Increasing our Understanding of the Association between Extreme Heat and
Hospital Admissions in Greater Sydney, Australia
Marissa Parry
A thesis in fulfilment of the requirements for the degree of
Doctor of Philosophy
School of Biological, Earth and Environmental Sciences
Faculty of Science
September 2018
1 2
3 4
5 Table of Contents
Acknowledgements 10
List of Tables 11
List of Figures 15
List of Abbreviations 16
Relevant Publications and Conference Presentations 17
Thesis Abstract 19
Chapter One: Introduction 21
1.1 Background 21
1.2 Objective and Aims 30
1.3 Thesis Structure 31
1.4 Thesis Scope 31
Chapter Two: Literature Review 33
2.1 Observed and Projected Changes in Ambient Temperature 33
2.1.1 Observed Changes in Average Temperature 33
2.1.2 Projected Changes in Average Temperature 34
2.1.3 Observed Changes in Temperature Extremes 35
2.1.4 Projected Changes in Temperature Extremes 37
2.1.5 Temperature Extremes and the Paris Agreement 39
2.2 Linking Extreme Heat and Human Health Outcomes 40
2.2.1 The Association between Temperature and Health Outcomes 40
2.2.2 The Association between Extreme Heat and Health Outcomes 40
2.2.3 Climate Change and Temperature-Related Health Outcomes 47
2.3 Linking Climate Change, Ambient Air Pollution and Health Outcomes 49
2.3.1 Ambient Air Pollutants: Ozone and Particulate Matter 49
6 2.3.2 Linking Meteorology and Ambient Air Pollution 50
2.3.3 The Impact of Climate Change on Ambient Air Pollution 52
2.3.4 Linking Ambient Air Pollution and Health Outcomes 54
2.3.5 Ambient Air Pollution and Health Outcomes under Climate Change 56
2.4 Ambient Air Pollution as a Confounder 58
2.5 Ambient Air Pollution as an Effect Modifier 59
2.5.1 Temperature, Ozone and Particulate Matter 60
2.5.2 Heat Waves, Ozone and Particulate Matter 63
5.2.3 Heat Waves, Ozone, Particulate Matter and Morbidity 64
Chapter Three: Methods 65
3.1 Study Setting 65
3.1.1 Greater Sydney, Australia 65
3.2 Data 66
3.2.1 Meteorological Data 67
3.2.2 Ambient Air Pollution Data 72
3.2.4 Health Data 74
3.3 Methods 76
3.3.1 Study Design 76
3.3.2 Statistical Analysis 77
3.4 Ethics 77
Chapter Four: Aim 1 78
4.1 Introduction 78
4.2 Data and Methods 82
4.2.1 Meteorological Data 82
4.2.2 Health Data 84
4.2.3 Study Design and Statistical Analysis 84
4.3 Results 87
7 4.4 Discussion 99
4.5 Chapter Conclusion 104
Chapter Five: Aim 2 105
5.1 Introduction 105
5.2 Data and Methods 108
5.2.1 Meteorological Data 108
5.2.2 Ambient Air Pollution Data 109
5.2.3 Health Data 110
5.2.4 Study Design and Statistical Analysis 111
5.3 Results 113
5.4 Discussion 123
5.5 Chapter Conclusion 129
Chapter Six: Aim 3 130
6.1 Introduction 130
6.2 Data and Methods 133
6.2.1 Meteorological Data 133
6.2.2 Ambient Air Pollution Data 134
6.2.3 Health Data 135
6.2.4 Study Design and Statistical Analysis 136
6.3 Results 138
6.4 Discussion 147
6.5 Chapter Conclusion 152
Chapter Seven: Conclusion 154
7.1 Summary of the Key Findings 154
7.2 Potential Strengths and Limitations 157
7.3 Potential Implications 161
7.4 Directions for Future Research 164
8 7.5 Conclusion 166
References 167
Appendix 210
9 Acknowledgements
I would like to express my sincere thanks and appreciation to my supervisors, Associate Professor Donna Green, Professor Andrew Hayen and Dr
Ying Zhang, for their support, advice and guidance throughout my PhD studies. I would also like to thank the Climate Change Research Centre and the ARC Centre of Excellence for Climate System Science.
I would like to thank and acknowledge Associate Professor Lisa
Alexander for her advice and insights regarding weather station data and temperature extremes; Professor Adrian Barnett for this assistance with the application of the ‘season’ package in the R Statistical Computing Environment;
James Goldie for his assistance and insights regarding public holiday data; Nicole
Mealing for her advice and assistance regarding the structure of this thesis; and
Jane Goodwin for her helpful comments and suggestions on an earlier version of this thesis.
I would like to thank and acknowledge several government agencies for providing the data used in this thesis. These include the Bureau of Meteorology for providing the meteorological data; the NSW Office of Environment and
Heritage for providing the air pollution data; the Centre for Epidemiology and
Evidence, NSW Ministry of Health, for providing the hospital admissions data from the Admitted Patient Data Collection (SAPHaRI); and the NSW Department of Education for providing the school holiday data.
Finally, I would like to express my deepest thanks and appreciation to my dear family and friends. I am ever so grateful for your continued support, encouragement and generosity over the past four years.
10 List of Tables
Table 2.1 Main population groups identified as being susceptible or vulnerable to the effects of extreme heat in Australia.
Table 3.1 Search criteria used to identify weather stations located within the SSD.
Table 3.2 Description of quality control flags provided by the Bureau of
Meteorology.
Table 3.3 NEPM standards and goals for pollutants used in this thesis.
Table 4.1 Descriptive statistics for selected heat-related EHAs during the warm season in the SSD, 2001 to 2013.
Table 4.2 Effect of heat waves first in season compared to heat waves not first in season on EHAs for acute renal failure in the SSD during the warm season, 2001 to 2013.
Table 4.3 Effect of heat waves first in season compared to heat waves not first in season on EHAs for dehydration in the SSD during the warm season, 2001 to
2013.
Table 4.4 Effect of heat waves first in season compared to heat waves not first in season on EHAs for fluid imbalance disorders in the SSD during the warm season, 2001 to 2013.
Table 5.1 Descriptive statistics for environmental variables in the SSD during the warm season, 2001 to 2013.
Table 5.2 Descriptive statistics for EHAs for three respiratory diseases in the SSD during the warm season, 2001 to 2013.
Table 5.3 Summary of heat wave characteristics for each heat wave definition used.
11 Table 5.4 The effect of heat wave days on EHAs for three respiratory diseases on days with high levels of ozone compared to days with low levels of ozone in the
SSD during the warm season, 2001 to 2013, for all ages. Effects are presented as odds ratios with their corresponding 95% confidence intervals.
Table 5.5 The effect of heat wave days on EHAs for three respiratory diseases on days with high levels of ozone compared to days with low levels of ozone in the
SSD during the warm season, 2001 to 2013, for specific age groups. Effects are presented as odds ratios with their corresponding 95% confidence intervals.
Table 6.1 Descriptive statistics for environmental variables in the SSD during the warm season, 2001 to 2013.
Table 6.2 Descriptive statistics for EHAs for six cardiovascular diseases in the
SSD during the warm season, 2001 to 2013.
Table 6.3 Summary of heat wave characteristics for each heat wave definition used.
Table 6.4 The effect of heat wave days on EHAs for six cardiovascular diseases on days with high levels of PM10 compared to days with low levels of PM10 in the SSD during the warm season, 2001 to 2013, for all ages. Effects are presented as odds ratios with their corresponding 95% confidence intervals.
