A detailed examination of this question led Schwartz et al (1996) to conclude that differences in measurement error were not likely to explain the tendency for coarse particles to have less significant effects than fine particles. This is still debated.
Long-Term Health Effects: Epidemiological Evidence from Cross-Sectional and Longitudinal Studies
117. Time series and panel studies provide information about the short-term impact of air pollutants on morbidity and mortality, but they do not allow full assessment of the burden of disease that is attributable to such pollutants over longer periods. For example, where peaks of mortality are observed in a population at times of increased pollution, some or all of the excess deaths may occur in individuals who are already severely ill and who therefore would have only a short life expectancy even in the absence of pollution episodes. Data on the longer-term impact of pollution come from cross-sectional and longitudinal comparisons of morbidity and mortality in populations that have differed in their average exposure to pollutants over a period of years. Several such studies have included measurements of more than one particulate fraction.
118. Ostro (1990) analysed the relation of acute respiratory morbidity to four alternative measures of particulate matter, using data from the US Annual Health Interview Surveys during 1979-81. Data on respiratory morbidity were ascertained for the two weeks leading up to each subject's interview and were summarised as the number of days of respiratory-related restricted activity. The four indices of particulate pollution were total suspended particulate, inhalable particulate (predominantly <15 µm), PM2.5 and sulphate, each of these being characterised by a two-week average. Inhalable particulate and PM2.5 were highly correlated (correlation coefficient = 0.81). The analysis was restricted to 7348 residents of 25 metropolitan areas who were aged 18-65 and currently working. It used a Poisson model to estimate the number of days of respiratory-related restricted activity that was associated with a given increase in the concentration of a pollutant, and adjusted for age, sex, race, education, family income, marital status, the existence of chronic health problems, outdoor temperature, population density and occupational class. Smoking was not included in the model, but showed minimal correlation with exposures to air pollution. In an analysis using pollution measures for the same two-week period as the health outcome, only sulphate was positively and significantly associated with respiratory disability. However, correlations were stronger and more significant when allowance was made for a two-week lag between exposure and illness. The regression coefficients for inhalable particles and PM2.5, measured in days restriction per µg/m3, were 1.29 (SE 0.11) and 0.90 (SE 0.07), respectively. When account is taken of the standard deviations of these two measures of particles (19.2 and 11.4), there appears to have been little difference in their ability to explain variation in health outcomes. It should be noted that the variation in pollution in this study arose both from geographical and from temporal differences.
119. As part of the Six Cities Study, Dockery et al (1989) examined the association of respiratory symptoms and lung function in white children aged 10-12 with different measures of particulate pollution. The results were expressed as odds ratios for the difference in pollution between the most and the least polluted city after adjustment for sex, age, parental education, maternal smoking and whether there was a gas stove in the home. In the analysis of symptoms, odds ratios for PM15 tended to be slightly higher than for PM2.5. For example, the values for chronic bronchitis were 2.5 (95% CI 1.1-6.1) and 2.1 (95% CI 0.8-5.9). Only TSP was associated with lower levels of pulmonary function. The power of this investigation to detect differences in the effects of specific particle fractions was, however, low.
120. Peters and colleagues (1999a and 1999b) compared the prevalence of respiratory illness and measures of lung function in children aged 9-16 years from 12 communities in southern California with varying patterns of air pollution. The data analysed came from the initial phase of a longitudinal study still in progress. Histories of respiratory illness were ascertained for 3676 children (response rate 76%) in 1993 through questionnaires completed with help from their parents. Measurements of lung function were carried out in 3293 children, again in 1993. Outdoor concentrations of PM10 and PM2.5 were monitored during 1994 (along with several other air pollutants), and were highly correlated (r = 0.90). The relation between health outcomes and average ambient levels of air pollutants was explored by two-stage regression methods that took account of a wide range of potential confounding variables. Neither PM10 nor PM2.5 was significantly associated with the prevalence of asthma (ever or in the past 12 months), bronchitis (in the past 12 months), cough (in the past 12 months) or wheeze (ever). For each diagnosis and symptom, the odds ratios for the two metrics, scaled according to their interquartile range, were similar. The four indices of respiratory function that were measured (FVC, FEV1, PEFR and MMEF) were all negatively related to both PM10 and PM2.5 in girls, but not in boys. After scaling to the interquartile range for each pollutant, the regression coefficients for PM10 and PM2.5 were generally similar, the biggest difference being for MMEF (-109 ml/sec per interquartile range (µg/m3) for PM10 and -130 ml/sec per interquartile range (µg/m3) for PM2.5).
