Health Canada
www.hc-sc.gc.ca
Common menu bar links
Institutional links
Priority Substances List Assessment Report for Respirable Particulate Matter
2.0 Summary of Information Critical to Assessment of "Toxic" Under CEPA 1999
2.1 Identity and physical/chemical properties
2.1.1 Identity
Particulate matter is defined for the purposes of this assessment as particles of less than or equal to 10 µm mass median aerodynamic diameter (MMAD) (PM10) that are emitted directly into the atmosphere or formed secondarily from precursor gases as a result of physical and chemical transformations.
Particles may range from approximately 0.005 µm to 100 µm in diameter, although the suspended portion is generally less than 40 µm. PM10 is generally subdivided into a fine fraction of particles 2.5 µm or less (PM2.5) and a coarse fraction of particles larger than 2.5 µm (PM10-2.5). It is further classified as primary (emitted directly into the atmosphere) or secondary (formed in the atmosphere through chemical and physical transformations). The principal gases involved in secondary particulate formation are sulphur dioxide (SO2), nitrogen oxides (NOx), volatile organic compounds (VOCs) and ammonia. Primary particles are present in both the fine and coarse fractions, whereas secondary particles, such as sulphates and nitrates, are present predominantly in the fine fraction. Both primary and secondary particulate matter can result from either natural or anthropogenic (human-made) sources.
Particulate matter is unique among atmospheric constituents in that it is not defined on the basis of its chemical composition. It may include a broad range of chemical species, including elemental carbon and organic carbon compounds, oxides of silicon, aluminum and iron, trace metals, sulphates, nitrates and ammonia.
The evaluation of the scientific information on particulate matter in this document focusses on particle size because the evidence indicates that particle size is important in influencing the site of deposition in the respiratory tract and the degree of toxicity. Particle size also reflects origin and formation of airborne particles, the larger sizes being often of crustal origin and the smaller sizes originating from combustion processes.
2.1.2 Physical/chemical properties
2.1.2.1 Particle size
Particle size is considered to be one of the most relevant parameters in characterizing the physical behaviour of particulate matter in the atmosphere. Extremely small ("ultrafine") particles less than 0.1 µm in diameter (the nuclei mode) are formed primarily from the condensation of hot vapours during high-temperature combustion processes and from the nucleation of atmospheric species to form new particles. While the greatest concentration of airborne particles is found in the nuclei mode, these particles contribute little to overall particle mass loading due to their tiny size. They are subject to random motion and to coagulation processes in which particles collide to quickly yield larger particles. Consequently, these tiny particles have short atmospheric residence times.
Particles in the size range of 0.1-2.0 µm (the accumulation mode) result from the coagulation of particles in the nuclei mode and from the condensation of vapours onto existing particles, which then grow into this size range. These particles account for most of the particle surface area and much of the particle mass in the atmosphere. The accumulation mode is so-named since atmospheric removal processes are least efficient in this size range. These fine particles can remain in the atmosphere for days to weeks. Dry deposition and precipitation scavenging are the primary processes by which these fine particles are eventually removed from the atmosphere. It is calculated that precipitation scavenging accounts for about 80-90% of the mass of particles in the accumulation mode removed from the atmosphere (Wallace and Hobbs, 1977).
Particles larger than 2.0 µm (the sedimentation or coarse mode) are typically associated with mechanical processes, such as wind erosion, breaking ocean waves and grinding operations. Grinding operations result in the physical breakdown of larger particles into smaller ones, such as windblown soil, sea salt spray and dust from quarrying operations. These particles are efficiently removed by gravitational settling and therefore remain in the atmosphere for shorter periods of a few hours to a few days. They contribute little to particle number concentrations but significantly to total particle mass.
2.1.2.2 Chemical composition
As a consequence of their different sources and mechanisms of production, fine and coarse particles have markedly different chemical composition also (Figure 1). Coarse particles consist primarily of particles derived from the earth's crust and are therefore rich in oxides of iron, calcium, silicon and aluminum and are typically basic in nature. Particles in coastal regions are enriched with sodium chloride from sea salt. Fine particles are composed mainly of sulphate, nitrate, ammonium, inorganic and organic carbon compounds, and heavy metals such as lead and cadmium, all of which are indicators of anthropogenic production processes (Seinfeld, 1986). Fine particles tend to be acidic in nature. Sulphate has repeatedly been shown to be the most abundant single component of fine particles (Keeler et al.,., 1990; Environment Canada, 1994). However, only a few of the numerous organic carbon compounds have been identified, and together these may comprise approximately 50% of the fine particle mass (van Houdt, 1990; Lowenthal et al.,., 1994).
Figure 1 Generalized chemical composition of particulate matter
Comparisons of urban and rural sites in close proximity to one another demonstrate that urban particulate matter concentrations are higher than rural ones, particularly for coarse particulate matter. This is mirrored by an enrichment in urban areas in the concentration of all inorganic elements and ions assayed. There are several elements/ions for which the urban-rural difference is disproportionately greater than the total mass difference, however, indicating that these constituents are particularly enriched in urban areas (calcium, silicon, nitrate, iron, aluminum, magnesium, zinc, titanium, manganese, vanadium, lead, nickel). This pattern is most likely attributed to the greater suspension of road dust and more intensive industrial and combustion activity in urban areas.
Estimates of the amount of fine and coarse particle mass attributable to carbonaceous material (organic and elemental carbon) were made using a mass reconstruction technique and data on inorganic species (Brook et al.,., 1997).
Depending upon site, only about 37-61% of the PM2.5 could be explained given the measured concentrations of several inorganic ions and elements. Thus, carbonaceous material, which was likely to have been predominantly organic in nature, was responsible for about half of the overall fine particle mass. This fraction was higher in Alberta and British Columbia (~65%) than it was on the east coast (40-45%). Sulphate, nitrate and ammonium dominate the identifiable components of the fine particulate matter mass, consistent with the results of many studies. Because of the increased importance of crustal material, a greater portion of the coarse particle mass (~70%) was explained by the inorganic constituents.
2.1.2.3 Other properties
Other physical characteristics that affect particle behaviour include particle shape and density and bulk properties such as chemical composition, vapour pressure, hygroscopicity (water-attracting nature), deliquescence and refractive index. Surface properties such as electrostatic charge, the presence of surface films and surface irregularities may also influence particle behaviour. Small particles are characterized by a larg e surface area relative to their mass, which, when combined with surface irregularities and internal pores, leads to greater reactivity of fine particles compared with coarse particles.
2.2 Sources of particulate matter
Particulate matter is a ubiquitous pollutant, reflecting the fact that it has both natural and anthropogenic sources.
Natural sources of primary particulate matter include windblown soil and mineral particles, volcanic dust, sea salt spray, biological material such as pollen, spores and bacteria, and debris from forest fires. By and large, these natural sources produce coarse particles, although high-temperature combustion sources such as wildfires will generate fine particulate matter.
Secondary particulate matter can be formed through reactions involving natural sources of the precursor gases. For example, VOCs are released from trees, and nitrogen oxides are released from soils.
Anthropogenic sources also produce both primary and secondary particulate matter and both coarse and fine particles. Windblown agricultural soil and dust from roads, construction sites and quarrying operations all contribute primarily to the coarse fraction. Smaller particles of more complex chemical composition are generated as a result of many industrial processes and through fossil fuel combustion (electrical power plants, gasoline and diesel vehicles, industrial boilers, residential heating, etc.), both directly and via the release of precursor gases (VOCs, sulphur dioxide and nitrogen oxides) (Table 1).
Current estimates of the magnitude of primary particulate matter emissions have been compiled as part of the Environment Canada Criteria Air Contaminants 1995 emission inventory. This inventory includes emission estimates for primary particulate matter designated as PART (total particulate matter), PM10 (particles sized from 0 to 10 µm) and PM2.5 (particles sized from 0 to 2.5 µm), as well as emissions of the principal precursor gases that contribute to the formation of particulate matter.
Environment Canada compiles emission inventories based upon three source types: point, area and mobile. Provinces and territories provide the point source information. Emissions from all three source types are estimated by applying some emission factor to a base quantity related to activity or production. Minimal information is available for directly measured emissions. For point sources, the U.S. Environmental Protection Agency PMCALC model was used to estimate emissions based upon Standard Classification Codes for each source type. Area and mobile source emissions are modelled using source-specific emission factors (principally from the U.S. Environmental Protection Agency AP42 Compilation of Air Pollutant Emission Factors) and information regarding activity level from Environment Canada, Statistics Canada, Natural Resources Canada and provincial/territorial government publications.
Table 2 provides the 1995 national Criteria Air Contaminants emission inventory by category. The provincial source sector breakdown is provided in Table 3. The sources included in each category in Table 3 are as listed in Table 2.
The available data clearly indicate that source contributions of primary particulate matter vary by province/territory and by region. In Yukon, the Northwest Territories, British Columbia and Saskatchewan, forest fires and prescribed burning are the largest estimated sources of particulate matter. Industrial sources are a major contributor to particulate matter emissions including precursor gases in all provinces except Prince Edward Island. In Prince Edward Island, non-industrial fuel combustion (primarily residential wood combustion) is a major source. The transportation sector is also a large contributor in Ontario and British Columbia, while non-industrial fuel combustion is significant in Alberta, Manitoba, Ontario, Quebec, New Brunswick and Nova Scotia. A further consideration is the seasonality of particulate emissions. Residential wood combustion is most prevalent during the winter months, while forest fires are generally limited to the summer months. Emissions from the industrial and transportation sectors occur year-round, although they are subject to many fluctuations.
Characterization of the sources and atmospheric processes that contribute to ambient particulate matter levels is a complex process due to the contribution of secondary formation from diverse biogenic and anthropogenic precursor sources. Primary particles, those emitted directly from sources, undergo few physical or chemical changes between source and receptor. Consequently, atmospheric concentrations are approximately proportional to their emission concentrations. Secondary particles undergo physical and chemical transformations, masking the chemical composition of the original source. Most sulphate and nitrate particles are of secondary origin, resulting from sulphur dioxide, nitrogen dioxide (NO2) and ammonia emissions. Some organic carbon is also of secondary origin, resulting from volatile organic gas emissions. There are limitations in the current ability to comprehensively and accurately assess particulate matter emissions due to a lack of any Canadian source apportionment information - i.e., information with which to estimate the magnitude of secondary particulate matter formed from precursor gases.