Table 6.5 The effect of heat wave days on EHAs for six cardiovascular diseases on days with high levels of PM10 compared to days with low levels of PM10 in the
SSD during the warm season, 2001 to 2013, for those aged 0-64 years and 65 years and over. Effects are presented as odds ratios with their corresponding 95% confidence intervals.
Table A1 Temperature thresholds calculated for the warm season in the SSD,
2001 to 2013.
12 Table A2 Fifteen definitions used to define single days of extreme heat.
Table A3 Definitions used for heat waves with maximum temperature as the metric.
Table A4 Definitions used for heat waves with mean temperature as the metric.
Table A5 Definitions used for heat waves with minimum temperature as the metric.
Table A6 Effect of heat waves first in season compared to heat waves not first in season on EHAs for direct heat-related conditions in the SSD during the warm season, 2001 to 2013.
Table A7 The average and peak intensity of heat wave days comprising of the first heat wave of the season and those heat wave days that do not during the warm season in the SSD, 2001 to 2013.
Table A8 The average duration of heat waves first in the warm season and those heat waves not first in the SSD during the warm season, 2001 to 2013.
Table A9 Characterisation of heat wave days as high and low level ozone days.
Table A10 The effect of heat wave days on EHAs for three respiratory diseases on days with high levels of ozone compared to days with low levels of ozone in the SSD during the warm season, 2001 to 2013 for all ages and specific age
groups at lag2.
Table A11 Characterisation of heat wave days as high and low level PM10 days.
Table A12 The effect of heat wave days on EHAs for six cardiovascular diseases on days with high levels of PM10 compared to days with low levels of PM10 in the
SSD during the warm season, 2001 to 2013, for all ages at lag2.
Table A13 The effect of heat wave days on EHAs for cardiovascular diseases on days with high levels of PM10 compared to days with low levels of PM10 in the
13 SSD during the warm season, 2001 to 2013, for those aged 0-64 years and 65 years and over at lag2.
14 List of Figures
Figure 3.1 An overview of the process used to clean the hospital admissions dataset.
Figure 4.1 The association between single days of extreme heat and EHAs for acute renal failure (ARF), dehydration (DEH) and fluid imbalance disorders
(FIDs) in the SSD during the warm season, 2001 to 2013.
Figure 4.2 The association between heat wave days and EHAs for acute renal failure (Figure 4.2a), dehydration (Figure 4.2b) and fluid imbalance disorders
(Figure 4.2c) in the SSD during the warm season, 2001 to 2013.
Figure 4.3 The association between heat wave days and EHAs for acute renal failure (Figure 4.3a), dehydration (Figure 4.3b) and fluid imbalance disorders
(Figure 4.3c) in the SSD during the warm season, 2001 to 2013, after controlling for daily average temperature (the ‘added heat wave effect’).
Figure 5.1 The association between heat wave days and EHAs for three respiratory diseases with, and without, controlling for daily average ozone at lag0 in the SSD during the warm season, 2001 to 2013.
Figure 6.1 The association between heat wave days and EHAs for six cardiovascular diseases with, and without, controlling for daily average PM10 at lag0 in the SSD during the warm season, 2001 to 2013.
Figure A1 The association between single days of extreme heat and EHAs for direct heat-related conditions in the SSD during the warm season, 2001 to 2013.
Figure A2 The association between heat wave days and EHAs for direct heat- related conditions in the SSD during the warm season, 2001 to 2013, after controlling for daily average temperature (the ‘added heat wave effect’).
15 List of Abbreviations
ARF: Acute renal failure
COPD: Chronic obstructive pulmonary disease
CSIRO: Commonwealth Scientific and Industrial Research Organisation
DAGs: Directed acyclic graphs
DEH: Dehydration
EHAs: Emergency hospital admissions
GHG: Greenhouse gas
FIDs: Fluid imbalance disorders
HWD01: Heat wave defined as two or more consecutive days were maximum temperature is greater than or equal to the 90th percentile of the warm season (1
November to 31 March) during 2001 to 2013.
HWD02: Heat wave defined as two or more consecutive days were mean temperature is greater than or equal to the 90th percentile of the warm season (1
November to 31 March) during 2001 to 2013.
HWD03: Heat wave defined as two or more consecutive days were minimum temperature is greater than or equal to the 90th percentile of the warm season (1
November to 31 March) during 2001 to 2013.
IPCC: Intergovernmental Panel on Climate Change
NEPM: National Environment Protection (Ambient Air Quality) Measure
NOAA: National Oceanic and Atmospheric Administration
NO2: Nitrogen dioxide
NOx: Oxides of nitrogen
NSW: New South Wales
PM10: Particles with an aerodynamic diameter less than 10m
16 PM2.5: Particles with an aerodynamic diameter less than 2.5m
PM0.1: Particles with an aerodynamic diameter less than 0.1m
RCPs: Representative Concentration Pathways
US EPA: United States Environmental Protection Authority
SSD: Sydney Statistical Division
Tmax: Maximum temperature
Tmean: Mean temperature
Tmin: Minimum temperature
VOCs: Volatile organic compounds
WSD: Weather Station Directory
17 Relevant Publications and Conference Presentations
The following conference presentations are based on work presented in this thesis.
Parry M*, Green D, Zhang Y and Hayen A. 2017. Heat Waves and Hospital
Admissions for Respiratory Diseases in Sydney, Australia: Effect Modification by
Ozone. 29th Annual Conference of the International Society for Environmental
Epidemiology, Sydney, Australia, 24-28 September 2017. Poster Presentation.
Parry M*, Green D, Zhang Y and Hayen A. 2016. When is Heat ‘Extreme’? The
Effect of Definition Choice on the Association between Extreme Heat and Cause-
Specific Morbidity in Sydney, Australia. 28th Annual Conference of the
International Society for Environmental Epidemiology, Rome, Italy, 1-4
September 2016. Poster Presentation.
Parry M*, Green D, Hayen A, Zhang Y. 2016. When is a Hot day, Hot? The
Importance of Definitions in Examining the Impact of Extreme Heat on Heat-
Related Hospital Admissions in Sydney, Australia. Joint National Conference of the Australian Meteorological and Oceanographic Society and ARC Centre of
Excellence for Climate System Science, Melbourne, Australia, 8-11 February
2016. Poster Presentation.
Note: *Denotes presenter
18 Thesis Abstract
A number of studies have examined the association between extreme heat and hospital admissions in Australia and other regions, but uncertainties and knowledge gaps remain. The overall objective of this thesis was to increase our understanding of the association between extreme heat and hospital admissions in
Greater Sydney, Australia. To achieve this, three epidemiological studies were conducted. The respective aims of these studies were to: examine and compare the impact of single and consecutive days of extreme heat, including high minimum temperatures, on heat-related hospital admissions for a suite of extreme heat definitions; examine whether ambient ozone modifies the short-term association between heat waves and hospital admissions for certain respiratory diseases; and examine whether ambient particulate matter modifies the short-term association between heat waves and hospital admissions for certain cardiovascular diseases. A time-stratified case-crossover study design with conditional logistic regression was used to estimate the association between extreme heat and emergency hospital admissions for specific diseases during the warm season (1 November to
31 March) for 2001 to 2013, adjusting for potential confounders where relevant.
Positive associations were observed between single and consecutive days of extreme heat and emergency hospital admissions for a range of heat-related conditions, although the strength and statistical significance of these associations differed across extreme heat definitions. For certain conditions, the strongest associations were observed when extreme heat was defined using minimum temperature (i.e. single and consecutive warm nights). Some evidence that ozone and particulate matter modifies the association was also found, with stronger heat wave effects observed on days with high compared to low-levels of ozone and
19 particulate matter for certain respiratory and cardiovascular diseases respectively.