121. In a further analysis of the same data (McConnell et al 1999), the prevalences of bronchitis and of phlegm were significantly associated with higher levels of particulate pollution among children with a doctor's diagnosis of asthma, although not in other children. Again, odds ratios scaled to the interquartile range were similar for PM10 and PM2.5.
122. Özkaynak and Thurston (1987) analysed mortality from all causes combined during 1980 in 113 Standard Metropolitan Statistical Areas of the United States and related it to different indices of particulate pollution (some measures were not available for all of the areas). The indices of pollution were summarised as means of 24-hour average concentrations recorded during 1979-82 in each area, and where inhalable and fine particulate had not been measured directly, they were estimated from other measures. Analysis was by multiple regression with adjustment for income, age, education and population density. No data were available on smoking but from experience in earlier studies the authors believed it was unlikely to be a major confounder. Nor was any allowance made for climate. Of the pollution variables examined, sulphate showed the most consistent and significant associations with mortality. The regression coefficients for inhalable particles and PM2.5, measured in deaths/105 persons per year per µg/m3, were 1.0 (SE 0.9) and 2.9 (SE 2.1), respectively. The adjustment for age in this analysis was relatively crude, and errors in indirectly inferred exposures to inhalable particles and PM2.5 may have weakened their associations with mortality.
123. Abbey et al (1995) studied 1868 Seventh Day Adventists who had lived near nine airports in California for more than 80% of the time since 1966, and who completed a questionnaire about respiratory symptoms in 1977 and again in 1987. Cumulative exposure to PM2.5 was estimated from airport visibility data and to PM10 from measurements of total suspended particulates. From the questionnaires, subjects were classified as having no, possible or definite obstructive airways disease, chronic bronchitis and asthma in each of 1977 and 1987. The risks of developing new or more severe symptoms over the 10-year study period were assessed in relation to PM10 and PM2.5 exposures by logistic regression with adjustment for sex, age and education. No clear differences emerged in the significance of associations with the two metrics, but they were highly correlated (R = 0.92 for 1977-87 subject means), making distinctions difficult. Errors in their indirect assessment from other measures may have further reduced the power of the study to detect differences.
124. As part of the Six Cities Study, Dockery et al (1993) analysed mortality prospectively over 14-16 years in random samples of adults (n = 8111). Mortality rate ratios were computed in relation to long-term measures of pollution, with adjustment for sex, age, smoking status, pack years, education and body mass index, and were estimated for the difference in pollution between the most and least polluted cities. Total mortality was significantly associated with both inhalable particulate (PM15/PM10) and PM2.5 with virtually identical rate ratios and 95% confidence intervals: 1.27 (1.08-1.48) and 1.26 (1.08-1.47), respectively. The associations were with deaths from lung cancer and cardiopulmonary disease but not with mortality from other causes.
125. In another cohort study, Pope et al (1995) followed 552,138 adults living in 151 US metropolitan areas during 1982-89. The subjects provided information at baseline about personal risk factors for mortality, while information about air pollutants was obtained from national databases. This was summarised as the mean sulphate concentration for 1980 in the participant's area of residence and the median concentration of fine particles during 1979-83 (data on fine particles were only available for 295,223 subjects). For the 47 metropolitan areas with both measures, the correlation coefficient was 0.73. After adjustment for sex, age, race, smoking habits, passive smoking, education, alcohol intake, body mass index and exposure to certain occupational hazards, associations with sulphates and fine particles (expressed as the risk ratio for the range of pollution across all areas) were similar for mortality from all-causes (1.15 and 1.17) and cardiopulmonary mortality (1.26 and 1.31). However, for lung cancer the relation with sulphates was stronger (1.36 v 1.03). This difference could not be explained by the restriction of the fine particle analysis to a smaller number of areas. A parallel cross-sectional analysis of total population mortality in the same areas during 1980 against sulphate and fine particulate concentrations produced similar findings.