- Notes:
- Table source is Deslauriers (2000).
- Numbers may not add to totals, due to rounding.
- The 1995 emission inventory was compiled with the latest technical and statistical information available; only the sulphur oxide emissions can be compared with previous emission inventories.
- The estimates for PART, PM10 and PM2.5 refer to primary particulate matter emissions only.
- PART = total particulate matter.
- Work is in progress to improve the road dust emission estimates, using the road dust study in British Columbia and additional measurements and statistics for selected provinces. These estimates provide an update to the mid-1998 emission figures released in Ontario and other provinces.
Table 3 Provincial/territorial distribution of particulate matter and precursor emissions (tonnes), 1995
Source apportionment, or source attribution, allows for the identification (qualitatively and quantitatively) of contributing sources to support the development of air quality management strategies. Source apportionment techniques use mathematical models that model the precursor gas emissions and complex atmospheric chemistry involved in the formation of secondary particulate matter. There are a variety of techniques of various degrees of complexity, from which several common features emerge:
- In the coarse fraction, which in some studies has been defined as PM10 or the range from PM2.5 to PM15, there is a predominance of crustal material (or road dust, which has components of crustal material). The exact contribution, of course, depends on the location, the season, etc. The contribution of crustal sources greatly diminishes in PM2.5 samples and is generally less than 5 -15%.
- The greatest contributions to PM2.5 come from organic compounds and secondary sulphates and nitrates. In urban areas, the sources of these compounds and their precursor gases (sulphur and nitrogen oxides and VOCs) are typically combustion processes - motor vehicles, industrial processes and vegetative burning. Even remote areas may be impacted by these sources.
- The early estimates for Canadian cities (EAG, 1984) are consistent with more recent analyses. The estimated motor vehicle source contribution ranges from 9 to 39% of PM2.5 from both REVEAL (Lowenthal et al.,., 1996) in the Lower Fraser Valley of British Columbia and EAG (1984) in a national overview. Chemical mass balance motor vehicle source apportionments within the Vancouver area were highly consistent, at about 40% of PM2.5. REVEAL (Pryor and Steyn, 1994) indicates a motor vehicle contribution of 15-20% to PM2.5. In Toronto, the motor vehicle contribution is approximately 50% (Lowenthal, 1997).
Long-range transport of particulate matter is also an important source of particulate matter in some regions of Canada. Back-trajectory analysis based on ambient monitoring provides insight into the portion of particulate matter contributing to local concentrations via long-range transport on a continental scale. Ambient concentrations of particulate matter for days on which back-trajectories identified air masses as originating from the north (Canada) or from the south (United States) are summarized in Table 4 for Kejimkujik, Nova Scotia, and Sutton, Quebec. Considering the average values, air masses originating from the north contain approximately two-thirds the PM10, half the PM2.5 and one-third the sulphate of air masses originating from the south. These results indicate the substantial contribution (50% or more for fine particles) from long-range transport events from industrial regions of the United States to ambient levels of particulate matter and are consistent with the fact that it is the smaller particles that are most likely to be transported long distances. Further analysis is required to determine how often long-range transport from the United States impacts Canada, and thus the total contribution of long-range transport to particulate matter concentrations in Canada.
2.3 Exposure characterization
2.3.1 Monitoring technology
Measurements of particulate matter for the purpose of current compliance monitoring are generally expressed in terms of mass. Mass measurements may be made directly or indirectly. Direct (or manual) measurements of particulate matter concentrations in the ambient air are made by collecting particles on a pre-weighed filter over a specified period of time, weighing the soiled filter and then dividing the gain in mass by the volume of air sampled. Samples are typically collected for a 24-hour period. Sampling inlets that remove particles larger than 10 µm may be selected so that particles in the PM10 size range are selectively retained on the filter. Particles may also be fractionated into a fine fraction (≤2.5 µm) and a coarse fraction (>2.5-10 µm), which are collected on separate filters for measurement and analysis using dichotomous samplers. Examples of manual samplers include high-volume (hi-vol) and dichotomous (dichot) samplers.
Table 4 Average, median and standard deviation of the 24-hour PM10, PM2.5 and sulphate measurements sorted by trajectory group
Indirect measurements are made using parameters other than mass, generally optical properties, which can then be converted to units of mass concentration based on known relationships between the two parameters. Examples of indirect methods include British Smoke, Tapered Element Oscillation Microbalance (TEOM) and beta-attenuation monitors; the latter two methods can provide near real-time measurements of particle concentrations.
Intercomparison studies indicate that size selective inlet (SSI) high-volume samplers and dichotomous samplers yield comparable results (Dann, 1994), although there are few data for PM2.5 samplers. With respect to indirect samplers, some data from co-located TEOMs and PM10 high-volume samplers have shown good agreement, with correlation coefficients of 0.977 for 24-hour averages (Meyer, 1993). However, there are indications that TEOM data are consistently lower (on the order of 25%) than data from manual samplers (Moore and Barthelmie, 1995).
2.3.2 Ambient levels of particulate matter
2.3.2.1 Background concentrations
"Background" particulate matter is generally defined as the distribution of particulate matter concentrations that would be observed in the absence of anthropogenic emissions of particulate matter and of VOC, nitrogen oxide and sulphur oxide (SOx) precursors. The actual magnitude of background concentrations of particulate matter for a given location is difficult to determine because of the influence of long-range transport of anthropogenic particles and precursor gases. The range of expected background concentrations on an annual or long-term basis is from 4 to 11 µg/m3 for PM10 and from 1 to 5 µg/m3 for PM2.5 for remote sites in North America (Trijonis, 1982; NAPAP, 1991; Malm et al.,., 1994). The range of expected background concentrations on a short-term basis is much broader given the episodic nature of such natural events as wildfires and prairie dust storms, which can result in short-term particulate matter levels comparable to those in polluted urban atmospheres.
2.3.2.2 Rural and urban concentrations
A national PM10 and PM2.5 monitoring program has been in operation since 1984 under the auspices of the National Air Pollution Surveillance (NAPS) network. This is primarily an urban network with few rural sites. Measurements are obtained from SSI high-volume samplers (PM10) and dichotomous samplers (PM10 and PM2.5). Since 1994, hourly PM10 data from TEOM instruments at NAPS sites have also been reported to the database. In addition to the national network, British Columbia, Ontario and Quebec operate particulate matter monitors. Particulate matter data are typically collected over a 24-hour sampling period on a one-day-in-six sampling regime. By operating on this schedule, given a long enough sampling period, each day of the week is equally well sampled, and hence all conditions during the week are represented. It should be noted, however, that this sampling frequency does not permit the extremes of the concentration distribution to be accurately quantified. The one-in-six-day schedule has the likelihood of underestimating the frequency and magnitude of high-concentration PM10 events (by 20-30%), because the nearest days to the event day, and/or the event day itself, may be excluded by the sampling schedule.
2.3.2.2.1 Twenty-four-hour average PM10 levels
Particulate matter data typically exhibit a skewed distribution dominated by a large number of low values. Particulate matter concentrations also typically exhibit variation on a number of temporal scales: diurnal, hebdomadal (day of week), seasonal and annual. The causes of these variations are multifaceted and are related both to emission variability and to variations in geophysical variables, such as mixed layer depth, wind speed and humidity levels.
Mean PM10 concentrations across Canada range from 11 to 42 µg/m3, with most sites in the range of 20-30 µg/m3 (Table 5). These levels are substantially above estimated background levels, indicating that anthropogenic activities make a significant contribution to ambient PM10 loadings. The highest PM10 concentrations recorded by the NAPS monitoring network were observed in Quebec (at a site in Montréal), Ontario (at sites in Windsor, Hamilton and Walpole Island) and Alberta (at a site in Calgary). However, even within cities, there may be sites that experience comparatively low ambient PM10 levels, as is the case in Montréal and Calgary. The three rural sites of Kejimkujik (Nova Scotia; 1992-1995), Sutton (Quebec; May-September 1993) and Egbert (Ontario; 1992-1995) recorded mean 24-hour PM10 concentrations of 11, 11 and 17 µg/m3, respectively.
The season of maximum PM10 concentrations is regionally variable, reflecting variations in dominant sources of PM10 (especially secondary aerosols) and synoptic meteorology. The sites that exhibit the highest degree of seasonality are in Windsor (a summertime maximum) and Victoria (a wintertime maximum). Many of the sites in British Columbia seem to exhibit a late winter/early spring maximum of both mean and median PM10 concentrations and the upper quartile of the distribution, indicating that both average and extreme PM10 concentrations are typically higher during the months of January, February and March. Sites in Ontario seem to exhibit summertime maximum PM10 concentrations, which may reflect the greater abundance of secondary aerosols in the Windsor-Québec corridor, where precursor concentrations are known to be high.
A hebdomadal cycle of PM10 concentrations is evident at most urban sites. Typically, weekend concentrations of PM10 are lower than those observed during the work week. This difference is magnified for roadway sites, where up to a 50% increase in PM10 was noted midweek relative to Sunday (all sites). This suggests a substantial contribution to PM10 concentrations from transportation sources.
There has been an apparent decrease in yearly variations in PM10 concentrations during the 1984-1995 sampling period at most sites with a complete data record. The largest percent decreases occurred at the Montréal-Duncan/Decarie, Edmonton and Vancouver sites. A trend analysis of annual PM10 data for 1984 through 1993 showed a statistically significant (p < 0.001) decreasing trend in PM10 concentrations on a national basis, averaging 2% per year (Dann, 1994).
Table 5 Frequency distributions of 24-hour average PM10 concentrations (µg/m3) from dichotomous sampler sites (1984-1995) and SSI sites (1988-1994) - national network
2.3.2.2.2 Twenty-four-hour average PM2.5 levels
Mean concentrations at the NAPS urban sites ranged from 6.9 to 20.2 µg/m3 (Table 6). PM2.5 concentrations are more spatially homogeneous than PM10 concentrations, but there are still significant site-to-site differences, even within the same urban area. The highest PM2.5 concentrations (in terms of means and 90th percentiles) were measured at sites in Montréal, Toronto, Hamilton, Windsor, Walpole Island and Vancouver. These were almost the same sites that recorded the highest PM10 concentrations. The three rural sites of Kejimkujik (1992-1995), Sutton (May-September 1993) and Egbert (1992-1995) recorded mean PM2.5 24-hour concentrations of 7.0, 7.7 and 10.5 µg/m3, respectively.