Such evidence, however, showed inconsistencies and differed across age group, lag and heat wave definition used. The findings of this thesis have important implications within the context of climate change, with the number of hot days and warm nights expected to increase in the future, as well as the frequency, intensity and duration of heat waves. Such localised work is vital to inform climate change adaptation planning in the health sector to prepare for a future of unprecedented extreme heat.
20 Chapter One: Introduction
This Chapter provides a background to this thesis; and outlines its objective, aims, structure and scope.
1.1 Background
The Earth’s climate is warming. Since 1880, global average surface temperature has warmed at an average rate of 0.07°C per decade, increasing
0.85°C (0.65°C to 1.06°C) during 1880 to 2012 (IPCC 2014; NOAA 2017). In
2016, global average surface temperature was 0.45° to 0.56°C above the 1981 to
2010 average, and more than 1°C (1.06° to 1.21°C) above the average of pre- industrial times (Sánchez-Lugo et al. 2017). This led to 2016 been declared as the warmest year on record, representing the fifth time since 2000 that global annual average temperature had reached a record high (NOAA 2017). A similar warming trend has also been observed for Australia, with national average surface air temperature increasing by approximately 1°C since 1910 (Bureau of Meteorology and CSIRO 2017). Australia had its warmest year on record in 2013, with national average temperature 1.20°C above the 1961 to 1990 average (Bureau of
Meteorology 2014).
Warming trends have also been observed for a range of hot temperature extremes. The number of hot days and warm nights has increased across most land areas globally, as has the frequency, intensity and duration of warm spells and summer heat waves (Alexander et al. 2006; Perkins et al. 2012; Donat et al.
2013). A similar pattern of change has also been observed for hot temperature extremes in Australia (Alexander and Arblaster 2009; Perkins and Alexander
2013; Bureau of Meteorology and CSIRO 2017). Globally, the magnitude of the
21 warming trend has generally been found to be larger for minimum temperature extremes than for maximum temperature extremes (Alexander et al. 2006; Perkins et al. 2012; Donat et al. 2013). In Australia, maximum and minimum temperatures have increased by approximately 0.8°C and 1.1°C since 1910 respectively
(Bureau of Meteorology and CSIRO 2014). During 2000 to 2014, the number of new hot temperature records set in Australia was twelve times greater than the number of new cold temperature records (Lewis and King 2015). Observations show that temperature records for high minimum temperatures have been broken at a faster rate than records for high maximum temperatures since the 1950s
(Lewis and King 2015).
The observed warming trends in average temperature and hot temperature extremes have been largely attributed to the increase in anthropogenic greenhouse gas (GHG) emissions since the pre-industrial era (IPCC 2014). Despite international efforts to reduce GHG emissions in recent decades, levels have continued to rise, with global average levels of carbon dioxide reaching 403.3 ppm in 2016 (World Meteorological Organization 2017). Projections show that such warming trends will continue in the future both globally and in Australia, with the magnitude and intensity of this warming expected to scale in accordance with future levels of GHG emissions (IPCC 2012; Sillmann et al. 2013; Cowan et al. 2014; IPCC 2014; Alexander and Arblaster 2017). In Australia, warming for certain hot minimum temperature extremes, including warm and tropical nights, is projected to be more pronounced than for certain hot maximum temperature extremes, including hot days, by the end of this century (Alexander and Arblaster
2017).
22 The 2015 landmark Paris Agreement aims to limit the increase in global average temperature to 1.5°C to 2°C above the pre-industrial average, acknowledging that achieving this target would reduce the effects of climate change (Paris Agreement 2015, Article 2.1(a)). However, recent studies have shown that an increase of 2°C in global average temperature is associated with much larger increases in regional hot temperature extremes across most land areas, including Australia (Seneviratne et al. 2016; Lewis et al. 2017). This is of paramount concern given that there is a 95 per cent chance that the global average temperature target of 2°C will be exceeded before the end of this century (Raftery et al. 2017). This means that both the global and Australian population face a future of exposure to unprecedented extreme heat.
Extreme heat is a significant public health concern both worldwide and in
Australia. It is Australia’s deadliest natural hazard, responsible for more deaths than all other natural disasters combined during 1900 to 2011 (Coates at el. 2014).
Heat waves, which are prolonged periods of extreme heat, can have adverse impacts on human health. For example, the 2003 European heat wave in August led to 40,000 excess deaths across the continent, with around 15,000 excess deaths occurring in France alone (Fouillet et al. 2006; García-Herrera et al. 2010).
Further, the 2010 Russian heat wave is estimated to have led to 55,000 excess deaths (Barriopedro et al. 2011). The 2009 south-east Australian heat wave led to
374 excess deaths in Victoria, representing a 62 per cent increase in total mortality
(Victorian Government Department of Human Services 2009). This heat wave was associated with more deaths than the 2009 Victorian Black Saturday bushfires and 2010/2011 Queensland floods, which resulted in 173 and 33 deaths respectively (Victorian Bushfires Royal Commission 2010; Queensland Floods
23 Commission of Inquiry 2012). Extreme heat exposure has also been associated with several other adverse health outcomes in Australia including hospital admissions, emergency department presentations and ambulance call-outs
(Victorian Government Department of Human Services 2009; Nitschke et al.
2011; Wilson et al. 2013; Jegasothy et al. 2017).
Extreme heat exposure can adversely affect an individual’s health by directly causing a heat-related illness, such as dehydration or heatstroke, or by indirectly exacerbating a pre-existing medical condition, including cardiovascular and respiratory conditions. When exposed to extreme heat, our bodies thermoregulate to maintain an internal core body temperature of approximately
37°C, with sweating the most efficient means by which our bodies shed heat
(Glazer 2005; Becker and Stewart 2011). The body can adapt to high temperatures after a number of weeks, undergoing a number of physiological responses including enhanced salt retention and increased rates of sweating (Glazer 2005).
Indeed, populations are known to live in areas with extreme hot temperatures across the globe, such as Dallol in Ethiopia (Middleton 2003). This would seem to suggest that humans have the ability to adapt to, and survive in, a world with more frequent and intense hot temperature extremes. However, the extent to which populations will adapt to warmer temperatures under climate change is not clear, although some studies have shown that the body’s ability to physiologically acclimatise is limited (Sherwood and Huber 2010; Hanna and Tait 2015). It is therefore vital that we continue to refine and enhance our understanding of the impact of extreme heat on human health to prepare for worsening extreme heat events under climate change.
24 There is no universal or standard definition of extreme heat. As such, studies examining the health impacts of extreme heat, including heat waves, have constructed and used a wide range of definitions to define heat exposure. There are, however, common elements to each definition and these include: the temperature metric (e.g. ambient temperature, apparent temperature, heat index), the temperature threshold (i.e. absolute or relative) and, for heat waves, the length or duration (e.g. ≥ 2, 3 or 4 consecutive days). In relation to the temperature metric, many studies have considered heat to be extreme when daily maximum or mean temperature (ambient or apparent) is greater than or equal to a specific temperature or threshold (e.g. Hansen et al. 2008b; Vaneckova and Bambrick
2013). As a result, much less is known about the association between extreme high minimum temperatures and health outcomes. This is concerning given the known relevance and importance of high overnight temperatures to human health.
The disrupted sleep patterns associated with exposure to such temperatures
(Okamoto-Mizuno et al. 2010; Lan et al. 2014), along with the need for the body to thermoregulate during the night, means that individuals are unable to naturally rest and recover from the day, placing our bodies under further stress.
