Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Public health impacts of naturally-derived particulate matter : a case study of Asian dust in southwestern… Bennett, Charmian Margaret 2005

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2005-0006.pdf [ 13.78MB ]
Metadata
JSON: 831-1.0091779.json
JSON-LD: 831-1.0091779-ld.json
RDF/XML (Pretty): 831-1.0091779-rdf.xml
RDF/JSON: 831-1.0091779-rdf.json
Turtle: 831-1.0091779-turtle.txt
N-Triples: 831-1.0091779-rdf-ntriples.txt
Original Record: 831-1.0091779-source.json
Full Text
831-1.0091779-fulltext.txt
Citation
831-1.0091779.ris

Full Text

PUBLIC H E A L T H IMPACTS OF N A T U R A L L Y - D E R I V E D P A R T I C U L A T E M A T T E R : A C A S E STUDY OF ASIAN DUST IN SOUTHWESTERN BRITISH C O L U M B I A by CHARMIAN MARGARET BENNETT B.Sc. (Hons), University of Canterbury, 1999 M.Sc., University of Canterbury, 2000  A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T OF T H E REQUIREMENTS FOR T H E D E G R E E OF M A S T E R OF ARTS in T H E F A C U L T Y OF G R A D U A T E STUDIES (Geography)  T H E UNIVERSITY OF BRITISH C O L U M B I A January 2005  © Charmian Margaret Bennett, 2005  ABSTRACT The adverse public health impacts of anthropogenically-derived particulate matter have been well documented, with measureable increases in both morbidity and mortality rates associated with high particulate matter pollution events. Most current research has focused on the health impacts of anthropogenically-derived particulate matter, and there is a distinct scarcity of literature that examines the role of naturally-derived particulate matter and adverse health impacts in the urban context. This study of a Gobi desert dust event in British Columbia, Canada, in spring of 1998 provided a unique opportunity to identify the adverse health effects related to naturallyderived particulate matter in a large urban setting. Respiratory and cardiac hospitalizations were examined for a three-year period (January 1997 to December 1999), with the Gobi dust event occurring in late April 1998. A meteorological analogue was identified for spring 1997 in order to identify the public health impacts associated with anthropogenically-derived particulate matter and those impacts associated with the presence of the Gobi desert dust. Results indicate that this Gobi dust event was not associated with an excess of hospitalizations for the Fraser Valley region. Peak particulate matter concentrations of Gobi desert dust in the airshed were only associated with an additional two or three hospitalizations for respiratory and cardiac illnesses, and these increases were not distinguishable from the 'normal' variability in hospitalization rates. Despite high particulate matter concentrations, fine particle size, presence of heavy metals in the dust and extended exposure periods, it appears that the Gobi desert dust event was not associated with significant risk to public health in the Fraser Valley, British Columbia. Therefore it is concluded that naturally-derived particulate matter is more benign than particulate matter of anthropogenic origin, and thus poses a low risk to health for the general public.  n  T A B L E OF CONTENTS  Abstract  ii  Table of Contents  iii  List of Tables  v  List of Figures  vi  Acknowledgements CHAPTER I  Introduction and Research Context  1.1  Introduction  1.2 1.3 1.4 1.5 1.6  Research Context: Dust in the Atmosphere The 1998 Gobi Dust Event in North America Identification of the Research Question Rationale Summary  CHAPTER II 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 CHAPTER III 3.1 3.2 3.3 3.4 3.5 CHAPTER IV 4.1 4.2 4.3  Review of Air Pollution and Health  viii 1 2 2 5 8 9 10 11  Introduction 12 Historical Accounts of Particulate Matter Pollution and Health Impactsl2 Overview of the Health Effects of Particulate Matter Pollution 14 The Effects of Particulate Matter on Health 16 Important Characteristics of Particulate Matter with Regard to Health Impacts 23 Gaps in the Current Literature 25 The Role of Naturally-Derived Particulate Matter 26 Summary 30 Research Setting Introduction Physical Setting and Air Pollution Meteorology : Pollutants and their Sources in the Fraser Valley Health Effects Associated with Air Pollution in the Fraser Valley Summary Methodology  32 33 33 36 40 42 44  Introduction 45 Epidemiological Methods Used to Investigate the Effects of Air Pollution 45 Assigning Causation and Possible Confounding in Ecological Studies. 49  4.4 4.5  Methodology of this Research Summary  CHAPTER V  52 60  Patterns of Particulate Matter Concentrations and Hospitalizations in the Fraser Valley  61  5.1 5.2 5.3 5.4  Introduction Annual Patterns of Particulate Matter Concentrations Annual Patterns of Respiratory Hospital Admissions Annual Patterns of Cardiac Hospital Admissions  62 62 66 70  5.5  Summary  73  C H A P T E R VI 6.1  Case Study of the 1998 Gobi Dust Event  74  Introduction  75  6.2 Respiratory Illnesses 6.3 Cardiac Illnesses 6.4 Summary CHAPTER V l l Conclusions  75 82 87 88  7.1 Introduction 7.2 . . Brief Surhmary of Chapters 7.3 Discussion and Final Conclusions 7.4 Future Research Directions Bibliography  '  89 89 90 91 ;  '.  93  LIST OF T A B L E S Table 2.1 Table 2.2  Table 4.1 Table 4.2 Table 4.3 Table 4.4  Table 6.1 Table 6.2  Summary of the combined estimated effects of daily mean particulate pollution on morbidity Examples.of human airborne pathogens commonly transported in dust or associated with exposure to dust clouds '. Hill's criteria used to determine causality between a pollutant exposure and an adverse health impact Major sources of confounding in ecological studies and techniques used in this research to minimize confounding Population by FSA postal code from the 1996 Statistics Canada Census.. Dates of the Gobi dust event and the corresponding dates of the meteorological analogue used in this research Calculations to obtain the average number of hospitalizations per day for respiratory illnesses Calculations to obtain the number of 10 ugm" increments above 20 |a.gm" for the 1-hour maximum P M i observed during the Gobi dust event Calculations to estimate the magnitude of effect on respiratory hospitalizations associated with the maximum 1-hour PMio observed during the Gobi dust event, based on the formula in Figure 6.1 Dates of the Gobi dust event and corresponding dates of the meteorological analogue used in this research Comparison of the estimates of the magnitude of effect on respiratory hospitalizations with actual rates of admissions during the Gobi dust event Calculations to obtain the average number of hospitalizations per day for cardiac illnesses Calculations to obtain the number of 10 ugm" increments above 20 ugm" for the 1-hour maximum PMio observed during the Gobi dust event Calculations to estimate the magnitude of effect on cardiac hospitalizations associated with the maximum 1-hour PMio observed during the Gobi dust event, based on the formula in Figure 6.1  18 30  50 51 56 57  76  3  3  0  Table 6.3  Table 6.4 Table 6.5  Table 6.6 Table 6.7  76  77 78  80 82  3  3  Table 6.8  82  83  v  LIST OF FIGURES Figure 1.1 Figure 1.2 Figure 1.3  Figure 2.1  Figure 2.2 Figure 2.3 Figure 2.4 Figure 2.5  Progression of the April 19 1998 Gobi dust cloud across the Pacific Ocean, derived from satellite data and imagery 5 Concentrations of selected ions and elements in the Gobi dust cloud for April 29 1998 (solid bars) at Rocky Point Park, Vancouver 6 Hourly P M | concentrations for stations east to west along the Fraser Valley axis for 0000 PST April 5 1998 to 0000 PST May 5 1998 7 0  Mortality rates around the time of the 1952 London Smogs, reproduced from the original report on the event by the U K Ministry of Health in 1954 13 Number of papers referenced in MEDLINE database for the search terms "air pollution" and "epidemiology" for 1970-2000 14 Schematic size distribution of particulate matter in the atmosphere and corresponding deposition in the human respiratory system 15 A summary of adverse health effects due to particulate matter exposure 17 Estimated percent changes in daily mortality associated with a 10 ugm" increase in PM\o (with 95% confidence intervals) for a number of cities 20 A summary of the effect of particulate matter pollution on public health 21 3  Figure 2.6  Figure 3.1  Satellite image of the Fraser Valley, British Columbia, showing patterns of urbanization 34  Figure Figure Figure Figure Figure  Design of a case-control study 46 Design of a cohort study 46 Design of a cross-sectional study 47 Major strengths of an ecological study design... 48 Structure of the B C L H D and context of the Hospital Separations Dataset used in this research 53 Map of southwestern British Columbia, Canada, showing the locations of urban centres aggregated into the study areas of Greater Vancouver and the Upper Fraser Valley 54 Allocation of urban areas in southwestern British Columbia to the three 'areas of interest' in this research 55 FSA postal codes allocated to the respective area of interest 55 Estimation of 'extra' hospitalizations due to PM from the Gobi dust event in southwestern British Columbia, as proposed by Vedal (1995). 57 Four-day time series of a/temperature, b/wind speed and c/wind direction for Chilliwack (Okanagan Valley) for the Gobi dust event (April 27-30 1998) and a meteorological analogue (May 12-15 1997) 58  4.1 4.2 4.3 4.4 4.5  Figure 4.6  Figure 4.7 Figure 4.8 Figure 4.9 Figure 4.10  vi  Figure 4.11  Four-day time series of a/ PMio at Chilliwack (Okanagan Valley), hi P M | at Abbotsford (Upper Fraser Valley), c/ PM . /PMio rations, d/ O3 and e/ CO concentrations for the Gobi dust event (April 27-30 1998) and a meteorological analogue (May 12-15 1997).. 59 0  Figure 5.1 Figure 5.2 Figure 5.3 Figure 5.4 Figure 5.5 Figure 6.1  2  5  Time-series of daily average PMio concentrations in the three study areas Time-series of PM2.5 for two of the study regions (PM2.5 data was unavailable for Metropolitan Vancouver) Time-series of hospital admissions for respiratory illnesses Age distribution of the population across the three study regions Time-series of hospital admissions for cardiac illnesses  63 65 67 69 71  Calculation used to estimate the impact on hospitalization rates associated with P M | , based on work by Vedal (1995) 75 Time-series of estimated extra hospital admissions for respiratory illnesses associated with the daily maximum 1-hour PMio for 1998. ... 78 Comparison of respiratory admissions during the Gobi dust event with admissions during the meteorological analogue from the previous year 79 Time-series of estimated extra hospital admissions for cardiac illnesses associated with the daily maximum 1-hour PMio for 1998. ... 84 Comparison of cardiac admissions during the Gobi dust event with admissions during the meteorological analogue from the previous year 85 0  Figure 6.2 Figure 6.3  Figure 6.4 Figure 6.5  vii  ACKNOWLEDGEMENTS First and foremost, grateful thanks to Prof. Ian McKendry for his endless encouragement, enthusiasm, guidance and support during the completion of my degree and this thesis. You are a role model for all great supervisors, and I have learnt a great deal from you imthe past 2 VJ years. I hope I can offer the same high quality supervision to students in the future. Sincere thanks to Dr. Shona Kelly, University of Nottingham, for her ongoing encouragement and guidance in the epidemiological aspects of this research. Your involvement has been greatly appreciated. I wish to thank Dr. Ken Denike for his assistance, particularly with the technical aspects of the analysis, and his willingness to act as second reader of my research. I would also like to acknowledge the participation of Dr. Tom Koch in the early stages of this research. Special thanks to my friends at Green College, who helped me maintain my sanity and provided lots of laughs throughout the last 2 I/2 years! Extra thanks are due to Kristina Llewellyn and Emma Cunliffe for their assistance with printing and submission of this thesis. Finally. I wish to thank my dear family for their endless love, support and encouragement as I pursued this degree half a world away. Thank goodness for phones, email and Messenger! I could not have gone^to UBC, and survived (!), without you all behind me, and for that I am immensely grateful. My time at U B C has been incredibly fulfilling and I have gained immeasureably from my experiences in Vancouver. I wish to dedicate this thesis to my Grandma, who.passed away during my'time at U B C . She was an inspiration to me, and always will be.  viii  CHAPTER I Introduction and Research Context  1.1 Introduction The long-distance transport of dust around the globe is a well documented phenomenon. North America is certainly not immune from the effects of these dust events on local air quality, with the southeastern states frequently experiencing elevated particulate matter concentrations due to dust from the African continent. However, a large dust event in April 1998 that originated in the Gobi desert, China, was the first documented case of Asian dust reaching the west coast of North America. This chapter will provide the contextual setting for this research. First, there is a brief discussion of dust in the atmosphere in general, with the two major dust sources of the world examined. Asian dust is, naturally, examined in more detail than African dust. The transPacific transport of Asian dust is also examined. Next, there is a description of the 1998 Gobi dust event, which is the focus of this research. Lastly, the research question is identified and discussed, and the rationale for this research is presented.  1.2 Research Context: Dust in the Atmosphere Frequencies of dust emissions and mechanisms of dust entrainment on a global scale are well addressed in Gillette (1999), Goudie & Middleton (1992) and Prospero et. al, (2002). In particular, Prospero et. al. (2002) examines more than 23 major dust generation regions and emission rates across the globe, with significant geophysical, climatological and anthropogenic factors taken into consideration. At the global scale, dust mobilization is dominated by natural sources. The largest and most persistent sources of atmospheric dust are in the Northern Hemisphere in a broad 'dust belt' across the west coast of Africa, the Middle East, Central and South Asia and into mainland China, with remarkably little major dust activity outside this region (Husar et. al., 1997; Prospero et. al., 2002). Some estimate that up to two billion metric tons of dust are lifted into the atmosphere every year (Griffin et. al., 2002). The African continent is a major global dust source, and often regarded as the strongest dust source in the world, contributing an estimated one billion tons of dust per year to the atmosphere (Griffin et. al., 2001 ; Prospero, 2001). African dust has been documented in the eastern United States (especially Florida) and the North Atlantic Ocean (Griffin et. al., 2001 ; Moulin et. al., 1997; Perry et. al., 1997; Prospero, 1999; Prospero, 2001), the Caribbean (Griffin et. al., 2001 ; Husar et. al, 1997; Prospero, 2001), Mediterranean (Caquineau et. al., 2002; Moulin et. al, 1997; Moulin et. al, 1998; Ozsoy et. al, 2001; Prospero, 2001), Europe (Ansmann et. al, 2003; Moulin et. al, 1998; Prospero, 2001), North and South America (Moulin el. al, 1998; Prospero, 2001), the Indian Ocean and Arabian Sea (Husar et. al, 1997), the Middle East (Alpert & Ganor, 2001; Moulin et. al, 1998; Prospero, 2001) and across the Pacific Ocean (Husar et. al, 1997). The characteristics and mechanisms of African dust entrainment and transport are examined in Swap et. al (1997) and Moulin et. al. (1997). a  a  b  2  1.2.1 Asian Dust The second major global dust source is the desert regions of Asia, especially China and Mongolia, with an estimated dust emission of 21.5 tons per square kilometer (Xuan et. al, 2004; Zhang et. al, 2003 ; Zhao et. al, 2003). Other desert regions in Tadzhikistan, Afghanistan, Iran, Pakistan and India are also significant contributors (Golitsyn & Gillette, 1993; Guo et. al., 2004; Middleton, 1986). c  The dominant season for Asian dust emissions is spring (March to May) (Duce et. al, 1980; Merrill et. al, 1989; Murayama et. al, 2001; Prospero et. al, 2002; Uematsu et. al, 1983; Zhao et. al, 2003), with some observations of Asian dust events in fall and winter as well (Kim & Park, 2001; Lee et. al, 2004). Prevailing westerly monsoon winds with intense frontal activity during spring provides a mechanism for the injection of surficial materials into the lower and middle troposphere (Zhao et. al, 2003). Annual dust emissions from central Asia are estimated to contribute some 43 million tons per year, with spring-time emissions accounting for half of this (Chen et. al, 2004 ). Merrill et. al. (1989) provides a comprehensive analysis of Asian dust storms and long-distance aerosol transport using data from 750 active weather stations in China in addition to sites scattered across the North Pacific region. Spatial and temporal patterns of dust emissions and meteorology are examined, and long-distance transport is assessed using back trajectories. Asian dust characteristics are examined in depth in Gao & Anderson (2001), Gomes & Gillette (1993); Sun et. al. (2004), Xuan et. al. (2004) and Zhang et. al. (1998), and climatic controls on Asian dust storm frequency are examined by Gong et. al. (2004), Littmann (1991) and Zhang et. al (1997). a  Asian dust events that have been transported across Asia have occurred with enough frequency to have acquired local names: 'huangsha' in China, 'whangsa' in Korea and 'kosa' in Japan. The Taiwanese refer to these dust episodes as 'yellow sand events' (owing to the characteristic yellow colour of the dust) (Husar et. al, 2001; Lin, 2001). China has some of the earliest written records of dust storm activity in ancient Chinese literature, which refers to 'dust rain', 'dust fog' and 'yellow fog' - the first documented 'dust rain' was recorded in 1150 BC (Goudie & Middleton, 1992). A study of dust sources and transport pathways in Asia during spring 2001 found that the five major dust transport pathways all pass over Beijing (Zhang et. al, 2003°), and the recorded instances of dust storms in Beijing has grown dramatically over the last 50 years "from 5 times in the 1950s, 8 in the 1960s, 13 in the 1970s, and to 14 in the 1990s. In 2000, more than 12 severe dust storms occurred in North China" (Lin, 2001, pp5873). Between 45% and 82%> of the observed mass balance calculations in the 2001 study was attributed to Asian soil dust particles, which exhibited elemental contents that were high in silica, calcium, iron and aluminium (Zhang et..al, 2003 ). Approximately 30% of the total dust loadings was larger than 16 urn and 1.7% was less than 2.5 pm (Zhang et. al, 2003 ). Frequent yellow rains in China suggest that much of this larger dust load is rained out of the atmosphere, or is removed by gravitational settling, relatively close to the source region (Zhao et. al, 2003), with only the finest fraction of the dust load being entrained in the atmosphere and thus subject to long-distance transport processes. c  c  Due to proximity to the major Asian source regions, transport of desert dust throughout eastern Asia is particularly well documented, with episodes of dust transport and characteristics in China examined by Guo et. al. (2004 ), Guo et. al. (2004 ), Sugimoto et. al. (2003), Sun et. al. (2004), Wang et. al. (2004), Zhang et. al. (2001), Zhang et. al. (2003 ) and Zhou et. al. (2002). Asian dust typically takes only 2-3 days to reach Taiwan, as discussed by Chen et. al. (2004 ), Chen et. al. (2004 ), Fang et. al. (2002) and Lin (2001). Korea frequently experiences Asian dust events, as evidenced by Chun et. al. (2001a), Chun et. al. (200 l ), Choi et. al. (2001), Chung et. al. (2003), In & Park (2002), Kim & Kim (2003), Kim & Park (2001), Kim el. al. (2003), Kwon et. al. (2002), Lee et. al. (2004) and Yeo & Kim (2002). Japan is also subject to regular Asian dust episodes (Ma et. al, 2001; Uematsu et. al, 2002; Uno et. al. , 2001; Zhang et. al, 2003 ). a  b  b  a  b  b  a  Transport of Asian dust across the North Pacific is routinely observed, particularly between February and May, and has been extensively documented and modeled over the last few decades (Braaten & Cahill, 1986; Duce et. al, 1980; Gong et. al, 2003; Holzer et. al, 2003; Husar et. al, 2001; Lin, 2001; Liu el. al, 2003; Merrill et. al, 1989; Parrington et. al, 1983; Prospero et. al, 2002; Shaw, 1980; Wilkening et. al, 2000; Zhao et. al, 2003).  1.2.2 Transport of Asian Dust to North America Air pollution in northwestern North America has generally been considered to be of regional scale and local origin (Hacker et. al, 2001). However, there is growing evidence that incursions of pollutants from the Asian continent are impacting air quality in this region as well (Berntsen et. al, 1999; Hacker et. al, 2001; Holzer et. al, 2003; Husar et. al, 2001; Jaffe et al. 1999: McKendry et. al, 2001; 'Zhao.% Hopke, 2004). Asian dust has also been observed in Alaska and the North American arctic (Cahill, 2003; Duce et. al, 1980), and as far away as New England in the northeastern United States (DeBell et. al, 2004). There now seems to be a growing concensus that episodic transport of pollutants from the Asian continent to the west coast of North America is a relatively regular occurrence, particularly during the springtime (Jaffe et. al, 2003 ; Sassen, 2002; Simpson et. al, 2003). Pollutant outflow from Asia is highest in the spring and winter, and trans-Pacific transit times during these seasons is typically 5-10 days, depending on altitude and meteorology (Wilkening et. al, 2000). a  There is significant diversity in these episodes, with respect to the pollutants involved (including mineral dust, industrial emissions and various mixes thereof), transport mechanisms and trans-Pacific pathways (Jaffe et. al, 2003 ). Six previously unreported episodes of pollutant transport from Asia to the west coast of North America between 1993 and 2001 have been examined by Jaffe et. al (2003 ). The 1998 Gobi dust event (examined in this research) and another Gobi dust event in 2001 were composed of mineral dust, yet other episodes documented in Jaffe et. al (2003 ) were primarily composed of industrial emissions. Anthropogenic species (carbon monoxide, ozone, peroxyacetyl nitrate, radon and non-methane hydrocarbons) from east Asia have previously been documented in North America (Jaffe el. al, 1999). a  a  a  4  13 The 1998 Gobi Dust Event in North America In April 1998, a dense cloud of Gobi desert dust was generated by frontal activity over the Gobi desert in western China during an abnormally strong springtime storm on April 19 and was transported across the Pacific Ocean in approximately five days, where it subsided along the western coast of North America (Fig 1.1). Strong subsidence over the interior of southern British Columbia carried the Gobi dust cloud towards the surface, where it was entrained in surface easterly winds that transported the dust across the interior regions towards the west coast (Husar et al.. 2001; McKendry et al. 2001; Murayama et. al, 2001; Tratt et al, 2001). More comprehensive description and analysis of the Gobi dust cloud can be found in Lin (2001), Murayama et al. (2001) and Uno et al. (2001). The meteorological aspects of this Gobi dust event are well documented in Husar et al. (2001) and McKendry et al. (2001), and modeling of the transport and subsidence of the Gobi dust cloud is well examined in Hacker et. al. (2001). McKendry et. al. (2001) and Uno et. al. (2001). C N N reported that this dust event killed at least 12 people in the Xinjiang region, China, and yellow muddy rain was reported across eastern China and Korea (Husar et. al, 2001).  Fig 1.1. Progression of the April 19 1998 Gobi dust cloud across the Pacific Ocean, derived from satellite data and imagery (Source: Husar et. at., 2001, ppl8324). This dust event was first detected in North America on April 24 by lidar in Salt Lake City, Utah, with the main dust cloud arriving the following day. The dust cloud covered the entire west coast of North America from California to British Columbia, with a wedge-shaped area extending well into the centre of the North American continent (Husar et. al, 2001). British Columbia. Washington. Idaho and Oregon issued air pollution advisory warnings to the general public and banned open prescribed burning as well (Husar et. al, 2001). Unusually high particulate matter concentrations (in the range of ~100p.gm" ) and an intense haze were recorded from California to British Columbia (Husar et al, 2001; McKendry et al, 2001; Tratt et al. 2001; Vaughan et. al, 2001). The dust layer severely impacted visibility in the boundary layer, reduced direct solar radiation by 30-40%, doubled the diffuse radiation and discoloured the sky along the west coast (Husar et. al, 2001; Tratt et. al, 2001). It was estimated that the dust event contributed between 38% and 55% of observed 3  5  PM]o concentrations at that time. Of particular note, analysis of the Gobi dust showed that the dust mode was 1-4 pm, with an.estimated 30-50% of the dust mass below 2.5 pm (Husar et. al, 2001; McKendry et. al, 2001). The Gobi dust cloud also contained high concentrations of crustal elements (particularly silica, aluminium, iron, calcium and potassium) (Husar et. al, 2001; McKendry et al, 2001; Tratt et. al, 2001), which is not frequently experienced in the context of North American urban air pollution (Fig 1.2). This event was the first documented example of the transport of mineral dust from Asia to British Columbia and the Pacific Northwest and the chemical signature of the dust cloud was detectable as far inland as Minnesota (Hacker et. al, 2001; Husar et al, 2001; McKendry et al, 2001).  a. C O A R S E M O D E - P M , . 10 2  5  8 7 6  3 2 I 0  jftL cx 2  o Vi  b . FINE M O D E - PM2.5 7 6 5  \  4  rt 3 2 I 0  O  <  0  Fig 1.2. Concentrations of selected ions and elements in the Gobi dust cloud for April 29 1998 (solid bars) at Rocky Point Park, Vancouver, compared to historical (1996-1998) 99 percentiles (extent of top whisker), means (horizontal bar) and standard deviation (extent of bottom whisker) for a) coarse mode particles (PIVh.s-io) and b) fine mode particles ( P M ) (Source: McKendry et. al., 2001, ppl8365). th  2 5  6  In British Columbia, PMio concentrations in the southern interior regions were dramatically elevated to above 100 ugm" on April 28 (Fig 1.3). Peaks in PMiq were recorded on April 29 further down the Fraser Valley in the Greater Vancouver region, and further west on Vancouver Island on April 30 (Husar et. al, 2001). This pattern is consistent with modeling of the event, which showed strong subsidence and downward mixing of the dust-laden airmass in inland regions, where it was then chanelled westwards through the Fraser Valley system (Hacker et. al, 2001; McKendry et. al, 2001). 3  140 120 100 ft 80 "s 60 n 40 20 0  \  K u m loops Kelowna  140 120 100 80 60 40 20 0  i  Chilliwack  1 1  j  \ (L  AA*  AJjfk  Rocky Point P a r k  IA  A  i i  Mrs]  I i  i  1  140 120 100 80 3, 60 40 20 0  i  i  Richmond Nanaimo  0:00 25/04/98  0:00 27/04/98  0:00 29/04/98  0:00 05/01/98  0:00 05/03/98  0:00 05/05/98  Date/Hour Fig 1.3. Hourly PM, concentrations for stations east (top) to west (bottom) along the Fraser Valley axis for 0000 PST April 25 1998 to 0000 PST May 5 1998 (dates are dd/mm/yy). Kamloops and Kelowna are in the Okanagan Valley study area for this research (also see Chapter 4), Chilliwack is in the Upper Fraser Valley, Rocky Point Park and Richmond are in Greater Vancouver, and Nanaimo is on Vancouver Island (which is not considered in this research) (Source: McKendry et. at., 2001, ppl8363). 0  7  1.4 Identification of the Research Question As discussed earlier in this chapter, there was a significant episode of naturally-derived particulate matter (PM) pollution in British Columbia in late April 1998, resulting from the long-distance transport of Gobi desert dust across the Pacific Ocean. This dust event was associated with elevated PM concentrations (Fig 1.3) and an intense haze in the urban regions of the Fraser Valley, British Columbia, which persisted for several days. Based on current literature that links P M exposure with adverse health effects, it is expected that this desert dust event will also be associated with adverse public health impacts. However, there has been no research to date to quantify this hypothesis. A connection could be made with the studies by Ostro et. al. (1999 & 2000), who found increases in mortality rates associated with P M exposure in an urban location that was dominated by coarse particles of geologic origin. It may be possible to identify a signal in hospital admissions for respiratory and cardiac illnesses that is associated specifically with P M exposure from natural sources. It is proposed to identify any such signal from the Gobi dust event in hospital admissions in the Fraser Valley, British Columbia, through a comparison of hospital admissions in April 1998 with a meteorological analogue from the previous year (a case study of anthropogenic P M pollution). Research to date has shown that despite a large body of literature on the health impacts of P M exposure, there is extremely little knowledge about the specific impacts of naturallygenerated PM. The;mean diameter of particles from the Gobi dust event was 2-4 um (Husar et. al, 2001; McKendry el. al, 2001). This strongly suggests that the dust event would be associated with adverse public health impacts, because the particles were in the fine fraction of PM that has been most strongly associated with adverse health effects (in comparison to coarser particles) - this will be discussed in more depth in Chapter 2. The Gobi dust event was also associated with a huge injection of crustal elements into the urban atmosphere, particularly silica, iron, aluminium and calcium (Fig 1.2), which could also have distinct signals with regard to adverse health impacts. This dust event was coincident with a period of photochemical smog in the Fraser Valley. As a result, attention has been drawn to disentangling the health impacts associated with the anthropogenic component of the pollution event - the photochemical smog - and the health impacts associated with the naturally-derived component - the Gobi dust. Ideally, the distinct spatial and temporal characteristics of the Gobi dust event will enable patterns of hospital admissions to be tied to the presence of the Gobi dust, rather than the more generalized, locally-generated photochemical pollution. The 1998 Gobi dust event period will be compared to a meteorological analogue in 1997, which also featured a major inversion and associated high urban air pollution concentrations of anthropogenic origin, in an attempt to disentangle the associated health impacts. It is also anticipated that the spatial and temporal dimensions of the dust event will be replicated in the analysis of adverse health impacts, with connections between the rates of hospitalizations and the progression of the dust cloud down the Fraser Valley.  8  Therefore, the 1998 Gobi dust event and the identification of a meteorological analogue in the previous year, presents a 'natural experiment' scenario and offers a unique opportunity to investigate the-adverse health impacts associated with a naturally-derived P M episode in an urban setting'.  1.5 R a t i o n a l e There are three major factors guiding the development of this research. Firstly, there is a distinct scarcity in the literature on the health effects of naturally-derived particulate matter. There is a significant body of research on the health effects of particulate matter in general that has predominantly been based on the examination of anthropogenically-derived particulate matter. This literature is extremely relevant, as urban air pollution is often controlled to a large extent by local-scale sources and processes, and based on this evidence, it is expected that the naturally-derived Gobi dust event will also be associated with similar adverse public health impacts as an anthropogenic pollution episode. However, naturallyderived particulate matter (from dust storms and bush fires, for example) can also have a significant effect on urban air quality and, potentially, public health. Secondly, although Gobi desert dust reaching North America is a relatively rare event, desert dust storms are a seasonal phenomenon in many arid regions of the world. The growing rate of industrialization in central Asia and increasing desertification across arid regions of the globe could result in more frequent long-range transport of desert dust and other pollutants in the future, thus potentially posing a significant threat to public health (Murayama et. al, 2001; Tratt et. al, 2001; Uno el. al, 2001). Thirdly", a unique natural event provided the opportunity to investigate the effects of naturally-derived particulate matter on public health. The 1998 Gobi dust event could be regarded as an ideal case study: the dust event was very clearly defined in both spatial and temporal dimensions, and was closely monitored from its origins in central Asia to its arrival at surface level in North America by a global research team (Husar et. al, 2001). The chemical composition of the dust was significantly different to typical North American particulate matter, containing a high concentration of crustal elements such as silica, calcium, aluminium and iron (McKendry et. al, 2001) and thus could be clearly differentiated against the backdrop of 'normal' particulate matter.  