126. More recently, Krewski et al (2000) have reanalysed data from the studies by Dockery et al (1993) and Pope et al (1995), in some cases incorporating additional air monitoring data that were not used in the original analyses. The findings for the Six Cities Study were similar to those reported previously. For the study by Pope et al, the risk estimates in relation to cardiopulmonary mortality were little altered, but the associations of all-cause mortality with PM2.5 and of lung cancer mortality with sulphate were reduced. Corresponding analyses were also carried out for PM15 and PM15-2.5, both of which showed a weaker relation to mortality than PM2.5.
127. Several factors limit the conclusions that can be drawn from the epidemiological studies described in this chapter, regarding the relevance to health of different measures of particulate matter. First and most importantly, the various metrics examined have often correlated strongly with each other, making it difficult to discriminate any differences in effect. Secondly, exposures have been classified on the basis of measurements made at fixed outdoor locations. Such measurements do not always accurately reflect personal exposures (see Personal Exposure chapter), and where they do not, risk estimates can be biased. The effects of such bias are difficult to assess, particularly in multivariate statistical analyses that include several measures of pollution, each of which may be misclassified (Carrothers and Evans 2000). Thirdly, in the cross-sectional and longitudinal studies looking at the long-term impact of air pollutants, the health outcomes investigated may have been influenced not only by contemporary exposures to particles, but also by exposures to particles and other pollutants earlier in life. However, data on the latter have not been available. Because of these limitations, interpretation of the epidemiological data that are currently available must be circumspect.
128. For the health outcomes that have been studied epidemiologically, the variation in risk across the range of ambient levels (e.g. 10th to 90th percentile) has generally been similar for PM10 and PM2.5. PM10-2.5 has shown less consistent associations. Sulphates, which form part of PM2.5, have also been associated with health effects, but probably do not account for all of the effects of PM2.5. Black Smoke has tended to show similar effects to those of PM10, and to be robust in multi-pollutant models.
129. These findings indicate that currently PM10 and PM2.5 would provide equally effective standards for the prevention of adverse health effects from particulate pollution in the general population of the UK. Although some studies suggest that the main toxicity of particulates lies in the finer fraction, they do not provide a basis on which to set an air quality standard for additional or alternative metrics that would protect the public more effectively than a standard based only on PM10. However, because of the uncertainties in the epidemiological data currently available, further research on this question would be helpful. Such research should focus on populations in which differences between the effects associated with different exposure metrics could most readily be discriminated - i.e. those for which the intercorrelation between the metrics is relatively low.
Abbey DE, Ostro BE, Petersen F, Burchette RJ. Chronic respiratory symptoms associated with estimated long-term ambient concentrations of fine particulates less than 2.5 microns in aerodynamic diameter (PM2.5) and other air pollutants. J Expos Anal Environ Epidemiol 1995; 5: 137-159.
Bremner SA, Anderson HR, Atkinson RW, et al. Short-term associations between outdoor air pollution and mortality in London 1992-94. Occup Environ Med 1999; 56: 237-244.
Burnett R, Dales RE, Raizenne ME, et al. Effects of low ambient levels of ozone and sulphates on the frequency of respiratory admissions to Ontario hospitals. Environ Res 1994; 65: 172-194.
Burnett RT, Cakmak S, Brook JR, et al. The role of particulate size and chemistry in the association between summertime ambient air pollution and hospitalization for cardiorespiratory diseases. Environ Health Perspect 1997; 105: 614-620.
Burnett RT, Dales R, Krewski D, Vincent R, Dann T, Brook JR. Associations between ambient particulate sulfate and admissions to Ontario hospitals for cardiac and respiratory diseases. Am J Epidemiol 1995; 142: 15-22.
Carrothers TJ, Evans JS. Assessing the impact of differential measurement error on estimates of fine particle mortality. J Air Waste Manag Assoc 2000; 50: 65-74.
Castillejos M, Borja-Aburto VH, Dockery DW, Gold DR, Loomis D. Airborne coarse particles and mortality. Inhal Tox 2000; 12: 61-72.
Delfino RJ, Becklake MR, Hanley JA. The relationship of urgent hospital admissions for respiratory illnesses to photochemical air pollution levels in Montreal. Environ Res 1994; 67: 1-19.