The impact of heat waves on health outcomes can differ in accordance with their characteristics, such as their intensity, duration and timing within the summer season. Studies have shown that those heat waves that are more intense and longer in duration can have a greater impact on mortality risk (Anderson and
Bell 2009; Anderson and Bell 2011; D’Ippoliti et al. 2010; Tong et al. 2014), although evidence for the impact of heat wave timing is somewhat less consistent
(Anderson and Bell 2011; Kent et al. 2014; Tong et al. 2014). Some studies have suggested that the impact of heat waves on mortality risk can be captured more
25 accurately by separating the effects of heat waves into two measures: the ‘main effect’ and the ‘added effect’ (Hajat et al. 2006; Gasparrini and Armstrong 2011;
Rocklöv et al. 2012).
Local meteorological conditions can influence the creation, accumulation and dispersion of ambient air pollutants (Fiore et al. 2012). As such, climate change is expected to directly affect future levels of ambient air pollutants.
Studies conducted in the Northern Hemisphere have projected an increase in concentrations of ambient ozone in Europe and North America under climate change, although geographical differences have been observed (Bell et al. 2007;
Meleux et al. 2007). The projections for ambient particulate matter are less certain, although some studies have projected that levels will increase in Europe and North America in specific areas (Jacob and Winner 2009; Huszar et al. 2011;
Shen et al. 2017). Longer and more intense air pollution episodes have also been projected under climate change in the United States, as well as more frequent and longer extreme ozone episodes in Europe (Mickley et al. 2004; Meleux et al.
2007). Few studies have examined how climate change will alter ambient air pollution levels in Australia (e.g. Cope et al. 2008). Cope et al. (2008) projected that peak ozone concentrations will increase in the Greater Sydney Metropolitan region under climate change, as well as the number of exceedances of the current
1-hour and 4-hour national health standards.
Exposure to ambient air pollution can have adverse impacts on human health. Positive, short-term associations have been observed between ambient ozone and respiratory mortality and hospital admissions (Gryparis et al. 2004;
Bell et al. 2005; Ji et al. 2011; Peng et al. 2013). Further, positive, short-term associations have been observed between ambient particulate matter (including
26 PM10 and PM2.5) and cardiovascular mortality and hospital admissions (Samet et al. 2000; Anderson et al. 2004; Barnett et al. 2006; Dominici et al. 2006; Peng et al. 2008; Stafoggia et al. 2013). Some studies have estimated the health impacts associated with increased exposure to ambient ozone and particulate matter concentrations under climate change at the global and regional level (e.g. Bell et al. 2007; Fang et al. 2013; Orru et al. 2013; Silva et al. 2017). These studies have largely projected an increase in mortality or morbidity associated with exposure to these pollutants, although the magnitude of this increase is expected to vary geographically.
Heat waves and extreme air pollution episodes can coincide (Schnell and
Prather 2017). This is because these two events occur under similar meteorological conditions including high temperatures, low wind speeds and low levels of precipitation (Fiore et al. 2012; Schnell and Prather 2017). For example, during the 2003 European heat wave, extreme levels of ambient ozone were also observed across most parts of Europe (Fiala et al. 2003; Solberg et al. 2008).
Despite the known independent effects of heat waves and air pollution on health outcomes, few studies have investigated the potential joint, or interactive effects of these two environmental exposures. This is concerning given that the joint effect of weather and air pollution on health outcomes is thought to be greater than the risk derived from their individual impacts (Zanobetti and Peters 2015).
Indeed, as Schnell and Prather (2017) note, the tendency for heat waves and extreme air pollution episodes to coincide is of significant concern for human health as populations become exposed to both the hottest temperatures and highest levels of air pollution simultaneously (Schnell and Prather 2017). There is some
27 suggestion that an interactive effect between air pollution and temperature may be plausible on a biological level (Gordon 2003).
Little is known about whether air pollution modifies the association between extreme heat, particularly heat waves, and adverse health outcomes. A few studies from the Northern Hemisphere have investigated whether ozone modifies the association between temperature and all-cause or cardiovascular mortality (Ren et al. 2007; Burkart et al. 2013; Breitner et al. 2014). These studies have generally observed stronger heat effects at higher levels of ozone, although such evidence of effect modification was not consistency statistically significant.
Further, a few studies from Europe, North America, Asia and New Zealand have also assessed whether particulate matter (PM10) modifies the association between temperature and mortality (Hales et al. 2000; Basu et al. 2008; Burkart et al. 2013;
Breitner et al. 2014; Li et al. 2015). The results of these studies have been inconsistent, with some finding evidence of effect modification (e.g. Li et al.
2015) and others not (e.g. Hales et al. 2000; Basu et al. 2008).
Much less is known about whether ambient air pollution modifies the association between heat waves and adverse health outcomes. A few studies have estimated the number or proportion of excess deaths attributable to ozone or particulate matter exposure during specific heat wave events and summer seasons, such as the 2003 European heat wave and summer, and the 2004 Brisbane heat wave (e.g. Fischer et al. 2004; Stedman 2004; Tong et al. 2010a). Others have assessed the potential interactive effects of ozone and high temperatures during the 2003 European heat wave (e.g. Dear et al. 2005; Filleul et al. 2006). Time- series studies examining whether ambient air pollution modifies the association between heat waves and health outcomes are limited. One European study
28 examined whether ozone and particulate matter (PM10) modifies the association between heat waves and total and cause-specific mortality for nine European cities (Analitis et al. 2014). Some evidence of effect modification was found, with stronger heat wave effects observed at high compared to low-levels of ozone for total and cardiovascular mortality, although such evidence was not statistically significantly (Analitis et al. 2014). Stronger heat wave effects were also observed at high compared to low-levels of particulate matter (PM10) on total, cardiovascular and respiratory mortality (Analitis et al. 2014).
One of two epidemiological study designs is often employed to examine the short-term association between extreme heat and adverse health outcomes, namely, the time-series design or the case-crossover design. The time-series design typically uses Poisson regression to explore the association between changes in daily temperature and daily counts of deaths or hospital admissions
(Bhaskaran et al. 2013). The case-crossover design is equivalent to a matched pair case-control design: it uses conditional logistic regression to compare a case’s exposure on the day of an adverse health event (e.g. hospital admission) to their exposure on control days (or referent times) that are selected before and/or after the event (Janes et al. 2005; Bell et al. 2008; Barnett and Dobson 2010). An important difference between these two designs in the manner in which time- dependent confounders are controlled for, including long-term trends and seasonality (Janes et al. 2005). The time-series design controls for these confounders by using statistical methods, while these confounders are controlled for by study design using the case-crossover approach (Janes et al. 2005).
Sydney, Australia’s most populous city, experienced its hottest summer season on record in 2016/2017, with mean temperature reaching 2.8°C above
29 average (Bureau of Meteorology 2017a). Several extreme temperature records were also broken at various monitoring stations across the city, with records broken for the number of hot days and warm nights (Bureau of Meteorology
2017a). The number of hot days is expected to increase for the city under climate change, as well as the frequency and duration of heat waves (Cowan et al. 2014;
NSW Office of Environment and Heritage 2015a). Future urban expansion and land use change is expected to enhance the warming of minimum temperatures under climate change (Argüeso et al. 2014). Levels of ambient ozone and particulate matter can exceed national health standards, particularly during the warm season (Linfoot et al. 2010; NSW Environment Protection Authority 2013).
Projections show that ozone levels are expected to increase under climate change in the city, increasing the ozone-related health burden (Cope et al. 2008; Physick et al. 2014).
1.2 Objective and Aims
The overall objective of this thesis was to increase our understanding of the association between extreme heat and hospital admissions in Greater Sydney,
Australia. To achieve this, three epidemiological studies were designed and conducted to address the knowledge gaps and uncertainties regarding the association identified in Section 1.1 of this thesis. The respective aims of these three studies are outlined below.