9  1.6 S u m m a r y Long-distance transport around the globe is a well-documented phenomenon, with the arid regions of Africa and Asia as the greatest sources of atmospheric dust. Asian dust incursions are frequently experienced in China and other nearby Asian countries, but the long-distance transport of Asian dust across the Pacific to the west coast of North America is a comparatively rare, and undocumented, event. The 1998 Gobi dust event in southwestern British Columbia provided a unique opportunity to investigate the long-range transport of Asian dust, and more specifically, to investigate the potential public health impacts of naturally-derived particulate matter.  10  CHAPTER II  2.1 Introduction Air pollution is commonly acknowledged as one of the 'diseases' of the modern city. Despite historical documentation of poor health epidemics at times of lowered air quality that date back to antiquity, little attention has been devoted to this aspect of air pollution until relatively recently. However, concensus between meteorological, environmental and epidemiological research suggests that air pollutants, especially particulate matter (PM), have a deleterious effect on health, and P M has been the focus of increasing interest and concern over recent decades. Air pollution by P M is considered to be primarily an urban phenomenon (Schwela, 2000), and, understandably, most urban air pollution studies have focussed on anthropogenicallygenerated PM. Yet many US counties exceed the EPA (Environmental Protection Agency) air quality standards due to episodes of wind-blown dust (Schwartz et al, 1999) - in other words, naturally-generated PM. This literature review will examine the adverse health effects widely attributed to anthropogenically-derived particulate air pollution, and introduce the possibility that naturally-generated PM may also have severe adverse health effects when it is transported into the urban environment by long-range transport mechanisms. This review examines the adverse health effects associated with particulate pollution, with a view to providing some tentative predictions of the possible adverse health effects associated with the 1998 dust event. To begin, there is a brief review of historical associations of particulate pollution and health. An overview of particulate pollution and health is provided before a description of the effects of particulate pollution on rates of morbidity and mortality, and the effects of P M on certain sub-groups of the general population. The characteristics of P M are then discussed with respect to their associations with adverse health impacts. The role of naturally-generated PM in the urban context is introduced and followed by a summary of gaps in the literature that are being addressed with current research.  2.2 Historical Accounts of Particulate Pollution and Health Impacts Records of adverse health effects due to P M in the air can be traced back to antiquity. Numerous human remains from ancient Egypt show evidence of pneumoconiosis and sinusitis from wind-blown sand, which caused scarring of the lungs and respiratory passages. Similar findings have been recorded in the U K from remains dating from the Bronze to Middle Ages. The first documented links between air quality and health were made by the Romans, who noticed epidemics of illness often coincided with periods of poor air quality (Brimblecombe, 1999). More modern records of air pollution and health invariably point towards the infamous London Smogs of December 1952. These smogs were the product of coal burning, mainly in domestic fireplaces (Anderson, 1999; Harrison & Yin, 2000). In the two weeks following the smog event, mortality rose by an estimated 4000 deaths, or a 160% increase above 'normal' mortality rates, thus making this event a landmark in air pollution epidemiology (Anderson, 1999; Ayres,. 1994; Dockery, 1994; Harrison & Yin, 2000). Mortality rates rose on the day pollution levels rose, but there was also a lag effect of several days between the onset of the  12  smog and the distinct peaks in mortality rates (Figure 2.1) (Schwartz et al, 1999). Most of the fatalities were thought to have been people who were likely to already be facing a shortened life span due to pre-existing. ..disease, but a number of previously healthy individuals also died, some within hours-of exposure (Brunekreef. & Hoek, 2000). In addition, since death represents the most 'acute' or 'extreme' adverse health effect, it is reasonable to assume that many more people experienced less severe (non-fatal) adverse health effects during this event (Vedal, 1995).  WEEKLY NUMBER OF DEATHS REGISTERED 5000  4000  160 Greal Towns excluding Greater Lon<Jon  1952-53  A '"•••/  / \ / \  /  3000  19S2-53\  2000 • Greater London  <xu  ^  v  Corresponding week 1951 - 52  ~?<^  1000  INFLUENZA  FOG  1953  1952  1 '  1 1  <-  LIT  Week 18 25 1 a 15 22 29 6 13 20 27 3 10 17 24 31 7 14 21 28 7 14 21 28 ended Ocl November December January February March ATMOSPHERIC POLLUTION (mean values)  0.30  Parts per million sulphur dioxide  /  0.20  1  1952 - 53  / '  0.10  '  \  V  /  '  -'  '  m  ~  \\  \\ \  v  v  /  \  t B n d a a  i"  i  II  ' /  A  \  / \ / /  \ /  Week  '  V\ yv  .A  '  Jl\  >  - VT .r ~ ' ^ S  /  J  i  i  \  \\  Corresponding week 1951-52  i —I—I—I—i—i—I—J  /  /i—' i  t  >  — /  /  y\ \ \ V \ \ \y  \  i  18 25 1 8 15 22 29 6 13 20 27 3 10 17 24 31 7 14 21 28 7 14 21 28 November December January Rihniam 0  e  l  Fig 2.1. Mortality rates around the time of the 1952 London Smogs, reproduced from the original report on the event by the UK Ministry of Health in 1954 (Source: Anderson, 1999, p467).  In London at the time of this smog episode, the winter average of emergency bed requests for acute (non-infectious) cases was approximately 150-180 admissions per day. During the smog, the peak daily emergency bed request rate was 492 on December 9, the fourth day of the smog episode and the day immediately after the peak black smoke concentrations. This was an increase of more than 150% above 'normal' emergency bed request rates. Most of 13  these additional admissions were for respiratory diagnoses, which increased by a factor of four, and cardiac disorders also tripled. In addition, the weekly average acute respiratory hospital admissions was 750 for the week preceeding the smog episode - this was elevated to 1110 admissions on December 9, of which 460 were for respiratory disease, and the weekly rates of respiratory admissions did not return to pre-smog-episode levels until December 25 that year (Lipfert, 1993). For comparison: 'high' urban pollution events today usually record peak concentrations of P M in the region of 100-300 ugm" . The London Smogs recorded peak concentrations of black smoke and SO2 of 4460ugm" , with average concentrations of around 1600u.gm" (approximately five times the normal level) over the four days of highest pollution (Anderson, 1999; Dockery, 1994; Lipfert, 1993). There is considerable evidence from contemporary studies to show that even the lower concentrations of particulate pollution that many modern-day cities experience are producing the same effects on health, only at much less dramatic scales than earlier P M episodes like this (Vedal, 1995). 3  2.3 Overview of the Health Effects of Particulate Matter Pollution There has been a significant increase in the number of studies published about particulate air pollution and health effects in the past decade. This is primarily due to the increasing realisation that adverse health effects can occur at relatively low levels of pollution, in combination with advances in statistical analysis techniques, software applicable to epidemiological studies, and better access to data on health, meteorology and P M (Figure 2.2) (Pope & Dockery, 1999; Samet, 2002).  400  V  1  350 300 250 i 200  -O zs IZ  150  /  i  100 50  0  Year of publication Fig 2.2. Number of papers referenced in MEDLINE database for the search terms "air pollution" and "epidemiology" for 1970-2000 (Source: Samet, 2002, pi 19).  P M is of major importance to health because the particles are small enough and present in sufficient concentrations to by-pass the body's natural defense mechanisms (such as nasal hair and mucous membranes) and be transported directly into the air passages and lungs. 14  PMio (particles less than 10 pm in diameter) has a high probability of being deposited in the bronchial regions of the lungs and upper airways (Figure 2.3). More importantly, PM2.5 (particles less than 2.5 pm in diameter) is small enough to penetrate deep into the alveolar region of the lung (Figure 2.3), where air exchange occurs, and experiences a prolonged retention time within the lungs before expulsion, thus increasing its relative toxicity. Some PM is so fine that once it is inhaled into the lungs, it cannot be exhaled (Griffin et al, 200l ; Griffin et al, 2002; Loomis, 2000; Schwartz et al, 1996 ; Schwartz et al, 1999; Vedal, 1995). Advances in technology now permit measurement of P M as fine as P M (particles less than 1 pm in diameter), and it is hoped that this measure will be used more routinely in the future. a  a  t  0. ")0l  •:). 1  0,01  *  Ultraikie  1  10  1  100  Coaise  Fin©  C  100 80-  1(,r;i«  y' -  UiiM<H =  60-  fr 4 0 20 •  u  '  \  vs  i  301  0.01  / """"""  0,1 Paris  hlm.^V, '  / f  '"^1  1  10  10  ml  Fig 2.3. Schematic size distribution of particulate matter in the atmosphere and corresponding deposition in the human respiratory system (Source: Fenger, 1999, p4884).  The main sources of PM in the urban environment are by-products of human activity - from industry, fuel combustion processes, vehicle emissions and home heating (especially wood burning and coal fires) (Beer, 2001; Fenger, 1999; Holman, 1999; Lighty et al, 2000; Vedal, 1995). The major fraction of PM10 is thought to be more attributable to natural origins, such as dust storms and sea spray, in addition to dust from roadways, with the finer fractions (PM2.5 and smaller) produced primarily by combustion processes (Fenger, 1999; Lighty et al, 2000; Schwela, 2000; Schwartz et al, 1996 ). a  15  PM is also of concern for reasons other than health effects in the urban environment. A n obvious effect of high particulate pollution is the reduction of visibility and the presence of a noticeable haze in the air. Events like this can often carry particles that are larger than P M | , which can be easily detected as gritty dirt and dust on urban surfaces, such as vegetation, vehicles and houses. PM may also play a role in modifying the urban radiation budget through the reduction of solar radiation, therefore affecting surface temperatures and humidity and, thus, potentially have an impact on local climate change and global warming processes (Fenger, 1999). 0  2.4 The Effects of Particulate Matter on Morbidity and Mortality Despite the profusion of studies produced in the past decade on P M and health, there have been remarkably consistent results shown, with increases in both morbidity and mortality rates in the general population during episodes of elevated P M pollution. In spite of all the differences between these studies, the size of the effect on health (statistically speaking) is similar. These increases also seem to be relatively independent of other pollutants (Loomis, 2000), with PM been shown to be responsible for more multiple adverse health effects than gaseous pollutants (Vedal, 1995). Although there are multiple adverse health effects associated with PM pollution, the majority of effects fall into the broad categories of respiratory and cardiac illnesses, and range from relatively minor irritation to hastened death. The weather (in "general) can have a confounding effect on health statistics in epidemiological studies (Vedal, 1997) and thus should be mentioned before the discussion of current research. For example, very cold temperatures can induce asthma attacks and exacerbate bronchitis, very warm temperatures can increase blood pressure and heart strain and, thus, heighten the risk of heart attacks, and windy weather can trigger numerous allergies from airborne particles. In addition, population health is also strongly affected by epidemics of common respiratory conditions, such as the common cold, influenza and viruses (Braga et al, 2000). But, a meta-analysis of air pollution and mortality rates showed that the relative mortality risk was identical in cities with above average annual temperatures, in colder climates, in drier and in more humid environments, and the risk was similar across a wide range of temperature and particle concentration correlations (Schwartz, 1994). This indicates that the observed relationship between pollutant and mortality may be partially, but not completely, explained by the general climate of the study site. In addition, Braga et. al. (2000) found that the association between air pollution episodes and daily mortality rates was not due to a failure by researchers to control for influenza or pneumonia epidemics.  16  2.4.1 Effects ofPMon  Morbidity  In most otherwise healthy individuals, effects on morbidity are small, sometimes almost immeasureably so, and relatively minor, such as coughing and shortness of breath, and increases in the typical duration of such illnesses. Figure 2.4 illustrates the expected impact of P M on a population, with the area of each segment roughly corresponding to the proportion of the population that would be expected to be adversely affected in that manner. Thus a large proportion of the healthy population would be expected to experience no or few adverse health effects due to PM exposure. As exposure increases, or P M concentrations rise, or an individual's health is compromised through pre-existing disease, the health effects experienced will likely become more adverse and thus correspond with higher segments of the pyramid. Mortality is regarded as the most adverse possible health outcome (Vedal, 1995). Mortality, and other more severe adverse effects, is usually associated with the exacerbation of pre-existing disease in vulnerable persons and can occur at relatively low levels of PM concentration (Ayres, 1994; Osunsanya et al, 2001).  Fig 2.4. A summary of adverse health effects due to particulate matter exposure. The area corresponding to a health effect roughly corresponds to the proportion of the population expected to be adversely affected (Source: Vedal, 1995, ppl2).  The effects of air pollution on the respiratory system include "acute and chronic changes in pulmonary function, increased incidence and prevalence of respiratory symptoms, sensitization [sic] of airways to allergens, and exacerbation of respiratory infections, such as rhinitus, sinusitis, pneumonia, alveolitis, and legionnaires' disease" (Schwela, 2000, pi8). The most common effects reported are shortness of breath, coughing, reductions in F E V (forced expiratory volume) and PEF (peak expiratory flow), and the exacerbation of symptoms of COPD (Chronic Obstructive Pulmonary Disease), asthma and bronchitis (Bates 17  et. al, 1990; Brauer et. al, 2001; Brunekreef et. al, 1997; Desqueyroux et. al, 2002; Gehring et. al, 2002; Jaffe et. al, 2003 ; Just et. al, 2002; Koenig, 1999; Koenig et. al, 1993; Lee et. al, 2002; Lipsett et. al, 1997; McGowan et. al, 2002; Mortimer et. al, 2002; Pope et. al, 1991) . Table 2.1 summarizes the most widely reported respiratory symptoms associated with PM exposure and the percentage increase above 'normal' reporting rates for each 10 ugm" increase in PMio. Effects on the cardiovascular system can strongly influence incidence rates of acute illness in particular, with increases reported in numerous cardiovascular diseases, cardiac arrythmia, myocardial infarction, and the consequent increase in mortality due to cardiovascular ill-health (Brauer et. al, 2001; Schwela, 2000; Tolberte/a/.,2000). b  3  Table 2.1. Summary of the combined estimated effects of daily mean particulate pollution on morbidity (Source: Dockery, 1994, p51). Health Measure (Number ofStudies)  "  "n ( lisiiigv for each I0|i«m increase in PM,,, J  Increase in hospital usage (respiratoi\) Admissions (3) Emergency Department Visits (3)  0.9 1.0  Exacerbation of Asthma *\ Asthmatic Attacks (3) Bronchodilator Use (3) Emergency Department Visits (1) Hospital Admissions (2)  3.0 2.9 3.4 1.9  Increase in Respiratory Symptoms Reported Lower Respiratory (6) Upper Respiratory (6) Cough (6)  3.0 0.7 2.5  .Decrease in Lung Function, :• Forced expired volume (4) Peak expiratory flow (6)  0.15 0.08  Irritation of the sinuses, eyes (especially conjunctivitus), nose, throat and other mucous membranes is frequently reported, as are allergic reactions resulting in skin irritations and the exacerbation of excema (Griffin et al, 200l ; Loomis, 2000; Schwela, 2000; Wolf, 2002). Exposure to particulate air pollutants have been shown to also adversely affect the immune system, sensory system, and the central and peripheral nervous system (Osunsanya et al, 2001; Pope et al, 1995; Schwela, 2000; Tolbert et al, 2000). Some recent studies have suggested PM pollution may be linked to some cancers of the respiratory system, as well as skin cancers, leukaemia, some auto-immune diseases and possibly even renal disease (Griffin et al, 200l ; Nyberg et. al, 2000; Schwela, 2000; Vedal, 1995). Links have also been made between PM exposure and pre-term births (Ritz et. al, 2000), low birth weights (Wang et. al, 1997), and infant mortality (between 1 month and 1 year of age) (Bobak & Leon, 1992; Loomis el. al, 1999; Woodruff et. al, 1997). However, the bulk of literature at present indicates that the most common adverse health effects of P M exposure concern the respiratory and cardiac systems and function. a  a  18  The research discussed so far has used hospital and emergency department admissions data to compile statistics on health impacts. But if the majority of the population, who is otherwise in good health, is not expected to be severely adversely affected by exposure to PM, then these studies may not be identifying the true magnitude of health impacts, especially in the primary health care sector. The first port-of-call for health care for someone who is unwell to a relatively minor extent will usually be their general practitioner, not the local hospital. A study by Hajat et al. (2002) examined the number of general practitioner consultations for correlations with PM pollution events in London, and found that consultation rates increased by 10.2% in persons 65 years and over, 5.7% in adults (aged 15-64) and 3.5% in children as PMio concentrations increased, and that the increases often correlated with relatively low ambient PMio concentrations. In addition, the results were largely attributed to wintertime consultation rates for the elderly and children, and summertime consultation rates for adults, which implies a relationship between seasonal fluctuations in the composition of P M with regard to the magnitude of adverse health impacts. Studies of the rates of prescription for respiratory drugs also provide valuable information, showing increases in the rates of asthma medication use (Pope et. al, 1991) and bronchodilator use (Delfino et. al, 1998). In addition, many people also cope with relatively minor adverse health effects by restricting their activities or taking sick leave from work or school (Stieb et. al, 2002 ; Vedal, 1995). b  2.4.2 Effects of PM on Mortality P M is the silent killer in modern cities - its effects are often not obvious or considered of enough magnitude to be important. The original report published on the London Smogs of 1952 mentions that the unusually high mortality rates were not immediately noticed, as opposed to the heavy demand for hospital beds. Thus the event was reported to be a "supreme example of the way in which a metropolis of eight and a quarter million people can experience a disaster of this size without being conscious all the while of its occurrence" (UK Ministry of Health, quoted in Brunekreef & Hoek, 2000, p449). Nowadays, however, the relative risks of PM exposure are in the order of a few percentage points increase in mortality rates (Brunekreef & Hoek, 2000). Current literature consistently shows that P M is positively and significantly associated with increases in mortality rates, particularly during episodes of extremely elevated P M concentrations, and PM exposure is considered to make a strong contribution to mortality rates (Dockery et al, 1993; Pope et al, 1995; Schwartz, 1995; Stieb'et. al, 2002 ). Increased mortality due to PM exposure seems to start at concentrations above 20ugm" , although there is no evidence to say that many adverse effects are not also occurring below this concentration in certain individuals (McClellan, 2002; Vedal, 1995). In fact, Schwartz et. al. (2002) found a linear relationship between PM2.5 and'daily deaths down to concentrations of just 2 ugm" , and the association showed no signs of a threshold. It is important to note that mortality in this context is not generally regarded as the death of people who were previously healthy prior to exposure to the pollution event. Respiratory symptoms from PM exposure by themselves are generally not fatal, but when acting in combination with pre-existing diseases like asthma or COPD, can result in hospitalisation and even death (Loomis, 2000). Therefore, air pollution is thought to hasten death in those who were already chronically or terminally ill, and is thought to hasten death in the order of days to weeks and perhaps months, rather a  3  19  than years (Brunekreef & Hoek, 2000; Schwartz, 2000). Effects on mortality rates are generally short-lived and may be offset from, or lag, peak PM concentrations by several days (Katsouyanni et. al, 1996; Rahlenbeck & Kahl, 1996; Wong el. al, 2001). PM generally has the greatest effect on mortality attributed to respiratory illness, as well as a measureable effect on mortality attributed to cardiopulmonary disease and lung cancer (Dockery et al, 1993; Pope et al, 1995). Interestingly, Goldberg et. al (2002) found no association between daily mortality from congestive heart failure and ambient air pollution. Most research has reported mortality rate increases range from 0.5% to 18%> increases over typical death rates. For every lOpgm" increase in PMio concentrations, mortality is thought to increase by approximately 1%, with studies reporting increases ranging from 0.18% to 1.6% per lOpgm" increase in P M , (Dockery, 1999; Harrison & Yin, 2000; Kaiser, 2000; Ostro el al, 1999; Pope et al, 1995; Simpson et al, 2000; Vedal, 1995). Mortality due to 3  3  0  3 *  respiratory illnesses is thought to increase by 3.4% for each lOpgm" increase in PMio, and mortality due to cardiac illnesses is thought to increase by 1.4% for each lOpgm" increase in PMio (Fig. 2.5) (Vedal, 1995). Levy el. al. (2000) investigated between-study variability that could have contributed to these varying estimations of effect and concluded that "mortality rates increase on average by 0.7% per 10 pg/m increase in PMio concentration, with greater effects at sites with higher ratios of particulate matter <2.5 pm in aerodynamic diameter (PM )"(pl09). 3  3  25  4 A t: o  3 A  TO 2  Q  •n U) co 0  • ii ii  • **  o  rt x± Oi e © =b o w - - 5u o - =_ u 3 | 3 * i f 3  <n  :  =  C. a V) >k c o c a i ! l a 8 » ! i ; §  o / t « J < D C 3 ( . —  g 's a | U « o  CO Q.  bo J  m  ^= en  _  E o (b S S "0 S m t. o tt 5 . E8 £ g s 2 a » o i < 14  1  O o  a o  to  «  ° 01  2  ^3  Fig 2.5. Estimated percent changes in daily mortality associated with a lOugm" increase in PM 95% confidence intervals) for a number of cities (Source: Pope and Dockery, 1999, pp681). 3  ]0  (with  20  Although increases in mortality appear to be slight, one study of 90 cities in the US found that an average increase in mortality of 0.5% for every 10(xgm" increase in PMio added up to an estimated 60.000 'extra* deaths in the US per year (Kaiser, 2000). A 1998 study in Sydney, Australia, estimated 400 premature deaths were caused by PM in the urban airshed, and 2400 such deaths annually in Australia (Beer, 2001). In British Columbia, increases in PMio are estimated to cause 82 extra deaths per year, of which 24 are due to lung disease and 27 due to heart disease (Vedal, 1995). 3  However, any analysis of mortality statistics should consider the 'harvesting effect". Harvesting refers to the enhancement and temporal displacement of mortality statistics by the deaths of people who were already likely to die within a few days in any case (Schwartz, 2000; Schwartz. 2001), independent of the pollution episode. For example, someone who is terminally ill is probably already confined indoors, most likely in a medical care facility, and so is not exposed to ambient particulate concentrations outdoors. Figure 2.6 illustrates the expected relationship between the general population and mortality during a pollution episode. On any given day. it is assumed that there is a flux of people moving between the 'general population" and the 'pool of susceptibles' as a result of'normal' health fluctuations. For example, most of the population is generally healthy, but if an individual develops bronchitis or has an asthma attack, they move into the pool of susceptibles until the condition has cleared and their health returns to 'normal'. During an episode of high particulate concentrations, we expect that any deaths due to the pollution will be sourced from the pool of susceptibles - the pollution episode is not expected to induce premature death in people who are otherwise healthy. Therefore, if a small proportion of the population is expected to be adversely affected to the point of mortality, then we also expect a correspondingly larger proportion of the population to be less adversely affected and move into the pool of susceptibles. If the resulting statistics show that the increase in mortality rates is not associated with a corresponding increase in the flow of people from the general population into the pool of susceptibles (that is, be adversely affected to a less severe degree than death), then it is likely that harvesting of mortality statistics has occurred (Schwartz, 2000; Schwartz, 2001).  GENERAL POPULATION  Fig 2.6. A summary of the effect of particulate matter pollution on public health. The greatest flux should be between the general population and the pool of susceptibles, with the flux in mortality only affecting those in the pool of susceptibles. If the latter flux is greater than the first, or occurs in the absence of the first, then harvesting of mortality statistics is likely to be occurring (Adapted from Schwartz, 2001, p56).  21  The extent of harvesting in mortality studies to date is a major point of controversy with a valid theoretical basis but which is difficult to assess. Attempts have been made to quantify the impact, if any. of harvesting on mortality statistics, using regression analysis and statistical models (for more detail of these methodologies, see Schwartz, 2000; Schwartz, 2001 and Zeger el. al, 1999), with one study estimating that results can be biased by 10-30% due to harvesting (Zeger el. al, 1999). But Zeger et. al. (1999) reassessed mortality data from a study in Philadelphia (Kelsall et. al, 1997) and found that "the previously reported associations between air pollution indicators and mortality cannot be attributed solely to harvesting" (pi75). Schwartz (2000) examined harvesting at different time scales and found that not all deaths associated with air pollution were due to harvesting - that is, air pollution advanced the mortality of some people by a non-trivial amount of time, such as weeks to months, rather than just days. A later study by the same author (Schwartz, 2001) again confirmed these conclusions, finding that air pollution events did increase the size of the 'pool of susceptibles', thus accounting for higher mortality rates due to a larger risk pool, with some deaths being advanced by months or even years. Thus it appears that harvesting, whilst an important consideration in statistical analysis, does not appear to totally explain the associations between air pollution and mortality, indicating that air pollution can, and does, kill.  2.4.3 Effects of PM on Certain Subgroups of a Population The vast majority of research to date has focussed on epidemiological studies of the general population and the results have been discussed in the previous two sections. More recently, studies have begun to isolate the health impacts on certain subgroups of the general population, particularly those thought to be more vulnerable to adverse health impacts. The main rationale behind this change in emphasis has been the growing realisation that the magnitude of adverse health impacts appears to be related to characteristics of the population under examination, such as population age, ethnicity, underlying health and concomitant exposure to other pollutants (Loomis, 2000; Zanobetti et. al, 2000). The elderly have been singled out as potentially being more susceptible to adverse health effects from PM exposure (Vedal, 1995). Schwartz (1994 ) found that increases in PMio were associated with small increases in daily hospital admissions for pneumonia and COPD in persons 65 years and over, in Detroit, USA. Similar results were found in another study of pneumonia and COPD hospital admissions for persons over 65 years by the same author (Schwartz, 1994 ) in Minneapolis-St. Paul, Minnesota, USA. Cardiovascular mortality in the elderly has also been significantly associated with PMio, PM2.5 and PM10-2.5 (Mar et. al, 2000). a  h  Children have also been used as a subject group for health effects research. One study in the US examined rates of respiratory disease in children during a period when a steel mill adjacent to a school closed for a winter due to a strike, and found that hospital admissions of children for bronchitis and asthma reduced by a factor of three during the winter of the steel mill strike, compared to non-strike winters (Dockery, 1994). Exposure to urban P M concentrations and the respiratory health of children was also studied by Schwartz & Neas (2000), who used diaries completed by parents to record respiratory symptoms on a daily basis (such as cough, wheezing, tightness in the chest and asthma). This study found the 22  association of respiratory symptoms and P M pollution was strongest for fine particle air pollution (PM2.5), especially for lower respiratory symptoms and cough. Significant increases in upper respiratory illness in children due to P M exposure have yet to be demonstrated (Hajat et al, 2002). Investigations of pediatric lung function and respiratory health with respect to PM have shown positive associations (Brunekreef et. al, 1997; Koenig et. al, 1993), and asthmatic children have been particularly well represented in the literature so far, with consistently positive associations found (Heinrich et. al, 2002; Just et. al, 2002; Lee et. al, 2002; Mortimer et. al, 2002). Infants have also demonstrated adverse effects due to P M exposure in a study by Gehring et. al. (2002). Respiratory infections have been shown to lower resistance to P M and enhance the susceptibility of a person to adverse health effects as a result of P M exposure. Zanobetti et. al (2000) found that hospital admissions associated with increases in air pollution for cardiovascular diseases (COPD, pneumonia, asthma, heart failure) almost doubled in patients with concurrent respiratory infections, with no variation by age, sex or ethnicity. Long et. al (1998) actually found that respiratory symptoms associated with P M were more pronounced in female subjects, but there appears to be scant supporting evidence to support and explain this finding.  2.5 Important Characteristics of Particulate Matter with Regard to Health Impacts Prior to the early 1990's, PM was thought to be relatively innocuous in relation to health effects, except when in the presence of S 0 . This theory was based on the assumption that P M only acted as a transport agent to transfer sulphur dioxide on its surface deep into the lung tissues (Harrison & Yin, 2000). The repeated demonstration of the adverse health effects of PM in the absence of SO2. and at concentrations previously considered 'safe', neccessitated a complete turnabout in the theories of air pollution toxicology from the early 1990's onwards (Harrison & Yin, 2000). 