The seasonal variability of PM2.5 is more pronounced than that of PM10; however, the seasonal patterns vary for different regions. The Montréal, Ottawa, Edmonton, Calgary and Vancouver/Victoria sites record higher PM2.5 concentrations in the winter months, particularly during January and February. Other Ontario sites record the highest daily concentrations in the summer months, with a peak median in August, reflecting long-range transport from the U.S. Midwest. For Maritime sites, there is variable seasonal variation in PM2.5 concentrations, with Saint John and Kejimkujik having a summer maximum (as a consequence of long-range transport from the east coast of the United States) and Halifax a winter maximum.
Minimum PM2.5 concentrations occur on Sunday and maximum concentrations during the middle of the week at most urban sites. Again, this difference is magnified for roadway sites, where up to a 60% increase in PM2.5 midweek relative to Sunday was noted for all sites. This indicates that there are large day-of-week differences in anthropogenic emissions and significant contributions from motor vehicles.
A trend analysis of PM2.5 data for the period 1984-1993 showed a statistically significant (p < 0.001) decreasing trend in PM2.5 on a national basis, averaging 3.3% per year (Dann, 1994). For the Ontario sites, there was no significant change in PM2.5 between 1987 and 1993.
2.3.2.2.3 One-hour average PM10 levels
In 1994, 10 sites (all but two in the Lower Fraser Valley) reported hourly PM10 concentrations to the NAPS network using TEOM instruments. A maximum one-hour PM10 concentration of 255 µg/m3 was measured at the Abbotsford site (in the Lower Fraser Valley), and a maximum one-hour concentration of 204 µg/m3 was recorded at the Edmonton site. Analysis of diurnal variations in PM10 has shown that a substantial increase in PM10 levels occurs during the morning rush hour, with a secondary peak during the late evening. Minimum values occur during the mid-afternoon and in the early hours of the morning (12:00-6:00 a.m.).
2.3.2.3 Relationships among TSP, PM10 and PM2.5 and inorganic constituents of particulate matter
Fourteen urban sites in the NAPS dichotomous sampler network operating from 1986 to 1994 made simultaneous measurements of total suspended particulates (TSP), PM10, PM2.5 and sulphate. This data set is valuable in that it allows comparison of the composition of these different particulate matter fractions at the 14 sites (Table 7). On average across the 14 sites, PM10 accounted for approximately 50% of TSP, while PM2.5 accounted for approximately 25% of TSP. Both fine and coarse particles accounted for approximately equal portions (about 50%) of the PM10. Most of the sulphate was present in fine particles, where it comprised on average approximately 17% of the fin e particulate matter. However, there was considerable variation within and among sites for these ratios. The relationships between TSP, PM10 and PM2.5 are dependent on concentration, with ratios of PM10 and PM2.5 t o TSP decreasing with increasing TSP concentration (i.e., more of the TSP mass is composed of very coarse particulate matter) (Brook et al.,., 1997).
Table 6 Frequency distributions of 24-hour average PM2.5 concentrations (µg/m3) from dichotomous sampler sites (1984-1995) - national network
Table 7 Combined summary statistics for 14 urban NAPS sites in operation between 1986 and 19941 (from Brook et al., 1997)
Other data from the NAPS network corroborate both the variability in PM2.5/PM10 ratios and the overall finding that on average across Canada, approximately 50% of PM10 is made up of fine particles (Figure 2). Based on data collected from 1984 to 1993 from 19 sites (16 locales), the median PM2.5/PM10 ratios for most sites fall within a fairly narrow range of 0.4-0.6; that is, at least half of the time, 40-60% of PM10 at a site is composed of fine particles. Although there is clearly temporal variability in PM2.5/PM10 ratios at a site, about 50% of the time the ratios do not vary by much more than ±10%, as indicated by the interquartile ranges (25th-75th percentiles) (Brook et al.,., 1997).
Figure 2 Distributions of the ratio of PM2.5 to PM10 mass at the NAPS dichotomous sampler sites. The box plots indicate the median, 5th and 95th and 25th and 75th percentiles (from Brook et al.,., 1997).
There are relatively strong correlations (r 2) between PM10 and PM2.5 at each of the 19 sites, which is consistent with the belief that temporal variations in fine particles have a significant influence on the observed variability in PM10. At a majority of the sites, the daily variability in fine particle mass had a stronger influence on the variations in PM10 than did the daily variability in coarse particle mass. This was most evident at the rural locations and at sites not heavily impacted by urbanization (i.e., traffic and construction). The exceptions to this pattern were the Prairie sites, where coarse mass dominated PM10, and a site in Montréal that is heavily impacted by traffic (Brook et al.,., 1997).
There are two key trends based on comparisons of TSP, PM10, PM2.5 and sulphate mass distributions at sites across Canada (Figure 3). First of all, sites in the three Prairie cities of Winnipeg, Calgary and Edmonton have large and variable TSP concentrations, but their PM2.5 and sulphate concentrations are small relative to the other sites and exhibit less variability. Much of the airborne particulate matter observed in these areas is considered to be mechanically derived and likely consists of local crustal material. Secondly, there is an obvious decrease in sulphate levels from the sites located east of the upper Great Lakes to those located west of the lakes. This pattern has been repeatedly observed and is a direct reflection of the magnitude and spatial density of sulphur dioxide emissions within and upwind of these two area (Brook et al., 1997).
2.3.3 Indoor and personal air
The vast majority of particulate monitoring data are available for ambient air. However, North Americans spend, on average, less than 10% of their time outdoors (U.S. EPA, 1995; Leech et al., 1996). In addition, individuals can spend time in a number of microenvironments that differ in their particulate matter concentrations during the course of a day. Consequently, indoor air and personal air levels of airborne particles have been examined in a number of studies.
Figure 3 Comparison of the distributions of TSP, PM10, PM2.5 and sulphate at 11 urban sites (1984-1993). The box plots indicate the median, 5th and 95th and 25th and 75th percentiles (from Brook et al., 1997).
2.3.3.1 Indoor air
Levels of particulate matter in indoor air are a function of indoor sources, outdoor particle levels, the fraction of ambient air penetrating indoors, filtration, air exchange rates, decay rates and resuspension rates (U.S. EPA, 1982, 1996; Clayton et al., 1993; Wallace et al., 1993; Thatcher and Layton, 1995). Fine particles readily penetrate buildings; penetration factors of 0.6-1 have been reported for PM2.5, with the coarser fraction (PM10-2.5) probably penetrating less effectively (Dockery and Spengler, 1981; Yocom, 1982; Lioy et al., 1990; Colome et al., 1992; Koutrakis et al., 1992; Özkaynak et al., 1993; Thatcher and Layton, 1995; U.S. EPA, 1996). Once indoors, ambient particles settle out quickly by gravity or electrostatic forces. Average decay rates, due to diffusion or sedimentation, were calculated as part of the Particle Total Exposure Assessment Methodology (pTEAM) study (Özkaynak et al., 1993) for sulphate (0.16 per hour), PM2.5 (0.39 ± 0.16 per hour), PM10 (0.65 ± 0.28 per hour) and the coarse fraction, PM10-2.5 (1.01 ± 0.43 per hour). Particles of size range 0.1-1 µm have negligible setting velocities, while particles >10 µm normally settle out of the air. Once deposited, particles greater than 5 µm are easily resuspended during indoor activities of occupants, while smaller particles (<1 µm) are "not" resuspended, and particles 1-5 µm in size can be resuspended with vigorous activity (Thatcher and Layton, 1995).
In three large-scale studies conducted in the United States - the Harvard Air Pollution Health Effects Study (also called the Six Cities Study), the New York State Energy Research and Development Authority (ERDA) Study, and the pTEAM study - the range of mean values for PM2.5 or Pm3.5 in various cities across the country was 20-47 µg/m3 in indoor air, compared with 13-50 µg/m3 in outdoor air (Dockery and Spengler, 1981a; Sexton et al., 1984; Spengler et al., 1985; Sheldon et al., 1989; Pellizzari et al., 1992). Concentrations in indoor air generally exceeded those outdoors at sites with low outdoor levels and were slightly less at sites with high outdoor concentrations (Wallace, 1996). Levels in indoor air were, however, within a factor of two of those outdoors in all of the available studies.
Source contributions to levels of respirable particles in indoor air were estimated from the pTEAM results using elemental analyses and a non-linear method of solving a mass balance model. Averaged over 244 homes containing no known indoor sources, outdoor sources accounted for 75% of indoor PM2.5 and 65% of PM10. In homes where there was tobacco smoking or cooking (the two main identified indoor sources of particles), each accounted for approximately 20-30% of the indoor particle concentrations on average; however, outdoor air still remained the largest source of indoor PM10 or PM2.5, accounting for approximately 60% of indoor air levels (Özkaynak et al., 1993, 1995a). Source apportionment analyses from the Harvard Six Cities and ERDA studies confirmed the importance of outdoor a ir and (in homes with smokers) tobacco smoking as sources of indoor particles (Santanam et al., 1990; Koutrakis et al., 1992). In Canada, where building construction emphasizes energy efficiency and, therefore, low air exchange rates, the fractions of fine and coarse particles of ambient origin that are found indoors under equilibrium should tend towards 50% or less, particularly in winter.
2.3.3.2 Personal air
Personal exposures to airborne particles have been measured in a small number of studies. In non-smoking households, mean levels of PM2.5, Pm3.5 or PM10 in personal air were roughly 1.5-2.5 times higher than particle concentrations in indoor air (Spengler et al.,., 1985; Lioy et al.,., 1990; Özkaynak et al.,., 1993; Neas et al.,., 1994; Thatcher and Layton, 1994; Wallace, 1996). The source of the difference between personal exposures and indoor concentrations, the "personal cloud," has not yet been determined. However, most of the excess personal exposure is likely due to generation or re-entrainment of particles during personal activities. In the pTEAM study, the increment in personal exposure levels over the corresponding concentrations of particles in indoor air was restricted to daytime monitoring, whereas levels were similar at night, when subjects were less active. In addition, personal exposure levels were significantly greater in subjects who were engaged in activities such as cooking, cleaning the house and smoking during the monitoring period (Pellizzari et al.,., 1992). The "personal cloud" may be made up mostly of coarse particles (PM10-2.5) (Bahadori et al.,., 1995), which is consistent with the evidence that coarse particles are more easily resuspended than fine particles (Thatcher and Layton, 1995).