30 Aim 1: To examine and compare the impact of single and consecutive days of extreme heat, including high minimum temperatures, on heat-related hospital admissions in Greater Sydney, Australia, for a suite of extreme heat definitions
(Chapter Four).
Aim 2: To examine whether ozone modifies the short-term association between heat waves and hospital admissions for certain respiratory diseases in Greater
Sydney, Australia (Chapter Five).
Aim 3: To examine whether particulate matter modifies the short-term association between heat waves and hospital admissions for certain cardiovascular diseases in
Greater Sydney, Australia (Chapter Six).
1.3 Thesis Structure
This thesis is divided into seven chapters. Chapter Two reviews the literature pertinent to this thesis. Chapter Three outlines the study setting, data and study design used in thesis. Chapters Four, Five and Six present and discuss the findings of the three respective thesis aims outlined in Section 1.2. Chapter Seven outlines the conclusion of this thesis.
1.4 Thesis Scope
A range of meteorological elements may affect human health, such as ambient temperature, humidity and wind. This thesis considers one of these elements, namely, ambient temperature. Specifically, it focuses on the association between extreme high ambient temperature and hospital admissions. It does not,
31 therefore, consider any potential combined effects of these meteorological elements, such as ambient temperature and humidity.
Similarly, a range of air pollutants may adversely affect human health, including ozone, particulate matter, sulphur dioxide, carbon monoxide and nitrogen dioxide. This thesis focuses on two of these pollutants, namely, ozone and particulate matter. This is because elevated levels of these two pollutants have been reported to coincide with heat waves (e.g. 2003 European heat wave); and due to their established association with adverse health outcomes.
32 Chapter Two: Literature Review
This Chapter is divided into five sections and reviews the literature pertinent to this thesis.
2.1 Observed and Projected Changes in Ambient Temperature
This section reviews the observed and projected changes in ambient temperature at the global, national and local level, with a particular focus on hot temperature extremes.
2.1.1 Observed Changes in Average Temperature
The Intergovernmental Panel on Climate Change (IPCC) has declared that the warming of the Earth’s climate is ‘unequivocal’ (IPCC 2014, p.2). Since 1880, global average surface temperature has warmed at an average rate of 0.07°C per decade, with this rate more than doubling to 0.17°C per decade since 1970
(NOAA 2017). In 2016, global average surface temperature was 0.94°C warmer than the twentieth century average, making it the hottest year on record (NOAA
2017). Global average annual land surface temperature was also at a record high in 2016, reaching 1.43°C above the twentieth century average (NOAA 2017).
Average land surface temperature has increased at a faster rate than ocean temperature since 1880, increasing 0.10°C and 0.06°C per decade respectively
(Dahlman 2017; NOAA 2018a; NOAA 2018b).
A similar warming trend has also been observed for Australia, with a number of temperature records broken in recent times. Average national surface air temperature has increased by approximately 1°C since national observational records commenced in 1910 (Bureau of Meteorology and CSIRO 2017;
Australian Government and CSIRO 2018). Australia had its warmest year on
33 record in 2013, with observations showing annual average surface air temperature to be 1.20°C above the 1961 to 1990 average (Bureau of Meteorology 2014).
Since 2005, Australia has experienced seven of its ten warmest years on record
(Bureau of Meteorology 2018a). The recent 2016/2017 summer season was the hottest on record for Australia’s largest city, Sydney, with mean temperature reaching 2.8°C above average (Bureau of Meteorology 2017a). Observations show that Sydney’s average temperature has been increasing since the 1960s, with much of this warming occurring in recent times (NSW Office of Environment and
Heritage 2015a).
The observed warming of the Earth’s climate has largely been attributed to the increase in anthropogenic GHG emissions since the pre-industrial era, which has led to unprecedented levels of carbon dioxide, methane and nitrous oxide in the atmosphere (IPCC 2014). Levels of atmospheric carbon dioxide reached 403.3 ppm in 2016, representing an increase of 145 per cent since 1750 or pre-industrial times (World Meteorological Organization 2017). Despite international efforts to reduce anthropogenic GHG emissions in recent decades, levels have continued to rise, but the rate of increase has somewhat slowed in recent times (den Elzen et al.
2017; Olivier et al. 2017). In 2016, total global GHG levels grew by around 0.5 per cent, marking the slowest growth since the beginning of the 1990s, apart from those years in global economic recession (Oliver et al. 2017).
2.1.2 Projected Changes in Average Temperature
Global average surface temperature is projected to continue to increase throughout this century, although the magnitude of this increase is expected to depend on past, present and future levels of anthropogenic GHGs, as well as natural variations in the Earth’s climate (IPCC 2014). Projections show that for all
34 Representative Concentration Pathways (RCPs), global average surface temperature is expected to increase by 0.3°C to 0.7°C during the period of 2016 to
2035, relative to the period of 1986 to 2005 (Kirtman et al. 2013). Projections for the period of 2081 to 2100 differ for each of the RCPs, with the greatest increases in global average temperature expected for those pathways that allow for the greatest increases in anthropogenic GHG levels (Collins et al. 2013). These projections include: 0.3°C to 1.7°C (RCP2.6); 1.1°C to 2.6°C (RCP4.5); 1.4°C to
3.1°C (RCP6.0); and 2.6°C to 4.8°C (RCP8.5) (Collins et al. 2013). Sanford et al.
(2014) observed that current GHG emission levels are largely following RCP8.5.
National average air temperature is also expected to continue to increase in the future in Australia, with projections showing an increase of 0.6°C to 1.3°C by
2030, relative to the period of 1986 to 2005, under RCP4.5 (CSIRO and Bureau of
Meteorology 2015). Similar to global patterns, national projections to the end of the twenty-first century also differ for each of the RCPs, with annual average temperature projected to increase by 2090 to between: 0.6°C to 1.7°C (RCP2.6);
1.4°C to 2.7°C (RCP4.5); and 2.8°C to 5.1°C (RCP8.5) (CSIRO and Bureau of
Meteorology 2015).
2.1.3 Observed Changes in Temperature Extremes
Changes have been observed for a wide range of both cold and hot temperature extremes at the global level. A significant decrease in the annual number of cool nights has been observed across most land areas since 1950, while a significant increase in the number of warm nights has been found (Alexander et al. 2006; Donat et al. 2013). A similar change in the annual number of cool and warm days has also been reported, with a reduction in the number of cool days and increase in the number of warm days observed (Donat et al. 2013). Further,
35 since 1950, the absolute value of the coldest day and coldest night has significantly increased over several regions, including Australia (Donat et al.
2013). A similar, but weaker trend has been observed for value the warmest day and warmest night, increasing around 1°C on average globally since 1950s (Donat et al. 2013). The intensity, frequency and duration of warm spells and heat waves has increased since 1950 (Perkins et al. 2012), while the frequency and duration of cold spells has generally decreased (Alexander et al. 2006; Donat et al. 2013).
Across these studies, the magnitude of the warming trend for minimum temperature extremes (e.g. single and consecutive warm nights) is generally much larger than for maximum temperature extremes (e.g. single and consecutive hot days), with minimum temperature extremes observed to be warming at a faster rate (Alexander et al. 2006; Perkins et al. 2012; Donat et al. 2013).