2  Research is now starting to focus on identifying which components of P M are most important when considering health impacts, as measures of ambient P M concentrations only represent the maximum possible dose, and may not effectively explain the observed health effects (Lighty et al, 2000). Numerous recent studies have established the link between P M pollution and health, but not the mechanism of effect (Harrison & Yin, 2000). As Kaiser (2000) states, "although many questions remain about how fine particles kill people,...there's no mistaking that P M is the culprit" (p22). It is currently unclear whether particle concentration, composition and/or size has the most deleterious effect on human health, and at what scales such an effect is observable. A common misconception is that minor adverse health effects only occur at the lowest P M concentrations, and the most severe or acute health impacts only occur at high concentrations (Vedal, 1995). In reality, the adverse health effects of P M pollution exposure depend to a significant degree on the individual's state of health at the time of exposure, with pre-existing respiratory or cardiac disease dramatically enhancing the likelihood of an individual being adversely affected to a significant degree by PM. Research has also shown that once a person 23  t  is sensitised to a known allergen, such as PM, small quantities of the allergen can elicit acute reactions (Griffin et al, 200l ). Thus a major concern is that there does not appear to be a threshold P M concentration, below which, adverse health affects disappear (Beer, 2001; McClellan, 2002; Ostro & Chestnut, 1998; Pope et al, 1995; Schwela, 2000). Therefore, adverse health effects may be occurring at relatively low P M concentrations and be occurring in places where air quality currently meets PM air quality standards. a  The chemical components of PMio are very diverse and "range from near neutral and highly soluble substances such as ammonium sulphates, ammonium nitrate and sodium chloride through sooty particles made up largely of elemental carbon coated in organic compounds, and essentially insoluble materials such as particles of clay" (Harrison & Yin, 2000, p87). However, these components represent different fractions of the P M load in different locations, and yet epidemiological studies show no obvious results that reflect this variation (Harrison & Yin, 2000). Particles produced by combustion processes seem to be more consistently pathogenic than crustal particles, but it is not clear whether this is due to the chemical characteristics of particles with relation to their generation, or whether it is due to the fact that combustion particles are typically in the fine fraction range of PM, whereas crustal particles are typically coarser (Schwartz et al, 1996 ; Vedal, 1997). Trace metals, particularly iron, copper and nickel, (present as metal ions in the form of salts and other compounds) can significantly influence the toxicology of P M and have been associated with increases in mortality (Burnett et. al, 2000; Harrison & Yin, 2000; Lighty et al, 2000). Most of the evidence comes from toxicological, rather than epidemiological, studies and is based on the premise "that metals are redox-active and can, therefore, induce or catalyse chemical change leading to the production of free radicals... which have a known ability to cause tissue inflammation" (Harrison & Yin, 2000, p87). In general, trace metal concentrations in the urban atmosphere have fallen dramatically in recent years, and more so than P M concentrations as a whole, but as yet there are no data to quantify if health impacts have correlated with the decreases in either trace metals or particle mass (Harrison & Yin, 2000). a  Epidemiological studies have shown that adverse health effects are primarily linked to PMio. The bulk of research on health impacts of P M has focussed on PMio simply because there are large datasets available (Harrison & Yin, 2000). Various studies have concluded that an increase in mortality of approximately 1% can be observed for each lOugm" increase in PMio concentrations (Harrison & Yin , 2000; Ostro et al, 1999; Vedal, 1995). Ostro et al. (1999 and 2000) and Villeneuve et. al. (2003) found that increases in mortality in urban areas were also associated with particles between 2.5um and lOum in diameter. However, most studies of urban P M | and PM2.5-10 are thought to be biased, as the fine particle fraction (<2.5um in diameter) often dominates P M | concentrations, and it is thought that fine particles have stronger adverse affects on health (in comparison to 'coarse' PMio particles). 3  0  0  In recent years, the case has been made that PM2.5 may be the component of PM responsible for more adverse health effects, primarily because this smaller size fraction is able to penetrate deeper into the respiratory system, and with far greater efficiency than particles of 2.5 - lOum in diameter (the remaining fraction of PMio) (Harrison & Yin, 2000; Vedal, 1997). Schwela (2000) stated that the observed effects of PMio on health are largely associated with the fine fraction of PMio, and less so with the coarse fraction. Mortality rates in particular are most strongly associated with fine P M (Dockery et al, 1993; Schwartz et al, 1996 ), and a study of coarse particle exposure in the US did not show any evidence of a  24  enhanced mortality rates during dust storms characterised • by high fractions of PMio, concluding that "the toxicity of coarse particles is substantially, less than that of fine particles" (Schwartz et al, 1999, p341). A study of emergency department visits for respiratory and cardiac conditions showed distinct increases due to P M exposure, with PM2.5 concentrations correlating with greater increases in visits for asthma, COPD and respiratory infections in comparison to the increases seen for P M | (Stieb et al, 2000). PM2.5 has also been shown to be more strongly associated with adverse health effects, including mortality, than PM2.5-10 (Burnett el. al, 2000; Loomis, 2000). In particular, a study by Schwartz et. al. (2002) stated that "the magnitude of association [between P M 5 and daily deaths] suggests that controlling fine particle pollution would result in thousands fewer early deaths per year" (pl025). 0  2  Toxicological studies have shown that ultrafine particles (less than 0.1pm) appear to have "considerably enhanced toxicity per unit mass and that their toxicity increases as particle size decreases" (Harrison & Yin, 2000, p88). This is thought to occur because ultrafine particles possess a greater surface area per unit mass for toxic components to reside on, and also have the ability to penetrate pulmonary organs at the cellular level, and enter the interstitial spaces and bloodstream (Donaldson & MacNee, 1998 in Harrison & Yin, 2000; Lighty et al, 2000; Osunsanya et al, 2001; Schwela, 2000). In particular, this theory is thought to explain increases in ischaemic heart disease symptoms during air pollution episodes, as ultrafine particles are able to penetrate the lung wall, inducing inflammation, which in turn stimulates the production of clotting factors in the blood and results in the exacerbation of ischaemic heart disease (Seaton et al, 1995 in Harrison & Yin, 2000). Of note is one study based on the exposure of rats to PM, which found that exposure to very small particles had a markedly greater impact on pulmonary inflammation than exposure to the same mass of P M of a larger size, and concluded that it was particle size that controlled the toxicity of P M pollution, rather than the inhaled mass or exposure concentration (Osunsanya et al, 2001).  2.6 Gaps in the Current Literature Future research on P M pollution and health effects seems to be heading in five major directions: 'personal' exposures vs. general population exposures, cumulative vs. one-off exposures, indoor vs. outdoor ambient concentrations, acute vs. lagged health effects, and transient vs. chronic health impacts. 'Personal' exposures to P M pollution depend on a person's individual lifestyle and activity patterns, rather than treating the exposure of an urban population to PM as homogeneous. We know that PM pollution levels fluctuate during the day as the result of factors like vehicle flow patterns, or cold weather spells increasing fuel consumption. Therefore, it is expected that a person who spends a large part of their day outdoors, especially during times of elevated P M concentrations, would experience a much greater dose of P M pollution (and, potentially, experience more adverse health effects), compared to someone who spends most of their day indoors, or far from sources of PM (Fenger, 1999; Schwela, 2000). Likewise, people who are already chronically or terminally ill, and therefore most at risk from P M exposure, are probably already inside, where ambient concentrations differ from ambient outdoor concentrations (Vedal, 1995). As such, outdoor ambient P M concentrations may not 25  be the most suitable measure for assessing personal exposures and thus, correlating exposures with health effects (Brauer et. al, 2002). This was a concern in this context, but recent research has not demonstrated any increased value from collecting individual-level exposure information, and so ambient P M concentrations were considered to be satisfactory representatives of exposure for this research (Brauer et. al, 2001). Cumulative exposures to PM pollution over a period of 1-2 months have been shown to be more harmful than shorter exposures of similar concentration, with respect to associated daily mortality rates (Brunekreef & Hoek, 2000; Vedal, 1995). This cumulative effect on health can also be reflected in the health of the general population, with more individuals being moved into the " 'pool of susceptibles' from which air pollution takes its toll" (Schwartz, 2000 in Brunekreef & Hoek, 2000, p450). There is also evidence emerging that long-term exposure to low concentrations of PM (starting at lOugm" ) is associated with reduced life expectancies, in the order of 2-3 years, in communities with consistently high P M concentrations (Schwela, 2000). 3  Indoor air pollution levels are currently under scrutiny, as indoor P M concentrations could be higher than outdoor concentrations, as 'dirty' air is trapped within the confines of the room or building. Indoor concentrations have been found to be approximately 80-90% of outdoor concentrations, and thus people spending long periods of time indoors experience much more sustained and repetitive exposures compared to outdoor exposure patterns (Schwartz, 1995). Indoor air pollution is also of considerable concern in developing countries where biomass fuels are burned for open-stove cooking and heating (Schwela, 2000). An important consideration is the temporal scale of adverse health effects - how long does it take for particles to react in the body to the point that adverse health symptoms are experienced? For example, mortality during the 1952 London Smogs did not peak until several days after the smog onset, even though mortality increases were seen from the first smog day (Schwartz et al., 1999). Therefore, new research is being aimed at detecting the lagged health effects of PM pollution, as well as the more obvious, acute, impacts. Finally, there is a need to examine the long-term effects of PM exposure on health. Are any, some or all adverse health effects transient and completely reversible, or is there some chronic health impact component? Schwela (2000) has already stated that P M pollution can cause permanent lung damage and pulmonary insufficiency in sensitised individuals. Heinrich et. al (2002) examined the epidemiology of respiratory symptoms in children in Germany over seven years of improving air quality, and based on the observed decrease in prevalence of symptoms, concluded that respiratory effects of P M in children are reversible.  2.7 The Role of Naturally-Derived Particulate Matter The vast majority of current research has focussed on the health impacts anthropogenically-sourced PM - P M that is generated in the urban environment activity. But there is another very significant contributor to urban air quality derived PM. Very few researchers have examined the role of naturally-derived P M health, and even fewer have considered its role in urban health impacts.  related to by human naturallyon human  26  A possible explanation for this scarcity of research is that large plumes of naturally-generated P M are invariably the result of extreme biochemical events, such as volcanic eruptions, largescale biomass burnings and dust storms, and as such, are relatively rare or infrequent occurances in large, urban areas. However, with increasing evidence of climate change, rapid deforestation and desertification of areas surrounding urban settlements, events of this type are becoming more and more frequent, and thus could pose a growing threat to urban areas, and associated public health, in the near future. Anecdotal evidence already indicates that desert dust clouds in Asian cities are associated with widespread ocular, dermatological and respiratory irritation, yet there has been little formal investigation of these effects to date. Volcanic eruptions in particular can produce vast quantities of extremely fine particles in a short space of time, which can spread over large distances rapidly, and can be highly acidic (and, thus, toxic to humans). For example, following the 1980 eruptions of Mount St. Helens, there was a four-fold increase in the number of emergency room visits for asthma, a two-fold increase in emergency room visits for bronchitis, and mild exacerbation of symptoms in people with other chronic lung disease (Baxter et al, 1983). Bush fires can impact urban P M concentrations due to advection of smoke plumes by the prevailing winds. Johnston et. al. (2002) found significant increases in asthma presentations at the Royal Darwin Hospital during the bushfire season of 2000 and concluded that PM from bushfires "should be considered as injurious to health as those from other sources" (p535). In contrast, Smith et. al. (1996) found that the 1994 Sydney bushfires were not associated with increases in asthma presentations to Sydney area hospitals. 'Naturally-derived' P M has also been defined as P M resulting from agricultural burnings, and health impacts studies have found the same symptoms have been reported following these events as those symptoms related to anthropogenic PM - cough, wheezing, shortness of breath, chest tightness and the exacerbation of asthma and bronchitis (Jacobs et. al, 1997; Long et. al, 1998). Quantities of naturally-derived dust in the atmosphere are significant. Estimates of dust transport into the atmosphere vary "from 500 million tons annually (from all the deserts combined), to as high as 1 billion tons annually from the Sahara and Sahel [deserts] alone" (Griffin el al, 200l , p22). The suspension of soil dust in the Northern Hemisphere is estimated to be around 150 million tons per year - and this amount doubles if contributions from the Sahara Desert are included. Salt is also an important contributor to P M in coastal areas (Pooley & Mille, 1999). a  2.7.1 Health Impacts, of Asian Desert Dust  - \  ,  There has been a huge body of literature published about dust storms -transport processes, dust volumes, dust cloud origins, deposition sites, and chemical composition, but remarkably few have investigated the potential adverse health effects associated with these dust events. Perhaps this is because relatively few large scale dust events impact large urban areas to a significant extent, or if they do, only very rarely. There is a small but growing body of literature that examines the impact of desert dust on urban areas in Taiwan and Korea, which are impacted on a seasonal basis by Asian desert dust. Surprisingly, there is a distinct scarcity of studies based in China, which, due to its proximity to the major source region of the Gobi desert, is impacted by significant quantities of desert dust every year. 27  Lei et. al. (2004) conducted a laboratory experiment that exposed rats to dust obtained from an Asian dust event in 2002, in order to assess the effect on lung inflammation. Exposed rats were found to have elevated lung inflammation markers in their blood stream (white blood cell counts and blood proteins), compared to control rats exposed only to room air, and these markers indicated a positive dose-response relationship. The mechanism of effect (particle size versus toxicity) was not investigated in this study. Chen et. al. (2004 ) studied the effects of Asian dust storms on daily mortality rates in Taipei City, Taiwan. 39 Asian dust events were recorded in Taipei for 1995-2000, with 24-hour average PMio levels of 126 ugm" +/- 34 ugm" across the dust event days. Comparison days, which were 7 days prior to, and 7 days following, a dust event, recorded average PMio 3 3 concentrations of 58 ugm" +/- 33 ugm" . O3 levels were also higher during dust events than on comparison days. Measurements of SO2, NO2 and CO were comparable for Asian dust event days and comparison days. A lag effect was determined, with lag length depending on the health effect: the highest positive association for respiratory mortality rate was one day following the dust event, with an observed increase of 7.66% above comparison day mortality rates. The highest associations for cardiac and non-accidental mortality appeared two days after the event, with increases of 2.59% above 'normal' rates on the second day. Increased risk for respiratory disease appeared one day after the event, and increased risk of cardiac disease was highest two days after the event. Overall, this research found that each 10 ugm" increase in P M | from an Asian dust event was associated with 0.72%o average allcause increase in the mortality rate, which is-eomparable to many previous papers which suggest an average 1% increase in total mortality per 10 ugm" increase. The methodology used in this paper, by design, controlled for long-term time trends, seasonal patterns and dayof-the-week effects. In addition, meteorological factors known to affect mortality, especially temperature, did not differ significantly between dust events and comparison days. Therefore, it was concluded that the patterns of mortality associated with the Asian dust were not confounded by the above factors (Chen et. al, 2004 ). Although none of the results were found to be statistically significant, the authors concluded that their work increased the likelihood that Asian dust was causally associated with increased daily mortality. d  3  3  3  0  3  a  Another Asian dust-focused study was published by Kwon et. al. (2002) for Seoul, Korea. This paper identified 28 Asian dust days between 1995 and 1998, with 24-hour average PMio concentrations during dust events of 101 ugm" , compared to 73 ugm" on control days (nondust days in spring where average pollutant levels and weather variables were comparable). Using Poisson regression analysis techniques, weak statistical associations were found for all-cause mortality rates within 3 days of a dust event, with slightly stronger associations for respiratory and cardiac-related mortality. Strongest associations were 3.7% increase in risk for cardiovascular and respiratory mortality on the same day as the dust event, and 2.1% increase in risk for the following two days. Associations were also found to be strongest for those people aged 65 and over, with 5.3%> increase in risk of mortality after a two day lag. Deaths for causes other than cardiovascular and respiratory conditions showed a negative association with the dust on the same day and with a one day lag. Again, the weak associations led to relatively inconclusive results, although the stronger associations with cardiac and respiratory mortality suggested a causal link between Asian dust and adverse health in Seoul.  28  One widely accepted theory states that naturally-derived-dust produces less adverse health effects than anthropogenic pollutants of the same size, on the basis that naturally-derived dust is composed of less toxic components. Naturally-generated dust is often very high in crustal elements, particularly silica. Silica, found in quartz sand and a common component of desert soils, is the causative factor of a respiratory disease called silicosis, which causes shortness of breath, fever, fatigue, tissue scarring and fibrosis in the lungs and can be fatal. While silicosis is usually associated with occupational exposure (mining or cement work for example), studies have shown that individuals in silicate-rich environments are at risk of developing the disease (Griffin et al, 2001' ). In the US, more than 200 deaths were attributed to silicosis in 1996, and the World Health Organisation estimates approximately 24,000 deaths in China are due to silicosis each year (Griffin et al, 2001 ). 1  a  Naturally-derived dust is commonly believed to primarily consist of coarse-sized particles PM2.5-10: Plumes of desert dust in urban areas, however, are typically very fine, due to distance from the source and deposition of coarser particles during transport. The mode of the Gobi dust event in southwestern British Columbia was between 2-4 pm, with up to 50% of PMio mass less than 2.5 pm (Husar et. al, 2001). Based on the current literature, this size range would suggest a strong association between P M concentrations and adverse health effects. Yet studies to date that specifically consider naturally-derived P M have identified varied associations. Ostro el. al. (1999) found that a 10 pgm" increase in PMio, which was dominated by crustal particles, was associated with an approximately 1% increase in mortality, which is in line with many other P M studies to date, and a follow-up study (Ostro et. al. 2000) found similar results with P M 5. Hefflin et. al. (1994) found small, yet positive, associations between natural dust storms in Washington State and emergency room visits for bronchitis and sinusitis. "Desert Storm Pneumonitis" and "Desert Lung Syndrome" are other human diseases linked to the inhalation of desert dust (Griffin et al, 200l ). In contrast, Laden el. al (2000) and Schwartz et. al (1999) found no associations between crustal PM2.5 and daily mortality, or windblown dust (PMio) and mortality risk, respectively. 3  2  a  Naturally-derived dust clouds also possess the ability to act as transport agents of other pollutants and biological pathogens into distant urban environments, such as herbicides, pesticides, bacteria, viruses and fungal spores. Herbicides and pesticides are frequently blamed for a swathe of adverse health effects ranging from allergic responses to birth defects (Griffin el al, 2001 ; Griffin et al, 2002). Table 2.2 illustrates some human airborne pathogens that are commonly transmitted in dust and/or associated with exposure to desert dust clouds. African dust events in the Caribbean produce a two-to-three-fold increase in the number of airborne bacteria and viruses (Griffin et al, 2001 ) and Asian desert regoliths are known to contain infectious micro-organisms. Fungal spores have been transported to Korea by Asian dust clouds (Yeo & Kim, 2002) and there is also anecdotal evidence of epidemics of flu-like illness in the Hawaiian Islands linked to the presence of Asian-sourced dust (Tratt et al, 2001). Naturally-derived dust may also contain high concentrations of heavy metals (which are toxic to human cells) or radioiostopes (known carcinogens) (Griffin et al, 200l ; Harrison & Yin, 2000). a  b  a  29  Table 2.2. Examples of human airborne pathogens commonly transported in dust or associated with exposure to dust clouds (Source: Griffin et al., 2001", pp25). Bacteria  I- t i l l " i  Viruses  Black Plague Anthrax Tuberculosis Legionnaires' Disease  Cryptococcosis Aspergillosis Coccidiomycosis Histoplasmosis  Whooping Cough Diphtheria Psittacosis Bacterial Influenza Bacterial Meningitis  Blastomycosis  Common Cold Viral Influenza Chicken Pox Hantavirus Syndrome Smallpox  Pulmonary  There is already some evidence that dust storms can have adverse impacts on health by acting as transport agents for biological pathogens. African dust events have been shown to have direct impacts on health, with outbreaks of meningococcal meningitis often following local or regional scale dust events, and there are some researchers currently investigating the link between an increase in African dust transport and high rates of asthma in the Caribbean (Griffin et al, 2001 ; Griffin el al, 2002). b  At present, the transmission of most airborne biological pathogens by a dust cloud has only been shown to occur within the confines of a continent. Future research should examine if desert dust clouds can transport viable biological pathogens across great distances, such as the Pacific Ocean (Griffin et al, 200l ), and therefore have a potential impact on distant, large, urban populations. a  2.8 S u m m a r y There is significant evidence in the literature to suggest that urban air pollution, and in particular, P M pollution, has been associated with adverse health effects in the urban population. The vast majority of literature published to date focusses on the health impacts of anthropogenically-derived PM pollution, which is of significant concern as industrialisation and urbanisation processes continue to accelerate. The majority of persons who are recorded as 'adversely affected' after exposure to P M pollution seem to be affected to a relatively minor degree if they are otherwise healthy. The more severe and acute health impacts are primarily seen in those who are already affected by respiratory or cardiac illnesses, and exposure to P M pollution is commonly seen to exacerbate symptoms and illnesses that are already present. Mortality rates consistently seem to show an increase during periods of high pollution, although these increases appear to be closely linked to persons who are already chronically or terminally ill, and exposure to P M pollution has prematurely hastened their death, rather than been the primary cause of death.  30  At this time, there has been very little research published that examines the relation between naturally-derived P M events and human health impacts. Current literature does suggest that adverse health affects are also associated with naturally-derived dust, as well as anthropogenically-derived PM, but there is little conclusive evidence at this stage, and little indication of the magnitude of these effects.  31  CHAPTER III Research Setting  3.1 Introduction The Fraser Valley of British Columbia is commonly perceived to have episodes of poor air quality by both residents and scientists, particularly in the summer months due to photochemical smogs. This is the result of complex interactions between various pollutants (of both natural and anthropogenic origin), a coastal setting and complex topography. This review summarises what is currently known about air pollution in the Fraser Valley. In this chapter, the physical setting and character of the Fraser Valley is introduced, and related to the air pollution meteorology of the area. Each of the major air pollutants of concern in the Fraser Valley is discussed in turn, including what is known of their spatial and temporal variability, and pollutant sources. Lastly, a review of health impacts studies that have been conducted in the context of Fraser Valley air pollution is discussed. A variety of research methodologies employed in research the Fraser Valley context is also included in these discussions.  3.2 Physical Setting and Air Pollution Meteorology The Fraser Valley is a large valley in a North American west coast mid-latitude setting (roughly 49°N latitude, 121-123°W longitude), oriented WNW-ESE and bordering the Georgia Strait and Pacific Ocean at its western end. The valley floor is nearly flat and no more than a few hundred metres above sea level. The Fraser Valley is bounded by mountain ranges - the Coast Mountain Ranges to the north and the Cascade Ranges to the south - and the valley walls rise steeply to over 2000m on the north side and to about 1000m on the south side, with many tributary valleys. (McKendry, 2000; McKendry et. al, 2001; Salmond and McKendry, 2002; Steyn el. al, 1997). There is a large, rapidly growing urban population within the Fraser Valley, with the northwestern end the most heavily urbanized (the population of the Greater Vancouver area is approximately 2 million). The valley becomes increasingly more agricultural and less urbanised towards the eastern end of the valley, with more scattered urban centers (Fig 3.1) (McKendry et. al, 2001).  33  Fig 3.1. Satellite image of the Fraser Valley, British Columbia, showing patterns of urbanization. North is to the top of the image. Vancouver is at the western end of the valley (left side of image), with the more rural regions of the Upper Fraser Valley and Okanagan Valley to the east. The Coast Mountains are to the north and Cascade Mountains to the southeast, with the Fraser River clearly visible (Source: Advanced Satellite Productions Inc., 2003).  The Fraser Valley has a predominantly easterly wind regime due to topographic channeling, although there are significant thermal air flows that play an important role in air pollution (McKendry, 2000; McKendry et. al, 2001; Salmond and McKendry, 2002; Steyn et. al, 1997). During the daytime, pollutants generated in the Greater Vancouver urban region are advected inland and up-valley by on-shore sea breezes from the coastal plains into the main valley as well as adjoining valleys. At night, the flow regime reverses, with down-valley land breezes dominating air flow and transporting pollutants out of the valley system (Banta et. al, 1997; McKendry et. al. 1998 ; Pryor and Steyn, 1995). This cycle is particularly pronounced in summer, with an average wind speed from July to September of less than 3 ms , with the mountains to the north and south channeling air flows along the valley axis (Pryor and Steyn, 1995). a  Apart from degrading the air quality of regions some distance from the generating area, this diurnal reversal of circulation patterns in combination with the complex topography of this region creates the potential for re-circulation of pollutants within the valley system, although this process in the Fraser Valley is poorly understood at present (Banta et. al, 1997; McKendry et. al. 1998 ; Pryor and Steyn, 1995). In particular, the presence of inversions in the summer months limits vertical mixing of the pollutant layer, and so the Fraser Valley is frequently incompletely ventilated (Pryor and Steyn, 1995), which could contribute substantially to the re-circulation of air pollutants. a  The most polluted regions of the Fraser Valley are to the east of Vancouver due to the upvalley advection of air masses under the influence of on-shore sea breezes. As the urbanized region spreads eastwards in the valley, the plume of pollution is also shifting eastwards and beginning to affect regions even further upwind (Steyn et. al, 1997). Studies that looked at visibility also observed a west-to-east gradient of declining visibility (Pryor et. al, 1997). Volatile organic compound (VOC) and nitrogen oxide (NO ) emissions decreased to the east of Vancouver as visibility decreased in the same direction, indicating that advection of pollutants plays an important role in visibility in the Fraser Valley (Pryor et. al, 1997). The x  34  conversion of primary pollutants into secondary pollutants also occurs during advection. For example, N O emissions are highest in the western end of the Fraser Valley. They are emitted into warm, moist airmasses which are advected to the east as N O is converted primarily into nitric acid (HNO3) and peroxyacetyl nitrates (PAN's) (Pryor et. al, 1997). N  x  The main meteorological conditions that lead to elevated air pollution concentrations in the Fraser Valley are calm or light wind conditions, summer anticyclones and winter 'gap wind' events (McKendry, 2000). Low wind speeds aid the development of a stable nocturnal boundary layer, which creates periods of reduced dispersion and can also lead to sudden peaks in pollutant concentrations in the evening or early in the night as the flow regime changes from day-time up-valley sea breezes to night-time down-valley land breezes. Summertime anticyclonic conditions result in warm, still weather, which is ideal for the production of photochemical smogs. Anticyclones reduce atmospheric dispersion of pollutants, and concentrations of photochemical pollutants and P M rise (PM concentrations can become elevated to concentrations of 50-75 ugm" ), although peak concentrations are lower than those produced by nocturnal boundary layer development. Advection of pollutants up-valley to the east of Vancouver is also predominant during the summer under anticyclonic conditions. Finally, wintertime 'gap wind' events in the eastern end of the Fraser Valley can lead to extended periods of elevated P M concentrations (around 100 p.gm" for several days), primarily due to the localised suspension of crustal materials (such as soil and agricultural dust) and road dust (McKendry et. al, 1998 ; McKendry, 2000). 