Correlations between ambient particulate matter data obtained from fixed ambient monitors (FAMs) and personal exposure data obtained from personal exposure monitors have been examined in a number of studies. Most of these reveal poor correlations for data collected at one point in time, particularly during the day, and indicate, not surprisingly, that personal exposures are usually greater than indoor or outdoor ambient concentrations. When personal exposure is longitudinally regressed against levels at the nearest outdoor site, the correlations improve (Lioy et al.,., 1990; Wallace, 1996). The improvement in correlation coefficients suggests that, for individuals who are not exposed to important microenvironmental sources of particles (e.g., smoking) and whose day-to-day activities are fairly repetitive, ambient levels of particles may more directly reflect their exposure to particles. The correlation also increases when the mean of the personal exposures from various studies is related to the FAM (Mage and Buckley, 1995). Thus, ambient fine particles measured at the FAM can serve as an indicator of community (population) exposure.
Personal and population exposure models have been developed that combine ambient measurements of pollutants with information on age-specific time-activity relationships and estimates of microenvironmental pollutant concentrations. A probabilistic PM10 exposure model was applied to Canadian ambient PM10 measurements, demographics and smoking prevalences by region to produce estimated distributions of 24-hour average personal, indoor, outdoor and in-transit PM10 concentrations (Özkaynak et al.,., 1995b). The model reproduced the empirical findings that personal exposures to PM10 are substantially greater than ambient concentrations at FAMs (the predicted mean personal air concentration was 39 µg/m3, compared with an average of 28 µg/m3 measured at the ambient sites used as input to the model), and that personal exposures were quite variable (e.g., the 95th percentile, at 93 µg/m3, was nearly 2.5 times greater than the mean).
2.4 Effects characterization
2.4.1 Humans
2.4.1.1 Controlled human exposure studies
Almost all of the human clinical studies have been based on observations of pulmonary function changes and reports of subjective symptoms. All of the exposures were conducted for very short periods of time (i.e., most of them between 40 and 120 minutes). Controlled human exposures to acidic and inert particles have not caused significant alterations in respiratory function and symptoms in healthy individuals, even at levels higher than those in the environment (450-1000 µg/m3) (Utell et al.,., 1983a, 1984). The clinical studies have identified asthmatics as a susceptible population for acidic aerosols (Utell et al.,., 1983b), but not persons with chronic obstructive pulmonary disease (COPD) (Morrow et al.,., 1994) or the elderly (Koenig et al.,., 1992, 1993; Morrow et al.,., 1994). In controlled studies, asthmatics, especially children and adolescents, have experienced adverse effects on airway function at concentrations encountered on occasion in ambient air (exposure to sulphuric acid at ~35 µg/m3 for 40 minutes) (Koenig et al.,., 1989, 1992; Hanley et al.,., 1992).
Few data are available on particle-induced airway inflammatory responses in humans. In one study (Frampton et al.,., 1992), brief exposures to sulphuric acid aerosols at 1000 µg/m3 had only minor effects on the airway defence system in healthy subjects - i.e., increased antibody-mediated cytotoxicity of alveolar macrophages, a trend (not statistically significant) to a decreased percentage of T lymphocytes in bronchoalveolar lavage fluid, and no effects on the number of polymorphonuclear leukocytes in the bronchoalveolar lavage fluid or on the release of superoxide anion or inactivation of influenza virus in vitro when compared with sodium chloride.
No data on changes to the cardiovascular system in controlled human studies were identified.
There has been very little work comparing the effects of different particle sizes on human health. In a panel study, Peters et al.,. (1997) reported that symptoms (cough) and decrements of peak expiratory flow in asthmatic subjects (n = 27) were more strongly associated with the five-day mean of the number or mass of ultrafine particles (MMAD 0.01-0.1 µm) than with other discrete fractions of the fine particles (MMAD 0.1-0.5, 0.5-2.5 or 0.1-2.5 µm).
The particles to which volunteers were exposed in the human clinical studies do not adequately reflect the complexity of ambient particles. Based on the extremely limited clinical database available, acidic aerosols produce the most significant bronchoconstriction, while the toxicity of sulphate is related to acidity per se (Utell et al.,., 1983b, 1989). Nitrates did not exert effects on lung function at concentrations below 1000 µg/m3 in clinical studies (U.S. EPA, 1989; Aris et al.,., 1993), while inert particles appear to have no effect on lung function in either healthy or asthmatic volunteers (Anderson et al.,., 1992). In one study (Sandstrom and Rudell, 1991), very fine particle diesel exhaust (3 x 106 particles/cm3) affected neutrophil production and macrophage clearance of microorganisms from the lung, but exposure to formaldehyde and other combustion gases (nitrogen dioxide, nitric oxide and carbon monoxide [CO]) was also elevated, which might confound any effects from particulate matter.
Few clinical studies have been conducted to investigate the effects of air pollutant mixtures on humans. Pre-exposure of normal and asthmatic adults to sulphuric acid aerosol potentiated reductions in lung function and increases in airway resistance induced by exposure to ozone (O3) in the asthmatic group only (Frampton et al.,., 1995). Respiratory symptoms were increased in healthy and asthmatic children by a mixture of sulphuric acid, ozone and sulphur dioxide (Linn et al.,., 1997). Concurrent exposure to ozone and sulphuric acid at concentrations typically observed in air pollution episodes produced no changes in lung function or symptoms greater than those associated with ozone alone in normal or asthmatic adults and children (Linn et al.,., 1994, 1995).
It should be noted that decrements in pulmonary function measured in most clinical studies may not be a sensitive indicator for particle-induced lung injury. Relatively high levels of acidic aerosols produce only small decrements in lung function, even in susceptible subpopulations. Moreover, if reductions in pulmonary function serve to protect the lungs from receiving further insults in the deep airways, failure of ce rtain subjects (such as COPD patients) to respond to particles in this manner might render them more vulnerable to pulmonary injury.
2.4.1.2 Epidemiological studies
That high levels of ambient particulate matter from combustion sources could have severe adverse effects on health was noted in the air pollution episodes of the 1940s to 1960s. Indeed, one such episo de in London, England, in 1952 was responsible for several thousand premature deaths within a week. However, until the publication of new studies beginning in the early 1990s, there were no data to suggest that relatively low concentrations of particulate matter, as currently experienced in urban areas of North America and Western Europe, had effects on human health.
Many of these recent studies on air pollution, including all the mortality and hospitalization studies, have been based on the time-series analysis of associations between daily variations in ambient concentrations and daily variations in adverse health outcomes, with data obtained from large administrative databases. In this longitudinal type of study, confounding due to population differences is much less likely than in a cross-sectional analysis, since the population remains the same over a short time (typically one or a few days) and acts as its own "control." However, the potential for confounding by seasonal variations and weather remains a problem in the time-series analysis, and much recent effort has been directed to addressing this, with considerable success. A major advantage of the time-series study is that it usually provides many more units of observation (typically 1000 days and 10 000 adverse effects) than the cross-sectional study (2-150 communities), and thus its power to detect effects of low magnitude is usually greater than that of even the most sensitive cross-sectional study. While time-series studies do not provide information on exposures at the individual level (i.e., they are considered to be "ecological" with respect to their exposure component), they can provide important evidence for the assessment of relationships between levels of air pollution and health at a community level.
Several cohort studies, in which the effects of longer-term exposures to particulate matter were examined and where potentially confounding risk factors were taken into account in the analysis, have also been completed. However, exposure remained on a community basis in the available studies, which limits the inferences to be drawn from these studies to groups of individuals with a given set of measured risk factors.
Studies on adverse effects other than premature mortality and hospitalizations, such as effects on lung function, have been completed using a variety of methods.
2.4.1.2.1 Acute effects: mortality
Nineteen time-series studies in which the relationship of daily or short-term variations in particulate matter with mortality was investigated are summarized in Table 8. The studies were conducted in cities across North and South America and Europe, and almost all of them demonstrated associations between particulate air pollution and acute mortality. The associations of mortality with particulate matter could not be explained by the influence of weather, season, yearly trends, day-to-day variations or variations due to holidays, epidemics or other non-pollutant factors, since all the studies investigated these potential biases and adjusted for them in various ways in the analyses.
For 21 out of 23 cities, the observed increases in relative risk (RR) in relation to elevated PM10 concentrations were statistically significant or close to significance, as measured by the 95% confidence interval (CI). The exceptions were Salt Lake County, Utah (Styer et al.,., 1995) and Kingston-Harriman, Tennessee (Dockery et al.,., 1992). Negative results in the latter were explicable on the basis of a combination of poor exposure assessment and inaccurate methodology (Dockery et al.,., 1992). The negative results for Salt Lake County (Styer et al.,., 1995) are not easily explained except as a lack of statistical power to detect an effect due to small population and possibly also due to overcompensation for weather and seasonal factors in the method used to analyse the results.
For PM10 at ambient concentrations averaging 18-115 µg/m3 (from 23 reported cities), a 100 µg/m3 increase was associated with a mean (unweighted) RR of mortality from all causes except accidents of 1.082 ± 0.056 [mean ± standard deviation (SD)] and a median of 1.08 (Table 8). The results indicate that each 10 µg/m3 of daily increase in PM10 is associated with an unweighted and weighted mean increase in daily mortality of 0.8% and 0.5%, respectively. The weighted RRs were calculated based on the standard deviation of each study; RRs with smaller standard deviations were weighted more heavily than those with large standard deviations (see Schwartz, 1994e, for detailed method of weighting).
For Black Smoke (a finer particle, approximately PM5) with average concentrations of 12-84 µg/m3 in six locations, the observed increase in RR with increasing Black Smoke concentration was statistically significant in all but one of the six cities (Paris, France), where it was marginal (Dab et al.,., 1996). Across all studies, the unweighted mean RR for mortality was 1.096 ± 0.050 (mean ± SD), with a median at 1.08, for a 100 µg/m3 increase. This indicates that a 10 µg/m3 increase in Black Smoke is associated with a daily increase in mortality of approximately 1% on average, with the median being 0.8%.