Similar warming trends have also been observed for temperature extremes in Australia. Since 1910, national maximum and minimum temperatures have increased by 0.8°C and 1.1°C respectively (Bureau of Meteorology and CSIRO
2014). The frequency of warm nights and hot days has increased across most parts of Australia, as has the intensity, duration and frequency of heat waves
(Alexander and Arblaster 2009; Perkins and Alexander 2013; Steffen et al. 2014;
Bureau of Meteorology and CSIRO 2017). The first heat wave of the summer season has also been shown to now occur earlier (Steffen et al. 2014). During the period of 2000 to 2014, the number of new hot temperature records set in
Australia was twelve times greater than the number of new cold temperature records (Lewis and King 2015). Observations show that temperature records for high minimum temperatures have been broken at a faster rate than records for high maximum temperatures since the 1950s (Lewis and King 2015).
36 A few Australian studies have examined the extent to which the occurrence of specific temperature record-breaking events, such as the unprecedented temperatures during 2013 and the summer of 2012/2013, can be attributed to anthropogenic influences on the climate (e.g. Lewis and Karoly
2013; Lewis and Karoly 2014; Perkins et al. 2014). Lewis and Karoly (2013) showed that the extreme summer temperatures of 2013 were between 2.5 to 5 times more likely to occur due to anthropogenic influence. The risk of heat wave intensity and frequency experienced during the summer of 2012/2013 was made 2 and 3 times more likely, respectively, due to human influence (Perkins et al.
2014). Lewis and King (2015) showed that recent Australian high temperature records would not have been broken without anthropogenic influence.
2.1.4 Projected Changes in Temperature Extremes
The 2012 Special Report of the IPCC on extreme events showed how changes in the temperature distribution, including the mean, variance and shape of the distribution, can affect temperature extremes (IPCC 2012). It is expected that such changes in the temperature distribution will alter the frequency, intensity, duration and timing of future temperature extremes (IPCC 2012).
Temperature extremes are expected to continue to warm throughout this century (IPCC 2012). At the global scale, a decrease in the number of cold days and cold nights is projected across all land areas by the end of this century, while an increase in the number of warm days and warm nights projected (Sillmann et al. 2013). An increase in the value of the coldest night and hottest day is projected during the twenty-first century, with projections showing an increase of 1.8°C and
1.4°C (RCP2.6), 3.4°C and 2.7°C (RCP4.5) and 6.7°C and 5.4°C (RCP8.5) respectively (Sillmann et al. 2013). The projected changes for minimum
37 temperature extremes (e.g. cool nights and warm nights) are generally more pronounced than those for maximum temperature extremes (e.g. cool days and hot days) (Sillmann et al. 2013). Warm spells and heat waves are also projected to increase in intensity, duration and frequency across most regions (IPCC 2012).
A similar pattern of change is also projected for temperature extremes in
Australia. The number of warm nights and hot days is projected to increase
(Alexander and Arblaster 2017; Bureau of Meteorology and CSIRO 2017).
Relative to 1981 to 2010, the number of hot days (> 35°C) in Sydney is projected to increase from 3.1 (1995) to 4.5 (2090, RCP 2.6); 6.0 (2090, RCP4.5) and 11
(2090, RCP8.5) (CSIRO and Bureau of Meteorology 2015). This increase is not expected to be uniform across the city, with Western Sydney and the Hawkesbury region projected to experience the greatest increases in the number of hot days
(NSW Office of Environment and Heritage 2015a). Most of these additional hot days are expected to occur during the months of summer and spring, but will also likely occur in autumn towards the end of this century (NSW Office of
Environment and Heritage 2015a).
The frequency, intensity and duration of summer heat waves is projected to increase in the future across Australia, with the largest changes projected under
RCP8.5 (Cowan et al. 2014). The changes in these heat wave characteristics are expected to be more pronounced for the northern tropical areas of Australia, compared to the southern areas (Cowan et al. 2014). Under RCP4.5(RCP8.5),
Sydney is projected to have an additional ~14 (> 42) heat wave days per summer season by the end of the century (Cowan et al. 2014). Heat waves are also projected to be ~7 (12 to 14) days in duration under RCP4.5(RCP8.5) (Cowan et al. 2014). The expected changes in heat wave intensity are less pronounced, with
38 little warming under RCP4.5 and between ~1°C to 2°C warming under RCP8.5
(Cowan et al. 2014).
2.1.5 Temperature Extremes and the Paris Agreement
The 2015 landmark Paris Agreement is one of the most significant legal instruments in international climate change law to date. This is due, in part, to the
Agreement’s aim to limit the warming of global average temperature to between
1.5°C to 2°C above the pre-industrial average (Paris Agreement 2015, Article
2.1(a)). In limiting the increase in global average temperature to this degree, the
Agreement acknowledges that this would reduce the effects of climate change
(Paris Agreement 2015, Article 2.1(a)).
Even if this ambitious temperature target is met, it does not ensure that temperature extremes will also warm to this degree, particularly over land at the regional level (Orlowsky and Seneviratne 2012; Seneviratne et al. 2016). For example, Seneviratne et al. (2016) showed that an increase of 2°C in global average temperature is associated with much larger increases in regional temperature extremes across most land areas. An increase of 2°C in global average temperature is associated with an increase of 5.5°C in cold temperature extremes in the Arctic region, and a 3°C increase in hot temperature extremes in the Mediterranean region (Seneviratne et al. 2016). This is also true for projected changes in hot temperature extremes in Australia. Lewis et al. (2017) showed the potential for future maximum temperatures to reach 50°C in Sydney and
Melbourne under 2°C of warming in global average temperature.
39 2.2 Linking Extreme Heat and Human Health Outcomes
The link between weather and human health is well known. This section reviews the association between extreme heat and human health outcomes, with a particular focus on heat waves.
2.2.1 The Association between Temperature and Health Outcomes
There is a well-established association between ambient temperature and adverse health outcomes, particularly mortality. A non-linear ‘U’ shaped relationship has been reported in several studies, where an increased risk of death is often observed above (high temperature) and below (cold temperature) a certain temperature threshold (Anderson and Bell 2009; Gasparrini et al. 2015a).
Temperature thresholds for heat effects differ across regions, countries and cities, with higher absolute temperature thresholds often observed in locations with warmer summer temperatures or close to the equator (Gosling et al. 2007; Guo et al. 2014). The observed differences in these thresholds indicate that populations can adapt and acclimatise to different climates, although such differences are also likely to be influenced by a range of demographic, behavioural, technological and cultural factors specific to each geographic location (Hajat and Kosatky 2010). A short lag effect has also been observed in the association, where heat effects have generally been found to occur on the day of exposure or between 1 to 3 days after heat exposure (Yu et al. 2011; Bao et al. 2016).
2.2.2 The Association between Extreme Heat and Health Outcomes
Extreme heat, including heat waves, can have profound, adverse impacts on human health. Several studies have estimated the number of excess deaths or hospital admissions associated with major heat wave events. For example, the
40 2003 European heat wave is estimated to have led to 40,000 excess deaths across the continent, with around 15,000 excess deaths estimated to have occurred in
France alone (Fouillet et al. 2006; García-Herrera et al. 2010). The 2010 Russian heat wave led to around 55,000 excess deaths (Barriopedro et al. 2011). The 1995
Chicago heat wave resulted in an estimated 692 excess deaths and 1,072 excess hospital admissions (Semenza et al. 1999; Kaiser et al. 2007).
Extreme heat is Australia’s deadliest natural hazard. An estimated 4,555 heat-related deaths occurred in Australia between 1900 and 2011, constituting 55 per cent of the total number of deaths attributable to natural hazards during the period (Coates et al. 2014). Recent heat waves in Australia have also had significant, adverse impacts on human health. For example, the January 2009 south-east Australian heat wave led to 374 excess deaths in Victoria, representing an increase of 62 per cent in total mortality (Victorian Government Department of
Human Services 2009). A 25 per cent increase in ambulance emergency cases was also observed during this event, as well as a 12 per cent increase in emergency department presentations (Victorian Government Department of Human Services
2009). In Adelaide, 33 excess deaths were estimated to have occurred, as well as a
16 per cent increase in ambulance call-outs, 8.2 per cent increase in hospital admissions and 2.4 per cent increase in emergency department presentations
(Nitschke and Tucker 2009). The 2004 Brisbane heat wave is estimated to have led to 75 excess deaths (Tong et al. 2010a).
