2 Health Impacts of Particles
Our understanding of the health impacts of particles in the air is based largely on the results of epidemiological studies carried out in numerous locations throughout the world. Historically, much of these data came from studies carried out in the United States. However, in more recent years health effects studies have also been carried out in other countries. These studies range from extensive research in areas such as Europe to single city time-series studies in smaller urban areas such as Christchurch.
The health effects associated with particles in the air range from minor nose and throat irritations to more severe impacts such as hospital admissions and premature mortality. The population base susceptible to the different impacts varies by effect (Figure 2.1) with a smaller subsection comprising of the elderly and those with pre-existing conditions most at risk of impacts such as premature mortality.
There are limitations associated with using epidemiological studies to determine health impacts, particularly in understanding the extent of impact in assessing causality and in determining biological mechanisms. The most common methodology for assessing the relationships has been the standard time series epidemiological studies. Historically these have shown increases in mortality of around 0-1% per 10 µgm-3 increase in the 24-hour average PM10 concentration. However, recent reanalysis of the National Morbidity, Mortality and Air Pollution Study (NMAPS) suggests that the size of the effect has been overestimated in a number of the US studies owing to the use of the default convergence criteria in the S-Plus, which were inadequate to produce optimal estimates and could result in an upward bias (Samet et al, 2000). The statistical technique is associated with the general additive model (GAM) method of estimating an effect of one or more variables on an outcome. Many other studies, including those carried out in Australia, have used the GAMs models but have not encountered the same issues as those observed in the US studies (NEPC, 2002). The implications for New Zealand studies are minimal, as risk assessments have typically been based on studies such as Hales et al (1999) and Kunzli et al (2000), which incorporated different software and methodologies respectively.
Although the most common method of study, the time series studies provide only an indication of the acute health impacts that are documented immediately after exposure. This is because they are based on the temporal relationship between pollution episodes and the recording of health endpoints. An alternative study methodology, which is not subject to these limitations or potential underestimates associated with the GAMs models, is the prospective cohort design used in two US studies, the Harvard Six Cities study (Dockery et al, 1993) and the American Cancer Society study (Pope et al, 1995). These studies followed groups of 8111 and 552,138 adult subjects for 14-16 and seven years respectively. Higher concentrations of particles were found to be associated with increased mortality in both studies. In response to criticisms of the health impacts assessments, these studies were thoroughly reviewed for the United States Health Effects Institute (Krewski, 2000). The review was in support of the original findings.
Kunzli et al (2000) used the results from these studies to estimate the impact of PM10 concentrations on mortality in Switzerland, France and Austria. This assessment was based on an increase of 4.3% in mortality per 10 µgm-3 increase in annual average PM10 concentrations assuming a no effects threshold of 7.5 µgm-3. The basis was primarily the Dockery et al (1993) study which indicated an increase in mortality of around 26% from the least polluted to the most polluted city, giving a 4.3% increase per 10 µgm-3 increase in annual average PM10. A higher ratio of 8.3% was found in for PM2.5. Similar results were found in the Pope et al (1995) study, which indicated a 6.9% increase in mortality per 10 µgm-3 increase in annual average PM2.5. The dose-response estimates used of 4.3% are conservative relative to the WHO (2002) recommendations, which indicate a 10% increase in mortality per 10 µgm-3 increase in PM10 for the impacts of long-term exposure.
Although a significant study methodology, cohort studies of the impact of particle pollution are limited. In addition to the Dockery et al (1993) and Pope et al (1995) studies, a study was conducted by Abbey et al (1999) and more recently a cohort design study was carried out outside of the USA, in the Netherlands. This latter study examined the relationship between cardiopulmonary mortality and living near to a major road (Hoek et al, 2002). Results indicated increased relative risk of cardiopulmonary mortality and all cause mortality for those living near a major road but no increase risk for non-cardiopulmonary and non-lung cancer deaths.