For the fine fraction of particles (PM2.5), at concentrations averaging 11.2-21 µg/m3, RRs for mortality and PM2.5 were increased in all nine cities, although the increase was marginally significant in three cities (St. Louis, Missouri, in Dockery et al.,., 1992; Steubenville, Ohio, and Portage, Wisconsin, in Schwartz et al.,., 1996) and was not significant in two cities (Kingston-Harriman, Tennessee, in Dockery et al.,., 1992; Topeka, Kansas, in Schwartz et al.,., 1996). The weak or negative results from St. Louis and Kingston-Harriman were considered likely due to the poor placement of monitors and generally less developed methodology in this relatively early study (Dockery et al.,., 1992). In Portage and Topeka, PM2.5 concentrations were very low and the population was small, which resulted in relatively few "events" and lack of statistical power. The marginal increase in RR in Steubenville was considered to be due to a high correlation with coarse particulate (see below) (Schwartz et al.,., 1996). Across all studies, the unweighted RRs were also elevated, with a mean ± SD of 1.15 ± 0.05 (n = 9) and a median of 1.14, for a 100 µg/m3 increase. The results indicate an average increase in daily mortality of 1.5% (unweighted) for each 10 µg/m3 daily increase in PM2.5. No quantitative results have been reported for multivariate analyses of weighted data, except the combined (weighted) RR of 1.15 (95% CI 1.11-1.19) for the Six Cities Study (Schwartz et al.,., 1996). The increase in PM2.5-related risk of mortality is thus about twice that for PM10.
Table 8 Summary of relative risks for particulate matter from time-series studies: Total mortality
The relationship between sulphate concentrations and mortality was investigated in only two studies (Dockery et al.,., 1992; Schwartz et al.,., 1996). In only one study (Schwartz et al.,., 1996) was there a significant association, with a mean RR of 1.22 for six cities combined (95% CI 1.13-1.33) for a 100 µg/m3 increase, which can be interpreted as an average 2.2% increase in mortality per 10 µg/m3 increase in sulphate. There was no association of mortality with sulphate concentrations in two other cities (Dockery et al.,., 1992), although, as noted above, the negative results in these cities (St. Louis, Missouri, and Kingston-Harriman, Tennessee) may have been due to poor placement of monitors and limitations in the methodology in this early study.
The coarse fraction of PM10 above 2.5 µm in diameter was not assoc iated with mortality in the overall analysis of six U.S. cities (0.4% increase in mortality; 95% CI 0.01-1.0%) or in five individual cities (Schwartz et al.,., 1996). In the sixth city (Steubenville, Ohio), the positive association was explained by the high correlation coefficient between coarse particles and PM2.5 (r = 0.7), which resulted in attribution of the increased risk of mortality to the coarse fraction instead of to PM2.5.
There is no clear evidence of a level of particulate matter that is without effect on mortality; instead, analyses suggest some increase at even the lowest ambient levels studied. The RR of mortality was observed to increase monotonically with increasing concentration of PM10 in the concentration range below 80-100 µg/m3, in both non-parametric analyses (Schwartz, 1993; Ostro et al.,., 1996; Pope and Kalkstein, 1996) and quartile or quintile analyses (Pope et al.,., 1992; Schwartz, 1993; Saldiva et al.,., 1995). The quintile results for St. Louis, Missouri, and Kingston-Harriman, Tennessee (Dockery et al.,., 1992) were also roughly monotonic when analysed together, but not separately, due to relatively poor information on exposure and (in Tennessee) to a small study population. A curvilinear response was observed in Santiago, Chile, with high slope at low PM10 concentrations and levelling off at concentrations above 100 µg/m3 (Ostro et al.,., 1996). There is also an apparent curvilinear response in the Birmingham, Alabama, data at PM10 concentrations approximately above 50 µg/m3 (Schwartz, 1993). The quintile analysis of Utah Valley data by Lyon et al.,. (1995) indicated no effects at the mean concentration of 47 µg/m3; however, discrepancies were evident in the quintile means, and caution was indicated in interpretation of these data.
A number of investigators examined the specificity of effect for the causes of mortality. In six out of nine studies, there was a moderate to strong elevation of mortality from respiratory disease when compared with death from all causes (Schwartz, 1993; Ostro et al.,., 1996; Schwartz et al.,., 1996; Sunyer et al.,., 1996; Zmirou et al.,., 1996). In three studies (Lyon et al.,., 1995; Anderson et al.,., 1996; Ballester et al.,., 1996), there was a lower RR for respiratory compared with total mortality. In the Six Cities Study (Schwartz et al.,., 1996), RRs for both COPD and pneumonia mortality were higher than for total mortality. Compared with total mortality, the RR for COPD was almost doubled, while the RR for pneumonia was 2.7 times the RR for total mortality for a 10 µg/m3 increase in PM2.5, which suggests an increased risk for people with pre-existing respiratory diseases.
The association between particulate matter and cardiovascular disease (CVD) was also examined in time-series analyses of mortality. In seven out of nine studies, there was a modestly (13-33%) (Ballester et al.,., 1996; Schwartz et al.,., 1996; Sunyer et al.,., 1996) to substantially (up to 300%) (Schwartz, 1993; Lyon et al.,., 1995) higher RR for CVD than for total mortality. Nevertheless, the association with CVD was often of lesser magnitude than the association with respiratory disease.
Elderly people have been suggested to be at increased risk from exposure to particulate matter (Schwartz, 1993, 1994d; Lipfert, 1994), but, overall, there is only a modest increase in RR for the elderly compared with the whole population.
Because of the strong intercorrelations often demonstrated between co-pollutants, such as particulate matter, sulphur dioxide, nitrogen dioxide, carbon monoxide and ozone, it has been difficult to attribute effects of air pollution to any single one of these agents to the exclusion of others. Overall, particulate matter retained its association with acute mortality in analyses that adjusted for other pollutants, although the RRs for particulate matter were slightly reduced during bivariate analyses (Dockery et al.,., 1992; Kinney et al.,., 1995; Saldiva et al.,., 1995; Ito and Thurston, 1996; Ostro et al.,., 1996; Pope and Kalkstein, 1996; Touloumi et al.,., 1996; Verhoeff et al.,., 1996; Zmirou et al.,., 1996) and in one multivariate analysis for Toronto (Özkaynak et al.,., 1995c). The relative risk for mortality and PM10 in nine bivariate analyses (unweighted) was 1.07 ± 0.05 (mean ± SD), with a median of 1.05 (Table 8).
In all of the analyses that examined one or more air pollutants together in the same model with particulate matter, the association of particulate matter with daily mortality was remarkably robust. This was the case for all four of the normally considered gaseous pollutants -sulphur dioxide, nitrogen dioxide, carbon monoxide and ozone. Moreover, in most locations, the magnitude of the particulate matter association was greater than for any of the other air pollutants considered, the exceptions being ozone in London, U.K. (Anderson et al.,., 1996) and sulphur dioxide in Lyon, France (Zmirou et al.,., 1996) and Barcelona, Spain (Sunyer et al.,., 1996). The magnitude, robustness and consistency of this association across so many locations with differing air pollutant mixtures indicate that particulate matter, especially PM10 and PM2.5, is the best indicator of the effects of air pollution on mortality, possibly acting together with other air pollutants.
2.4.1.2.2 Acute effects: hospitalizations and emergency department visits
A summary of the findings for particulate matter from the time-series analyses of hospitalizations for cardiorespiratory disease is presented in Table 9.
In all of the 16 studies using univariate analyses, there were significant associations with PM10 after adjustment for potential confounders and covariates, at concentrations averaging 25-55 µg/m3. Each 10 µg/m3 increase in PM10 was associated with elevated respiratory and/or cardiac hospitalizations or emergency department visits, ranging between 0.35% and 7.3%, with a median of 1.7%.
Studies varied in quality. There were methodological limitations in handling of potential confounders and covariates in the majority of the older studies (1994 and earlier). In the most reliable studies, results for larger cities over a period of several to many years were reported, thereby increasing the number of data points and the statistical power to detect adverse effects. In several studies, data from only one city for a limited time, as short as one summer, were examined (Delfino et al.,., 1997), which limits the statistical power. Certain studies involving larger cities with directly measured PM10 were carefully done and included data over several years (Schwartz, 1994a,b,c, 1996; Dab et al.,., 1996; Burnett et al.,., 1997); however, all were lacking in one or more respects. For example, only 388 days of data over three years were considered in the Toronto, Ontario, study by Burnett et al.,. (1997). Schwartz and co-workers (Schwartz et al.,., 1994a,b,c, 1995, 1996; Schwartz and Morris, 1995) included only the elderly, aged 65 or more (due to limitations in the U.S. medicare database), and did not always consider the possibility of effects from other co-occurring gaseous air pollutants. The most reliable results with respect to estimation of increased hospitalization risks associated with PM10 are considered to be those of Burnett et al.,. (1995) for sulphate pollution for all of southern Ontario (population 8.7 million), over a period of six years (2192 days). Although PM10 itself was not directly measured, site-specific conversion factors were available for estimation of PM10 from measured sulphate. An increase in respiratory hospitalizations of 0.7% (95% CI 0.5-1.0%) per 10 µg/m3 increase in PM10 was calculated based on the results of this study. This figure is within the same range as the risk observed from the combined studies in which PM10 was directly measured.
Table 9 Summary of results of time-series studies of hospital admissions and emergency department visits in relation to particulate matter
The results for Black Smoke are also presented in Table 9. In six of eight studies, there was a significant association between concentrations of Black Smoke and respiratory hospitalizations. The increase in respiratory hospitalizations in the positive studies varied widely from 0.4% (95% CI 0.07-0.75%) to 12.3% (95% CI 5.8-18.2%) per 10 µg/m3 increase in Black Smoke, at the mean concentrations ranging between 12.7 and 75 µg/m3. In the two studies in which th ere was no association (Schouten et al.,., 1996 in Amsterdam, Netherlands; Ponce de Leon et al.,., 1996 in London, U.K.), the ambient Black Smoke concentrations were quite low (<15 µg/m3 versus >26 µg/m3). The authors of the London study (Ponce de Leon et al.,., 1996) explained their negative results with respect to a Black Smoke-respiratory association as misclassification of exposure due to poor placement of Black Smoke monitors in the city. A strong and consistent association with ozone was also noted in this study, as well as a somewhat less consistent association with nitrogen dioxide, with which Black Smoke was highly correlated. In the case of the Amsterdam results, there were only 6.5 respiratory admissions per day. The authors repeated their analysis for the entire population of the Netherlands, and there was an increased risk of 1.01% (95% CI 0.2-1.8%) for a 10 µg/m3 increase in Black Smoke. Thus, these two studies do not detract from the consistency of the positive association observed between increases in Black Smoke and increased hospitalizations.