3  3  a  Ozone in particular is strongly linked to meteorology. It is produced in the presence of U V (ultraviolet) radiation from sunlight, with light or no wind, clear skies and warm temperatures. In the Fraser Valley, ozone concentrations peak under the influence of a strong ridge of high pressure, which produces up-valley air flows with subsidence inversions, which significantly reduce the mixing depth and trap the pollutants close to the surface (McKendry, 1994; McKendry et. al, 1998 ; McKendry, 2000). However, exceedances of air quality standards in the Fraser Valley cannot be totally accounted for by meteorology (Pryor et. al, 1995) - variability in emissions is also important. a  There has been a concentrated research effort focused on investigating the structure of the Fraser Valley atmosphere and the identification of layers in which pollutants may be trapped and/or advected within the Fraser Valley, such as McKendry et. al. (1997), Pisano et. al. (1997) and Haydn et. al. (1997). There has also been significant research investigating how pollutants, particularly ozone, move within the atmosphere vertically and the structure and behavior of the mixing layer within the Fraser Valley. This field of research is well reviewed in McKendry and Lundgren (2000). A variety of methodologies have been employed in this area of research, including the use of instrument towers (Salmond and McKendry, 2002), aircraft flights (Haydn et. al, 1997; Hoffef. al, 1997; McKendry et. al, 1997; McKendry et. al, 1998 ; Price et. al, 2003), lidar (Hageli et. al, 2000; Hoff et. al, 1997; Maletto et. al, 2003; McKendry et. al, 1997; Steyn et. al, 1999), tethered balloons (Haydn et. al, 1997; McKendry et. al, 1997; Pisano et. al, 1997: Salmond and McKendry, 2002), high altitude balloon sondes and other vertical profiling techniques (Haydn et. al, 1997; McKendry et. al, 1997). In addition, there have been various attempts to model the air flow regimes of the Fraser Valley for use in models to predict the photochemistry of the atmosphere, model pollutant dispersion and thus, predict air quality. Modelling so far has attempted to simulate the physical forcing effects of the Fraser Valley's coastal setting and complex topography on b  35  air flow, as well as mixing layer depths, surface energy budgets, wind fields and coupling between atmospheric conditions aloft and at the surface (Cai and Steyn, 2000; Cai et. al, 2000; Hedley and Singleton, 1997; Pottier el. al, 1997; Steyn and McKendry, 1988). The Fraser Valley is also subject to pollutant incursions from long-distance sources. Midlatitude westerlies dominate flow over the North Pacific from fall through to spring, when trans-Pacific pollutant transport from Asia to North America is most common (Price et. al, 2003). A small but significant body of recent research has been focused on trying to identify how the dust is entrained and transported across the Pacific, but more importantly, how the dust, which is entrained in the higher levels of the atmosphere, is transported down to surface level environments (Gong et. al, 2003; Hacker et. al, 2001; Holzer et. al, 2003; Jaffe et. al, 1999; McKendry et. al, 2001; Zhang et. al, 2003; Zhao et. al, 2003). As well as sample analysis from ground-based aerosol sampling stations across Asia and northwestern North America and Canada, this research utilizes various modeling techniques, including mesoscale and global atmospheric models (Gong et. al, 2003; Hacker et. al, 2001; Holzer et. al, 2003; Jaffe et. al, 1999; McKendry et. al, 2001; Zhao et. al, 2003), trajectory analysis (Jaffe et. al, 1999; Jaffe et. al, 2003 ; Zhang et. al, 2003) and aerosol analysis from satellite observations (Gong et. al, 2003; Husar et. al, 2001; Jaffe et. al, 2003 ). a  a  3.3 Pollutants and their Sources in the Fraser Valley The pollutants • • • • • •  of most concern in the Fraser Valley context are: Particulate Matter (PM) Ozone (O3) Hydrocarbons (HCO's) Sulphur Oxides (SO ) Nitrogen Oxides (NO ) Cardon Monoxide/Dioxide (CO/C0 ) X  X  2  The current knowledge on the nature of these pollutants, their spatial and temporal variability and their sources is discussed in turn.  3.3.1 Particulate Matter (PM) Monitoring of P M in the Fraser Valley is based on a 24-hour rolling average of 50 pgm" , and there are few exceedances of this air quality objective throughout the year, although individual sites may experience peaks of more than 200 pgm" over short periods (of several hours) (McKendry, 2000). However, overall P M measurements are relatively low in contrast to other large urban areas in North America and Europe (McKendry, 2000). 3  Maximum concentrations are observed during winter months in the Vancouver area, with the rest of the Fraser Valley recording peak concentrations in the summertime (McKendry, 2000). The wintertime increase cannot be explained by an increase in emissions, thus it is most likely due to changes in meteorological conditions during winter (Brook et. al, 1997). 36  Further research has established that wintertime low-level atmospheric stability reduces pollutant dispersion in the more urbanized areas of the Fraser Valley. Research to date suggests that P M in the Fraser Valley is predominantly sourced from vehicle exhaust and road dust, with an estimated 78% of fine fraction P M (PM less than 2.5 u.gm~ , PM2.5) coming from the transportation sector (McKendry, 2000), especially in the winter months. Other major particulate sources include soil dust and secondary aerosols (Pryor, et. al, 1997). Coarse-fraction wind-blown dust of crustal origin (suspended road dust and soil dust from agricultural activities) tends to dominate summer P M readings, especially in the eastern-end of the Fraser Valley (McKendry, 2000), which is more rural in nature and less heavily urbanized. More than 65% of the overall fine particle mass in the Fraser Valley was found to be of organic origin, compared to 40-45%) of the particle mass measured on the east coast of Canada (Brook et. al, 1997). 3  Due to its meteorology and topography, the Fraser Valley is rarely affected by the long-range transport of PM, and most P M is generated within the Fraser Valley region (McKendry, 2000). Some point sources of PM pollution from the US have been identified in the Fraser Valley (McKendry, 2000), but as the Asian dust event of April 1998 demonstrated, the Fraser Valley is not immune from the effects of long-distance P M transport (Husar et. al, 2001; McKendry, 2000). Another similar dust event in April 2001 was also observed to produce elevated levels of P M (Jaffe et. al, 2003 ; Price et. al, 2003). Increasing industrialization and desertification in the arid regions of Asia and Africa suggest that long-range sources of dust may become increasingly important in the Fraser Valley in the future. a  3.3.2 Ozone (Od Ozone is a major pollutant in the Lower Fraser Valley and has been widely investigated. O3 is a secondary pollutant produced in the presence of sunlight, and thus is most pronounced in the Fraser Valley atmosphere during the summer months in the form of photochemical smogs (McKendry et. al, 1998b). The Canadian ambient air quality hourly objective for O3 is 82 ppb, and this guideline is exceeded within Fraser Valley on approximately 8-10 days per summer. These exceedances are typically only single-day events and the elevated O3 concentrations do not persist (Hedley et. al, 1997; McKendry et. al, 1998 ; Salmond and McKendry, 2002; McKendry, 1994). Ambient O3 concentrations as high as 200 ppb have been observed on rare occasions (Li et. al, 1997). Observed daily maximum O3 concentrations in downtown Vancouver have increased slightly over the period 1984 - 1991, whilst outside the urban core, daily maximum concentrations have typically decreased (Pryor, 1998). b  As O3 is a secondary pollutant requiring sunlight for generation, it is expected to display peak concentrations during the day, and low concentrations at night (as the O3 is deposited on surfaces or chemically removed from the atmosphere). However, one study found nocturnal increases in surface O3 concentrations (Salmond and McKendry, 2002), although the nocturnal peaks rarely exceeded 50 ppb, well below the ambient air hourly guideline of 82 ppb. There is a growing body of thought that this phenomena is linked to complex surface terrain.  37  There is significant advection of O3 pollution from the generation region of metropolitan Vancouver. The highest measured surface O3 levels are typically in Pitt Meadows, to the east of the metropolitan Vancouver area. This advection is due to complex up-valley thermodynamic airflow regimes, during which in-transport chemical transformation of the precursor pollutant species (into O3) also occurs (Joe et. al, 1996; McKendry et. al, 1998 ). b  The primary source of precursor species for O3 generation, particularly N O and hydrocarbons, is vehicle emissions (McKendry et. al, 1998 ). As such, O3 levels in the Fraser Valley are particularly sensitive to fluctuations in these species, particularly VOC's (volatile organic compounds) (Pryor, 1998). O3 variations are also closely related to solar radiation (especially U V radiation), air temperature, wind, atmospheric stability and inversion height (McKendry, 1994). O3 concentrations are typically higher in weekends than on week-days (Pryor and Steyn, 1995). O3 levels have also been shown to increase due to long-range dust transport events, to a maximum of 90 ppb (Jaffe et. al, 2003 ; Price et. al, 2003), but, as mentioned above, elevated O3 is not always linked to such events. x  b  a  3.3.3 Hydrocarbons  (HCO\s)  Hydrocarbons are an important ozone precursor species, and are present in the Fraser Valley in various forms, including VOC's (including benzene, propane, methane and 1,3-butadiene), non-methane hydrocarbons (NMHC's), polycyclic aromatic hydrocarbons (PAH's), PAN's, polychlorinated biphenyls (PCB's), chlorofluorocarbons (CFC's), dioxins and furans. Emission of these precursor species can be biogenic or anthropogenic (Steyn et. al, 1997). Approx 80% of the total mass of emissions is from vehicle exhausts (Gertler et. al, 1997; McKendry el. al, 1998"; Steyn et. al, 1997). Industrial emissions also contribute to hydrocarbon loadings in the Fraser Valley, particularly at the western end (which is more heavily urbanized and industrial) (Hoff et. al, 1997). Drewitt et. al (1998) investigated hydrocarbon emission rates for various vegetation species in the Fraser Valley (including agricultural crops and forest tree species), and found considerable variations in emission species and rates related to plant species, species distribution, environmental conditions, nutrient supply and other variables. Ozone precursor species have also been transported across the Pacific Ocean to North America from Asia, leading to substantial ozone enhancements in the Pacific Northwest, including the Fraser Valley (Jaffe et. al, 2003 ). This effect is not thought to commonly occur during events of mineral dust transport from Asia (Jaffe et. al, 2003 ), although an Asian dust event in April 2001 was shown to produce elevated N M H C and PAN levels (Price et. al, 2003). Propane levels were measured at 85%> above observed springtime median propane values, indicating that anthropogenic pollutant sources in Asia also played an important role in this long-distance dust transport event. a  a  Researchers are currently investigating the chemistry of ozone production in order to establish scientifically sound air quality regulations, as various hydrocarbons have different ozone-production potentials. O'Brien et. al. (1997) investigated a series of hydrocarbon oxidation products (including alkyl nitrates, hydroxy nitrates and carbonyl compounds) 38  involved in photochemical smogs to investigate their importance as a sink for N O and the relative importance of direct emission versus photochemical production of these reagents. Li et. al (1997) completed similar research and showed that the contributions of photochemistry towards organic compound production (PAN's, ketones, aldehydes) were more important than daily emission rates. Bottenheim et. al. (1997) examined the reactivity of non-methane hydrocarbons, Biesenthal et. al. (1997) examined isoprene and its relationship to ozone, and Jiang el. al. (1997) looked at the sensitivities of ozone concentrations to V O C and N O emissions. x  x  3.3.4 Sulphate Oxides (SOy) Acid aerosols, including SO , have been found to be particularly harmful to human health. Sulphate particles have been found to be present in relatively low concentrations in Vancouver, with an average concentration of 16 nmolm" . In contrast, studies in the rest of Canada have averaged concentrations of 40-70 nmolm" , particularly in Ontario (Brook et. al., 1997). x  3  3.3.5 Nitrate Oxides (NOJ N O is one of the most important precursor species for ozone generation in the Fraser Valley and acts as a catalyst for the production of ozone (Pisano et. al., 1997). N O concentrations of 20 ppbv have been observed aloft in the Fraser Valley atmosphere, although surface concentrations were substantially lower at around 2 ppbv (Pisano et. al, 1997). Although N O is a criteria pollutant that is recognized to have adverse effects on human health, it rarely reaches concentrations at the surface in the Fraser Valley that are deemed hazardous with regards to health (Pisano et. al, 1997). x  x  x  N O emissions have been investigated in some depth due to their key role in photochemical ozone formation. N O in the Fraser Valley is primarily from vehicle exhausts, with more than 77% of total N O emissions assigned to mobile sources (Gertler et. al, 1997; McKendry et. al, 1998 ). Concentrations are typically higher mid-week than during the weekend - this is probably due to the absence of rush-hour traffic and reduced industrial activity during weekends (Pryor and Steyn, 1995). Li et. al. (1997) found that direct emissions were more important than photochemistry for the generation of HNO2, but for ITNO3, photochemistry was dominant-over direct emissions. ' x  x  x  a  Models have also been used to try and predict vehicle contributions to N O emission inventories. However, modeling has not proved to be accurate enough for predictive use at present, with model outputs under-estimating N O emissions by up to 23%), or overestimating emissions by up to 13% (when compared to real-world observations) (Gertler et. al, 1997). x  x  39  3.3.6 Carbon Monoxide/Dioxide  (CO/CO?)  Carbon monoxide (CO) and carbon dioxide (CO2) have hardly been investigated in the Fraser Valley context. Most emissions are from vehicles (Gertler et. al, 1997), and an attempt to model vehicular emissions (in comparison to real world observations) resulted in modeled outputs under-estimated by up to 36%, or over-estimated by up to 24% (Gertler et. al, 1997). An Asian dust event in April 2001 was shown to produce elevated C O levels (Price et. al, 2003), with concentrations of 36% above median springtime C O levels, thus indicating that anthropogenic sources of CO in Asia could also be important contributors to C O and C 0 budgets in the Fraser Valley. 2  3.4 Health Effects Associated with Air Pollution in the Fraser Valley There has been surprisingly little research to date on links between public health and air pollution in the Fraser Valley considering its long history of poor air quality. Burnett et. al. (1997 ; 1997 ; 1998; 2000) investigated the associations between various air pollutants and hospital admissions across major cities in Canada, including Vancouver. Increases in ambient ozone, CO, NO2 and S 0 were associated with increases in hospital admissions for respiratory diseases (Burnett et. al, 1997 ), admissions for congestive heart failure in elderly persons were associated with C O , NO2, SO2 and haze (Burnett et. al, 1997 ), and increases in risk for premature mortality were attributed to a mix of gaseous pollutants in large cities across Canada (Burnett et. al, 1998). A fourth paper (Burnett et. al., 2000) examined 8 Canadian cities (including Vancouver) and attempted to associate the observed adverse health effects with either particulate or gas-phase air pollutants, with interesting results. They found that size-fractioned P M explained 28%> of the total observed health effects of the air pollution mixture (of PM and gaseous pollutants), with gas-phase pollutants accounting for the remainder (of health effects). In particular, they found that sulphate ions, iron, nickel and zinc present in the fine fraction of P M were most strongly associated with mortality - the Gobi dust particulate event was also found to contain significant levels of iron. These studies indicate that P M is associated with adverse effects on public health in the Canadian urban context. a  b  2  a  b  However this group of studies analysed the patterns of hospital admissions and air pollutants by considering a group of large Canadian cities as one 'whole', and little analysis was focused at the individual city level, such as Vancouver. One early study by Bates et. al. (1990) looked at hospital visits for respiratory ailments, particularly asthma, and tried to correlate peaks in admissions with changes in various air pollutants and meteorological variables. They found no correlation between hospital visits and temperature, ozone or N 0 levels, but there was some correlation between hospital admissions and S O levels during summer and especially winter months. 2  x  Whilst P M has received considerable coverage in health-and-air-pollution literature, extremely little research has been conducted on the health effects of this pollutant in the Fraser Valley context. The closest case study was by Vedal et. al. (1998), who investigated 40  the effects of increases in P M on lung function and respiratory symptoms in children on Vancouver Island for 1990-1992. This study found that increases in P M were associated with reductions in lung function and increased reporting of coughs, phlegm and sore throats. Children with asthma were found to be more susceptible to these effects than the rest of the study group. The most comprehensive study to date on the effects of PMio on health in the Fraser Valley is by Vedal (1995), who offers some estimates of the magnitude of effect of P M on public health in B C . Adverse health effects of PMio pollution were estimated on the basis of hospital admissions data, activity restriction and absenteeism, and Vedal concluded that increases in PMio pollution in BC "cause 82 extra deaths 69 extra hospitalizations for lung disorders, 60 for heart disorders and 17 for asthma" (Vedal, 1995, p3) every year. For emergency room visits, P M | was estimated to cause "283 extra visits for asthma and 71 extra visits for chronic bronchitis or emphysema" (Vedal, 1995, p4). Much greater impacts were estimated for activity restriction, school absenteeism and respiratory symptoms. In addition, each 10 pgm" increase in PMio concentrations was estimated to account for 0.8% increase in hospitalizations, 1.0% increase in emergency room visits for respiratory illnesses, 9.5% increase in days of restricted activity (due to respiratory or cardiac symptoms), 4.1% increase in school absenteeism and 1.2% increase in the reporting of coughs (Vedal, 1995). Villeneuve et. al. (2003) also looked at PM in Vancouver, and found that the coarse fraction (PM2.5-10) was associated with increased cardiovascular mortality (in the order of approximately 6%), but found no association between fine P M (PM2.5) and mortality. Thus it is expected that there will be a signal of hospital admissions for respiratory and cardiac illnesses associated with the Gobi dust event in April 1998. An application of Vedal's work to the case of the Gobi dust event is presented in Chapter 6. 0  Vedal et al (2003) published an interesting study which found that small increases in low concentrations of air pollution were associated with increases in daily mortality in Vancouver. Mortality statistics (all-cause, and respiratory-and-cardiac-related) and PMio, ozone, sulphur dioxide, nitrogen dioxide and carbon monoxide were examined for a three year study period. Over this three-year period, pollutant concentrations were consistently relatively low (the 50 and 90 percentiles of daily average PMio concentrations were 13 pgm" and 23 pgm" respectively, and the 50 and 90 percentiles of daily ozone 1-hourmaximum were 27ppb and 39 ppb). Increases in total mortality, and respiratory-and-cardiacrelated mortality, were both associated most strongly with summertime ozone increases (when photochemical smogs dominate air quality in the Fraser Valley), with wintertime increases in N 0 being associated with increases in total mortality. There was some suggestion of links between PMio and mortality as well. This study is particularly interesting as it shows measureable adverse health effects at relatively low pollutant concentrations. The Gobi dust event being investigated in this research produced hourly PMio concentrations in excess of 100 pgm" in the Fraser Valley - if effects are observable at very low P M concentrations, it is reasonable to assume that similar, if not greater, effects will also be observable at the levels experienced during the Gobi dust event. th  3  3  lh  th  th  2  Questions have been raised about the use of ambient pollutant concentrations to assess exposures and make links to adverse health effects. As such, some studies are now specifically addressing these issues in their methodologies. Brauer et. al. (2001) used Vancouver to compare personal exposures to ambient P M concentrations with respect to a 41  range of respiratory and cardiac health effects and found that the "use of personal exposures did not improve the strength of any associations [between air pollution and adverse health outcomes] or lead to increased effect estimates" (Brauer et. al, 2001, p490). Brauer and Brook (1997) studied personal exposures to ozone in the Fraser Valley and found a negative association between ozone exposure and lung function. This association was apparent the day following exposure as well, suggesting a persistent adverse ozone effect. More information on the personal exposure measurement techniques can be found in Brauer and Brook (1995), and a more detailed account of this study is also presented by Brauer et. al. (1996). One of the most important findings of these papers was that the personal ozone exposures studied (daily maximum ambient ozone concentrations of 40 ppb), which displayed an adverse effect on lung function, were all well below current Canadian Air Quality Objective ozone concentrations (82 ppb), suggesting that current air quality objectives are not necessarily appropriate guidelines for assessing health-related pollutant impacts. Despite a number of studies focusing on health effects in this airshed, none have considered the role of natural dust as a contributor to the pollutant load, nor attempted to identify what impacts this particulate source may, or may not, be associated with. The Asian dust events of April 1998 had significant adverse health effects in China, where at least 12 deaths were reported, but the impacts on public health in the Fraser Valley, if any, are unclear at present. However, based on the mass of literature suggesting an adverse link between health and dust exposure, health advisory agencies in British Columbia, Washington, Idaho and Oregon issued warnings to the general public when the dust arrived as a precautionary measure (Husar et. al, 2001).  3.5 S u m m a r y Air pollution meteorology in the Fraser Valley is complex. Whilst the pollutants themselves are problematic with regards to health, air quality in the Fraser Valley is also strongly influenced by complex topographic and thermal influences on air flow and meteorology. There is currently a very incomplete picture of air pollution in the Fraser Valley, with the spatial and temporal behavior of various pollutants unknown and the interactions between pollutants unclear. There is also little evidence of research into some air pollutants, such as heavy metals, which may or may not be a problem in this context. However, certain aspects of air pollution in the Fraser Valley have received considerable research attention, particularly PM, ozone and hydrocarbons, the structure of the atmosphere, the transport of pollutants within the Fraser Valley and, more recently, from long-distance pollutant sources. The Fraser Valley has a large resident population and is acknowledged as having occasional episodes of poor air quality. Yet there has been a surprising lack of research into the health impacts of air pollution in the Fraser Valley context. Research to date indicates the wealth of literature from other settings is consistent with findings in this context, with air pollution being associated with increased rates of hospitalization and mortality due to respiratory and cardiac-related conditions. It is highly likely that the health impacts associated with Fraser Valley air quality extend significantly beyond the need for hospital care, with a much larger proportion of the population probably experiencing less severe health impacts. Vedal (1995) was the first and only paper to date to attempt to quantify these less severe effects in the • •• '  42  Fraser Valley context, finding significant associations between air pollution, activity restriction and school absenteeism. This study alone indicates a strong need to investigate this area further, and the studies published since then certainly support those early air quality and health associations.  43  CHAPTER IV Methodology  4.1 Introduction Epidemiological methods have been widely used since the 1950's to investigate the health impacts of air pollution (Samet & Jaakkola, 1999). The infamous London Smogs of 1952 in particular drew significant attention to this issue. Since then, there has been a growing interest and importance in exposure-centred (as opposed to disease-centred) environmental epidemiology with the increasing realization that many chronic, degenerative diseases (like asthma and bronchitis) are multifactorial, with no single hazard as a necessary cause (Terracini, 1992). In addition, many environmental hazards, like air pollution, are causally associated with more than one disease or health outcome. There has also been considerable concern from the public, health authorities and the scientific community about environmental agents and exposures for which the potential adverse health effects are unknown or poorly understood (Terracini, 1992). Various epidemiological methods have been used to investigate the links between P M and health, and these are outlined in Section 4.2. This research uses an ecological study design, which is discussed in Section 4.2.3., and the particular methods used in this research are outlined in more detail in Section 4.3. The chapter concludes with a brief summary.  4.2 Epidemiological Methods Used to Investigate the Effects of Air Pollution Various epidemiological methodologies have been used to investigate the associations between air pollution exposures and adverse health impacts. Most methodologies focus on the individual, rather than studies at the group level, and Samet and Jaakkola (1999) provide a particularly good discussion of epidemiological methodologies, including study designs, limitations, interpretation, exposure assessment and health outcome measures. The three main epidemiological study designs that have been used in this area are 1) cohort, 2) cross-sectional and 3) ecological studies. Case-control studies have also been used in air pollution epidemiology, where exposures for people with the health outcome of interest are compared with those of control cases to measure the association between exposure and outcome (Fig 4.1). However it is difficult to identify suitable 'control' cases that are comparable to the cases of interest. As such, these time-consuming studies require epidemiological expertise, detailed planning and significant resources, and they have been less frequently used than the other methodologies for investigating air pollution (Cuzick & Elliott, 1992; Samet & Jaakkola, 1999).  45  Design of a Case-Control Study  Exposed  Not  Expoa&d  \Z  Exposed  Not Exposed  \Z  Disease  No Disease  "Cases"  "Controls"  Fig 4.1. Design of a case-control study (Source: Gordis, 2000, ppl80).  4.2.1 Cohort Studies Cohort studies assess exposures and then follow two groups of people, one of exposed individuals and another of nonexposed individuals, for the development of the health outcome of interest over time (Fig. 4.2). If there is a positive association between the exposure and disease, then it is expected that the incidence of disease will be greater in the exposed population than in the nonexposed population. These studies can be prospective (the health outcome will occur sometime in the future) or retrospective (the exposure and health outcome have already occurred prior to the study beginning) (Gordis, 2000; Samet & Jaakkola, 1999).  Design of a Cohort Study  Exposed  7\  Develop Disease  Do Not Develop Disease  Not Exposed  7 \  Develop Disease  Do Not Develop Disease  Fig 4.2. Design of a cohort study (Source: Gordis, 2000, ppl80).  46  4.2.2 Cross-Sectional  Studies  In a cross-sectional study, both exposure and disease outcome are measured simultaneously in an individual -like a health snapshot in time. The most important limitation in these studies is the inability to establish a temporal correlation between exposure and health outcome in order to define an association or causal relationship from the findings (Gordis, 2000; Samet & Jaakkola, 1999).  Begin with:  Defined Population  7TT\ 1  Gather Data on Exposure and • la ease  Four Groups Are Possible:  Exposed; Have Disease  Exposed; Do Not Have Disease  Not Exposed; Have Disease  Not Exposed; Do Not Have Disease  Fig 4.3. Design of a cross-sectional study (Source: Gordis, 2000, ppl54).  4.2.3 Ecological  Studies  This research utilizes an ecological study design, which will be discussed in more detail in Section 4.3. Ecological studies focus on groups of people rather than individuals - there is no individuallevel data in the analysis. This methodology is particularly useful when information on individual exposures is unavailable (English, 1992; Gordis, 2000; Morgenstern, 1995). Ecological studies have been used by epidemiologists for more than a decade, particularly to identify the first potential associations between disease and environmental exposures (English, 1992; Morgenstern, 1995). The major aim of ecological studies is thus to formulate hypotheses about disease aetiologies, which can then be more rigorously tested with other epidemiological techniques, such as cohort and case-control studies. Ecological studies are also known as 'observational studies', as the topic of interest is being investigated in an observational and retrospective manner, rather than using an experimental methodology. This methodology has been particularly useful for studying air pollution, as air pollution generally displays a large degree of spatial and temporal variation (Samet & Jaakkola, 1999). Morgenstern (1995) provides a detailed assessment of ecological study designs, methods and analysis. A summary of the strengths of this methodology is presented in Figure 4.4.  47  STRENGTHS  '"  '  ,  ,  7  '.  •'  ,  '  •  Low cost and convenient - many secondary data sources, such as population and socio-economic information, are available and easily linked at the aggregate level (for example: province, city, census subdivision or postal code).  •  Measurement limitations of individual-level studies - // is difficult to accurately measure exposures for large study groups and/or over large study areas. Ecological studies are a very practical way to assess large-scale exposures.  •  Design limitations of individual-level studies - if exposure varies little within the study group, individual-level assessments may not be practical or useful. Ecologic studies can indicate a much wider range of exposures and associated effects.  •  Interest in ecological effects - the target level of interest may be ecological, such as investigating the effects of population-level interventions (programs, policies and/or legislation).  •  Relative simplicity of analysis - large studies based at the individual-level may be Conceptually and logistically difficult. It can often be simpler, and more effective, to treat the data ecologically and focus on ecological-level results.  Fig 4.4. Major.strengths of an ecological study design (Based on English (1992) and Morgenstern (1995)).  