Delfino RJ, Murphy-Moulton AM, Burnett RT, Brook JR, Becklake MR. Effects of air pollution on emergency room visits for respiratory illnesses in Montreal, Quebec. Am J Respiratory Crit Care Med 1997; 155: 568-576.
Dockery DW, Pope CA, Xu X, Spengler JD, Ware JH, Fay ME, Ferris BG, Speizer FE. An association between air pollution and mortality in six US cities. N Engl J Med 1993; 329: 1753-9.
Dockery DW, Schwartz J, Spengler JD. Air pollution and daily mortality: associations with particulates and acid aerosols. Environ Res 1992; 59: 362-373.
Dockery DW, Speizer FE, Stram DO, Ware JH, Spengler JD, Ferris BG. Effects of inhalable particles on respiratory health of children. Am Rev Respir Dis 1989; 139: 587-594.
Gold DR, Damokosh AI, Pope CA, et al. Particulate and ozone pollutant effects on the respiratory function of children in southwest Mexico City. Epidemiol 1999; 10: 8-16.
Gwynn RC, Burnett RT, Thurston GD. A time-series analysis of acidic particulate matter and daily mortality and morbidity in the Buffalo, New York, region. Environ Health Perspect 2000; 108: 125-133.
Koenig JQ, Larson TV, Hanley QS, et al. Pulmonary function changes in children associated with fine particulate matter. Environ Res 1993; 63: 26-38.
Krewski D, Burnett RT, Goldberg MS, Hoover K, Siemiatycki J, Abrahamowicz M, White WH, et al. Investigators' Report: Reanalysis of the Havard Six Cities Study and the American Cancer Society Study of particulate air pollution and mortality. Health Effects Institute, 2000; pp 98.
Levy JL, Hammitt JK, Spengler JD. Estimating the mortality impacts of particulate matter: what can be learned from between-study variability. Environ Health Perspect 2000; 108: 109-117.
Liao D, Creason J, Shy C, Williams R, Watts R, Zweidinger R. Daily variation of particulate air pollution and poor cardiac autonomic control in the elderly. Environ Health Perspect 1999; 107: 521-525.
Lipfert FW, Wyzga RE. Air pollution and mortality: the implications of uncertainties in regression modelling and exposure measurement. J Air Waste Manag Assoc 1997; 47: 517-523.
Loomis D, Castillejos M, Gold DR, McDonnell W, Borja-Aburto VH. Air pollution and infant mortality in Mexico City. Epidemiol 1999; 10: 118-123.
Mar TF, Norris GA, Koenig JQ, Larson TV. Association between air pollution and mortality in Phoenix, 1995-1997. Environ Health Perspect 2000; 108: 347-353.
McConnell R, Berhane K, Gilliland F, Lodon SJ, Vora H, Avol E, Gauderman WJ, Margolis HG, Lurmann F, Thomas DC, Peters JM. Air pollution and bronchitic symptoms in southern Californian children with asthma. Environ Health Perspect 1999; 107: 757-760.
Neas LM, Dockery DW, Burge H, Koutrakis P, Speizer FE. Fungus spores, air pollutants, and other determinants of peak expiratory flow rate in children. Am J Epidemiol 1996; 143: 797-807.
Neas LM, Dockery DW, Koutrakis P, Tollerud DJ, Speizer FE. The association of ambient air pollution with twice daily peak expiratory flow rate measurements in children. Am J Epidemiol 1995; 141: 111-122.
Ostro BD. Fine particulate air pollution and mortality in two Southern California counties. Environ Res 1995; 70: 98-104.
Ostro BD. Associations between morbidity and alternative measures of particulate matter. Risk Anal 1990; 10: 421-427.
Ostro BD, Hurley S, Lipsett MJ. Air pollution and daily mortality in the Coachella Valley California: A study of PM10 dominated by coarse particles. Environ Res 1999; 81: 231-238.
Özkaynak H, Thurston GD. Associations between 1980 US mortality rates and alternative measures of airborne particle concentration. Risk Anal 1987; 7: 449-461.