A summary of findings for PM2.5 and other fine particle metrics is presented in Table 9. Only three hospitalization studies, two in Toronto (Thurston et al.,., 1994; Burnett et al.,., 1997) and one in Montréal (Delfino et al.,., 1997), directly examined the association between PM2.5 and hospitalizations or emergency department visits. Significantly positive associations of PM2.5 (mean concentrations 12.2-18.6 µg/m3) with respiratory effects were seen in univariate analyses in all three studies. Each 10 µg/m3 increase in PM2.5 was associated with a 3.3-9.6% increase in respiratory hospitalizations and emergency department visits and a 2.8% increase (not statistically significant) for cardiac hospitalizations. The association was slightly reduced in Toronto (Burnett et al.,., 1997) when ozone was included in a bivariate regression analysis and was no longer significant in an earlier Toronto study (Thurston et al.,., 1994) or in Montréal (Delfino et al.,., 1997) after inclusion of ozone. The high correlation between ozone and PM2.5 in these two studies (r = 0.62-0.7) makes it difficult to single out the individual effects of PM2.5 from ozone. However, in the larger Toronto study by Burnett et al.,. (1997) in which the correlation coefficient was relatively low (r < 0.34), the association with PM2.5 remained significant after including ozone. A study by Burnett et al.,. (1995) also provided additional information on the magnitude of the risk of respiratory hospitalizations associated with PM2.5 based on measurements of sulphate in southern Ontario. An overall site-specific conversion factor for sulphate to PM2.5 was determined using the equation PM2.5 = 6.973897 + 1.917519 sulphate. The data indicate that a 1.1% (95% CI 0.7-1.4%) increase in respiratory hospitalization and a 1.0% (95% CI 0.5-1.5%) increase in cardiac hospitalization were associated with an estimated 10 µg/m3 increase in PM2.5 (adjusted for ozone).
In all seven studies in which sulphate was examined, there were positive associations with respiratory endpoints (Table 9). Increases ranged from 0.8% to 9.6% in respiratory hospitalizations (n = 5), and up to 18% for emergency department visits (n = 1), associated with every 10 µg/m3 increase in ambient sulphate, at mean concentrations of 4.4-11.8 µg/m3. As noted previously, reliance has been placed on the study by Burnett et al.,. (1995) for southern Ontario, because of the very large database drawn from a population of 8.7 million over 2192 days and the advanced methodological treatment of confounders and covariates. In two studies (Burnett et al.,., 1995, 1997), associations between cardiac effects and sulphate air pollution were also reported. The mean ozone-adjusted increase in cardiac hospitalizations was 2.5% (95% CI 1.3-3.7%) per 10 µg/m3 elevation of sulphate in the large southern Ontario yearly study (Burnett et al.,., 1995). However, in the much smaller summertime study in Toronto (Burnett et al.,., 1997), the increase in RR was not significant (mean 0.4%; 95% CI -1.1% to 0.9%). The associations were weakened (not statistically significant) in both studies when ozone was included in the regression with sulphate. Sulphate forms a part of the fine particle fraction and is usually <1.0 µm in size. Because it correlates well with PM2.5 (correlation coefficients >0.7 in southern Ontario), it appears to serve well as a fine particle indicator in the absence of direct measurements of PM2.5. However, the respective roles of sulphate and non-sulphate fine particles remain unclear.
While the results of the available studies suggest that it is largely the fine fraction of particulate matter that is associated with cardiorespiratory morbidity, significant associations were observed between the coarse fraction (PM10-2.5) and respiratory and/or cardiac admissions in two studies in Toronto (Thurston et al.,., 1994; Burnett et al.,., 1997) and between PM10 and increased respiratory emergency department visits in two U.S. studies in which PM2.5 was an unusually small fraction of PM10 (Hefflin et al.,., 1994; Gordian et al.,., 1996). In those studies in which account was taken of the effects of co-occurring pollutants, the association of the coarse fraction with hospitalizations was robust to the inclusion of gaseous pollutants in some instances (Burnett et al.,., 1997, for respiratory admissions), but not in others (Thurston et al.,., 1994; Burnett et al.,., 1997, for cardiac admissions with sulphur dioxide or nitrogen dioxide); however, the high correlations with one or more gaseous pollutants, and among the particle metrics, complicated attribution of the observed effects to the coarse fraction.
Overall increases in respiratory hospitalizations were associated with particulate matter even at the low concentration ranges (10-100 µg PM10/m3) examined. Curves appeared to increase monotonically, with steep slopes at low concentrations and some suggestion of curvilinear responses (lower slopes) at higher concentrations. The curve for COPD admissions in Detroit, Michigan (Schwartz, 1994b) was the only one showing an anomalous response, since hospitalizations for quartile 2 (at 30 µg/m3) were higher than for quartile 3 (at 50 µg/m3).
The specificity of the association between particulate matter and the causes of disease was investigated. In a series of studies on the elderly in four U.S. cities (Schwartz, 1994a,b,c, 1996), the association between PM10 and COPD was strong, 2.0-5.7% per 10 µg/m3 increase. However, in these studies, the authors did not compare COPD admissions for the elderly with total respiratory admissions or admissions of any other age categories. The study conducted in southern Ontario (Burnett et al.,., 1995) on the entire age range indicates that the increased hospitalization risk for COPD (4.8%) was somewhat higher than that for total respiratory admissions (excess admissions 3.7%) per 13 µg/m3 increase, using sulphate as the particle metric. The southern Ontario studies also demonstrate an effect of sulphate pollution on asthma onset (Burnett et al.,., 1994, 1995). The excess asthma admissions were higher than the total excess respiratory admissions for all ages (7.1% versus 5.8% for a 5.3 µg/m3 increase in sulphate + 100 µg/m3 increase in ozone), the highest rate being for infants ≤1 year (13%).
There was a positive, but less strong, association of CVD with particulate matter pollution in southern Ontario, with an excess hospitalization rate of 2.5% (95% CI 1.3-3.7%; adjusted for ozone) for a 10 µg/m3 increase in sulphate, in comparison with a rate of 2.7% (95% CI 1.8-3.6%) for respiratory disease (Burnett et al.,., 1995). In a much smaller study carried out in Toronto with a mean PM2.5 concentration of 16.8 µg/m3, the risks for CVD hospitalizations were lower by up to one-half the RR for respiratory disease and were not significant (2.8% increase in CVD versus 3.3% increase in respiratory disease per 10 µg/m3 increase in PM2.5 without ozone adjustment) (Burnett et al.,., 1997). Similarly, in Detroit, Michigan (Schwartz, 1994b; Schwartz and Morris, 1995), admission rates were higher for respiratory diseases (2.0% for COPD and 1.2% for pneumonia per 10 µg/m3 increase in PM10) than for CVD (0.56% for ischemic heart disease and 1.0% for heart failure) at PM10 concentrations averaging 48 µg/m3.
The effects of age on hospitalizations or emergency department visits were examined in several locations. For cardiovascular hospitalizations, there was a 3.5% increase (95% CI 1.9-5.0%) in the elderly compared with a 2.5% increase (95% CI 0.5-4.8%) for those less than 65 years of age for a 13 µg/m3 increase in sulphate (Burnett et al.,., 1995). In several studies, elderly people (>64 years) were also at higher risk of respiratory hospitalizations due to particulate matter pollution than were younger people (Schouten et al.,., 1996; Delfino et al.,., 1997). In Amsterdam, Netherlands, the elderly appeared to be at greatest risk for respiratory disease associated with Black Smoke; however, most risks were not significantly elevated, likely due to the small numbers in subdivided groups (Schouten et al.,., 1996). In Montréal, Quebec, mean increases of 10 µg/m3 for PM10, PM2.5 and sulphate were associated with increases of emergency department visits in the elderly of 7.3% (95% CI 1.95-12.7%), 9.6% (95% CI 1.9-17.3%) and 18.2% (95% CI 2.4-34.3%), respectively (Delfino et al.,., 1997). Results of other studies are not consistent with the above observations. In the southern Ontario studies, after adjustment for ozone and temperature, warm-season sulphate was associated with smaller increases in respiratory hospitalizations in the elderly (4.3% increase per 5.1 µg sulphate/m3 increase) than in those between ages 2 and 64 (5.5-7.2%) (Burnett et al.,., 1994), whereas increases for all-year sulphate were similar for younger (3.7% increase) and older people (65+ years) (3.8% increase per 13 µg/m3 increase in sulphate) (Burnett et al.,., 1995). In Rotterdam, Netherlands, the association between Black Smoke and all respiratory admissions was strongest for the 15- to 64-year age group, with an RR of 1.37 compared with an RR of 0.97 for those aged 65 or older (Schouten et al.,., 1996). In London, U.K., there were no consistent differences in RRs of respiratory hospitalizations for those 65 and over compared with those between 15 and 64 years (Ponce de Leon et al.,., 1996). With asthma as the endpoint, the elderly were at less risk than those under age 65 years in Seattle, Washington (Schwartz et al.,., 1993), Anchorage, Alaska (Gordian et al.,., 1996) and southern Ontario (Burnett et al.,., 1994).
Young children appear to be a high-risk group with respect to respiratory hospitalizations. In the Utah Valley study, bronchitis and asthma admissions were twice as high for children (1-5 years) when a local steel mill was open as when it was closed; for all ages combined, the rates were only 1.4 times as high for bronchitis and 1.2 times as high for pneumonia (Pope, 1991). In Anchorage, Alaska (mean PM10 45.5 µg/m3), upper respiratory tract infections were most strongly associated with PM10 in children (<10 years) and in older adults (≥45 years) compared with adolescents and younger adults (Gordian et al.,., 1996). In southern Ontario, the association between summertime ozone (mean one-hour maximum 64-140 µg/m3) and sulphate (mean 3.1-8.2 µg/m3) combined and respiratory hospitalizations in infants and children was greater (14.8%) than for adults (4.3-7.2%) (Burnett et al.,., 1994), whereas associations between all-year sulphate (2.0-7.7 µg/m3) and respiratory hospitalizations in children (mean excess admission 2.7%) was less than in adults (3.7-3.8% for 13 µg/m3 increases in sulphate) after adjusting for the effects of ozone (Burnett et al.,., 1995). In the London, U.K., respiratory hospitalization study (Ponce de Leon et al.,., 1996), children (aged 0-14 years) were not at increased risk for respiratory admissions due to Black Smoke or other air pollutants, at least not at the concentrations experienced in the 1980s and early 1990s.