In ecological studies, time and place of residence are typically used as surrogate measures of exposure (English, 1992). Thus, large groups of people are assigned the same level of exposure and then compared to the rates of disease in a population. This technique is considered to be less subject to the random effects of measurement of exposure, in comparison to analytical studies where exposure and disease are measure in the same individual, and which have been shown to be subject to greater error in exposure measurement than population-based studies (English, 1992). However, the validity of this methodology depends on how well the surrogate measure of exposure (time and place of residence) represents the actual exposure for an individual who develops the disease of interest (English, 1992). Exposure in ecological studies is assessed as an average exposure for a community, yet the true range of exposure could be quite large. However, the smaller the spatial unit used to group individuals, the more likely it is that the grouped data (community exposure) will also apply to the individual (English, 1992; Morgenstern, 1995). The associations drawn from an ecological study (between disease and environmental exposure) are, by design, only applicable to groups. The 'ecological fallacy' is to assume that the relationship between environmental exposure and disease at the population level will also hold at the individual level - ecological studies cannot be used to predict outcomes at the individual level (Cusick & Elliott, 1992; English, 1992; Gordis, 2000; Samet & Jaakkola, 1999).  48  4.3 Assigning Causation and Possible Confounding in Ecological Studies 4.3.1 Assigning Causation in an Ecological  Study  Statistical measures may show a significant association between a pollutant and a health outcome, yet this association may not necessarily assign causality - the association may be due to coincidental or masked confounding factors and not related to the pollutant, exposure or health outcome at all - "although a significant association between [pollutant] level and response may be established, cause cannot be claimed because a confounder could well be present" (Zidek & Bates, 2002, pi61). For example, Zidek & Bates (2002) quote an interesting example where fallacious interpretation of association as causation resulted in the conclusion that 75% of the variation in human birth rates in Oslo could be explained by the number of storks in the city. In fact, the actual association between these two seemingly unrelated factors was the annual population count in Oslo. Steady population increase over the years had naturally resulted in growing birth rates and, also, the amount of garbage. As storks are scavengers, the increase in garbage had encouraged the stork population to grow at the same time! Health responses can be compared before and after an 'episode' in a time series of an environmental risk factor, and if the adverse health response is higher after the episode than before it, then the environmental factor can be implicated as a potential cause of that adverse effect. However, temporal confounders (as discussed further in the next section) allied with these 'episodes' may still remain, and thus, it is widely concluded that "causality cannot be established beyond doubt in an observational study" (Zidek & Bates, 2002, pi61). An eminent epidemiologist of the 1960's, Sir Austin Bradford Hill, proposed a set of criteria that could be used in making a judgement about causality in observational studies. It was stressed that these criteria did not operate as 'laws', but were simply "aspects of the association [between the environmental factor and an adverse health impact] that were of assistance in deciding the tenability of the causality hypothesis" (Zidek & Bates, 2002 p 155). Hill's criteria are outlined in Table 4.1. It is important to remember that uncertainty about a study factor, such as pollutant exposure, can weaken the strength of an association, but does not necessarily mean that the effect of the pollutant is also 'weak' (Zidek & Bates, 2002). For example, exposure of a population may not be accurately assessed by measures from an ambient air pollutant monitor, thus the association between the pollutant and the adverse health effect at the population level may appear weak, yet the association between the same pollutant and adverse health effects in an individual may still be strong (with personal exposure to the pollutant producing a measureable adverse health effect).  49  Table 4.1 Hill's criteria used to determine causality between a pollutant exposure and an adverse health impact (Based on Zidek & Bates, 2002, ppl55). Causality Criteria •  Strength of the Association  Generally measured with statistical techniques  •  Consistency  •  Specificity  Observed by different people? in different places? at different times? in different circumstances? For example: occupational exposures  •  Temporality  Exposure predates the outcome  •  Biological Gradient  Dose-response curve  •  Biological Plausability  Toxicological corroboration of observed health outcomes  •  Experiment  Examination of effect of any preventative actions  •  Analogy  Argument of known effects from another agent  Hill's criteria to assign causality seem to hold well under the weight of current research evidence on the association between P M exposure and adverse health impacts. The 'strength of the association' is well supported by the results from numerous methodologies and analysis methods, including rigorous statistical techniques, with good 'consistency' in findings by many different researchers, in different locations, over several decades of focused research and using a variety of research techniques. 'Specificity' has been shown to hold in various settings, including occupational and residential contexts, and from very localized to regional scale analyses, where exposure predates the measured outcome. Similarly, the association between P M exposure and adverse health has been shown to persist in the absence of additional pollutants, which may, in themselves, have similar impacts on public health. Current research is starting to examine the effect of preventative actions, such as the issue of public health advisories and classifications of air quality measures into more general terminologies for public use. The 'biological gradient' criteria, however, is the one factor that has not been satisfied by current research in this area - as discussed in Chapter 2, there does not appear to be a dose-response relationship between P M exposure and adverse health impacts, despite significant toxicological evidence of 'biological plausability'. Despite this one anomaly, Hill's criteria to assign causality seem to be well satisfied by the wealth of current research on PM exposure and adverse health effects. 50  4.3.2 Possible Confounding Factors in Ecological  Studies  Ecological studies must be interpreted carefully, as there are many other factors, apart from environmental exposures, that can influence the spatial variations in the frequency of disease. In particular, differences in the quality of diagnosis and classification of ill-health, rates of reporting of disease, population estimates per spatial unit, and ethnic, genetic and socioeconomic factors could potentially cause significant levels of bias in ecological results (English, 1992; Samet & Jaakkola, 1999). Some major sources of potential confounding are presented in Table 4.2, along with steps taken in this research to minimize their impact on results.  Table 4.2. Major sources of confounding in ecological studies and techniques used in this research to minimize confounding (Based on English (1992) and Samet & Jaakkola (1999)). Potential Confounders Quality of diagnosis & classification can vary from place to place Variations in reporting of disease Population estimates can vary in quality  Control Techniques Used Use of diagnosis codes assigned by a trained professional Use of provincially-administered data, data set confined to one province Use of official Census data for all study analysis, same spatial measures (FSA postal codes) used for all analysis  Temporal confounding is also an important consideration. Pollutant levels can be inextricably linked to factors like vehicle usage or industrial emissions, which may be higher on weekdays than weekends, or during certain hours of the day (such as 'rush hours'). Utilisation of hospital services often show a weekly temporal pattern due to factors like availability of physicians, with increased emergency room visits and decreased hospital admissions during the weekends, when many surgeons and general physicians may not be working. Mondays often display a small increase above 'normal' weekday admission rates owing to this weekend 'backlog' of unseen cases (Lipfert, 1993; Schwartz et. al, 1996 ). b  Seasonal variations can also confound findings in ecological studies. Most obviously, respiratory illnesses invariably display a wintertime increase due to influenza episodes and the spread of bacterial and viral infections (Lipfert, 1993). Coincident wintertime increases in air pollutants could then be falsely associated with this increase in respiratory symptoms, yet in reality, the pollutants may not be the causal factor in this health outcome. Likewise, seasonal variations in temperature and humidity can result in seasonal patterns of hospitalizations that are unrelated to variations in air pollutants. One method to reduce such confounding is to examine deviations from 'normal', or expected, rates of hospitalizations for that day of the week, month or season (Lipfert, 1993). Thus Chapter 6 presents some estimates of expected hospitalizations associated with the daily average rate of admissions and the level of PM experienced at that time, and compares these estimates of 'normal' rates of hospitalizations with the actual observed pattern of hospital admissions.  51  There is, however, little current knowledge about the effects of confounding factors at the group level. Despite these potential limitations, and the variety of methodologies used, ecological studies have proved to be highly informative, and displayed remarkable concensus, on the adverse effects of P M pollution on health (Samet & Jaakkola, 1999). Lipfert (1993) provides a detailed discussion of approximately 100 air pollution and health studies that range from case studies of major air pollution episodes that lasted only a few days (such as London, 1952) to time-series, cross-sectional and other miscellaneous studies. These investigations used a variety of study designs, diagnoses examined, lag periods considered, spatial areas used and the ways in which confounding variables were controlled. Yet despite these methodological differences, Lipfert (1993) found that there was good concensus on a statistically significant association between air pollution and hospital use.  4.4 Methodology of This Research 4.4.1 Data Hospital Separations Data was obtained for 1997, 1998 and 1999 for the province of British Columbia, Canada from the British Columbia Linked Health Database (BCLHD) (Fig. 4.5), administered by the Centre for Health Services and Policy Research (CHSPR). The B C L H D is a population-based data resource on health care and health care service utilization in British Columbia. The B C L H D is compiled from hospital discharge data, thus a person is only registered in the database once they are discharged from hospital care - 74.4% of cases in this study were discharged from hospital within one week of admission, and 97.9%> of cases were discharged within 30 days of admission. The set of abstracted data used in this research contained information on age, sex, partial postal code, hospital admissions and discharges, and diagnoses. Individuals could not be identified from the data provided.  52  THE  BC L I N K E D  HEALTH  DATABASE  P O P U LAT30 N  Omiks itk msp R«££fttf*4i0n Fifes  SOCIAL  INVESTMENT/  SAFETY N E T Q  BC  mlrrri*'TnmpmMinw Dsjiud  \ HEALTH  CARE  SYSTEM  SOCUi/PHYStCAL CONTEXT  Social  f'upMi.niut)'. IK' Ambulance Servi**  Q M S P Practitioner fife  I ' j j u . L CeiKU* t*ik !9ti. 2t*i  (cm nuiinst (are  P.I; O P L . F population eHimatei  ihssiiijl Separations  B l S U U data wt* am) profile*  Medial Service* Pl*»  rrffjTijiii < I C t t M M of health professional*  Menu! Weak*  I Settlement pattern data .*«»/ I Travel time and network data  • Fig 4.5. Structure of the BCLHD and context of the Hospital Separations Dataset used in this research (Source: CHSPR, 2004).  4.4.2 Identification of Study Cases Firstly, extended care hospitals were removed from the dataset. These hospitals provide longterm personal and/or palliative care, assessment, treatment and rehabilitation, and do not treat the acute admissions and illnesses which are of interest in this study. From an initial set of 144 hospitals across British Columbia, 17 were designated as extended care facilities, and thus removed from the analysis dataset. Secondly, the level of care variable was used to isolate those admissions that were classed as 'acute* (61% of all remaining cases). In addition, cases classed as 'DPU/GEAR' (Discharge Planning Unit/Geriatric Evaluation and Assessment for Rehabilitation) were retained in the analysis dataset. These beds are assigned to patients who are ready or near ready to leave acute care but are awaiting the organisation of additional services to provide post-hospital care, and comprised just 0.3%> of all remaining cases. Day surgeries, long term care patients and rehabilitation cases were thus removed from the analysis. Next, cases were selected for analysis based on the primary diagnosis code. This variable provided groups of cases based on the principal diagnosis as determined by the Canadian Institute for Health Information standards. These codes are assigned by a trained professional  53  and are considered to be a reliable and consistent classification of illness. Based on the wealth of literature to date that suggests air pollution is primarily associated with respiratory and cardiac symptoms, cases were selected if they were coded as 'respiratory' (for 'Diseases of the Respiratory System') or 'cardiac' ('Diseases of the Circulatory System'). Each case in the B C L H D also contained up to 16 possible diagnostic codes, so the use of this primary diagnosis variable did not guarantee that all relevant cases were selected for this analysis. On the other hand, significant expertise is required to evaluate the 16 diagnostic codes and assess which were most important. Thus the primary diagnosis code was regarded as the most reliable method by which to select cases of interest for this research. Lastly, three 'areas of interest' were identified that reflected the spatial characteristics of the Gobi dust event (as discussed in Chapter 3) - the Okanagan Valley, the Central/Upper Fraser Valley and the Greater Vancouver Metropolitan Area. These areas follow the movement of the dust through the Fraser Valley, starting inland and moving coastwards towards Vancouver Island. Figure 4.6 shows a map of the general study area and Figure 4.7 illustrates the aggregation of urban settlements into the three 'areas of interest'. Hereinafter, these population aggregations will be referred to as the Okanagan Valley, the Upper Fraser Valley and the Greater Vancouver region.  Fig 4.6. Map of southwestern British Columbia, Canada, showing the locations of urban centres aggregated into the study areas of Greater Vancouver and the Upper Fraser Valley. The urban centers aggregated into the Okanagan Valley study area are located to the east of Hope, at the right end of the map (North is to the top of the image).  54  •'•Okahagah V a l l e y ' , • •'• • * [ Vernon, Kelowna, Penticton, Enderby, Summerland, Oliver, Keremeos, Kamloops, Salmon Arm, Merritt, Princeton, Armstrong, Osoyoos ' U p p e r Fraser V a l l e y ' ~ i Langley. Chilliwack. Mission. Abbotsford. Maple Ridge. Hope, Pitt Mead(>\\s_ ' G r e a t e r Vancouver* __ Vancouver, North Vancouver, West Vancouver, New Westminster, Surrey, Richmond, Delta, Burnaby, White Rock, Port Moody, Coquitlam, Port Coquitlam Fig 4.7. Allocation of urban centres in southwestern British Columbia to the three 'areas of interest' in this research.  In accordance with an ecological study design, postal codes for the patient's place of residence were used to aggregate individual's home addresses into study groups based on geographic location. This was based on the assumption that the majority of exposures occurred at or near the place of residence. The B C L H D dataset contained Forward Sortation Area (FSA) postal codes for each admission "The first three characters of the postal comprise the FSA or Forward Sortation Area. A n FSA (Forward Sortation Area) provides the general area where the mail is going The first character in the FSA identifies any one of ten provinces, three territories, and six districts or geographic regions across Canada....The second and third characters in the FSA help to identify the exact area in a city or town (or other geographic location) where mail will be delivered." (extract from Canada Post Corporation website, 2004).  The FSA postal codes were identified and relevant cases were allocated to the respective area of interest (Fig. 4.8). O k a n a g a n Valley VOE, VOH, VI B, V I E , V1H, V1K, VIP, VIS, V1T, V1V, V I W , V1X, V 1 Y , V I Z , V2A, V2B, V2C, V2E, V2H, V4T, V4V U p p e r F r a s e r Valley V O M , VOX, V I M , V2P, V2R, V2S, V2T, V2V, V2W, V2X, V2Y, V2Z, V3A, V 3 G , V3Y, V4R. V4S, V4W, V4X, V4Z G r e a t e r Vancouver, V3B, V3C, V3E, V3H, V3J, V3K, V3L, V3M, V3N, V3R, V3S, vT|\"V3\\~Y3\\ , V V V V4A, V4B, V4C, V4E, V4G, V4K, V4L, V 4 M , V4N, V4P, V5A, V5B, V5C, V5E, V 5 G , V5H, V5.T, V5K, V5L, V 5 M , V5N, V5P, V5R, V5S, V5T, V5V, V5W, V5X, V5Y, V5Z, V6A, V6B, V6C, V6E, V6G, V6H, V6J, V6K, V6L, V 6 M , V6N, V6P, V6R, V6S, V6T, V6V, V6W, V6X, V6Y, V6Z, V7A, V7B, V7C, V7E, V7G, V7H, V7J, V7K, V 7 L , V7M, V7N, V7P,V7R, V7S, V7T, V7V, V7W, V7X, V 7 Y Fig 4.8. FSA postal codes allocated to the respective area of interest.  55  4.4.3 Methodology for Chapter 5- Patterns Hospitalizations in the Fraser Valley  of Particulate  Matter  Concentrations  and  Once the data was allocated to the three areas of interest, frequencies of the date of admission to hospital were calculated. However, frequencies are only applicable within a study area, and rates of admission per head of population are necessary to draw conclusions across the three areas of interest in this study. Thus data from the May 1996 Statistics Canada 'Census of Canada' was obtained, organised by FSA postal codes. Population counts for the three areas of interest in this research are presented in Table 4.3. This allowed rates of hospitalizations to be calculated for respiratory illnesses and cardiac illnesses. This analysis provided an important indicator of the seasonal and inter-annual influences on hospitalizations across the three areas of interest.  Table 4.3. Population by FSA postal code from the 1996 Statistics Canada Census. Stuclv Area Okanagan Valley Upper Fraser Valley Greater Vancouver  Population (1996) 370,624 persons 402,359 persons 1,630,701 persons  Time series of PMio for 1997 - 1999 were also created with data from the B C Ministry of Water, Land and Air Protection. Daily average PMio concentrations were calculated from hourly measurements for the 24-hour period from midday to midday. These time series permitted the visualization and identification of seasonal variations and potential correlations between PM and hospitalizations for respiratory and cardiac illness in BC.  4.4.4 Methodology for Chapter 6 - Case Study of the 1998 Gobi Dust Event This chapter examines the 1998 Gobi dust event in more detail, and includes a comparison of this event with a meteorological analogue from the previous year in order to control for possible meteorological confounding. The initial step was to attempt to identify the magnitude of the signal from the Gobi dust event that was expected to appear in hospital admissions as a result of elevated P M levels. Vedal (1995) completed a significant study that established that air pollution in British Columbia is associated with adverse public health impacts and which also assessed the magnitude of impact of air pollution on rates of hospitalizations. In that study, Vedal proposed that each 10 ugm" increase in the 24-hour average (or 'daily') P M above a concentration of 20 pgm" was associated with a 0.8% increase in hospitalizations due to respiratory illness and a 0.6%) increase in hospitalizations due to cardiac illness in British Columbia. Figure 4.9 illustrates Vedal's procedure to estimate the number of 'extra' hospitalizations due to elevated PM. The specific calculations used in this research are explained in detail in Chapter 6, section 6.2 and 6.3. 3  i 0  56  l-\lra rcspiniion admissions associated v iili I'M lor Gobi e\eni  Average number of respirator, admissions per dj\ lor 190S  X  "•'() increase in respirator} admission?* per 10 ugm"' increase in above 20 ugm '  Number of 10 ugm increments lor maximum observed l'\I|„ concentration 0  \  Fig 4.9. Estimation o f extra' hospitalizations due to PM from the Gobi dust event in southwestern British Columbia, as proposed by Vedal (1995).  The days of the Gobi dust event, and those immediately afterwards (as the literature suggests that health impacts can lag the exposure by several days), were examined in more depth than that provided in Chapter 5. The meteorological conditions during the Gobi dust event in April 1998 were quite unusual for that time of year, with high temperatures, low wind speeds and a strong inversion along the Fraser Valley, with the elevated particulate matter concentrations due to the dust also enhancing a pre-existing photochemical smog event. A spring analog was identified for May 1997, when synoptic conditions, temperatures, wind speeds and photochemical smog components were remarkably similar to the 1998 dust event (Table 4.4, Figs 4.10 and 4.11). The major difference between the 1997 and 1998 cases was the traces of particulate matter concentrations, which is argued to be the direct contribution of the 1998 Gobi dust event (Fig 4.1 la and b). The meteorological analogue was employed to disentangle the health effects due to meteorological influences and the presence of ambient photochemical air pollution from those effects associated with the additional burden of the Gobi dust in the airshed. With the estimates of effect based on work by Vedal (1995), frequencies of hospital admissions were used to identify the actual magnitude of effect of the Gobi dust event on hospitalizations for respiratory and cardiac illness. Table 4.4. Dates of the Gobi dust event and the corresponding dates of the meteorological analogue used in this research. 1  C v o h i V.\ cnt  I)a> 1 l)a> 2 l)a> 3 I)a\ 4  April April April April  27 28 29 30  199S 1998 1998 1998  Analogue May 12 1997 May 13 1997 May 14 1997 May 15 1997  57  a. Temperature  mS May b. W i n d Speed  8i  c. Wind Direction 360 T  Fig 4.10. Four-day time series of a/temperature, b/wind speed and c/wind direction for Chilliwack (Upper Fraser Valley) for the Gobi dust event (April 27-30 1998) and a meteorological analogue (May 12-15 1997). Note: a second meteorological analogue of May 27-30 1995 is also displayed in these figures, although it is not used in this research (Source: McKendry et. al., 2001, pp!8364).  58  Fig 4.11. Four-day time series of a/ P M at Chilliwack (Okanagan Valley), b/ P M at Abbotsford (Upper Fraser Valley), c/ PIVl, /Plvl ratios, d/ 0 and e/ CO concentrations for the Gobi dust event (April 27-30 1998) and a meteorological analogue (May 12-15 1997). Note: a second meteorological analogue of May 27-30 1995 is also displayed in these figures, although it is not used in this research (Source: McKendry et. a/.,2001,ppl8364). 1 0  S  ln  I 0  3  59  4.5 S u m m a r y  Although a variety of epidemiological methodologies have been used to investigate the links between PM and adverse health impacts, there has been remarkable concensus in the findings. Ecological study designs have been widely used and are thought to be particularly important in the initial establishment of links between P M exposures and health impacts, such as in this research. This analysis focused on the allocation of population groups to residential postal codes (on the assumption that most exposures occurred at or near a person's place of residence) and the spatial aggregation of urban centres into three major 'areas of interest' (which reflected the spatial characteristics of the Gobi dust event). Analysis was divided into two sections - first, the establishment of the health and P M context in which this dust event occurred, and secondly, a case study of the Gobi dust event days and those immediately following the event. A meteorological analogue was employed to distinguish the health impacts due to ambient photochemical smog and meteorological effects, from those effects due to the presence of the Gobi dust in the airshed.  60  CHAPTER V  Patterns of Particulate Matter Concentrations and Hospitalizations in the Fraser Valley  61  5.1 Introduction In order to identify the true magnitude of effect of the Gobi dust event on hospitalizations, it is necessary to set the context of P M and health impacts in southwestern BC. A false signal can easily be assigned to the Gobi dust event if it is considered in the absence of other environmental factors in this setting, such as seasonal and meteorological influences on hospitalizations. Meteorological factors are accounted for with the use of a meteorological analogue, which is discussed further in the next chapter. Seasonal factors, however, can only be determined through the analysis of a longer period of time than just the four days of the Gobi dust event. Thus this chapter focuses on the three year period of January 1 1997 to December 31 1999, which spans both sides of the Gobi dust event in late April 1998. The aim of this chapter is to establish if there is any initial correlation between ambient P M concentrations and hospital admissions for respiratory and cardiac illnesses in southwestern BC. Three-year time series of PMio and PM2.5 are examined. Following this, patterns of respiratory hospitalizations are discussed, and the Gobi dust event in particular is identified. Cardiac admissions are then examined in the same manner.  5.2 Annual Patterns of Particulate Matter Concentrations 5.2.1 Patterns of PMw  Concentrations  Hourly P M | measurements were obtained from the B C Ministry of Water, Land and Air Protection, and 24-hour average PMio concentrations were calculated for noon each day. The three year time-series of these daily averages for the Okanagan Valley is displayed in Figure 5.1(a), the Upper Fraser Valley in Fig 5.1(b), and Greater Vancouver in Figure 5.1(c). Note that the scale in this series of graphs differs. 0  62  j,,,  jDaily Average F Okanai  1Z  '  2* S! ?  J  SJ S?~S?  2 3 8 ) 6? 'fj  Q2 SJ 2 l  £ 5 8 ! g> <»i g  £5 ' S ! S! S* S  9J  g  ffl  5  | | — :  ) s>  a>^2; >s ;  <-3 55 2! e*  ,  ?  32!  s  s>  y  ejt^  Kensington - — Bumaty - — Delta  e;  Port Moody  s;'i?  Richmond  s*  s>  Surrey  s. <s;'  YVR  s  c  e  e; - "3  s;  s  1  s>  Kitsilar  Fig 5.1 (a-c) Time-series of daily average PM, concentrations in the three study areas. Note the different scale for the Okanagan Valley (a). The Gobi dust event is marked by a vertical green dashed line. 0  63  Data from two PMio stations was available for the Okanagan Valley, five stations in the Upper Fraser Valley, and eight stations in Greater Vancouver - all stations were graphed to highlight the significant variation in the P M signal between, and within, regions. As discussed in Chapter 3, complex topography and pollutant advection play an important role in this variation. Yet there is still a fairly distinct seasonal pattern that can be determined from these complex time-series. All three regions tend to have daily PMio averages of less than 30 pgm" throughout the year. Daily averages are higher during the summer months.(late June to September), at around 2025 pgm , whilst the winter months typically display daily averages of around 5-10 pgm" . The summer months in particular seem to be highly variable, with the daily average PMio reaching 35-40 pgm" on approximately 10 days each summer. 0  3  3  The Okanagan Valley pattern is rather less distinct than that of the other two regions due to the significantly different scale of the vertical axis to accommodate a large P M event at the end of 1998, which peaked at over 160 pgm" (probably due to an episode of prescribed burning). When this scale is decreased, the intra-annual pattern becomes clearer and is comparative to that displayed in the other two regions. The Gobi dust event in late April 1998 is reflected in clearly identifiable peaks in P M i across all three study areas. In addition, this peak is also reflected by each station within each study area, which indicates the Gobi dust event was associated with a widespread elevation of PMio across southwestern BC. 0  However, in the Okanagan Valley, the peak produced by the Gobi dust event is not unique in magnitude when compared to the three year time-series. Here, there are five other PMio 'events' where daily average PMio concentrations exceeded those measured during the Gobi dust event. Yet in the Upper Fraser Valley and Greater Vancouver, the Gobi dust event produced significant elevations that were not exceeded for the three years examined here. This suggests that variations in emissions and meteorological influences may be more important factors influencing elevations of PMio above 'normal' daily average P M j concentrations in southwestern BC, in addition to the long-range transport of Asian desert dust. .  0  5.2.2 Pall ems QJPMT^  Concentrations  Hourly PM2.5 measurements were obtained from the B C Ministry of Water, Land and Air Protection, and 24-hour average PM2.5 concentrations were calculated for noon each day. The three year time-series of these daily averages for the Okanagan Valley is displayed in Figure 5.2(a), and for the Upper Fraser Valley in Fig 5.2(b). PM2.5 observations were incomplete for the Okanagan Valley monitoring site and were unavailable for the Greater Vancouver region.  64  D.iil/ A v i M . K j p P M 2 5 TUIIP-SUMOS Ol'ciii.ici.in Valley  a)  •i  - g  20.0  flip i • *' 4  lilts S .-«B ..S v CO 26 O  r;j  fsl r^'"  *  Kamloops i:  iDaily Average PM2.5 Time-Series: J ^ Upper Fraser Valley  b)  co S ' « S O ? : frj ^ ^ ro i f U> S N to cn o ~£ ?S ^ i . . •  i  Fig 5.2 (a-b) Time-series of P M for two of the study regions (PM data was unavailable for Metropolitan Vancouver). The Gobi dust event is marked by a vertical green dashed line. 25  25  PM2.5 tends to display a similar signal to PMio, but with a smaller magnitude. Daily average PM2.5 concentrations are relatively low year-round, with a summer-early fall average of around 10-15 ugm* and wintertime averages around 5-10 ugm" . In Figure 5.2 (a) and (b), the Gobi dust event is clearly visible, displaying a sudden and dramatic increase in PM2.5 concentrations on 29 April 1998. Concentrations were relatively low in comparison to the signal of the Gobi dust event in the PMio observations, with the Okanagan Valley recording a daily average PM2.5 of approximately 28 ugm" , and the Upper Fraser Valley, approximately 36 ugm"~\ For the Upper Fraser Valley, this event produced the highest daily average PM2.5 concentrations across the three years examined - almost triple the typical PM2.5 average for that time of-year-.. The Okanagan Valley, which is further east than the Upper Fraser Valley 3  3  3  65  region and thus would be expected to have displayed the highest concentrations, actually recorded slightly lower PM2.5 concentrations during the Gobi dust event than the Upper Fraser Valley. This anomaly is possibly due to the influence of complex topography between these two regions that could have significantly affected the downward transport of the dust layer towards the surface in this general area, thus leading to significant spatial differences in surface concentrations. P M | concentrations in the Okanagan Valley were similar to concentrations recorded in the Upper Fraser Valley, and so it is possible that the spatial difference in PM2.5 was due to extended entrainment aloft compared to the larger fraction of PM, and so the finer fraction traveled further westwards before being transported down to the surface. 0  5.3 Annual Patterns of Respiratory Hospital Admissions 5.3.1 Pal terns of Respiratory  Admissions  Rates of hospitalizations for respiratory illness were calculated and graphed as a time-series for all three study areas from 1 January 1997 to 31 December 1999. There is considerable variability in rates of hospitalizations, which makes it difficult to discern a distinct pattern. Hence a seven-day running average was used to define a basic signal from the highly variable rates that would ideally also detect any significant impact on hospitalizations due to the Gobi dust event (Figures 5.3 (a,b and c)).  66  a)  r.w.-z  ,  Date of Hospital Admission  b)  ,  ^t-  ~ " Dato of Hospital Admission f  *  f  *  V<* (  1  \  f  > '  c)  j  I I'l 11'! 