Populations live in locations with extreme climates throughout the globe, including those with extreme hot temperatures, such as Dallol, Ethiopia
(Middleton 2003). This would seem to suggest that populations have the ability to adapt to, and survive in, a world with more frequent and intense hot temperature
41 extremes. Studies have shown, however, that our capacity to physiologically acclimatise to the projected changes in temperature under climate change is limited (Sherwood and Huber 2010; Hanna and Tait 2015). Indeed, Hanna and
Tait (2015) argue that much of our adaptation to climate change must come from reducing our exposure to heat through changes in social, cultural, technological and behavioural norms.
2.2.2.1 Heat Wave Characteristics
Time-series studies have also examined the association between heat waves and adverse health outcomes (Anderson and Bell 2011; Guo et al. 2017).
Such studies have allowed researchers to consider how the impacts of heat waves might differ in accordance with their characteristics. Heat waves that are more intense and longer in duration have been found to have greater impacts on mortality in Europe, the United States and Australia (Anderson and Bell 2009;
Anderson and Bell 2011; D’Ippoliti et al. 2010; Tong et al. 2014). Research considering the impact of heat wave timing has been less consistent (Anderson and Bell 2011; Tong et al. 2014). Anderson and Bell (2011) observed that heat waves that occur first or earlier within the season have a greater impact on mortality in the United States, while Tong et al. (2014) found that those heat waves that occur later in the season have a greater impact on mortality in
Brisbane, Australia. A recent international study observed a decrease in mortality risk throughout the summer season (Gasparrini et al. 2016). Fewer time-series studies have considered how heat wave characteristics, particularly heat wave timing, affect the association between heat waves and morbidity.
42 2.2.2.2 The Added Heat Wave Effect
Some studies have suggested that the impact of heat waves on mortality risk can be captured more accurately by separating the effects of heat waves into two measures: the ‘main’ effect and the ‘added’ effect (Hajat et al. 2006;
Gasparrini and Armstrong 2011; Rocklöv et al. 2012). The main effect estimates the mortality risk associated with single days of high temperatures, while the added effect estimates the mortality risk associated with prolonged exposure to extreme heat over several consecutive days (Gasparrini and Armstrong 2011;
Zeng et al. 2014). Some evidence of an added heat wave effect on mortality risk has been found, although the magnitude of this effect has differed across studies
(Gasparrini and Armstrong 2011; Barnett et al. 2012; Dong et al. 2016). For example, Barnett et al. (2012) found a small, significant added heat wave effect on mortality risk in the United States of 0.5 and 1.6 per cent for heat waves defined at the highest temperature thresholds (98th and 99th percentile, respectively). Dong et al. (2016) reported a large significant effect of 10 to 18 per cent on cardiovascular mortality risk in Beijing, China, for heat waves defined at lower thresholds, although longer in duration. Fewer studies have investigated whether there is an added heat wave effect in the association between heat waves and morbidity (e.g. Gronlund et al. 2014; Chen et al. 2017). Gronlund et al. (2014) found a significant, added heat wave effect in the association between heat waves and hospital admissions for renal diseases among those aged 65 years and over in the United States.
43 2.2.2.3 Extreme Heat Definitions
There is no standard definition of extreme heat. As such, epidemiological studies have used different definitions to define extreme heat exposure when examining its impact on adverse health outcomes. There are, however, common elements to each definition and these include the: temperature metric (e.g. ambient temperature, apparent temperature, heat index), temperature threshold
(i.e. an absolute or relative threshold), and for heat waves, duration (≥ 2, 3 or 4 consecutive days). The reasons for selecting and using specific extreme heat definitions are not often clear in studies; although most definitions appear to be selected on the basis of their application in previous work, or because they are official definitions of local meteorological or health authorities, or are reflective of local meteorological conditions.
Studies have shown that the choice of definition can affect both the magnitude and statistical significance of the estimated association (Tong et al.
2010b; Kent et al. 2014; Xu et al. 2016). For example, Tong et al. (2010b) used 10 different heat wave definitions to examine the association between heat waves and adverse health outcomes in Brisbane, Australia. Odds ratios for the association between heat waves and emergency hospital admissions were found to range from
1.03 to 1.18 (Tong et al. 2010b). The lack of a standard definition can make it difficult to compare health risk estimates across studies, although given the climatological, demographic and socioeconomic differences between regions, countries and cities, it may be more appropriate to develop and use localised definitions.
Despite the lack of a standard definition of extreme heat, it is common for studies to use derivatives of maximum and mean temperature as the temperature
44 metric (e.g. Hansen et al. 2008b; Vaneckova and Bambrick 2013). As such, much less is known about the association between extreme high minimum temperatures and health outcomes. Exposure to elevated overnight temperatures adversely impacts the quality and quantity of an individual’s sleep, leading to increased wakefulness, and decreased duration of rapid eye movement and slow wave sleep
(Okamoto-Mizuno et al. 2010; Lan et al. 2014). This disrupted sleep, along with the need for the body to thermoregulate during the night, means that individuals are unable to naturally rest and recover from the day. Some studies have found strong, positive associations between high minimum temperatures or hot nights and mortality, with some evidence suggesting that such temperatures may have a greater impact on mortality than maximum temperatures, particularly in urban areas (Laaidi et al. 2012; Heaton et al. 2014; Royé, 2017). For example, Laaidi et al. (2012) found that consecutive nights of high minimum temperatures had a greater impact on mortality among the elderly in Paris during the 2003 European heatwave, than daily maximum temperatures. Little is known about the association between extreme high minimum temperatures and morbidity.
2.2.2.4 Vulnerable Populations
Table 2.1 summarises the main population groups that have been identified as being susceptible or vulnerable to extreme heat in Australia.
45 Table 2.1 Main population groups identified as being susceptible or vulnerable to the effects of extreme heat in Australia.
Population Group Reference
Elderly Victorian Government Department of Human Services 2009 Individuals living alone Zhang et al. 2013; Zhang et al. 2017 Individuals with pre-existing medical conditions including: - Chronic heart disease - Renal diseases Hansen et al. 2008a; Hansen et al. 2008b; Williams et al. 2012; - Cardiovascular diseases Wilson et al. 2013; Zhang et al. 2013; Zhang et al. 2017 - Respiratory diseases - Diabetes - Mental disorders Individuals with socio-economic disadvantage Zhang et al. 2013 Individuals that do not have private health insurance Zhang et al. 2013 Infants and children Xu et al. 2013 Outdoor workers Xiang et al. 2014 Migrants Hansen et al. 2013 Living in remote areas Xiao et al. 2017
46 2.2.3 Climate Change and Temperature-Related Health Outcomes
Heat-related mortality is generally projected to increase in the future under climate change in most regions, while cold-related mortality is expected to decrease, although consideration regional variation has been observed
(Vardoulakis et al. 2014; Guo et al. 2016; Gasparrini et al. 2017; Weinberger et al.
2017). The magnitude of these projected changes differs under the four RCPs, with larger increases in heat-related mortality projected under those RCPs that allow for greater increases in GHG emissions (Gasparrini et al. 2017). The increase in heat-related mortality is generally projected to outweigh the decrease in cold-related mortality, resulting in an overall net-increase in heat-related mortality in some areas, particularly in those regions currently with warm or tropical climates (Gasparrini et al. 2017). An increase in mortality associated with heat waves under climate change has also been projected in the United States
(Peng et al. 2011; Wu et al. 2014). A few studies from Europe and North America have also projected an increase in heat-related morbidity under climate change, specifically respiratory morbidity (Lin et al. 2012; Åström et al. 2013).