In most of the studies, potential confounding by temperature and season was well addressed using a variety of methods. However, there has been concern that the effects observed might be due to the fluctuation of other unmeasured pollutants in concert with particulate matter. Most of the hospitalization studies on particulate matter examined at least one or two other air pollutants as potential covariates (Table 9). Indeed, in many studies, there was a strong correlation between particulate matter and gaseous pollutants; yet for respiratory hospitalizations or emergency department visits, gaseous pollutants in bivariate or multivariate analyses reduced but did not abolish the significance of the risk for particulate matter. With respect to cardiac hospitalizations, the results are inconsistent. In studies conducted by Schwartz and Morris (1995) in Detroit, Michigan, and by Burnett et al.,. (1997) in Toronto, Ontario, the association between PM10 and cardiac hospitalizations remained significant in the bivariate or multivariate regression with ozone, carbon monoxide and sulphur dioxide. In the southern Ontario study (Burnett et al.,., 1997), co-regression with ozone also did not reduce the significance of the risk of sulphate on cardiac hospitalizations. However, in the Toronto study (Burnett et al.,., 1997), co-regression of PM2.5 or sulphate with ozone reduced the cardiac risk to insignificance.
2.4.1.2.3 Acute effects: respiratory health (lung function, symptoms, restricted activity and days absent from work or school)
Increases in ambient particulate matter have been associated with small, reversible decrements in lung function in normal asymptomatic children and in both adults and children who have some form of pre-existing respiratory conditions, particularly asthma, at PM10 concentrations ranging from 10 to 174 µg/m3 (Pope et al.,., 1991, 1992; Spektor et al.,., 1991; Hoek and Brunekreef, 1993, 1994; Roemer et al.,., 1993; Peters et al.,., 1996a,b). For every 10 µg/m3 increase in PM10, average decrements in peak expiratory flow rate were 0.09-0.4%. These changes were often accompanied, especially in adults, by increases in, for example, chronic bronchitis or cough. Each 10 µg/m3 increase in PM10 was associated with increases of 0.6-2.2% in respiratory symptoms. Based on limited data, associations between PM2.5 pollution and increased respiratory symptoms have been observed at concentrations averaging 18-22 µg/m3 (Ostro and Rothschild, 1989; Ostro et al.,., 1991; Schwartz et al.,., 1994; Peters et al.,., 1996a,b). Each 10 µg/m3 increase in PM2.5 was associated with increases of 0.9-2.2% in respiratory symptoms. For every 10 µg/m3 increase in sulphate, average decrements in peak expiratory flow rate were 1.2%, while the risk for increased respiratory symptoms was 4.6-16.4% (Ostro et al.,., 1991; Schwartz et al.,., 1994; Peters et al.,., 1996a,b).
Respiratory-related restrictions in activity severe enough to result in an increased number of days lost to work in adult workers and in school absences in children were also associated with elevated ambient PM10 (mean concentrations 41-51 µg/m3) (Ransom and Pope, 1992) and PM2.5 (mean concentrations 20-25 µg/m3) (Ostro, 1987, 1990; Ostro and Rothschild, 1989) or with other fine particulate components such as sulphate (Ostro, 1990). Each 10 µg/m3 increase in PM10, PM2.5 and sulphate was associated with increases in respiratory symptom-related activity restriction of 9.0%, 12.9-15.8% and 17.4%, respectively.
2.4.1.2.4 Long-term and chronic effects
Relatively few studie s are available in which the effect of long-term or chronic exposure to particulate matter on health outcomes has been examined. Such exposures, varying in duration between one and 16-20 years, were associated with increases in mortality, respiratory disease symptoms, decrements in lung function and, possibly, increases in lung cancer in both cross-sectional and prospective cohort studies.
In these studies, the concentrations of PM10, PM2.5 and sulphate were 18-47 µg/m3, 11-30 µg/m3 and 4.8-13 µg/m3, respectively. In two prospective cohort studies (Dockery et al.,., 1993; Pope et al.,., 1995), average increases in mortality from all causes of 10%, 7-14% and 7.5-32% were observed for each 10 µg/m3 increase in PM10, PM2.5 and sulphate, respectively, after adjustment for a number of potential confounders and covariates. In two descriptive studies (Özkaynak and Thurston, 1987; Pope et al.,., 1995), the odds for annual mortality were increased by 4.3-9.8% and 8.2-12.6% for each 10 µg/m3 increase in PM2.5 and sulphate, respectively. Based on the mean particulate matter levels across six cities, lifespan was estimated to have been reduced by about two years over a 14-year period (Dockery et al.,., 1993), an observation incompatible with suggestions that most or all of the observed deaths were due to "harvesting" or accelerating the death of persons already ill by a few days or a few weeks.
increased odds ratios for bronchitis were associated with chronic exposure of children to particulate matter for all or most of their lives. In the most powerful of the available studies, each 10 µg/m3 increase in PM10, PM2.5 and sulphate was associated with decrements of forced vital capacity of 1.4%, 2.2% and 4.5%, respectively (Raizenne et al.,., 1996). The odds ratios for bronchitis ranged from non-significant to 4.3% for each 10 µg/m3 increase in PM2.5 and from non-significant to 9.3% for each 10 µg/m3 increase in sulphate (Özkaynak and Thurston, 1987; Dockery et al.,., 1996a).
In a prospective study, an 18% increase in new cases of chronic bronchitis in older adults was associated with each 10 µg/m3 increase in PM2.5 for the 10 years of the observation period (Abbey et al.,., 1995a). Development of new cases of chronic bronchitis in association with exposure to particulate matter has also been observed in children (Dockery et al.,., 1989, 1996a). Increases in the severity of respiratory symptoms of airway obstructive disease, chronic bronchitis and asthma have been associated with exposure to TSP, PM10, PM2.5 and sulphate for 10 years (Abbey et al.,., 1995a).
There is also some evidence for an association between long-term exposure to fine particulate air pollution and lung cancer. In two recent cohort studies, exposure to average annual concentrations of 11-29.6 µg/m3 for PM2.5 (Dockery et al.,., 1993) or 3.6-23.5 µg/m3 for sulphate (Pope et al.,., 1995a) was associated with an increased risk of lung cancer mortality (19.9% or 18.1% for an increase of 10 µg/m3 of PM2.5 or sulphate, respectively), after adjustment for potential confounders or covariates. The associations were weak compared with those for other lifestyle factors such as smoking. In a cohort study conducted by Abbey et al.,. (1995b) on non-smoking Californian Seventh-Day Adventists (n = 6340), exposure to PM10 concentrations above 100 µg/m3 for 42 days per year was associated with a marginally increased incidence of all cancers combined in females (RR 1.15, 95% CI 0.97-1.38) and a non-significant increase in respiratory cancers, but these results were based on relatively few cases of total (n = 175) and respiratory (n = 17) cancers. The association with lung cancer at this time is inconclusive.
2.4.2 Experimental animals and in vitro
Given the extensive epidemiological evidence of an association between acute health effects and ambient exposure to particulate matter (Section 2.4.1.2), information on particulate matter-induced effects in animals is primarily of interest with respect to the extent to which it provides insight into the biological plausibility for the association. Consequently, this section only briefly summarizes the results of studies in animals exposed to particulate matter, with emphasis on the target tissues, susceptible subpopulations, the toxicity of various particle size fractions and compositions, and plausible biological mechanisms for particulate matter-induced effects identified in these studies. These include severe damage of the alveolar interstitium tissue by the highly reactive surfaces of ultrafine particles and metal components, causing acute pulmonary edema and inflammation and/or particulate matter-induced alteration (enhancement or impairment) of the respiratory immune system.
2.4.2.1 Acute and short-term exposure
Acute exposures (four- to six-hour single exposures) of laboratory animals to a variety of types of particles, almost always at concentrations well above those occurring in the ambient environment (particles >1 mg/m3, acid aerosols >50 µg/m3), confirm that the cardiorespiratory system is a target for particle-induced effects.
Effects on the respiratory system include decreased ventilatory function and airway hyperresponsiveness (a hallmark of human asthma) in guinea pigs or rabbits (Chen et al.,., 1990, 1991b, 1992a; El-Fawal and Schlesinger, 1994), altered mucociliary clearance in rabbits (Chen and Schlesinger, 1983; Grose et al.,., 1985; Naumann and Schlesinger, 1986) and a range of histological and cellular (Callis et al.,., 1985; Wiessner et al.,., 1989, 1990; Guilianelli et al.,., 1993) and biochemical (Lindenschmidt et al.,., 1990; Chen et al.,., 1991a,b,c, 1992b; Mohr et al.,., 1992; Kobzik et al.,., 1993; Kodavanti et al.,., 1997) alterations in the lung, including the production of proinflammatory cytokines and other mediators by pulmonary alveolar macrophages.
The pulmonary immune system is also affected by exposure to particulate matter, seen as increased numbers of alveolar macrophages and polymorphonuclear leukocytes in the alveoli of mice, hamsters and rats (Brain and Cockery, 1977; Adamson and Bowden, 1981; Lehnert et al.,., 1985), decreased mobility of alveolar macrophages in rabbits (Schlesinger, 1987) and altered (increased or decreased) ability of macrophages to phagocytize particles in mice and rats (Fisher and Wilson, 1980; Tabata and Ikada, 1988; Schlesinger et al.,., 1990; Warheit et al.,., 1991; Chen et al.,., 1992b). (The latter effect is often related to particle composition, with silica and quartz causing a decrease in macrophage activity and iron oxide causing an increase in macrophage activity.) Particulate matter also modifies immunological responses, including airway defence mechanisms against microbial infections. This effect appears to be related to composition and not the particle effect, since particles with known cytotoxic properties, such as metals, affect the immune system to a significantly greater degree than other particles (Hatch et al.,., 1985; Chen et al.,., 1989, 1992b; Oberdörster et al.,., 1992a, 1994b; Zelikoff and Schlesinger, 1992; Berg et al.,., 1993; Nadeau et al.,., 1995, 1996).