1^11 |I 1IIIIJ  1-1-1111 lil 111111  Fig 5.3 (a-c) Time-series of hospital admissions for respiratory illnesses. The Gobi dust event is marked by the green dashed line.  67  There is a very clear and consistent pattern to respiratory hospitalizations that is replicated in all three study areas for the three years studied. Respiratory admissions display a distinct peak during the late winter months of January, February and March each year, with approximately 5 admissions per 100,000 residents in the Okanagan Valley, 4-5 admissions per 100,000 in the Upper Fraser Valley and 3-4 admissions per 100,000 in Greater Vancouver. These rates compare to summer-time rates of approximately 2-3 admissions per 100,000 residents in the Okanagan Valley, 2 admissions per 100,000 in the Upper Fraser Valley and 1-2 admissions per 100,000 in Greater Vancouver. Lipfert (1993) found that a winter-time increase in respiratory illnesses was largely due to the spread of influenza and other viral infections. Previous studies identified an autumn increase in asthma admissions (Bates et. al. (1990) and Dales et. al. (1996)), which does not seem to correspond to this analysis, which consistently displays a decreasing trend in respiratory admissions from July to October, with only small, short-lived increases present. The aetiology of this autumn increase was not known, although potential influences were suggested, including meteorology, viral infections, aeroallergens (such as pollen and house mites) and poorer indoor air quality due to increasing home heating at this time of year. However, both these studies examined asthma admissions in particular, rather than respiratory illnesses in general, and so an autumn increase in asthma cases, as compared to asthma rates over summer, could easily be masked by a stronger signal in other more common respiratory illnesses, such as bacterial and viral infections. The strong seasonal signal in respiratory admissions indicates that meteorological influences could be an important factor influencing admission rates for respiratory conditions. Thus the use of a meteorological analogue in the next chapter is important in order to remove these strong seasonal effects from any signal associated with the Gobi dust event. Interestingly, there is a significant difference in the magnitude of rates between the three areas of interest. This could be a reflection of the spatial pattern of air quality, as discussed in Chapter 3, with the poorest air quality to be found in the eastern-most regions of the Fraser Valley. The admission rates in Greater Vancouver are effectively half those of the Upper Fraser Valley and Okanagan Valley. In other words, the highest admission rates are found to the east of Vancouver, slightly lower rates in the Upper Fraser Valley area, which is west of the Okanagan and closer to Vancouver, and the lowest rates are found in Greater Vancouver, the western-most study area. This is a very generalist assumption, and the spatial pattern in admissions could just as conceivably be due to differences in health service provision in these study areas. For example, a shortage of general practitioners per capita in the more rural regions of southwestern British Columbia (the Upper Fraser Valley and Okanagan Valley) could prompt more hospital admissions as people use the emergency department as their main health care facility, whether their symptoms require hospital-level care or not. In comparison, more general practitioners per capita in Greater Vancouver could reduce hospital admissions for less serious illnesses and thus lower the overall rate of admissions for respiratory illness. However, the most likely explanation for the differences in hospitalization rates between the three study regions is the age distribution of each population (Figure 5.4). The Okanagan Valley and Upper Fraser Valley have a greater proportion of children and elderly persons, 68  compared to Greater Vancouver. These two age groups have higher healthcare demands than adults, thus the rates of hospitalizations are higher in these two regions.  Age Distribution of the Population A c r o s s the Three Study Areas 30 0 |  —  0-14  15-29  30-44  45-59  60-74  75+  Age (years) • Okanagan Valley • Upper Fraser Valley • Greater Vancouver  Fig 5.4 Age distribution of the population across the three study regions.  Figure 5.3 (a, b and c) displays a sudden and distinctive drop in admissions for 27-31 March 1999, followed by a sudden increase back to 'normal' admission rates on April 1 1999. This could potentially reflect a strike of unionized health care workers in southwestern BC, which could result in the sudden decrease of hospital admissions and sudden pick-up only days later. However, since the sudden pick-up following 31 March 1999 did not rise above the 'normal" level, and thus reflect the treatment of a 'back-log' of cases due to a cessation of health care provision for a period of time, it is assumed that this feature is not due to a strike of health care personnel. A more likely explanation is related to the procurement of the original data set from CHSPR. When the data was first obtained, only admissions data until 31 March 1999 was currently available, and the remaining eight months of data to 31 December 1999 was obtained at a later date. These two data sets were then merged before analysis began. It is assumed that as the end of March 1999 was at the end of the original data set. and because the additional data did not 'repeat' this week of admissions, it is possible that these last few days of the original data set were incomplete. Since the pattern after March 31 1999 is consistent with the previous two years at the same time of year, and comparative across the three study areas, it is presumed that this anomaly is due to incomplete data for approximately one week at the end of March 1999.  69  5.3.2 The Effect of the Gobi Dust Event on Respiratory  Admissions  The Gobi dust event is barely detectable in the pattern of respiratory admission rates. The average rate of respiratory admissions for April and May (based on the three years studied) was 4 admissions per 100,000 residents in the Okanagan Valley, 3 admissions per 100,000 residents in the Upper Fraser Valley and 2 admissions per 100,000 residents in Greater Vancouver. Only the Upper Fraser Valley and Greater Vancouver seem to reflect a slight increase in the 7-day running average immediately following the Gobi dust event, yet even these small peaks represent less than one admission per 100,000 above 'normal' for that time of year, and are matched or exceeded by numerous other 'peaks' in admissions during the three-year study period. Such a small signal in late April 1'998 could easily be attributed to multiple other potential influences on respiratory admissions that have no connection to the presence of the Gobi desert dust, such as physical exertion triggering an asthma attack,- or a lingering cough following a cold. Thus it is not possible to conclude that the Gobi dust event 'caused' any extra respiratory admissions, based on the evidence in Figure 5.3.  5.4 Annual Patterns of Cardiac Hospital Admissions 5.4.1 Patterns of Cardiac  Admissions  As for respiratory admissions, the rates of hospitalization for cardiac illness were calculated and graphed as a three-year time-series for all three study areas. A seven-day running mean was then calculated to smooth out the admissions signal and also detect any signal due to the Gobi dust event (Figure 5.5 (a, b and c)).  70  Cardiac A d m i s s i o n s 1997-1999 Okanagan Valley  2!  3!  5.  U ij  .f4- .,,:  P  1  IT 2! 2  \  1  . .  >  5  ft  "B £  So 5  3 ' '  S? £3 2' 2!  * (£  t? §  D3tc of Hospital A d m i s s i o n  m  S  £2?  «3 =2  5  s  c«J  §  ,  l  S  j  § , g>  ,f'  8" P> t?  ,  S£ 9? 2» <!?  ^  § i  '  Cardiac A d m i s s i o n s 1997-19 Upper Fraser Valley  ft ' "  S  S  "3 £  tB< 5> 3  ^  ?3- t= fa, S  5  in  £  Date of Hospital A d m i s s i o n ' '  Cardiac A d m i s s i o n s 1997-1999:1 Greater Vancouver  is>H  isle! 1  Date of Hospital A d m i s s i o n  Fig 5.5 (a-c) Time-series of hospital admissions for cardiac illnesses. The Gobi dust event is marked the green dashed line.  The pattern of cardiac admissions is far less variable than for respiratory admissions, with the 7-day running average displaying a fairly consistent rate of cardiac admissions throughout the year. Average daily admission rates for cardiac illness are approximately 6 admissions per 100,000 residents in the Okanagan Valley, 4.5 admissions per 100,000 in the Upper Fraser Valley, and 3.5 admissions per 100,000 in Greater Vancouver. Cardiac admissions do not seem to display a seasonal pattern, with intra-annual variations in the order of only 1-2 admissions per 100,000 across the study period and all study areas. There are numerous noticeable departures from the average admission rate, although even the largest variations only represent 3 or 4 admissions more than the average on any one day. As with respiratory admissions, there are multiple other factors, apart from the presence of elevated P M levels, which could account for such admissions. The lack of a seasonal signal in cardiac illnesses is likely related to fewer environmental triggers (compared to respiratory illnesses) and such triggers are mostly over-ridden by treatment effects. Again, the magnitude of the admissions signal is significantly lower in Greater Vancouver than further up-valley. Although spatial variations in access to health care could be influencing the rates of admission, it is most likely that age distribution is the most influential factor, with a larger proportion of young and elderly persons in the Okanagan Valley and Upper Fraser Valley utilizing healthcare services more frequently than adults. This pattern of age-driven healthcare utilization enhances the hospitalization rates in the up-valley regions, as compared to Greater Vancouver. As discussed in Section 5.3.1, there is an anomalous sudden drop in admissions rates at the end of March 1999. Again, as this feature is replicated across all three study areas on the exact same days, it is presumed that this is the result of incomplete data from CHSPR, and does not reflect any other environmental influence on admission rates.  5.4.2 The Effect of the Gobi Dust Event on Cardiac  Admissions  As with respiratory admissions, there is a barely discernable signal in cardiac admissions associated with the Gobi dust event in late April 1998. Average cardiac admission rates during April and May (based on the three years studied) were in the order of 6-7 admissions per 100,000 residents in the Okanagan Valley, 4-5 admissions per 100,000 residents in the Upper Fraser Valley and 3-4 admissions per 100,000 residents in Greater Vancouver. Small increases in admission rates are visible immediately after the Gobi dust event in all three regions, although these 'peaks' represent less than 0.5 admissions above the average, and are thus inconclusive. In addition, these small peaks are equivalent, or smaller, in magnitude with the numerous other intra-annual and inter-annual variations in admission rates that are not related to the Gobi dust event. Therefore, it does not appear that the Gobi dust event had a significant impact on cardiac hospitalizations based on this analysis.  72  5.5 S u m m a r y Both PMio and PM2.5 displayed a strong seasonal pattern, with peak concentrations observed during the summer months. The Gobi dust event was associated with a distinct and consistent signal in both PMio and PM2.5, and across all PM monitoring stations in all three study areas. It was presumed that the greatest impact on health as a result of the Gobi dust event would be observed in the Upper Fraser Valley, as that region experienced the greatest elevations of P M during the Gobi dust event, and that the western-most study area (Greater Vancouver) would display the smallest signal. The southern half of Vancouver Island was originally the western-most study area, but on closer analysis, the signal associated with the Gobi dust event was even more dilute than in the regions examined here, and thus the region was removed from the final analysis. However, across all three study regions, the PM time series revealed numerous other periods of elevated PM during 1997 to 1999, some of which were greater in magnitude than the Gobi dust event, and which, theoretically, should also have resulted in increased rates of hospitalizations. However, none of the time series of hospital admission rates reflected either the Gobi dust event, or any of the other significant elevations in PM. In fact, there appears to be an inverse relationship between P M and hospitalizations in southwestern B C , with P M peaks observed during the summertime, and hospitalizations for respiratory illness highest in the winter months. Cardiac symptoms did not appear to display any consistent seasonal signal. Thus it appears that, despite the significant body of literature that suggests a distinct correlation between PM and hospitalizations, the association in southwestern B C is strongly influenced by other factors, such as climate and socio-economic influences on health care access.  73  CHAPTER VI Case Study of the April 1998 Gobi Dust Eyent  6.1 Introduction When trying to disentangle the health impacts of the naturally-derived Gobi dust event from the effects of anthropogenically-derived ambient air pollution and meteorology, it is useful to employ an analogue. The use of an analogue period potentially allows the effects due to the Gobi dust event to be isolated, as the analogue provides control for the coincidental and confounding influences of other pollutant exposures and meteorological influences on morbidity. This chapter is a case study of,the Gobi dust event in southwestern B C as compared to an analogue from the previous year. The chapter is split into two major sections, dealing with the impacts of the Gobi dust event on rates of hospital admissions for respiratory illnesses and then cardiac illnesses. First, estimates of the magnitude of effect on hospital admissions of the Gobi dust event, based on the P M concentrations observed, are made, based on work by Vedal (1995). Then, an analysis of the Gobi dust event and comparison with the analogue period is presented. Evaluation of the estimations of effect are also discussed.  6.2 Respiratory Illnesses The first analysis step was to estimate the magnitude of the effect on hospital admissions as a result of the elevated PM levels from the Gobi dust event. Vedal (1995) estimated that P M is associated with 86 extra hospital admissions for lung disorders and asthma per year for the entire province of British Columbia - an average of just 0.2 extra admissions per day. Thus the signal of respiratory hospital admissions associated with PM is very subtle in British Columbia as a whole, and the signal in just the southwestern part of the province is even smaller. For each study area, an estimate of the number of extra hospital admissions for respiratory illnesses due to P M from the Gobi dust event was calculated using the formula in Fig. 6.1, based on calculations by Vedal (1995) (abbreviations used in the rest of this chapter are in brackets).  1.xlra rcspiiuloiy ad minion-. associated with PM lor Gobi event ( " e x t r a  \\ciage number of ivspii\ilor\ admissions per da> lor IW8 ("(/VlTc/•„'<.'  "o increase in respiratory admissions per 10 pgm"' increase in P\l|<> above 20 pgm"'  Number of 10 pum"' increments for maximum observed PM concentration  tii/ini\\ion\')  ("%  ('no  ith  imrcment  rcasc )  of  increments')  Fig 6.1. Calculation used to estimate the impact on hospitalization rates associated with PIVI , based on work by Vedal (1995). 10  75  As defined by Vedal (1995), 'extra admissions' are hospital admissions above the 'normal' or average admission rate. The 'average admissions' was obtained from the total respiratory admissions for the study area for 1998, to get an average number of daily admissions for respiratory illness (Table 6.1). For '% increase per increment', Vedal (1995) calculated the increase of respiratory admissions (as a percentage above average admission rates) associated with each 10 pgm".. increase in PMio above 20 pgm" (on the basis that there is little evidence of adverse health effects below a concentration of 20 pgm" ). Thus for British Columbia, Vedal (1995) estimated that each 10 pgm" ('incremental') increase in daily/24-hour average PMio was associated with a 0.8% increase in respiratory hospitalizations. Lastly, the number of increments for the maximum observed PMio concentration in the study area during the Gobi dust event was calculated (Table 6.2). The maximum one-hourly P M concentration observed was used to represent the maximum possible exposure in each study population, and thus would be expected to be associated with the largest signal of associated hospital admissions. 3  3  Table 6.1. Calculations to obtain the average number of hospitalizations per day for respiratory illnesses. 'average admissions' Okanagan Vallc} ..Upper FraserValley Met. \ ancouv or  1 oul Number of Respirator} Vlmis.Mons in 1998 4134 3667 11543  calculation  4134/365 3667/365 11543 /365  Average Hail} Admissions for Rcspiralorv Illness 11.3/day 10.0/day 31.6/day  Table 6.2. Calculations to obtain the number of 10 ugm increments above 20 ugm for the 1-hour maximum P M observed during the Gobi dust event. As discussed in Chapter 5, Section 5.2.2, the Upper Fraser Valley region recorded higher peak P M i concentrations than the Okanagan Valley, despite the general west-to-east movement of the dust cloud from the Okanagan Valley to the Upper Fraser Valley. 3  3  1 0  0  Maximum Uhoui 'no. of increments' PMio During Gobi • Event Observedm This Study Area Okanagan Valley, 127 pgm" ' Upper.Fraser Valley, 162 pgm" ** i-.-> -J **** Met. Vancouver 12J pgm 3  3  nik-ulation  (127-20)/ 10 (162-20)/ 10 (123-20)/ 10  Number of 10 pgm" Increments above 20 pgm"" for Maximum PM,,, 10.7 14.2 10.3  1  recorded in Kamloops on 30 April 1998 at 7pm recorded in Chilliwack on 1 May 1998 at 10pm recorded in Burnaby (Kensington station) on 1 May 1998 at 8am  Therefore, the estimated magnitude of effect of the Gobi dust event on hospital admissions for respiratory illnesses was calculated, as set out in Table 6.3.  76  Table 6.3. Calculations to estimate the magnitude of effect on respiratory hospitalizations associated with the maximum 1-hour PM observed during the Gobi dust event, based on the formula in Figure 6.1. 10  "average  "'!<,  "no. of increments"  ...Okanagan Valley  11.3 /day  increase - increment' 0.8 %  "UpperFraser • Ynllev.  10.0/day  0.8 %  14.2  Metropolitan Vancouver  31.6/day  0.8 %  10.3  ailmisMon-s"  10.7  calculation  "1 X TRA ADMISSIONS"  11.3 \ 0.008 x 10.7 10.0 x 0.008 x 14.2 31.6 x 0.008 x 10.3  0.97 admissions  1.1 admissions  2.6 admissions  Based on these calculations of the estimated magnitude of effect, it appears that even the maximum observed PMio concentration from the Gobi dust event was likely to be associated with relatively few extra hospitalizations. Furthermore, based on the times these maximum concentrations were recorded (Table 6.2), it is likely that relatively few people in each of the study populations were even exposed to these high PMio concentrations. Therefore, it is estimated that Gobi dust event will not be associated with significant increases in rates of hospitalizations in southwestern BC. However, using the same techniques applied to the whole of 1998, and using daily maximum PMio concentrations from the same locations as above, it is estimated that there could be as many as 66.9 extra admissions for respiratory illness in Greater Vancouver per year, 87.5 extra respiratory admissions in the Upper Fraser Valley per year, and 152.8 extra respiratory admissions in the Okanagan Valley per year. A time series of these extra admissions for 1998 is presented in Figure 6.2. As can be readily observed, the estimated influence of the Gobi dust event is clearly identifiable, with simultaneous increases in estimated extra admissions in all three areas. On the other hand, despite the visibility of the Gobi dust event in the time series of estimated effect, there are also other peaks in this estimate of extra admissions later in the year that are of an equivalent, or even greater, magnitude, and thus the Gobi dust event does not appear to be particularly unusual in terms of signal magnitude in this context. It is important to note here that the graph in Figure 6.2 (and Figure 6.4 later in this chapter) illustrates the estimated number of extra respiratory admissions and is based on an average daily admission rate. If the actual rate of admissions for a particular day falls below this average rate, then any 'extra' admissions estimated by Figure 6.2 will not necessarily appear as 'extra' admissions above the average, in a time-series of actual admissions. Thus the signal shown by the Gobi dust event in this figure may not necessarily replicate itself in further analysis presented in this chapter, due to normal daily fluctuations in hospitalization rates.  77  i Estimated Additional Hospital Admissions for Respiratory Illnesses Associated with Daily Maximum PM10 for 1998  ififlfllJIB ''^ZZT ,• Me,  V a n c o u v e r  Upper Fraser Valley  Okanagan Val .  Fig 6.2. Time-series of estimated extra hospital admissions for respiratory illnesses associated with the daily maximum 1-hour PMm for 1998. The Gobi dust event in late April is marked in green.  6.2.1 Comparison of Respiratory Admissions Purine Admissions Pur ins a Meteorological Analogue  the Gobi Dust Event with Respiratory  The four days of the Gobi dust event in southwestern B C were compared with a meteorological analogue from the previous year in order to control for meteorological influences on morbidity. This analogue is discussed in more detail in Chapter 4, and the relevant dates are presented again in Table 6.4. Again, results are separated by study region and are presented in Figure 6.3.  Table 6.4. Dates of the Gobi dust event and corresponding dates of the meteorological analogue used in this research.  :  l)a\ 3 l)a\ 4  Gobi April April April April  I \enl 27 1998 28 1998 29 1998 30 1998  Analogue May 12 1997 May 13 1997 May 14 1997 May 15 1997  78  a)  R e s p i r a t o r y A d m i s s i o n s for t h e G o b i E v e n t (1998) v s . M e t e o r o l o g i c a l A n a l o g u e Okanagan  (1997):  Valley  • 1997 M1998  II  Dayi  Day2  Day3  Day4  Day5 D a / 6 D a y 7  DayS  Day 9 Day 10 Doy 11 Day 12 Day 13 Day 14 Day -5 Day 16 Day 17 Day 18  Day of Admission  Gobi Ewnt  R e s p i r a t o r y A d m i s s i o n s f o r t h e G o b i E v e n t (1998) v s . M e t e o r o l o g i c a l A n a l o g u e  b)  I I  (1997):  Upper Fraser Valley  • 1997 B1998  Day 1  Day4  Day 5 0 s y 6  Day?  DayS Day9 Day 10 Day 11 Day 12 Day 13 Day 14 Day 15 Day 16 Day 17 Day 16 Oay of Admission  [._  R e s p i r a t o r y A d m i s s i o n s f o r t h e G o b i E v e n t (1998) v s . M e t e o r o l o g i c a l A n a l o g u e  c)  (1997):  Metropolitan Vancouver  • 1997 H1998  kill Dayi  Day2  Day 3 Day4  • Day5  Day6  Day 7 DayS  •  i t-  D a y 9 Day 10 Day 11 Day 12 Day 13 Day 14 Day 15 Day 16 Day 17 Day 18  Day o f Admission  Fig 6.3 (a-c). Comparison of respiratory admissions during the Gobi dust event with admissions during the meteorological analogue from the previous year. Days 1-4 are the Gobi dust event/meteorological analogue comparison. Admission rates for two weeks following the Gobi dust event are also depicted to identify any lagged morbidity effects that may have resulted (meteorology is no longer comparable during this period, hence the 1997 data is absent). The dotted line represents the average admission rate for April/May 1998 as a measure of 'normal' springtime admission rates. 79  A signal of hospital admissions associated with the Gobi dust event is very hard to distinguish in Figure 6.3. Across all three study regions, there does not appear to be a distinct increase in hospital admissions for respiratory illnesses during the four-day analogue period, when it is argued that the most significant difference between that and the Gobi dust event is the presence of the Gobi desert dust. It appears that any impact on hospitalization rates from the Gobi dust!event was very subtle and'barely distinguishable from 'normal' variability in hospitalization rates at that time of year. In the Okanagan Valley, there appears to be an increase in admission rates on Day 3 and Day 4 of the Gobi dust event, with 1.9 additional admissions per 100,000 (over the analogue day admission rate) on Day 4 (Fig 6.3(a)). This translates into 7.0 additional admissions associated with the Gobi dust event in the Okanagan Valley (population = 370,624 persons). This is in excess of the estimated magnitude of effect (as discussed in the previous section of this chapter) of 0.9 extra admissions based on the maximum PMio concentration observed. This suggests that the Gobi dust event did have an impact on public health in this region, but the magnitude of effect was so small (with potentially 7 additional admissions 'due to' the Gobi dust event) that it does not exceed normal variability in hospitalization rates for this time of year, and thus is not readily visible in the larger-scale analysis in Chapter 5. All three regions had at least one day of the 4-day comparison period in which admission rates during the Gobi dust event exceeded admission rates during the meteorological analogue. A summary of this 'signal' compared to the estimates of effect calculated in the previous section of this chapter is presented in Table 6.5. These calculations control for meteorological influences by considering the largest difference in hospitalization rates between the Gobi dust event days and the meteorological analogue days, thus assuming that any difference in admission rates is attributable to the presence of the Gobi desert dust. As this table indicates, it appears that there was a subtle signal in hospitalization rates associated with the Gobi dust event as compared to the number of hospitalizations estimated based on the highest PMio observations. But again, as these 'additional' admissions due to the Gobi dust event are not immense in number, this signal becomes lost when embedded in the annual scale of analysis in the previous chapter, thus indicating that this dust event was not associated with a signal of greater magnitude than normal variability in hospitalization rates. Table 6.5. Comparison of the estimates of the magnitude of effect on respiratory hospitalisations with actual rates of admissions during the Gobi dust event.  Population hsiimalcd Increase In Admission Rale (ireaiesi DilTcreiKe in Rales Between Gobi 1 v cm and Analogue l\lia \dmissions •Due to' Gobi Dust Event  OLinauan Yalle\ 370,624 0.9 extra admissions 1.9 more admissions per 100,000 on Day 4 of the Gobi event  7.0  I pper 1 raser Valle\ 402,359 1.0 extra admissions  Mel. \'ancou\ er 1,630,701 2.6 extra admissions  0.5 more admissions 0.7 more admissions per 100,000 on Day 1 per 100,000 on Day 2 of the Gobi event of the Gobi event  2.0  11.4  80  Maximum PMio concentrations associated with the Gobi dust event were recorded on Day 3 and Day 4, with all three study regions displaying a small increase in hospitalization rates (as compared to the analogue days) on Day 4. The replication of this pattern in Figure 6.3 suggests an association between the Gobi dust event and rates of hospitalizations for respiratory illnesses. In addition, the literature has shown that asthma attacks may occur within minutes or hours of an exposure, thus strengthening this close temporal linkage. The literature has also repeatedly demonstrated that there is often a lagged effect of health impacts (in the order of days) following a sustained period of exposure, such as the four days of the Gobi dust event, and all three study regions show an anomalous increase in hospitalization rates on Day 8, four days after the end of the Gobi dust event. This increase could be lagged indicator of the Gobi dust event. However, the very distinct nature of this increase on Day 8 in both the Okanagan Valley and the Upper Fraser Valley, with a much more dilute signal in Metropolitan Vancouver, and similar anomalous increases on Day 15 in the Okanagan Valley, and Day 1 and Day 16 in the Upper Fraser Valley, suggest that there is another factor influencing hospitalization rates. Day 8 in this analysis was a Monday, thus Day 1 and Day 15 were also Mondays. As these two study regions are much more rural in nature than Metropolitan Vancouver, it is possible that these increases are actually related to a 'day of the week' phenomenon - a small 'rush' of hospitalizations following the weekend when hospital services may be reduced (due to fewer staff available during the weekends). In contrast, larger metropolitan areas like Vancouver tend to have greater healthcare service provision during the weekends (due to greater population pressures), thus reducing this Monday 'rush' signal in such regions. Also, some ill effects associated with P M may be relatively mild and non-life-threatening, thus people may often wait until after the busy weekend period to seek treatment. Therefore, these Monday increases could be a 'normal' fluctuation in the more rural study regions and completely unrelated to the presence of the Gobi dust a few days previously. This pattern also lends weight to the argument that the increases in hospitalization rates on Day 3 and Day 5 are more likely related to the Gobi dust event than a 'day of the week' phenomenon, and perhaps indicate a lag period in public health impacts of only a day or two. In addition, there does not appear to be a sustained increase in hospitalization rates following the Gobi dust event, indicating that any public health impact was likely restricted to the days of, and immediately following, the Gobi dust event. It also appears that the 'signal' becomes even more diluted as the dust traveled down-valley (from the Okanagan Valley towards Metropolitan Vancouver). Vancouver Island, further west of the Metropolitan Vancouver region, was originally also considered in this analysis to illustrate the spatial progression of the dust and associated health impacts. However, the signal was indistinguishable in that context and thus was excluded from further analysis.  81  6.3 C a r d i a c I l l n e s s e s Using the same technique discussed in Section 6.2, estimates of extra admissions for cardiac illness associated with the Gobi dust event were calculated for each study area. Vedal (1995) estimated that P M is associated with 60 extra hospital admissions for heart disorders per year for the entire province of British Columbia - an average of just 0.12 extra admissions per day. Thus the signal of cardiac hospital admissions associated with P M is also very subtle (as with respiratory hospital admissions). Calculations of extra hospitalizations for cardiac illnesses were the same as outline in the previous section: 'average admissions' for cardiac illnesses are displayed in Table 6.6, and the calculation of 10 pgm" increments for the maximum observed 1-hour PMio is displayed in Table 6.7. 3  Table 6.6. Calculations to obtain the average number of hospitalizations per day for cardiac illnesses. 'average admissions'  1 oial Number of (. ardiac Admissions in 1998 , Okanagan Valley 8070 Upper Fraser Valley i 6728 Met. Vancouver *< 19007  Lulculalion 1  8070 / 365 6728 / 365 19007/365  ;  Average l)ail\ Vlmissions for Cardiac Illness 22.1 / day 18.4/day 52.1 / day  Table 6.7. Calculations to obtain the number of 10 ugm" increments above 20 ugm" for the 1-hour maximum PM| observed during the Gobi dust event. 3  3  n  Maximum I-hour 'no. of increments" • PM],. During Gobi , Event Observed.in This Study,Area Okanagan Valley 127 pgm" ' Upper! rascr Valley 162 pgm" ** '! * * * * Met. Vancouver 123 pgm" j¥  7  3  nilculalion  Number of 10 pgm"' Increments above 20 pgm"'lor Maximum P\1i<, 10.7 14.2 10.3  (127-20)/ 10 (162-20)/ 10 (123-20)/ 10  recorded in Kamloops on 30 April 1998 at 7pm recorded in Chilliwack on 1 May 1998 at 10pm recorded in Burnaby (Kensington station) on 1 May 1998 at 8am  Vedal (1995) estimates that each 10 pgm" increase above 20 pgm" is associated with a 0.6% increase in cardiac admissions in British Columbia. Therefore, the estimated magnitude of effect of the Gobi dust event on hospital admissions for cardiac illnesses was calculated, as set out in Table 6.8. 3  3  82  Table 6.8. Calculations to estimate the magnitude of effect on cardiac hospitalizations associated with the maximum 1-hour PM| observed during the Gobi dust event, based on the formula in Figure 6.1. 0  Okanagan , . Vallcv  22.1 / day  •"<, increase increment' 0.6 %  Upper Fraser ^. $alley-_  18.4/day  0.6 %  14.2  Metropolitan' Vancouver  52.1 / day  0.6 %  10.3  •average admissions'  "no. o increments  calculation  ' 1\1RA \l)MISs|i)\s  lu.7  22.1 \ ' 0.006 x •10.7 18.4 x 0.006 x 14.2 52.1 x 0.006 x 10.3  1.4 admissions  v  1.6 admissions  3.2 admissions  Thus, as with respiratory admissions, it appears that even the maximum 1-hourly PMio associated with the Gobi dust event was likely to be associated with only a few extra hospitalizations for cardiac illnesses. Using the same techniques but applied to the whole of 1998, and using daily maximum PMio concentrations from the same locations as above, it is estimated that there could be as many as 110.4 extra admissions for cardiac illness in Greater Vancouver per year, 144.3 extra cardiac admissions in the Upper Fraser Valley per year, and 251.9 extra cardiac admissions in the Okanagan Valley per year. A time series of these extra admissions for 1998 is presented in Figure 6.4. As with the estimates for respiratory illnesses, the estimated influence of the Gobi dust event on cardiac admissions is identifiable, with simultaneous increases in extra admissions in all three areas. But again, there are other increases in estimated extra admissions associated with PMio that are similar in magnitude to the Gobi dust event signal, thus the Gobi dust event signal is neither unique nor particularly significant in this context.  