Another recent publication of significance demonstrates the improvement in health impacts associated with regulations prohibiting the burning of coal in Dublin (Clancy et al, 2002). In the years following the ban, black smoke concentrations declined by 35.6 µgm-3, a 70% reduction, and there were about 116 fewer respiratory deaths and 243 fewer cardiovascular deaths. These equated to reductions of around 4% in respiratory mortality and around 3% reduction in cardiovascular mortality per 10 µgm-3 reduction in 24-hour average black smoke concentrations.
2.1 Current issues
Historically, one of the main concerns about the epidemiological studies has been establishing the causality of the relationship between particle pollution and the health impacts. That is, whether or not it was the particles themselves or some other factor that resulted in the observed relationship. Many researchers, policy makers and health experts in recent years have examined the case for and against causality. Generally the assessment is based on the weight of evidence approach and the causality of the association is considered using a set of established criteria, such as those proposed by Bradford Hill (1971). Of particular significance is a review by Dab et al (2001) which evaluated 15 assessments of causality conducted by leading researchers and concluded that the observed relationships were both valid and causal. Also significant in terms of causality are studies such as Clancy et al (2002) and Pope et al (1992) which demonstrate improvements in health indicators associated with reductions in particle concentrations.
This conclusion appears to be generally accepted and the focus of the PM10 health endpoint relationship is now on better characterising relationships and establishing biological mechanisms.
2.2 Biological mechanisms
The biological mechanisms responsible for the health impacts associated with particle concentrations have been considered in detail in recent years. A number of mechanisms have been demonstrated for the respiratory related impacts that focus on mechanisms of inflammation, tissue damage and repair (Brunekreef and Holgate, 2002). However, less is known about the cardiovascular impacts. Brunekreef and Holgate identify a number of factors such as blood clotting, increased fibrinogen and platelets and sequestration of red blood cells in the lung mass that appear to contribute but conclude that their significance remains to be established. Peters and Pope (2002) suggest that evidence points towards particle-induced pulmonary and systemic inflammation, accelerated atherosclerosis and altered cardiac autonomic function as possible pathways linking particle pollution and cardiovascular mortality.
Most toxicological studies investigating biological mechanisms have used either experimental animals or isolated cell systems and have generally been used to test hypothesis generated through epidemiological studies. Toxicity studies place emphasis on the composition of the particles. A number of mechanisms are proposed for different compositions, including a strong focus on transition metals (EPAQS, 2000). However, impacts are demonstrated for highly variable compositions, including pollen and fungal spore derived particles and non-transition metal particles. One of the complicating factors when considering toxicology is that it is unlikely that any one toxic fraction will be the cause of all types of impacts.
In explaining the biological mechanisms, present thinking focuses on there being something associated with the particle surface, most likely adsorbed transition metals but also possibly some other physical or chemical property that is able to initiate oxidative stress when it comes in contact with lung cells. This results in inflammation, which may have different consequences depending on individual susceptibility (EPAQS, 2000).
2.3 Issues of mortality advancement
The extent to which the associations between PM10 concentrations and mortality result in the advancement of death is not well addressed by most study designs. However, it is an important point in establishing the real life impact of the particle and mortality relationships.
The best information on the extent of mortality advancement stems from the cohort study designs (e.g. Dockery et al, 1993; Pope et al, 1995). These studies indicate life shortening by 1-2 years and that these effects may also depend on factors such as education and antioxidant vitamin status indicating a greater reduction in life expectancy for disadvantaged population groups (Brunekreef and Holgate, 2002).
The issue of life shortening is regularly raised in economic assessments, which attempt to put a cost on the value of the life lost. In that capacity, deaths are often characterised as acute harvest, acute non-harvest and chronic, with the acute harvest deaths referring to those that would have occurred within a period of a few weeks irrespective of the PM10 concentrations. Although the cohort designs provide some indication of the extent of advancement for the non-harvest effects, they do not assist in determining the proportion of deaths that fall within each category. The impact in terms of costs associated with the harvest versus non-harvest deaths is significant. For example Bicknell (2001) places a value of $14,195 on harvested deaths in New Zealand based on a one-month mortality advancement compared to a value of $172,000 per year of reduced life for non-harvest deaths.