Pekkanen J, Timonen KL, Ruuskanen J, Reponen A, Mirme A. Effects of ultrafine and fine particles in urban air on peak expiratory flow among children with asthmatic symptoms. Environ Res 1997; 74: 24-33.
Peters A, Dockery DW, Heinrich J, Wichmann HE. Medication use modifies the health effects of particulate sulfate air pollution in children with asthma. Environ Health Perspect 1997c; 105 (4): 430-435.
Peters A, Doring A, Wichmann HE, Koenig W. Increased plasma viscosity during an air pollution episode: A link to mortality? Lancet 1997b; 349: 1582-1587.
Peters A, Goldstein IF, Beyer U, et al. Acute health effects of exposure to high levels of air pollution in eastern Europe. Am J Epidemiol 1996; 144: 570-581.
Peters A, Wichmann HE, Tuch T, Heinrich J, Heyder J. Respiratory effects are associated with the number of ultrafine particles. Am J Resp Crit Care Med 1997a; 155: 1376-1383.
Peters JM, Avol E, Gauderman WJ, Linn WS, Navidi W, London SJ, Margolis H, Rappaport E, Vora H, Gong H, Thomas DC. A study of twelve southern California communities with differing levels and types of air pollution. II Effects on pulmonary function. Am J Resp Crit Care Med 1999b; 159: 768-775.
Peters JM, Avol E, Navidi W, Gauderman WJ, Lurmann F, Linn WS, Margolis H, Rappaport E, Gong H, Thomas DC. A study of twelve southern California communities with differing levels and types of air pollution. I Prevalence of respiratory morbidity. Am J Resp Crit Care Med 1999a; 159: 760-767.
Pope CA, Thun MJ, Namboodiri MM, Dockery DW, Evans JS, Speizer FE, Heath CW. Particulate air pollution as a predictor of mortality in a prospective study of US adults. Am J Resp Crit Care Med 1995; 151: 669-674.
Romieu I, Meneses F, Ruiz S, et al. Effects of air pollution on the respiratory health of asthmatic children living in Mexico City. Am J Resp Crit Care Med 1996; 154: 300-307.
Schwartz J, Dockery DW, Neas LM, et al. Acute effects of summer air pollution on respiratory symptom reporting in children. Am J Resp Crit Care Med 1994; 150 (5 Pt 1): 1234-1342.
Schwartz J, Dockery DW, Neas LM. Is daily mortality associated specifically with fine particles? J Air Waste Manag Assoc 1996; 46: 927-939.
Schwartz J, Norris G, Larson T, Sheppard L, Claiborne C, Koenig J. Episodes of high coarse particle concentrations are not associated with increased mortality. Environ Health Perspect 1999; 107: 339-342.
Schwartz J, Neas LM. Fine particles are more strongly associated than coarse particles with acute respiratory health effects in school children. Epidemiol 2000; 11: 6-10.
Sheppard L, Levy D, Norris G, Larson TV, Koenig JQ. Effects of ambient air pollution on non-elderly asthma hospital admissions in Seattle, Washington, 1987-1994. Epidemiol 1999; 10: 23-30.
Thurston GD, Ito K, Hayes CG, Bates DV, Lippmann M. Respiratory hospital admissions and summertime haze air pollution in Toronto, Ontario: consideration of the role of acid aerosols. Environ Res 1994; 65: 271-290.
Thurston GD, Ito K, Kinney PL, Lippmann M. A multi-year study of air pollution and respiratory hospital admissions in three New York State metropolitan areas: results for 1988 and 1989 summers. J Exp Anal Environ Epidemiol 1992; 2: 429-450.
Thurston GD, Lippmann M, Scott MB, Fine JM. Summertime haze air pollution and children with asthma. Am J Resp Crit Care Med 1997; 155: 654-660.
Tiittanen P, Timonen KL, Ruuskanen J, Mirme A, Pekkanen J. Fine particulate air pollution, resuspended road dust and respiratory health among symptomatic children. Eur Resp J 1999; 13: 266-273.
Verhoeff AP, Hoek G, Schwartz J, van Wijnen JH. Air pollution and daily mortality in Amsterdam. Epidemiol 1996; 7: 225-230.
Published 17 May 2001
Expert Panel on Air Quality Standards
Air and Envirnonmental Quality
Environmental Protection Index
Defra Home Page