83  Estimated Additional Hospital Admissions for Cardiac Illnesses Associated with Daily Maximum PM10 for 1998  } GobiDust Event | 4  M e l  Vancouver  Upper Fraser Valley  Okanagan Valley ,  Fig 6.4. Time-series of estimated extra hospital admissions for cardiac illnesses associated with the daily maximum 1-hour PM| for 1998. The Gobi dust event is marked in green. 0  6.3.1 Comparison of Cardiac Admissions Pur ins the Gobi Dust Event with Admissions During a Meteorological Analogue  Cardiac  As in the previous section, rates of admissions for the four days of the Gobi dust event were compared to a meteorological analogue from the previous year to control for meteorological influences on morbidity. The exact dates of this comparison were presented in Table 6.4. The results for cardiac admission rates are separated by study region and presented in Figure 6.5.  84  C a r d i a c A d m i s s i o n s f o r t h e G o b i E v e n t (1998) v s . M e t e o r o l o g i c a l A n a l o g u e (1997):  a)  Okanagan Valley  Dayi  Day2  Day3  av8  Dsy9  Day  Day  Day  Day  Day  Day  Day  Day  Day  10  11  12  13  m  15  16  17  18  Day of Admission  C a r d i a c A d m i s s i o n s for the G o b i E v e n t (1998) v s . M e t e o r o l o g i c a l A n a l o g u e (1997):  b)  Upper Fraser Valley  • 1997  •sv2 D a y 3  Day 4 Day 5 D a y 6  Day 7 Gay8  :•  DayS-  Day  Day  10  11  Day • • 12  Day  Day  Day  Day  rj •  Day  13  Day of Admission  C a r d i a c A d m i s s i o n s for the G o b i E v e n t (1998) v s . M e t e o r o l o g i c a l A n a l o g u e (1997):  c)  Greater Vancouver  • 1997 11998  Dayi  Day2 Day3 Day4  Day5  Day6  Day7  Day8 D a y 9  II  Day  Day  Day  Day  Day  Day  Day  Day  Day  10  11  12  13  14  15  16  17  18  Day of Admission^  Fig 6.5 (a-c). Comparison of cardiac admissions during the Gobi dust event with admissions during the meteorological analogue from the previous year. Days 1-4 are the Gobi dust event/meteorological analogue comparison. Admission rates for two weeks following the Gobi dust event are also depicted to identify any lagged morbidity effects that may have resulted (meteorology is no longer comparable during this period, hence the 1997 data is absent). The dotted line represents the average admission rate for April/May 1998 as a measure of 'normal' springtime admission rates.  85  The Upper Fraser Valley shows the strongest signal, with all four days of the Gobi dust event showing increased admission rates as compared to the analogue days. The signal is much less distinct in the Okanagan Valley and Metropolitan Vancouver. However, none of these rate increases are of a magnitude significantly greater than the average admission rate for that time of year, thus if this signal is indeed associated with the Gobi dust event, it is very subtle and it would be very difficult to assign causality with any certainty. Again, there does not appear to be a sustained increase in hospitalization rates for cardiac illnesses following the Gobi dust event, with only odd exceedances of the average admission rates for that time of year. The two weeks following the Gobi dust event appear to display a distinct weekly pattern of admission rates (with increases mid-week and lower rates over the weekends) and, as discussed in the previous section on respiratory admission rates, the anomalous increases are likely due to a 'day of the week' phenomenon and are thus unlikely to be related to a lagged effect of the Gobi dust event. In addition, the nature of cardiac illnesses, such as heart attacks, may typically require more immediate medical attention than many respiratory illnesses associated with P M exposure. Thus it is perhaps more likely for any public health impact to be closely linked to the PM event in the temporal scale, and be less likely to display a lagged impact. Also, someone with symptoms of a cardiac nature may attend a hospital emergency department during the weekend to get medical attention when their regular GP or pharmacist may not be open and in spite of longer waiting times due to fewer physicians on duty, whereas individuals may 'put up with' minor respiratory symptoms over the weekend until their GP re-opens on Monday. The overall magnitude of admission rates is greater for cardiac illnesses than for respiratory illnesses. It is proposed that this is closely related to the nature of cardiac illnesses and their treatment. The average GP is likely not a specialist in cardiac ailments, nor be open beyond business hours. An individual experiencing adverse cardiac symptoms is more likely to head for a hospital emergency department for prompt treatment, instead of waiting to see their regular GP. Thus admissions for cardiac illnesses are of a greater order of magnitude than admissions for respiratory illnesses, and this pattern is replicated across the three study regions in this research. The estimates of effect based on work by Vedal (1995) again appear to have underestimated the actual magnitude of effect of the Gobi dust event on cardiac illnesses. A summary of results is presented in Table 6.9 and, as further explained in Section 6.2.1, it appears that there is a signal in cardiac hospitalizations that is associated with the Gobi dust event. However, these extra admissions (as compared to the estimates based on PMio concentrations) are not significant in number and do not exceed the normal variability in hospitalization rates at this time of year, thus indicating that the signal is very subtle and not of major significance for public health planning.  86  Table 6.9. Comparison of the estimates of the magnitude of effect on cardiac hospitalisations with actual rates of admissions during the Gobi dust event. -Okanagan Valley 370,624 1.4 extra admissions  I'pper 1 raser \"alle\ 402,359 •1.6 extra admissions  Met. Vancouver 1.030.701 3.2 extra admissions  1,6 more admissions per 100,000 on Day 3 of the Gobi event  2.8 more admissions per 100,000 on Day 3 of the Gobi event  0.7 more admissions per 100,000 on Day 2 of the Gobi event  5.9  11.2  11.4  q  Population , Estimated Increase In Xdmission Rales (ireatesi Dit'lerenee in Rates 1 i d ween Gobi 1 \ cm and \naloeue f.xlra Admissions 'Due lo" (iobi Dusl 1 \enl ;  Maximum PMio concentrations were observed on Day 3 and Day 4 of the Gobi dust event, and the Okanagan Valley and Upper Fraser Valley both displayed the greatest increase in hospitalization rates for cardiac illness (as compared to the meteorological analogue) on Day 3. The literature suggests that heart attacks can occur within minutes to hours of a PM exposure, thus this close temporal linkage between high P M and increases in cardiac admission rates may indicate a causal linkage.  6.4 S u m m a r y The magnitude of results (rates of hospital admissions) was greater than expected based on the estimation of effect, but still not of enough magnitude to stand out from the annual timeseries of respiratory and cardiac admissions (in Chapter 5), nor did the signal appear distinctly in the closer analysis of the Gobi dust event presented in this chapter. Both respiratory and cardiac hospitalization rates appeared to increase on the last two days of the Gobi dust event, perhaps reflecting a temporal linkage and possible lag effect from the first day of the Gobi dust event. A sustained lag effect in the two weeks following the Gobi dust event was not evident in either respiratory or cardiac admission rates. The greatest increase in admissions (as compared to the analogue period) was in the order of only a dozen extra admissions, and when this handful of admissions is spread across half a dozen (or more) hospitals in each study region, the impact is barely negligible. Furthermore, the signal is not of sufficient magnitude to be identifiable in an annual time-series of hospital admissions (as in Chapter 5). Thus it appears that the Gobi dust event had a small impact on public health in this region. In addition, the estimates of the magnitude of effect taken from work by Vedal (1995) and based on the P M | concentrations observed, appear to have under-estimated the public health impact of the Gobi dust event. Thus, Vedal's estimates of the magnitude of effect of P M in general on respiratory and cardiac illnesses in BC may also be significantly under-estimated, and should be reviewed in the future. 0  87  CHAPTER VII Conclusions  7.1 Introduction This research was a unique opportunity to investigate the public health impacts of a naturally-derived P M event in southwestern B C , Canada. In particular, rates of hospital admissions for respiratory and cardiac illnesses were examined across the Fraser Valley region, with the Gobi dust event days being compared to an analogue period from the previous year in order to control for meteorological influences on morbidity and the impact of anthropogenically-derived ambient air pollution. This chapter will draw this research together with a very brief summary of each chapter, leading to a set of conclusions based on the research findings. To conclude, some future research directions stemming from this research are identified.  7.2 Brief Summary of Chapters Chapter 1 outlined the patterns of natural dust transport around the globe, with particular focus on Asian dust, its sources and characteristics. The 1998 Gobi dust event was described in this chapter too. The research question and rationale were then presented and discussed. The next two chapters set the context of this research. Chapter 2 then provided a comprehensive review of the science of particulate matter and public health, with some tentative predictions of the effect of the Gobi dust event based on this literature. The scarcity of literature focused on naturally-derived particulate matter was made apparent. Chapter 3 then discussed the physical setting of the study location and provided a summary of the air pollution meteorology of southwestern BC. The methodology of this research was then explained in Chapter 4, along with a short discussion of other epidemiological methodologies used in air pollution and health research. The problems of assigning causality and the effects of confounding in environmental health research were also discussed in this chapter. Results and discussion were presented in Chapters 5 and 6. First, the setting was established in Chapter 5 with an examination of the long-term trends in PM and rates of hospitalizations for respiratory and cardiac illnesses in this region. As was immediately apparent, the influence of the Gobi dust event in April 1998 did not appear in the long-term trends of hospitalization rates, despite being very obvious in the P M trends. Chapter 7 began with an estimation of the magnitude of effect of the Gobi dust event, based on work by Vedal (1995). Then the impact of the Gobi dust event was examined in more detail, with a day-by-day comparison of the Gobi dust event with an analogue period from the previous year to control for the influences of meteorology and ambient PM. Although a subtle signal in hospitalizations was detected, it was not significant in magnitude and would not have caused concern for public health authorities.  89  7.3 Discussion and Final Conclusions The estimates of impact magnitude, as based on work by Vedal (1995), indicated that even the peak PM concentrations experienced in southwestern BC were only likely to be associated with one or two 'extra' hospital admissions for respiratory illness due to P M exposure. With such a small estimate of effect, it was expected that it would be extremely difficult to detect a signal in hospital admissions that could be definitively linked to the presence of Gobi dust. For example, two 'extra' admissions could easily be associated with a car accident or workplace accident that was coincident, but not related to, the Gobi event. In contrast, a signal of perhaps twenty or more 'extra' admissions would be sufficiently in excess of 'normal' admissions variations to draw attention. This research actually identified a larger signal of hospitalizations potentially associated with the Gobi dust event than the signal estimated by Vedal's calculations, with up to 12 'extra' hospital admissions correlated with the Gobi dust event. As discussed in Chapter 4, it is extremely difficult to assign causality with confidence based on ecological study designs, and the signal of hospitalisations identified in this research is still relatively small, and does not appear in a long-term time-series of hospitalization rates. Thus it appears that there may have been a very subtle signal of hospitalizations associated with the. Gobi dust event that may have been apparent at the local scale, but was not of great significance at the regional scale. Most interestingly, this research brings particle toxicity into focus. The Gobi event produced highly elevated concentrations of respirable PM, particularly PM2.5, for several days - the literature suggests that these concentrations, particle sizes and duration of elevated P M should be associated with a distinct signal in adverse health effects. Yet there was barely an observable increase in demand for hospital treatment for respiratory or cardiac illness. The question is then raised about the importance of particle toxicity - perhaps particle toxicity is of more concern than particle size or concentration. Thus crustal material could be considered to be benign and of little consequence to healthy persons. However, as discussed in Chapter 2 and outlined in Table 2.2, there are many airborne pathogens that are commonly transported in desert dust and/or associated with exposure to desert dust, as well as heavy metals and other compounds that are known to have adverse toxicological effects. Therefore, it may not necessarily be the desert dust itself that is associated with adverse health effects, but instead, additional pathogenic agents that have been transported by the dust cloud may be of most concern to public health. African desert dust clouds have long been associated with outbreaks of meningococcal meningitis in parts of Africa, and large increases in the number of airborne bacteria and viruses in the Caribbean. More recently, Asian desert dust clouds have been found to contain viable infectious micro-organisms as well, and the 1998 Gobi dust event contained significant traces of heavy metals. As discussed in Chapter 1, each desert dust cloud carries its own distinct chemical signal owing to the particular origin and formation processes it experiences. Some Asian dust events to North America have been found to be largely crustal in composition, such as the 1998 event, yet other Asian dust events were primarily composed of industrial emissions. As a result, it may near-impossible to assess the risk posed by a future desert dust cloud to North America until it actually arrives and can be analysed for pathogenic hitch-hikers.  90  Thus, final conclusions of this research are as follows: •  The Gobi dust event was associated with a very subtle signal of increased hospital admission rates across southwestern BC during and immediately after the dust event.  •  There was scant evidence of a lagged effect of public health impacts in the two weeks following the Gobi dust event. '• '  •  It is reasonable to assume that many more people experienced minor adverse health impacts as a result of exposure to P M during the Gobi dust event, but utilized the medical services of their GP or pharmacist, and did not seek hospital-level treatment.  •  Crustal dust appears to be more benign than anthropogenically-derived PM.  •  There is little evidence based on the Gobi dust event of 1998 to suggest that further crustal dust events of Asian origin pose a significant public health risk at the hospitallevel for the population of southwestern BC.  •  There appears to be a low demand for specific public health planning and strategies for service provision to account for future Asian dust events dominated by crustal elements. Future Asian dust events dominated by industrial emissions may be of more concern with regards to public health.  7.4 Future Research Directions Ideally, similar research needs to be completed that utilizes data from GP's and pharmacists. Hospital admission data only captures the smallest proportion of a population that we would expect to experience adverse health effects as a result of exposure to a desert dust cloud. As illustrated by Vedal's triangle of expected health outcomes (Fig 2.2), hospitalization would be the outcome of a relatively few people in a population. By comparison, significantly more people would be expected to experience much less severe health outcomes, and would likely visit their family practitioner or local pharmacist to help alleviate symptoms they experience. Based on the inconclusive results of this research, it is expected that a significant proportion of the population in southwestern B C experienced some minor adverse effects as a result of exposure to the elevated PM concentrations of this Gobi event, but for various reasons, did not seek hospital-level treatment. Most commonly, people report eye irritation and a dry cough - symptoms which could commonly and frequently be alleviated with the use of overthe-counter medications from a pharmacy, or for more persistent irritations, perhaps a prescription medication from their GP. Other commonly reported symptoms include increased asthma inhaler and bronchodilator use, which would be reflected in GP prescription rates, rather than hospital admission rates. The lack of conclusive research on the effects of naturally-generated P M on human health also raises another important question: is it the dust itself, or something carried within the dust cloud, that makes people sick? Studies have documented the long-distance transport of 91  viable biological pathogens, including pollen, spores and bacteria, which have been shown to have had a measureable impact on public health. Perhaps the most influential factor in a dust cloud (with regards to public health) is not necessarily the actual elemental composition of the dust, or its concentrations, but moreover ^ what, other agents have used the dust particles as a transport agent over long distances. Dust particles can also act as transport agents for toxic anthropogenic pollutants, such as heavy metals, which may be more physiologically reactive and thus result in a more distinct signal in adverse health effects. Whether for health, economic or aesthetic reasons, the frequency of Asian dust events reaching the west coast of North America should be further investigated. Springtime dust storms are already annual events in many parts of mainland Asia, and indications are that desertification and industrialization are rapidly increasing on the Asian continent. As such, intense dust storms capable of entraining large amounts of dust into the troposphere, where it can be transported long-distance across the Pacific Ocean and carry toxic anthropogenic pollutants, may become more frequent events and thus be of more concern in this part of the world. Future Asian dust events may also draw multi-jurisdictional issues into focus, in terms of monitoring, research and analysis of public health policies and strategies. This could be particularly important if future analyses indicate that Gobi dust storms contain viable biological micro-organisms. Perhaps the dust itself is not of great significant with regard to health, but it could act as a transport agent for much more pathogenic elements.  92  BIBLIOGRAPHY  Advanced Satellite Productions Inc., Canada, 2003. ht tp:// www, ad sat .com Alpert, P. and E. Ganor, 2001. "Sahara mineral dust measurements from TOMS: Comparison to surface observations over the Middle East for the extreme dust storm, March 14-17, 1998" in Journal of Geophysical Research, v. 106 (D16), ppl 8275-18286 Anderson, H.R., 1999. "Health Effects of Air Pollution Episodes" in Holgate, S.T.,. J.M. Samet, H.S. Koren and R.L. Maynard (Eds.). Air Pollution and Health, Academic Press, London, pp461-484 Ansmann, A., .1. Bosenberg, A. Chaikovsky, A. Comeron, S. Eckhardt, R. Eixmann, V. Freudenthaler, P. Ginoux, L. Komguem, H. Linne, M . A . L . Marquez, V . Matthias, I. Mattis, V. Mitev, D. Miiller, S. Music, S. Nickovic, J. Pelon, L. Sauvage, P. Sobolewsky, M . K . Srivastava, A. Stohl, O. Torres, G . Vaughan, U . Wandinger and M . Wiegner, 2003. "Longrange transport of Saharan dust to northern Europe: The 11-16 October 2001 outbreak observed with EARLINET" in Journal of Geophysical Research, v. 108 (D24), ppl 2-1 - 1215 Ayres, J.G., 1994. "Air Pollution and Public Health in the United Kingdom" in Curtis, C.J., J.M. Reed, R.W. Battarbee and R . M . Harrison (Eds.). Urban Air Pollution and Public Health, Proceedings of a conference held at the Environmental Change Research Centre, University Colleg, London, Friday 23 September, 1994., pp45-48 Banta, R . M . , P.B. Shepson, J.W. Bottenheim, K . G . Anlauf, H.A. Wiebe, A. Gallant, T. Biesenthal, L.D. Olivier, C. Zhu, I.G. McKendry and D.G. Steyn, 1997. "Nocturnal Cleansing Flows in a Tributary Valley" in Atmospheric Environment, v. 31 (14), pp 21472162 Bates, D.V., M . Baker-Anderson and R. Si/to. 1990. "Asthma Attack Periodicity: A Study of Hospital Emergency Visits in Vancouver" in Environmental Research, v. 51, pp51-70 Baxter, P.J., R. Ing, H. Falk and B. Plikaytis, 1983. "Mount St. Helens Eruptions: The Acute Respiratory Effects of Volcanic Ash in a North American Community" in Archives of Environmental Health, v.38 (3), ppl38-143 Beer, T., 2001. "Air Quality as a Meteorological Hazard" in Natural Hazards, v.23, ppl57169 Berntsen, T., S. Karlsdottir and D.A. Jaffe, 1999. "Influence of Asian emissions on the composition of air reaching the northwestern United States" in Geophysical Research Letters, v. 26 (14), pp2171-2175 Biesenthal, T.A., Q. Wu, P.B. Shepson, H.A. Wiebe, K . G . Anlauf and G.I. Mackay, 1997. "A Study of Relationships Between Isoprene, Its Oxidation Products and Ozone, in the Lower Fraser Valley, BC" in Atmospheric Environment, v. 31 (14), pp2049-2058 Bobak, M and D.A. Leon, 1992. "Air pollution and infant mortality in the Czech Republic, 1986-88" in The Lancet, v.340, ppl010-1014 94  Bottenheim, J.W., P.C. Brickell, T.F. Dann, D.K. Wang, F. Hopper, A.J. Gallant, K . G . Anlauf and H.A. Wiebe, 1997. "Non-methane Hydrocarbons and C O during PACIFIC '93" in Atmospheric Environment, v. 31 (14), pp 2079-2087 Braaten, D.A. and T.A Cahill, 1986. "Size and composition of Asian dust transported to Hawaii" in Atmospheric Environment, v. 20 (6), ppl 105-1109 Braga, A.L.F., A. Zanobettr and J. Schwartz, 2000. "Do respiratory epidemics confound the association between air pollution and daily deaths?" in European Respiratory Journal, v. 16, pp723-728 Brauer, M . , .1. Brumm, S. Vedal and A.J. Petkau, 2002. "Exposure Misclassification and Threshold Concentrations in Time Series Analyses of Air Pollution Health Effects" in Risk Analysis, v.22(6), ppl 183-1193 Brauer, M., S.T. Ebelt, T.V. Fisher, J. Brumm, A.J. Petkau and S. Vedal, 2001. "Exposure of chronic obstructive pulmonary disease patients to particles: Respiratory and cardiovascular health effects" in Journal of Exposure Analysis and Environmental Epidemiology, v. 11, pp490-500 Brauer, M . and J.R. Brook, 1997. "Ozone Personal Exposures and Health Effects for Selected Groups Residing in the Fraser Valley" in Atmospheric Environment, v. 31 (14), pp 21132121 Brauer, M . , J. Blair and S.Vedal, 1996. "Effect of Ambient Ozone Exposure on Lung Function in Farm Workers" in American Journal of Respiratory and Critical Care Medicine, v. 154, pp 981-987 Brauer, M . and J.R. Brook, 1995. "Personal and Fixed-Site Ozone Measurements with a Passive Sampler" in Journal of the Air and Waste Management Association, v. 45, pp 529537 Brimblecombe, P., 1999. "Air Pollution and Health History" in Holgate, S T . , J.M. Samet, H.S. Koren and R.L. Maynard (Eds.). Air Pollution and Health, Academic Press, London, pp5-20 Brook, J.R., T.F. Dann and R.T. Burnett, 1997. "The Relationship Among TSP, P M , , P M . , and Inorganic Constituents of Atmospheric Particulate Matter at Multiple Canadian Locations" in Journal of the Air and Waste Management Association, v. 47, pp2-19 0  2  5  Brunekreef, B., N.A.H. Janssen, J. de Hartog, H. Harssema, M . Knape and P. van Vliet, 1997. "Air Pollution from Truck Traffic and Lung Function in Children Living near Motorways" in Epidemiology, v. 8(3), pp298-303 Brunekreef, B. and G. Hoek, 2000. "Beyond the Body Count: Air Pollution and Death" in American Journal of Epidemiology, v. 151 (5), pp449-451  95  Burnett, R.T., J.R. Brook, W.T.Yung, R.E. Dales and D. Krewski, 1997 . "Association between Ozone and Hospitalization for Respiratory Diseases in 16 Canadian Cities" in Environmental Research, v. 7,2, pp24-31 a  Burnett, R.T., R.E. Dales, J.R. Brook, M . E . Raizenne and D. Krewski,. 1997 . "Association between Ambient Carbon Monoxide Levels and Hospitalizations for Congestive Heart Failure in the Elderly in 10 Canadian Cities" in Epidemiology, v. 8(2), ppl62-167 b  Burnett, R.T., S. Cakmak and J.R. Brook, 1998. "The Effect of the Urban Ambient Air Pollution Mix on Daily Mortality Rates in 11 Canadian Cities" in Canadian Journal of Public Health, v. 89(3), ppl 52-156 Burnett, R.T., .1. Brook, T. Dann, C. Delocla, O. Philips, S. Cakmak, R. Vincent, M.S. Goldberg and D. Krewski, 2000. "Association Between Particulate- and Gas-Phase Components of Urban Air Pollution and Daily Mortality in Eight Canadian Cities" in Inhalation Toxicology, v.-12 (Suppl. 4), ppl 5-39 Burrows, W.R., M . Benjamin, S. Beauchamp, E.R. Lord, D. McCollor and B. Thomson, 1995. "CART Decision-Tree Statistical Analysis and Prediction of Summer Season Maximum Surface Ozone for the Vancouver, Montreal, and Atlantic Regions of Canada" in Journal of Applied Meteorology, v.34, ppl 848-1862 Cahill, C.F., 2003. "Asian aerosol transport to Alaska during ACE-Asia" in Journal Geophysical Research, v. 108 (D23), pp32-l - 32-8  of  Cai, X . M . and D.G. Steyn, 2000. "Modelling study of sea breezes in a complex coastal environment" in Atmospheric Environment, v. 34, pp 2873-2885 Cai, X . M . , R.A.S. Hourston and D.G. Steyn, 2000. "A Numerical Study of Meteorological Conditions During PACIFIC '93" in Atmosphere-Ocean, v. 38 (3), pp457-479 Canada Post Corporation, 2004. http://www.canadapost.ca/cpc2/addrnT/hh/doc/faq-e.asp Caquineau, S., A. Gaudichet, L. Gomes and M . Legrand, 2002. "Mineralogy of Saharan dust transported over northwestern tropical Atlantic Ocean in relation to source regions" in Journal of Geophysical Research, v. 107 (D15), pp4-l - 4-14 Chen, Y-S., P-C. Sheen, E-R. Chen, Y-K. Liu, T-N. Wu and C-Y. Yang, 2004 . "Effects of Asian dust storm events on daily mortality in Taipei, Taiwan" in Environmental Research, v. 95,ppl51-155 a  Chen, S-J., L-T. Hsieh, M-J. Kao, W-Y. Lin, K - L . Huang and C - C . Lin, 2004 . "Characteristics of particles sampled in southern Taiwan during the Asian dust storm periods in 2000 and 2001" in Atmospheric Environment, v. 38, pp5925-5934 b  Choi, J.C., M . Lee, Y. Chun, J. Kim and S. Oh, 2001. "Chemical composition and source signature of spring aerosol in Seoul, Korea" in Journal of Geophysical Research, v. 106 (D16), ppl8067-18074 96  CHSPR (Centre for Health Services http://vvAvvv.ch.spr.ubc.ca/Bclhd/aboutbclhd.htm  and  Policy  Research),  2004.  Chun, Y., J. Kim, J.C. Choi, K.O. Boo, S.N. Oh and M . Lee, 2001 . "Characteristic number size distribution of aerosol during Asian dust period in Korea" in Atmospheric Environment, v. 35,pp2715-2721 a  Chun, Y., K-O. Boo, J. Kim, S-U. Park and M . Lee, 2001 . "Synopsis, transport, and physical characteristics of Asian dust in Korea" in Journal of Geophysical Research, v. 106 (D16), ppl8461-18469 b  Chung, Y-S., H-S. Kim, J. Dulam and J. Harris, 2003. "On heavy dustfall observed with explosive sandstorms in Chongwon-Chongju, Korea in 2002" in Atmospheric Environment, v. 37, pp3425-3433 Cuzick J. and P. Elliott, 1992. "Small-area studies: purpose and methods" in Elliott, P., J. Cuzick, D. English and P. Stern (Eds.), 1992. Geographical and Environmental Epidemiology: Methods for Small-Area Studies, Oxford University Press, New York, ppl421 Dales, R.E., I. Schweitzer, J.H. Toogood, M . Drouin, W. Yang, J. Dolovich and J. Boulet, 1996. "Respiratory infections and the autumn increase in asthma morbidity" in European Respiratory Journal, v. 9, pp72-77 DeBell, L.J., M . Vozzella, R.W. Talbot and J.E. Dibb, 2004. "Asian dust storm events of spring 2001 and associated pollutants observed in New England by the Atmospheric Investigation, Regional Modeling, Analysis and Prediction (AIRMAP) monitoring network" in Journal of Geophysical Research, v. 109 (DO 1304), 16pp Delfino, R.J., R.S. Zeiger, J.M Seltzer and D.H. Street, 1998. "Symptoms in Pediatric Asthmatics and Air Pollution: Differences in Effects by Symptom Severity, Antiinflammatory Medication Use and Particulate Averaging Time" in Environmental Health Perspectives, v. 106 (11), pp751-761 Desqueyroux, H., J-C. Pujet, M . Prosper, F. Squinazi and I. Momas, 2002. "Short-Term Effects of Low-Level Air Pollution on Respiratory Health of Adults Suffering from Moderate to Severe Asthma" in Environmental Research Section A, v. 89, pp29-37 Dockery, D.W., 1994. "Acute Respiratory Effects of Particulate Air Pollution" in Curtis, C.J., J.M. Reed, R.W. Battarbee and R:M. Harrison (Eds.). Urban Air Pollution and Public Health, Proceedings of a conference held at the Environmental Change Research Centre, University College, London, Friday 23 September, 1994., pp49-52 Dockery, D.W., C A . Pope,X. Xu, J.D. Spengler, J.H. Ware, M^E. Fay, B.G. Ferris and F.E. Speizer, 1993. "An association between air pollution and mortality in six U.S. cities" in The New England Journal of Medicine, v.329 (24), ppl 753-1759  97  Drewitt, G.B., K. Curren, D.G. Steyn, T.J. Gillespie and H. Niki, 1998. "Measurement of Biogenic Hydrocarbon Emissions from Vegetation in the Lower Fraser Valley, British Columbia" in Atmospheric Environment, v. 32 (20), pp 3457-3466 Duce, R.A., C.K. Unni, B.J. Ray, J.M. Prospero and J.T. Merrill, 1980. "Long-Range Atmospheric Transport of Soil Dust from Asia to the Tropical North Pacific: Temporal Variability" in Science, v. 209, ppl522-1524 English, D., 1992. "Geographical epidemiology and ecological studies" in Elliott, P., J. Cuzick, D. English and P. Stern (Eds.), 1992. Geographical and Environmental Epidemiology: Methods for Small-Area Studies, Oxford University Press, New York, pp3-13 Fang, G - C , C-N. Chang, Y-S. Wu, S-C. Lu, P.P-C. Fu, S-C. Chang, C-D. Cheng and W-H. Yuen, 2002. "Concentration of atmospheric particulates during a dust storm period in central Taiwan, Taichung" in The Science of the Total Environment, v. 287, ppl41 -145 Fenger, J., 1999. "Urban air quality" in Atmospheric Environment, v.33, pp4877-4900 Gao, Y. and J.R. Anderson, 2001. "Characteristics of Chinese aerosols determined by individual-particle analysis" in Journal of Geophysical Research, v. 106 (D16), ppl803718045 Gehring, U., J. Cyrys, G. Sedlmeir, B. Brunekreef, T. Bellander, P. Fischer, C P . Bauer, D. Reinhardt, H.E. Wichmann and J. Heinrich, 2002. "Traffic-related air pollution and respiratory health during the first 2yrs of life" in European Respiratory Journal, v. 19, pp690698 Gertler, A.W./D.'N. Wittorff, R. McLaren, W. Belzer and T. Dann, 1997. "Characterization of Vehicle Emissions in Vancouver BC During the 1993 Lower Fraser Valley Oxidants Study" in Atmospheric Environment, v. 31 (14), pp 2107-2112 Gillette, D.A., 1999. "A Qualitative Geophysical Explanation for 'Hot Spot' Dust Emitting Source Regions" in Contributions to Atmospheric Physics, v. 72 (1), pp67-77 Goldberg, M.S., R.T. Burnett, M-F. Valois, K. Flegel, J.C. Bailar, J. Brook, R. Vincent and K. Radon, 2003. "Associations between ambient air pollution and daily mortality among persons with congestive heart failure" in Environmental Research, v. 91, pp8-20 Golitsyn, G and D.A. Gillette, 1993. "Introduction: A Joint Soviet-American Experiment for the Study of Asian Desert Dust and Its Impact on Local Meteorological Conditions and Climate" in Atmospheric Environment, v. 27A (16), pp2467-2470 Gomes, L. and D.A. Gillette, 1993. "A Comparison of Characteristics of Aerosol from Dust Storms in Central Asia with Soil-Derived Dust from Other Regions" in Atmospheric Environment, v. 27A (16), pp2539-2544  98  Gong, S.L., X . Y . Zhang, T.L. Zhao, I.G. McKendry, D.A. Jaffe and N . M . Lu, 2003. "Characterization of soil dust aerosol in China and its transport and distribution during 2001 ACE-Asia: 2. Model simulation and validation" in Journal of Geophysical Research,, v. 