2.4 Health indicators
One of the more current questions regarding health effects of particles is the selection of an appropriate indicator of impact. For the past decade monitoring has focused on the PM10 size fraction and consequently the majority of the epidemiology relates to this indicator. However, in more recent years there has been a tendency to focus on the smaller PM2.5 size fraction, because these particles penetrate deeper into the lung. Increased monitoring of PM2.5 has occurred within the last five years in both New Zealand and overseas with an additional focus on particle speciation to assist in studies on biological mechanisms and the impact of composition.
The relative impact of the PM2.5 size fraction versus the particles in the coarse PM10-PM2.5 size fraction has been considered in a number of ways. Based on the epidemiology, the variation in risk across a range of ambient concentrations has generally been similar for PM10 and PM2.5. While the PM10-2.5 size fraction has tended to show less consistent associations, associations between the coarse size fraction and adverse heath effects are shown to occur in a few studies. In particular the Expert Panel on Air Quality Standards (EPAQS, 2000) report examples of associations between the coarse size fraction and a number of health endpoints including:
- all cause mortality (from Dockery 1992, Mar et al, 2000, Castillejos et al, 2000 and Schwartz et al, 1996)
- cardiovascular mortality (from Mar et al, 2000)
- cough without other symptoms (from Schwartz and Neas, 2000)
- cough symptoms in children with chronic respiratory symptoms (from Tiittanen et al, 1999)
- hospital admissions for cardio-respiratory disease (from Burnett et al, 1997).
Other researchers (e.g. Brunekreef and Holgate, 2002) indicate that the bulk of the evidence suggests that mortality is associated with the finer PM2.5 size fraction, although the coarse fraction may be responsible for other effects such as hospital admissions for asthma.
Studies of the specific impact of the coarse fraction are clearly varied. For example, Schwartz et al (1999) found no association between coarse particles of crustal origin on mortality in the city of Spokane, WA, whereas Ostro et al (1999) found an association between PM10 and daily mortality in a desert area of California where coarse particles comprise the larger part of PM10.
Of particular interest in areas of New Zealand are the health implications of particles derived from sea spray emissions. While these particles typically reside in the PM10 size fraction, measurements in the finer PM2.5 size range have been measured in New Zealand. The extent to which studies indicating effects associated with coarse particles apply to sea spray is uncertain. It could be argued that particles from sea spray would dissolve in the lungs and therefore that health effects are unlikely. However, in the absence of literature specific to the health implications of sea spray particle emissions, the precise health implications remain uncertain.
While these data suggest some health impacts are associated with the coarser particle fraction, it is possible that another measure (e.g. a smaller size fraction, compositional analysis or particle count data) may provide a better indicator for assessing health effects and understanding biological mechanisms. Consideration of an alternative indicator was assessed for the United Kingdom by the Expert Panel on Air Quality Standards (EPAQS, 2000). The report concluded that based on the present evidence, the measurement of PM10 provided the most appropriate basis for an air quality standard for the United Kingdom. However, they also indicated that further research may lead to alternative metrics such as PM2.5 or counts of ultra fine particles but that current data on these measures was insufficient to derive standards.
In the United States a revised particle standard based on the PM2.5 size fraction was introduced in 1997. Other countries such as Australia and Canada have also moved towards standards for the PM2.5 size fraction. Some researchers argue that an even smaller size fraction (e.g. PM1) or particle count measurements may be a more suitable indicator of health impacts. It is unlikely, however, that standards for these alternative indicators would be adopted without further data on the relationships between such measurements and health impacts.