Finally, the cardiovascular system can also be affected; acute exposure to particulate matter induces electrocardiographic abnormalities in rats and dogs (Sakakibara et al.,., 1994; Campen et al.,., 1996; Nearing et al.,., 1996).
The results of studies in animals also confirm that adverse effects of exposure to particulate matter, including mortality, morbidity and bronchial hypersensitivity to non-specific stimuli, are much more pronounced in individuals with pre-existing cardiorespiratory diseases (Slauson et al.,., 1989; Raabe et al.,., 1994; Godleski et al.,., 1996; Gilmour et al.,., 1997; Killingsworth et al.,., 1997). For example, mortality was increased following exposure of rats with acute pulmonary inflammation or chronic bronchitis to 250 µg PM2.5/m3 (three days, six hours per day), while no deaths occurred in healthy rats (Godleski et al.,., 1996). Similarly, Killingsworth et al.,. (1997) reported that inhalation of fuel oil ash (approximately 580 µg/m3, six hours per day for two days) caused effects only in the rats with pre-existing cardiorespiratory injury induced by monocrotaline, including acute mortality (40%), inflammatory cell infiltration in pulmonary interstitium and blood vessel walls, and increases in expression in lung and heart of proteins and messenger RNA of several chemokines involved in inflammatory cell recruitment.
The particle types most likely to induce acute adverse effects include metals, organics, acids and acidic sulphates of the fine particle mode, possib ly occurring as coatings on fine or even ultrafine carrier particles (Chen et al.,., 1991a,b, 1992a). It appears that the ultrafine particle mode (≤0.1 µm in size) may be of significant toxicological importance due to its large number and slow clearance rate from pulmonary interstitium (Oberdörster et al.,., 1992b, 1994d). Ultrafine particles were shown to result in pulmonary inflammation and death after about 30 minutes in rats or guinea pigs exposed to a number concentration of 700 000 to 1 million particles (median diameter 26 nm) at near-ambient concentrations of ultrafine particles (9-60 µg/m3) (Warheit et al.,., 1990; Chen et al.,., 1992b; Oberdörster et al.,., 1995). There are indications that surface-complexed iron and other metals on particles are involved in pulmonary injury (Berg et al.,., 1993; Guilianelli et al.,., 1993; Carter et al.,., 1997; Kodavanti et al.,., 1997).
While the mechanisms for the cardiorespiratory effects observed following exposure to particulate matter are not clear, there is emerging evidence from the studies in animals that may ultimately explain the acute effects observed in the epidemiological studies. It is possible that the increase in respiratory diseases associated with particulate matter in the epidemiological studies is the result of severe damage of the alveolar interstitium tissue by the highly reactive surfaces of ultrafine particles (Ferin et al.,., 1991) and metal components (Berg et al.,., 1993; Guilianelli et al.,., 1993; Carter et al.,., 1997; Kodavanti et al.,., 1997), causing acute pulmonary edema and inflammation (Oberdörster et al.,., 1992a, 1994c); alternatively, these outcomes may reflect the particulate matter-induced alteration (enhancement or impairment) of the respiratory immune system (Brain and Cockery, 1977; Adamson and Bowden, 1981; Lehnert et al.,., 1985; Killingsworth et al.,., 1997). With respect to particulate matter-initiated acute cardiovascular effects, it has been postulated that oxidized low-density lipoprotein may play a pathological role, since diesel exhaust particles have been reported to cause oxidative modification of low-density lipoprotein in vitro (Ikeda et al.,., 1995). Oxidized low-density protein has been found to cause endothelial damage, the proliferation of smooth muscle cells, monocyte-endothelial interactions, platelet aggregation and inhibition of endothelial-derived relaxation of vascular smooth muscles (Morel et al.,., 1983; Ocana, 1989; Berliner et al.,., 1990; Cushing et al.,., 1990; Ezaki et al.,., 1994; Ichinose et al.,., 1995; Ikeda et al.,., 1995). These biochemical and pathological changes may all lead to coronary vasospasm, lesion and blockade, and hypertension. There are also reports that chemokines, which are among the mediators induced by exposure to particulate matter, are involved in myocardial dysfunction, decreased contractility and vasoconstriction in vitro and in vivo (DeMeules et al.,., 1992; Abe et al.,., 1993; Yokoyama et al.,., 1993; Mann and Young, 1994).
2.4.2.2 Subchronic and chronic exposure
In the available subchronic and chronic exposure studies, animals were repeatedly exposed to very high (>1 mg/m3) concentrations of a wide variety of types of particles, often on a schedule that mimicked workplace conditions (e.g., six hours per day, five days per week). These exposures resulted in a wide range of effects on the lung, including compromised lung functions in guinea pigs and rats (Wiester et al.,., 1980; Ellakkani et al.,., 1987; Mauderly et al.,., 1988; Begin et al.,., 1989; Heinrich et al.,., 1989), hyperresponsiveness in rabbits (Gearhart and Schlesinger, 1986), impaired airway clearance function in guinea pigs, donkeys and rabbits (Gearhart and Schlesinger, 1989; Nagai et al.,., 1991; Schlesinger et al.,., 1992; Samet and Cheng, 1994) and impairment of immune functions in the lungs of rats and mice by reducing macrophage phagocytosis and bactericidal capability (Spiegelberg et al.,., 1984; Gilmour et al.,., 1989a,b; Kleinman et al.,., 1995).
Long-term exposure to particulate matter results in histopathological and cytological changes in the lung, including chronic pulmonary inflammation, hyperplasia of the alveolar epithelium and pulmonary fibrosis, regardless of particle type, mass concentration, duration of exposure or species examined (Shami et al.,., 1984; Henderson et al.,., 1988; Gearhart and Schlesinger, 1989; Schlesinger et al.,., 1992; Kawabata et al.,., 1993; Kleinman et al.,., 1995). The development of lung tumours has also been observed following chronic exposure of rats to high levels of a wide variety of particle types (Heinrich et al.,., 1986a,b; Kawabata et al.,., 1993; Pott et al.,., 1994).
A particularly relevant subchronic study was conducted by Kleinman et al.,. (1995), in which rats were exposed to ammonium sulphate (20 or 70 µg/m3, 0.2 µm MMAD), ammonium nitrate (90 or 350 µg/m3, 0.6 µm MMAD), resuspended road dust (300 or 900 µg/m3, 4 µm MMAD) or purified air, four hours per day, four days per week, for eight weeks. Decreases in alveolar macrophage function and increased pulmonary permeability were observed following exposure to nitrate and sulphate (low and high doses) and road dust (high dose only). Based on quantitative histopathological analyses, there were moderate to substantial changes following exposure to particles, in the order of sulphate > nitrate > road dust. Thus, in this study, the lungs were adversely affected by repeated exposure to particles of a size, mass concentration and composition relevant to ambient exposure conditions, and the fine fraction of PM10 was more toxic than the coarse fraction.
2.5 Toxicokinetics
Particle size is believed to be the most important characteristic influencing deposition in the three anatomical regions of the respiratory system (Lippmann, 1977; Anderson et al.,., 1990; Dockery and Pope, 1994). In the extrathoracic region (nose and mouth) in human airways, virtually all particles >10 µm in diameter, when inhaled through the nose, are deposited in the nasal region, whereas during mouth breathing this drops to approximately 65% (U.S. EPA, 1982). In the tracheobronchial region, only particles ≤10 µm in diameter are deposited. However, owing to the bypass of the nasal cavity during oral breathing, up to 10% deposition of particles up to 15 µm in diameter can occur in the tracheobronchial region (Miller et al.,., 1979). The deposition of particles in the pulmonary region in humans is probably the most critical with regard to the health effects associated with particulate matter, since this appears to be the target tissue. Lung deposition in this region is highest for small particles of submicrometre size and is markedly reduced for particles of 2 µm and above (ICRP, 1994). Churg and Brauer (1997) examined the upper lobe apical segment parenchyma of autopsy lung tissue for long-term residents of Vancouver, British Columbia, using analytical electron microscopy and found that 96% of the particles deposited had aerodynamic diameters less than 2.5 µm.
While deposition of fine particles on a mass per unit alveolar surface area is not different between rats and humans, based on the calculations per ventilatory unit or per alveolus, humans receive much greater numbers of particles than do rats when exposed to the same mass concentration of particulate matter, particularly for particles 0.1-0.3 µm in size. This difference has been observed to be even more pronounced for individuals with compromised lungs (smokers and patients with respiratory diseases) than for normal subjects (Kim and Kang, 1997). It was estimated that rats exposed to 1-1.5 mg particles/m3 may actually have received a level of particles equivalent to 120-150 µg/m3 in humans (Miller et al.,., 1995).
Clearance of particles from the extrathoracic region occurs by mechanical processes; in the nasal area, by blowing, wiping or sneezing; and in the more anterior regions, either through swallowing (in mucus) or by expectorating. Clearance from the extrathoracic region may take up to one or more days in humans (Proctor and Wagner, 1965). In the conducting airways of the tracheobronchial region, the most prominent mechanism for elimination of particles is via the action of the mucociliary escalator into the gastr ointestinal tract. This is a very fast clearance pathway, which is mainly completed within 24 hours after deposition of particles in this region. The half-time of clearance in humans ranges from half an hour in the larger airways to five hours in the smaller airways (U.S. EPA, 1982), although there is some evidence of a long-term component to tracheobronchial retention (approximately 500 days) (Stahlofen et al.,., 1986a,b; Smaldone et al.,., 1988). In the pulmonary region (the site of effects in the epidemiological studies), insoluble particles are rapidly cleared through phagocytosis by alveolar macrophages. Following phagocytosis, the macrophages can migrate such that they are cleared via mucociliary flow; however, particle-laden alveolar macrophages are found in the lung up to many hundreds of days post-exposure, indicating that some macrophages do not migrate and that particles are reingested by other generations of alveolar macrophages. Particles may also enter the interstitium via endocytosis by alveolar epithelial cells, specifically Type I cells. Ultrafine particles (<~50 nm) have a much greater propensity to penetrate into the pulmonary interstitium and escape phagocytosis by alveolar macrophages (Ferin et al.,., 1991).