108 (D9),pp4-1-4-19 Gong, S.E., X . Y . Zhang;' T.L. Zhou and.L.A Barrie, 2004. "Sensitivity of Asian dust storm to natural and anthropogenic factors" in Geophysical Research Letters, v. 31, ppL07210 1-4 Gordis, L., 2000. Epidemiology (Second Edition), W.B. Saunders Company, Philadelphia Goudie, A.S. and N.J. Middleton, 1992. "The Changing Frequency of Dust Storms Through Time" in Climatic Change, v. 20, ppl97-225 Griffin, D.W., C A . Kellogg and E.A. Shinn, 2001 . "Dust in the wind: Long range transport of dust in the atmosphere and its implications for global public and ecosystem health" in Global Change and Human Health, v.2 (1), pp20-33 a  Griffin, D.W., V . H . Harrison, J.R. Herman and E.A. Shinn, 2001 . "African desert dust in the Caribbean atmosphere: Microbiology and public health" in Aerobiologia, v. 17, pp203-213 b  Griffin, D.W., C A . Kellogg, V . H . Garrison and E.A. Shinn, 2002. "The Global Transport of Dust" in American Scientist, v.90, pp228-235 Guo, J., K . A . Rahn and G. Zhuang, 2004 . "A mechanism for the increase of pollution elements in dust storms in Beijing" in Atmospheric Environment, v. 38, pp855-862 a  Guo, Z.G., J.L. Feng, M . Fang, H.Y. Chen and K . H . Lau, 2004 . "The elemental and organic characteristics of PM2.5 in Asian dust episodes in Qingdao, China, 2002" in Atmospheric Environment, v. 38, pp909-919 b  Hacker, J.P., I.G. McKendry and R.B. Stull, 2001. "Modeled Downward Transport of a Passive Tracer over Western North America during an Asian Dust Event in April 1998" in Journal of Applied Meteorology, v. 40, ppl 617-1628 Haegeli, P., D.G. Steyn and K.B. Strawbridge, 2000. "Spatial and Temporal Variability of Mixed-Layer Depth and Entrainment Zone Thickness" in Boundary-Layer Meteorology, v.97,pp47-71 Hajat, S., H.R. Anderson, R.W. Atkinson and A. Haines, 2002. "Effects of air pollution ongeneral practitioner consultations for upper respiratory diseases in London" in Occupational and Environmental Medicine, v.59, pp294-299 Harrison, R.M. and J. Yin, 2000. "PM in the atmosphere: which particle properties are important for its effects on health?" in The Science of the Total Environment, v.249, pp85101  99  Hayden, K.L., K . G . Anlauf, R . M . Hoff, J.W. Strapp, J.W. Bottenheim, H.A. Wiebe, F.A. Froude, J.B. Martin, D.G. Steyn and I.G. McKendry, 1997. "The Vertical Chemical and Meteorological Structure of the Boundary Layer in the Lower Fraser Valley During PACIFIC '93" in Atmospheric Environment, v. 31 (14), pp2089-2105 Hedley, M . and D.L. Singleton, 1997. "Evaluation of an Air Quality Simulation of the Lower Fraser Valley - I. Meteorology" in Atmospheric Environment, v. 31 (11), pp 1605-1615 Hedley, M . , R. McLaren, W. Jiang and D.L. Singleton, 1997. "Evaluation of an Air Quality Simulation of the Lower Fraser Valley - II. Photochemistry" in Atmospheric Environment, v. 31 (11), pp 1617-1630 Hefflin, B.J., B. Jalaludin, E. McClure, N . Cobb, C A . Johnson, L. Jecha and R.A. Etzel, 1994. "Surveillance for Dust Storms and Respiratory Diseases in Washington State, 1991" in Archives of Environmental Health, v. 49(3), ppl 70-174 Heinrich, J., B. Hoelscher, C. Frye, I. Meyer, M . Pitz, J. Cyrys, M . Wjst, L. Neas and H-E. Wichmann, 2002. "Improved Air Quality in Reunified Germany and Decreases in Respiratory Symptoms" in Epidemiology, v. 13(4), pp394-401 Hoff, R.M., M . Harwood, A. Sheppard, F. Froude and J.B. Martin, 1997. "Use of Airborne Lidar to Determine Aerosol Sources and Movement in the Lower Fraser Valley (Fraser Valley), BC" in Atmospheric Environment, v. 31 (14), pp2123-2134 Holman, C , 1999. "Sources of Air Pollution" in Holgate, S.T., J.M. Samet, H.S. Koren and R.L. Maynard (Eds.). Air Pollution and Health, Academic Press, London, ppl 15-148 Holzer, M . , I.G. McKendry and D.A. Jaffe, 2003. "Springtime trans-Pacific atmospheric transport from east Asia: A transit-time probability density function approach" in Journal of Geophysical Research, v. 108 (D22), pp 11-1-11-17 Husar, R.B., J.M. Prospero and L . L Stowe, 1997. "Characterization of tropospheric aerosols over the oceans with the N O A A advanced very high resolution readiometer optical thickness operational product" in Journal of Geophysical Research, v. 102 (D14), ppl6889-16909 Husar, R.B., D . M . Tratt, B.A. Schichtel, S.R. Falke, F. L i , D. Jaffe, S. Gasso, T. Gill, N.S. Laulainen, F. Lu, M . C Reheis, Y. Chun, D. Westphal, B.N. Holben, C. Gueymard, I. McKendry, N. Kuring, G . C Feldman, C. McClain, R.J. Frouin, J. Merrill, D. DuBois, F. Vignola, T. Murayama, S. Nickovic, W.E. Wilson, K. Sassen, N. Sugimoto and W . C Malm, 2001. "Asian dust events of April 1998" in Journal of Geophysical Research, v. 106 (D16), ppl8317-18330 In, H-J. and S-U. Park, 2002. "A simulation of long-range transport of Yellow Sand observed in April 1998 in Korea" in Atmospheric Environment, v. 36, pp4173-4187 Jacobs, J., R. Kreutzer and D. Smith, 1997. "Rice Burning and Asthma Hospitalizations, Butte County, California, 1983-1992" in Environmental Health Perspectives, v.105 (9), pp980-985 100  Jaffe, D., T.Anderson, D. Covert, R. Kotchenruther, B.Trost, J. Danielson, W. Simpson, T. Berntsen, S. Karlsdottir, D. Blake, J. Harris, G. Carmichael and I. Uno, 1999. "Transport of Asian Air Pollution to North America" in Geophysical Research Letters, v. 26 (6), pp711714 Jaffe, D., I. McKendry, T. Anderson and H. Price, 2003 . "Six 'new' episodes of transPacific transport of air pollutants" in Atmospheric Environment, v. 37, pp391-404 a  Jaffe, D.H., M . E Singer and A . A . Rimm, 2003 . "Air pollution and emergency department visits for asthma among Ohio Medicaid recipients, 1991-1996" in Environmental Research, v.91,pp21-28 b  Jiang, W., D.L. Singleton, M . Hedley and R. McLaren, 1997. "Sensitivity of ozone concentrations to V O C and N O emissions in the Canadian Lower Fraser Valley" in Atmospheric Environment, v. 31 (4), pp627 - 638 x  Joe, H., D.G. Steyn and E. Susko, 1996. "Analysis of Trends in Tropospheric Ozone in the Lower Fraser Valley, British Columbia" in Atmospheric Environment, v. 30 (20), pp34133421 Johnston, F.H., A . M . Kavanagh, D.M.J.S. Bowman and R.K.Scott, 2002. "Exposure to bushfire smoke and asthma: an ecological study" in Medical Journal of Australia, v. 176, pp535-538 Just, J., C. Segala, F. Sahraoui, G. Priol, A. Grimfeld and F. Newkirch, 2002. "Short-term health effects of particulate and photochemical air pollution in asthmatic children" in European Respiratory Journal, v.20, pp899-906 Kaiser, J., 2000. "Evidence Mounts That Tiny Particles Can Kill" in Science, v.289, pp22-23 Katsouyanni, K., J. Schwartz, C. Spix, G. Touloumi, D. Zmirou, A. Zanobetti, B. Wojtyniak, J.M. Vonk, A. Tobias, A. Ponka, S. Medina, L. Bacharova and H.R. Anderson, 1996. "Shortterm effects of air pollution on health: a European approach using epidemiologic time series data; the A P H E A protocol" in Journal of Epidemiology and Community Health, v.50 (Suppl. 1), ppS12-S18 Kelsall, J., J. Samet and S. Zeger, 1997. "Air pollution and mortality in Philadelphia, 19741988" in American Journal of Epidemiology, v. 146, pp750-762 Kim, B-G. and S-U. Park, 2001. "Transport and evolution of a winter-time Yellow sand observed in Korea" in Atmospheric Environment, v. 35, pp3191-3201 Kim, K - H . , and M - Y . Kim, 2003. "The effects of Asian dust on particulate matter fractionation in Seoul, Korea during spring 2001" in Chemosphere, v. 51, pp707-721 Kim, K - H . , G-H. Choi, C-H. Kang, J-H. Lee, J.Y. Kim, Y . H . Youn and S.R. Lee, 2003. "The chemical composition of fine and coarse particles in relation with the Asian Dust events" in Atmospheric Environment, v. 37, pp753-765 101  Koenig, J.Q., 1999. "Air pollution and asthma" in Journal Immunology, v. 104(4, part 1), pp717-722  of Allergy  and  Clinical  Koenig, J.Q., T.V. Larson, Q.S. Hanley, V. Rebolledo, K. Dumler, H. Checkoway, S-Z Wang, D. Lin and W.E. Pierson, 1993. "Pulmonary Function Changes in Children Associated with Fine Particulate Matter" in Environmental Research, v.63, pp26-38 Kwon, H-J., S-H. Cho, Y. Chun, F. Lagarde and G. Pershagen, 2002. "Effects of the Asian Dust Events on Daily Mortality in Seoul, Korea" in Environmental Research Section A, v. 90, ppl-5 Laden, F., L . M . Neas, D.W. Dockery and J. Schwartz, 2000. "Association of Fine Particulate Matter from Different Sources with Daily Mortality in Six U.S. Cities" in Environmental Health Perspectives, v. 108(10), pp941-947 Lee, J-T., H. Kim, H. Song, Y - C Hong, Y-S Cho, S-Y Shin, Y-J Hyun and Y-S Kim, 2002. "Air Pollution and Asthma Among Children in Seoul, Korea" in Epidemiology, v. 13(4), pp481-484 Lee, B-K., N-Y. Jun and H.K. Lee, 2004. "Comparison of particulate matter characteristics before, during, and after Asian dust events in Incheon and Ulsan, Korea" in Atmospheric Environment, v. 38, ppl535-1545 Lei, Y - C , C-C. Chan; P-Y. Wang, C-T. Lee and T-J. Cheng, 2004. "Effects of Asian dust particles on inflammation markers in peripheral blood and bronchoalveolar lavage in pulmonary hypertensive rats" in Environmental Research, v. 95, pp71-76 Levy, J.I., J.K. Hammitt and J.D. Spengler, 2000. "Estimating the Mortality Impacts of Particulate Matter: What Can Be Learned from Between-Study Variability?" in Environmental Health Perspectives, v. 108(2), ppl09-l 17 Li, S., K.G. Anlauf, H.A. Wiebe, J.W. Bottenheim, P.B. Shepson and T. Biesenthal, 1997. "Emission Ratios and Photochemical Production Efficiencies of Nitrogen Oxides, Ketones, and Aldehydes in the Lower Fraser Valley During the Summer PACIFIC 1993 Oxidant Study" in Atmospheric Environment, v. 31 (14), pp2037-2048 Lighty, J.S., J.M. Veranth and A.F. Sarofim, 2000. "Combustion Aerosols: Factors Governing Their Size and Composition and Implications to Human Health" in Journal of the Air and Waste Management Association, v.50, ppl565-1618 Lin, T.H., 2001. "Long-range transport of yellow sand to Taiwan in Spring 2000: observed evidence and simulation" in Atmospheric Environment, v.35, pp5873-5882 Lipfert, F.W., 1993. "A Critical Review of Studies of the Association between Demands for Hospital Services and Air Pollution" in Environmental Health Perspectives Supplements, v. 101 (Supp. 2),pp229-268  102  Lipsett, M . , S. Hurley and B. Ostro, 1997. "Air Pollution and Emergency Room Visits for Asthma in Santa Clara County, California" in Environmental Health Perspectives, v. 105(2), pp216-222 Littman, T., 1991. "Dust Storm Frequency in Asia: Climatic Control and Variability" in InternationalJournal of Climatology, v. 11, pp393-412 Liu, M . , D.L. Westphal, S. Wang, A. Shimizu, N. Sugimoto, J. Zhou and Y. Chen, 2003. "A high-resolution numerical study of the Asian dust storms of April 2001" in Journal of Geophysical Research, v. 108 (D23), pp21 -1 - 21 -20 Long, W., R.B. Tate, M . Neuman, J. Manfreda, A.B. Becker and N.R. Anthonisen, 1998. "Respiratory Symptoms in a Susceptible Population Due to Burning of Agricultural Residue" in Chest, v.113 (2), pp351-357 Loomis, D., 2000. "Sizing Up Air Pollution Research" in Epidemiology, v.l 1 (1), pp2-4 Loomis, D., M . Castillejos, D.R. Gold, W. McDonnell and V . H . Borja-Aburto, 1999. "Air Pollution and Infant Mortality in Mexico City" in Epidemiology, v. 10(2), ppl 18-123 Ma, C-J., M . Kasahara, R. Holler and T. Kamiya, 2001. "Characteristics of single particles sampled in Japan during the Asian dust-storm period" in Atmospheric Environment, v. 35, pp2707-2714 Maletto, A., I.G. McKendry and K.B. Strawbridge, 2003. "Profiles of particulate matter size distributions using a balloon-borne lightweight aerosol spectrometer in the planetary boundary layer" in Atmospheric Environment, v. 37, pp 661-670 Mar, T.F., G.A. Norris, J.Q Koenig and T.V. Larson, 2000. "Associations between Air Pollution and Mortality in Phoenix, 1995-1997" in Environmental Health Perspectives, v. 108(4), pp347-353 McClellan, R.O., 2002. "Setting ambient air quality standards for particulate matter" in Toxicology, v. 181-182, pp329-347 McGowan, J.A., P.N. Hider, E. Chacko and G.I.Town, 2002. "Particulate air pollution and hospital admissions in Christchurch, New Zealand" in Australian and New Zealand Journal of Public Health, v.26 (1), pp23-29 McKendry, I.G., 2000. "PMio Levels in the Lower Fraser Valley, British Columbia, Canada: An Overview of Spatiotemporal Variations and Meteorological Controls" in Journal of the Air and Waste Management Association, v. 50, pp443-452 McKendry, I.G. and J. Lundgren, 2000. "Tropospheric layering of ozone in regions of urbanized complex and/or coastal terrain: a review" in Progress in Physical Geography, v. 24 (3), pp329-354  103  McKendry, I.G., 1994. "Synoptic Circulation and Summertime Ground-Level Ozone Concentrations at Vancouver, British Columbia" in Journal of Applied Meteorology, v. 33, pp627-641 McKendry, I.G., J.P. Hacker, R. Stull, S. Sakiyama, D. Mignacca and K. Reid, 2001. "Longrange transport of Asian dust to the Lower Fraser Valley, British Columbia, Canada" in Journal of Geophysical Research, v.106 (D16), ppl 8361-18370 McKendry, I.G., D.G. Steyn, R . M . Banta, W. Strapp, K. Anlauf and J. Pottier, 1998 . "Daytime Photochemical Pollutant Transport over a Tributary Valley Lake in Southwestern British Columbia" in Journal of Applied Meteorology, v. 37, pp393-404 a  McKendry, I.G., D.G. Steyn, S. O.Kane, P. Zawar-Reza and D. Heuff, 1998 . "Lower Tropospheric Ozone Measurements by Light Aircraft Equipped with Chemiluminescent Sonde" in Journal of Atmospheric and Oceanic Technology, v. 15 (1), ppl 36-143 b  McKendry, I.G., D.G. Steyn, J. Lundgren, R.M. Hoff, W. Strapp, K. Anlauf, F. Froude, J.B. Martin, R.M. Bantas and L.D. Olivier, 1997. "Elevated Ozone Layers and Vertical DownMixing Over the Lower Fraser Valley, BC" in Atmospheric Environment, v. 31 (14), pp21352146 Merrill, J.T., M . Uematsu and R. Bleck, 1989. "Meteorological Analysis of Long Range Transport of Mineral Aerosols Over the North Pacific" in Journal of Geophysical Research, v. 94 (D6), pp8584-8598 Middleton, N.J., 1986. "A Geography of Dust Storms in South-West Asia" in Journal of Climatology, v. 6, ppl83-196 Morgenstern, H., 1995. "Ecologic Studies in Epidemiology: Methods" in Annual Review of Public Health, v. 16, pp61 -81  Concepts, Principles, and  Mortimer, K.M., L . M . Neas, D.W. Dockery, S. Redline and LB. Tager, 2002. "The effect of air pollution on inner-city children with asthma" in European Respiratory Journal, v. 19, pp699-705 ' .- • ;  Moulin, C , C.E. Lambert, U. Dayan, V. Masson, M . Ramonet, P. Bousquet, M . Legrand, Y.J. Balkanski, W. Guelle, B. Marticorena, G. Bergametti and F. Dulac, 1998. "Satellite climatology of African dust transport in the Mediterranean atmosphere" in Journal of Geophysical Research, v. 103 (Dl 1), ppl.3137-13144 Moulin, C , C.E. Lambert, F. Dulac and U.Dayan, 1997. "Control of atmospheric export of dust from North Africa by the North Atlantic Oscillation" in Nature, v. 387, pp691-694 Murayama, T., N . Sugimoto, I. Uno, K. Kinoshita, K. Aoki, N . Hagiwara, Z. Liu, I. Matsui, T. Sakai, T. Shibata, K. Arao, B.J. Sohn, J.G. Won, S.C. Yoon, T. L i , J. Zhou, H. Hu, M . Abo, K. lokibe, R. Koga and Y. Iwasaka, 2001. "Ground-based network observation of Asian dust events of April 1998 in east Asia" in Journal of Geophysical Research, v. 106 (D16), ppl8345-18359 104  Nyberg, F., P. Gustavsson. L . Jarup, T. Bellander, N . Berglind, R. Jakobsson and G. Pershagen, 2000. "Urban Air Pollution and Lung Cancer in Stockholm" in Epidemiology, v. 11(5), pp487-495 O'Brien, J.M., P.B, Shepson, Q. Wu, T. Biesenthal, J.W. Bottenheim, H.A. Wiebe, K . G . Anlauf and P. Brickell, 1997. "Production and Distribution of Organic Nitrates, and their Relationship to Carbonyl Compounds in an Urban Environment" in Atmospheric Environment, v. 31 (14), pp 2059-2069 Ostro, B. and L. Chestnut, 1998. "Assessing the Health Benefits of Reducing P M Air Pollution in the United States" in Environmental Research, v.16, pp94-106 Ostro, B.D., S. Hurley and M.J. Lipsett, 1999. "Air Pollution and Daily Mortality in the Coachella Valley, California: A Study of PM10 Dominated by Coarse Particles" in Environmental Research, v.81, pp231-238 • Ostro, B.D., R. Broadwin and M.J. Lipsett, 2000. "Coarse and fine particles and daily mortality in the Coachella Valley, California: a follow-up study" in Journal of Exposure Analysis and Environmental Epidemiology, v. 10, pp412-419 Osunsanya, T., G. Prescott and A. Seaton, 2001. "Acute respiratory effects of particles: mass or number?" in Occupational and Environmental Health, v.58(3), ppl 54-159 Ozsoy, E., N. Kubilay, S. Nickovic and C. Moulin, 2001. "A hemispheric dust storm affecting the Atlantic and Mediterranean in April 1994: Analyses, modeling, ground-based measurements and satellite observations" in Journal of Geophysical Research, v. 106 (D16), ppl8439-18460 Parrington, J.R., W.H. Zoller and N.K. Aras, 1983. "Asian Dust: Seasonal Transport to the Hawaiian islands" in Science, v. 220, ppl95-l 97 Perry, K.D., T.A. Cahill, R.A. Eldred, D.D. Dutcher and T.E. Gill, 1997. "Long-range transport of North African dust to the eastern United States" in Journal of Geophysical Research, v. 102 (D10), ppl 1225-11238 Pisano, J.T., I.G, McKendry, D.G. Steyn and D.R. Hastie, 1997. "Vertical Nitrogen Dioxide and Ozone Concentrations Measured from a Tethered Balloon in the Lower Fraser Valley" in Atmospheric Environment, v. 31 (14), pp2071-2078 Pooley, F.D. and M . Mille, 1999. "Composition of Air Pollution Particles" in Holgate, S T . , J.M. Samet, H.S. Koren and R.L. Maynard (Eds.). Air Pollution and Health, Academic Press, London. pp619-634 Pope, C A . and D.W. Dockery, 1999. "Epidemiology of Particle Effects" in Holgate, S T . , J.M. Samet, H.S. Koren and R.L. Maynard (Eds.), 1999. Air Pollution and Health, Academic Press, London, pp673-705  105  Pope, C.A., D.W. Dockery and J. Schwartz, 1995. "Review of epidemiological evidence of health effects of particulate air pollution" in Inhalation Toxicology, v.7, ppl-18 Pope, C.A., D.W. Dockery, J.D. Spengler and M . E . Raizenne, 1991. "Respiratory health and PMio pollution: a daily time series analysis" in American Review of Respiratory Disease, v.l44,pp668-674 Pottier, J.L., S.C. Pryor and R.M. Banta, 1997. "Synoptic Variability Related to Boundary Layer and Surface Features Observed During PACIFIC '93" in Atmospheric Environment, v. 31 (14), pp 2163-2173 Price, H.U., D.A. Jaffe, P.V. Doskey, I. McKendry and T.L. Anderson, 2003. "Vertical Profiles of O3, aerosols, C O and NMHCs in the Northeast Pacific during the TRACE-P and ACE-ASIA Experiments" in Journal of Geophysical Research, v. 108 (D20), pp G T E 20-1 20-18 Prospero, J.M., P. Ginoux, O. Torres, S.E. Nicholson and T.E. Gill, 2002. "Environmental Characterization of Global Sources of Atmospheric Soil Dust Identified with the NIMBUS 7 Total Ozone Mapping Spectrometer (TOMS) Absorbing Aerosol Product" in Reviews of Geophysics, v. 40 (1), pp2-l - 2-31 Prospero, J.M., 2001. "African Dust in America: Trade winds carry North African dust storms into the Caribbean and over North America, possibly affecting human health" in Geotimes, November 2001, p24-27 Prospero, J.M., 1999. "Long-term measurements of the transport of African mineral dust to the southeastern United States: Implications for regional air quality" in Journal of Geophysical Research, v. 104 (D13), ppl5917-15927 Pryor, S . C , 1998. "A Case Study of Emission Changes and Ozone Responses" in Atmospheric Environment, v. 32 (2), pp 123-131 Pryor, S.C. and D.G. Steyn, 1995. "Hebdomadal and Diurnal Cycles in Ozone Time Series from the Lower Fraser Valley, B.C." in Atmospheric Environment, v. 29 (9), ppl007-1019 Pryor, S . C , R.J. Barthelmie, R.M. Hoff, S. Sakiyama, R. Simpson and D.G. Steyn, 1997. " R E V E A L : Characterizing Fine Aerosols in the Fraser Valley, B.C." in Atmosphere-Ocean, v. 35 (2), pp 209-227 Pryor, S . C , I.G. McKendry and D.G. Steyn, 1995. "Synoptic-Scale Meteorological Variability and Surface Ozone Concentrations in Vancouver, British Columbia" in Journal of Applied Meteorology, v. 34, ppl 824-1833 Pryor, S . C , D.G. Steyn and S.N. Rogak, 1994. "Source Apportionment of VisibilityDegrading Aerosols in the Lower Fraser Valley, B.C." in Atmosphere-Ocean, v.32 (4), pp663-683  106  Rahlenbeck, S.I. and H. Kahl, 1996. "Air Pollution and Mortality in East Berlin during the Winters of 1981-1989" in International Journal of Epidemiology, v.25(6), ppl 220-1225 Ritz, B., F. Yu, G. Chapa and S. Fruin, 2000. "Effect of Air Pollution on Preterm Birth Among Children Born in Southern California Between 1989 and 1993" in Epidemiology, v.ll(5), pp502-511 Roemer, W., G. Hoek, B. Brunekreef, J. Clench-Aas, B. Forsberg, J. Pekkanen and A . Schultz, 2000. "PM10 elemental composition and acute respiratory health effects in European children (PEACE project)" in European Respiratory Journal, v. 15, pp553-559 Salmond, J.A. and I.G. McKendry, 2002. "Secondary ozone maxima in a very stable nocturnal boundary layer: observations from the Lower Fraser Valley, BC" in Atmospheric Environment, v. 36, pp5771-5782 Samet, J.M., 2002. "Air Pollution and EPIDEMIOLOGY: 'Deja Vu All Over Again?' " in Epidemiology, v. 13 (2), ppl 18-119 Samet, J.M. and J..T.K. Jaakkola, 1999. "The Epidemiologic Approach to Investigating Outdoor Air Pollution" in Holgate, S.T., J.M. Samet, H.S. Koren and R.L. Maynard (Eds.), 1999. Air Pollution and Health, Academic Press, London, pp431-460 Sassen, K., 2002. "Indirect climate forcing over the western US from Asian dust storms" in Geophysical Research Letters, v. 29 (10), ppl03-1 - 103-4 Schwartz, J., F. Laden and A. Zanobetti, 2002. "The Concentration-Response Relation between PM2.5 and Daily Deaths" in Environmental Health Perspectives, v.l 10(10), ppl0251029 Schwartz, J., 2001. "Is There Harvesting in the Association of Airborne Particles with Daily Deaths and Hospital Admissions?" in Epidemiology, v. 12(1), pp55-61 Schwartz, J., 2000. "Harvesting and Long Term Exposure Effects in the Relation between Air Pollution and Mortality" in American Journal of Epidemiology, v. 151(5), pp440-448 Schwartz, J. and L . M . Neas, 2000. "Fine Particles Are More Strongly Associated that Coarse Particles with Acute Respiratory Health Effects in Schoolchildren" in Epidemiology, v. 11 (1), PP6-10 Schwartz, J., G. Norris, T. Larson, L . Sheppard, C. Claiborne and J. Koenig, 1999. "Episodes of High Coarse Particle Concentrations Are Not Associated with Increased Mortality" in Environmental Health Perspectives, v. 107 (5), pp339-342 Schwartz, J., Dockery, D.W. and Neas, L . M . , 1996 . "Is Daily Mortality Associated Specifically with Fine Particles?" in Journal of the Air and Waste Management Association, v.46, pp927-939 a  107  Schwartz, J., C. Spix, G. Touloumi, L. Bacharova, T. Barumamdzadeh, A le Tertre, T. Piekarksi, A. Ponce de Leon, A. Ponka, G. Rossi, M . Saez and J.P Schouten, 1996 . "Methodological issues in studies of air pollution and daily counts of deaths or hospital admissions" in Journal of Epidemiology and Community Health, v. 50 (Suppl. 1), ppS3-Sl 1 b  Schwartz, J., 1995. "Air Pollution and Hospital Admissions for Respiratory Disease" in Epidemiology, v.7 (1), pp20-28 Schwartz, .1., 1994 . "Air Pollution and Hospital Admissions for the Elderly in Detroit, Michigan" in American Journal of Respiratory and Critical Care Medicine, v. 150, pp648655 a  Schwartz, J., 1994 . "PMIO, Ozone, and Hospital Admissions for the Elderly in MinneapolisSt. Paul, Minnesota" in Archives of Environmental Health, v.49 (5), pp366-374 b  Schwela, D., 2000. "Air Pollution and Health in Urban Areas" in Reviews on Environmental Health, v. 15 (1-2), ppl3-42 Shaw, G.E., 1980. "Transport of Asian Desert Aerosol to the Hawaiian Islands" in Journal of Applied Meteorology, v. 19, ppl254-1259 Simpson, J.J., G.L. Hufford, R. Servranckx, J. Berg and D. Fieri, 2003. "Airborne Asian Dust: Case Study of Long-Range Transport and Implications for the Detection of Volcanic Ash" in Weather and Forecasting, v. 18, ppl 21-141 Simpson, R., L. Denison, A. Petroeschevsky, L . Thalib and G. Williams, 2000. "Effects of ambient particle pollution on daily mortality in Melbourne, 1991-1996" in Journal of Exposure Analysis and Environmental Epidemiology, v. 10, pp488-496 Smith, M.A., B. .Talaludin, .I.E. Byles, L . Lim and S.R. Leeder, 1996. "Asthma Presentations to Emergency Departments in Western Sydney during the January 1994 Bushfires" in International Journal of Epidemiology, v. 25(6), ppl227-1236 Steyn, D.G., M . Baldi and R.M. Hoff, 1999. "the Detection of Mixed Layer Depth and Entrainment Zone Thickness from Lidar Backscatter Profiles" in Journal of Atmospheric and Oceanic Technology, v. 16, pp953-959 :  Steyn, D.G., J.W. Bottenheim and R.B. Thomson, 1997. "Overview of Tropospheric Ozone in the Lower Fraser Valley, and the PACIFIC '93 Field Study" in Atmospheric Environment, v. 31 (14), pp2025-2035 Steyn, D.G. and I.G. McKendry, 1988. "Quantitative and Qualitative Evaluation of a ThreeDimensional Mesoscale Numerical Model Simulation of a Sea Breeze in Complex Terrain" in Monthly Weather Review, v. 116, ppl914-1926 Stieb, D.M., S. Judek and R.T. Burnett, 2002 . "Meta-Analysis of Time-Series Studies of Air Pollution and Mortality: Effects of Gases and Particles and the Influence of Cause of Death, Age and Season" in Journal of the Air and Waste Management Association, v.52, pp470-484 a  108  Stieb, D.M., M . Smith-Doiron, J.R. Brook, R.T. Burnett, T. Dann, A. Mamedov and Y. Chen, 2002 . "Air Pollution and Disability Days in Toronto: Results from the National Population Health Survey" in Environmental Research Section A, v. 89, pp210-219 b  Stieb, D . M . , R.C. Beveridge, J.R. Brook, M . Smith-Doiron, R.T. Burnett, R.E. Dales, S. Beaulieu, S. Judek and A. Mamedov, 2000. "Air pollution, aeroallergens and cardiorespiratory emergency department visits in Saint John, Canada" in Journal of Exposure Analysis and Environmental Epidemiology, v. 10, pp461-477 Sugimoto, N., I. Uno, M . Nishikawa, A. Shimuzu, I. Matsui, X. Dong, Y. Chen and H. Quan, 2003. "Record heavy Asian dust in Beijing in 2002: Observations and model analysis of recent events" in Geophysical Research Letters, v. 30 (12), pp42-l - 42-4 Sun, Y., G. Zhuang, Y. Wang, L. Han, J. Guo, M . Dan, W. Zhang, Z. Wang and Z. Hao, 2004. "The air-borne particulate pollution in Beijing - concentration, composition, distribution and sources" in Atmospheric Environment, v. 38, pp5991-6004 Swap, R.""S. Ulanski, M . Cobbett'and .M. . Garstang,' 1996. ' "Temporal and spatial characteristics of Saharan dust outbreaks" in Journal of Geophysical Research, v. 101 (D2), pp4205-4220 :J  Terracini, B., 1992. "Environmental epidemiology: a historical perspective" in Elliott, P., J. Cuzick, D. English and P. Stern (Eds.), 1992. Geographical and Environmental Epidemiology: Methods for Small-Area Studies, Oxford University Press, New York, pp253263 Tolbert, P.E., M . Klein, K.B. Metzger, J. Peel, W.D. Flanders, K. Todd, J.A. Mulholland, P.B. Ryan and H. Frumkin, 2000. "Interim results of the study of particulates and health in Atlanta (SOPHIA)" in Journal of Exposure Analysis and Environmental Epidemiology, v. 10, 446-460  pp  Tratt, D.M., R..1. Frouin and D.L. Westphal, 2001. "April 1998 Asian dust event: A southern California perspective" in Journal of Geophysical Research, v.106 (D16), ppl 8371 -18379 Uematsu, M . , A. Yoshikawa, H. Muraki, K. Arao and I. Uno, 2002. "Transport of mineral and anthropogenic aerosols during a Kosa event over East Asia" in Journal of Geophysical Research, v. 107 (D7), pp3-l - 3-6 Uematsu, M . , R.A. Duce, J.M. Prospero, L. Chen, J.T. Merrill and R.L. McDonald, 1983. "Transport of Mineral Aerosol From Asia Over the North Pacific Ocean" in Journal of Geophysical Research, v. 88 (C9), pp5343-5352 Uno, I., H. Amano, S. Emori, K. Kinoshita, I. Matsui and N . Sugimoto, 2001. "Trans-Pacific yellow sand transport observed in April 1998: A numerical simulation" in Journal of Geophysical Research, v. 106 (D16), ppl8331-18344 Vaughan, J.K., C. Claiborn and D. Finn, 2001. "April 1998 Asian dust event over the Columbia Plateau" in Journal of Geophysical Research, v. 106 (D16), ppl 8381-18402 109  Vedal, S., 1997. "Ambient Particles and Health: Lines that Divide" in Journal of the Air and Waste Management Association, v.47, pp551-581 Vedal, S., 1995. Health Effects of Inhalable Particles: Implications for British Columbia, Air Resources Branch, Ministry of Environment, Lands and Parks, British Columbia, Canada Vedal, S., M . Brauer, R. White and J. Petkau, 2003. "Air Pollution and Daily Mortality in a City with Low Levels of Pollution" in Environmental Health Perspectives, v. 111(1), pp4551 Vedal, S., J. Petkau, R. White and J. Blair, 1998. "Acute Effects of Ambient Inhalable Particles in Asthmatic and Nonasthmatic Children" in American Journal of Respiratory and Critical Care Medicine, v. 157, pp 1034-1043 Villeneuve, P.J., R.T. Burnett, Y. Shi, D. Krewski, M.S. Goldberg, C. Hertzman, Y. Chen and J. Brook, 2003. "A time-series study of air pollution, socioeconomic status, and mortality in Vancouver, Canada" in Journal of Exposure Analysis and Environmental Epidemiology, v. 13,pp427-435 Wang, Y . Q . , X . Y . Zhang, R. Arimoto, J.J. Cao and Z . X . Shen, 2004. "The transport pathways and sources of P M | pollution in Beijing during spring 2001, 2002 and 2003" in Geophysical Research Letters, v. 31, ppL14110 1-4 0  Wang, X., H. Ding, L. Ryan and X . Xu, 1997. "Association between Air Pollution and Low Birth Weight: A Community-based Study" in Environmental Health Perspectives, v. 105(5), pp514-520 Wilkening, K.E., L.A. Barrie and M . Engle, 2000. "Trans-Pacific Air Pollution" in Science, v. 290 (5489), pp65-67 Wolf, C , 2002. "Urban air pollution and health: an ecological study of chronic rhinosinusitis in Cologne, Germany" in Health and Place, v. 8, ppl 29-139 Wong, C - M , S. Ma, A..T. Hedley and T - H . Lam, 2001. "Effect of Air Pollution on Daily Mortality in Hong Kong" in Environmental Health Perspectives, v. 109(4), pp335-340 Woodruff, T.J., J. Grillo and K.C. Schoendorf, 1997. "The Relationship between Selected Causes of Postneonatal Infant Mortality and Particulate Air Pollution in the United States" in Environmental Health Perspectives, v. 105(6), pp608-612 Xuan, J., I.N. Sokolik, J. Hao, F. Guo, H. Mao and G. Yang, 2004. "Identification and characterization of sources of atmospheric mineral dust in East Asia" in Atmospheric Environment, v. 38, pp6239-6252 Yeo, H - G . and J-H. Kim, 2002. "SPM and fungal spores in the ambient air of west Korea during the Asian dust (Yellow sand) period" in Atmospheric Environment, v. 36, pp54375442  110  Zanobetti, A., J. Schwartz and D. Gold, 2000. "Are There Sensitive Subgroups for the Effects of Airborne Particles?" in Environmental Health Perspectives, v. 108(9), pp841-845 Zeger, S.L., F. Dominici and J. Samet, 1999. "Harvesting-Resistant Estimates of Air Pollution Effects on Mortality" in Epidemiology, v. 10 (2), ppl 71-175 Zhang, D., Y. Iwasaka, G.'Shi, J. Zang, A. Matsuki and D. Trochkine, 2003 . "Mixture state and size of Asian dust particles collected at southwestern Japan in spring 2000" in Journal of Geophysical Research, v. 108 (D24), pp9-l - 9-7 a  Zhang, D., J. Zang, G. Shi, Y. Iwasaka, A. Matsuki and D. Trochkine, 2003 . "Mixture state of individual Asian dust particles at a coastal site of Qingdao, China" in Atmospheric Environment, v. 37, pp3895-3901 b  Zhang, X.Y., S.L. Gong, Z.X. Shen, F.M. Mei, X . X . X i , L.C. Liu, Z.J. Zhou, D. Wang, Y . Q . Wang and Y. Cheng, 2003 . "Characterization of soil dust aerosol in China and its transport and distribution during 2001 ACE-Asia: 1. Network observations" in Journal of Geophysical Research, v. 108 (D9), pp3-l - 3-13 c  Zhang, X . Y . , R. Arimoto, J.J. Cao, Z.S. An and D. Wang, 2001. "Atmospheric dust aerosol over the Tibetan Plateau" in Journal of Geophysical Research, v. 106 (D16), ppl 8471-18476 Zhang, X . Y . , R. Arimoto, G.H. Zhu, T. Chen and G.Y. Zhang, 1998. "Concentration, sizedistribution and deposition of mineral aerosol over Chinese desert regions" in Tellus, v. 50B, pp317-330 Zhang, X . Y . , R. Arimoto and Z.S. An, 1997. "Dust emission from Chinese desert sources linked to variations in atmospheric circulation" in Journal of Geophysical Research, v. 102 (D23),pp28041-28047 Zhao, W. and P.K. Hopke, 2004. "Source apportionment for ambient particles in the San Gorgonio wilderness" in Atmospheric Environment, v. 38, pp5901-5910 Zhao, T.L., S.L. Gong, X . Y . Zhang and I.G. McKendry, 2003. "Modeled size-segregated wet and dry deposition budgets of soil dust aerosol during ACE-Asia 2001: Implications for trans-Pacific transport" in Journal of Geophysical Research, v. 108 (D23), pp33-l - 33-9 Zhou, J., G. Yu, C. Jin, F. Qi, D. Liu, H. Hu, Z. Gong, G. Shi, T. Nakajima and T. Takamura, 2002. "Lidar observations of Asian dust over Hefei, China, in spring 2000" in Journal of Geophysical Research, v. 107 (D15), p5-l - 5-8 Zidek, J. and D.V Bates, 2002. "Statistical Issues and Causality" in Bates, D.V. and R.B Caton (Eds.), 2002. A Citizen's Guide to Air Pollution (2 Edition), David Suzuki Foundation, Vancouver, ppl 53-176 nd  111  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0091779/manifest

Comment

Related Items