The evidence for Australia is somewhat unclear. Gasparrini et al. (2017) projected a net-decrease in temperature-related mortality for Australia under climate change; although under RCP8.5, this would revert to a net increase sometime during the twenty-first century. Guo et al. (2016) projected a net increase in temperature-related mortality for Sydney and Brisbane, and a net decrease for Melbourne. A study undertaken by Pricewaterhouse Coopers
Australia (2011) projected an increase in the number of deaths associated with extreme heat waves under climate change in all major metropolitan regions, with the total number of deaths expected to double nationally by 2030.
47 A critical question that remains is the extent to which populations or individuals will adapt to changes in temperature in the future. Temporal changes in the association between high temperature and mortality have been observed, with studies reporting a decrease in mortality risk or rate due to heat exposure over time in several regions (Barnett 2007; Bobb et al. 2014a; Gasparrini et al.
2015b; Chung et al. 2017). A recent review examining this topic reported that 10 out of the 11 studies considered found some evidence of a decline in susceptibility to high temperatures among populations over time (Arbuthnott et al. 2016). In
Australia, a decrease in the number and rate of heat-related deaths since 1907 has been observed (Coates et al. 2014). The exact reason for this observed change remains unclear. However, studies have largely attributed this finding to a range of factors including physiological acclimatisation, changes in demographic and socio-economic factors, an increased prevalence and use of air conditioning and the development and implementation of extreme heat-health plans (Barnett 2007;
Gasparrini et al. 2015b).
A similar temporal pattern has also been observed for heat waves. An estimated 2,065 excess deaths occurred during the 2006 European heat wave in
July in France, 4,387 less than expected (Fouillet et al. 2008). An estimated 167 excess deaths occurred during the 2014 south-east Australian heat wave in
January, much less than the 374 excess deaths during the 2009 heat wave
(Department of Health 2014). The Victorian Government attributed the reduction in the number of deaths to the implementation of Victoria’s heatwave plan (2011) following the 2009 heat wave (Department of Health 2014). Similarly, the reduction in deaths during the 2006 European heat wave in France was attributed to the population’s decreased susceptibility to heat, particularly as a result of an
48 increased understanding of the health risks associated with heat waves and the implementation of a heat wave warning system (Fouillet et al. 2008).
2.3 Linking Climate Change, Ambient Air Pollution and Health Outcomes
This section reviews the relationship between ambient air pollution and meteorology, and the projected impact of climate change on levels of ambient air pollution. It also reviews the association between ambient air pollution and adverse health outcomes, and how climate change is expected to affect this association. This section will focus on two pollutants, ambient ozone and particulate matter.
2.3.1 Ambient Air Pollutants: Ozone and Particulate Matter
Ambient ozone, also known as tropospheric ozone, is a photochemical ambient air pollutant. It is created when the two precursor pollutants, oxides of nitrogen (NOx) and volatile organic compounds (VOCs), undergo a series of chemical reactions in the presence of sunlight (Linfoot et al. 2010; US EPA
2017). The major sources of anthropogenic NOx in Sydney are on-road mobile, industrial and off-road mobile sources, while domestic commercial and on-road mobile sources are the major sources of VOCs (Department of Environment and
Climate Change NSW 2007).
Particulate matter is comprised of a mixture of both solid particles and liquid droplets from natural and anthropogenic sources (US EPA 2016). The pollutant is classified according to the size of the particles and there are three main classifications including: particles with an aerodynamic diameter less than
10m, PM10; particles with an aerodynamic diameter less than 2.5m, PM2.5; and
49 particles with an aerodynamic diameter less than 0.1m, PM0.1 (Hime et al. 2015).
The NSW Government’s Office of Environment and Heritage maintains an air quality monitoring network that covers several regions throughout NSW including
Greater Sydney (NSW Office of Environment and Heritage 2018a). This network records daily values of particulate matter, specifically, PM2.5 and PM10.
2.3.2 Linking Meteorology and Ambient Air Pollution
Local meteorological conditions can influence the creation, accumulation and dispersion of ambient air pollutants, including ozone and particulate matter
(Fiore et al. 2012). Maximum temperature has been identified as the main meteorological driver of ambient ozone concentrations (Otero et al. 2016). A strong, positive relationship between ozone and maximum temperature has been observed in several regions, including Australia, where elevated ozone levels are associated with increased maximum temperature due to its strong effect on the emissions of ozone precursors and photochemical reactions (Camalier et al. 2007;
Dawson et al. 2007a; Linfoot et al. 2010; Pearce et al. 2011; Otero et al. 2016).
Several other meteorological variables have also been identified as having an influential role, and include relative humidity, wind speed and direction, solar radiation and precipitation (Camalier et al. 2007; Pearce et al. 2011; Otero et al.
2016). Particulate matter is also correlated with meteorological variables in several regions, including Australia (Dawson et al. 2007b; Pearce et al. 2011). A positive correlation between temperature and PM10 and PM2.5 concentrations has been observed in some areas (Galindo et al. 2010; Tai et al. 2010; Pearce et al.
2011).
Heat waves and extreme air pollution episodes can coincide (Schnell and
Prather 2017). This is because these events occur under similar meteorological
50 conditions including high temperatures, low wind speeds and low levels of precipitation (Foire et al. 2012; Schnell and Prather 2017). Hou and Wu (2016) found that heat waves are more likely than other weather events, including atmospheric stagnation and temperature inversion, to lead to elevated ozone concentrations during the summer in the United States. Heat waves, or consecutive days of high temperatures, have also been found to have more adverse effects on air quality than single, non-consecutive days of high temperatures (Hou and Wu 2016). The 2003 European heat wave is an important example of these two events coinciding, with extreme levels of ozone observed across most of Europe throughout the heat wave (Fiala et al. 2003; Solberg et al.
2008). Elevated levels of air pollution, specifically ozone, have also been observed during recent heat waves in Sydney (McInnes 2018).
2.3.2.1 Linking Meteorology and Ambient Air Pollution in Greater Sydney
Drainage flows and the coastal sea breeze largely influence the formation, circulation and dispersion of ambient ozone across Sydney (Linfoot et al. 2010;
Jiang et al. 2017). Elevated levels of ozone are often observed across the region when local weather synoptics are dominated by a high-pressure system over the
Tasman Sea, and northerly to north-westerly gradient winds over the state (Jiang et al. 2017). These local weather patterns are typically associated with light winds, high solar radiation and high temperatures across the city (Jiang et al. 2017).
Ozone levels can vary spatially across Sydney, with the progression of the typical afternoon sea breeze across the city influencing the location of elevated ozone concentrations (Jiang et al. 2017). Elevated levels of PM10 across Sydney are often observed when local weather synoptics are dominated by the movement of frontal or westerly trough systems over south-eastern Australia and elevated gradient
51 winds across the region (Jiang et al. 2017). These synoptic patterns are associated with low levels of precipitation and humidity, and high solar radiation (Jiang et al.
2017).
2.3.3 The Impact of Climate Change on Ambient Air Pollution
As a result of the known relationship between air pollutant concentrations and meteorology, several studies have examined if, and how, anthropogenic climate change will affect ambient air pollutants.
Climate change alone is generally expected to increase average and peak levels of ozone across most parts of Europe and North America, as well as the number of regulation exceedances, although spatial variation across regions has been found, with decreases observed in some areas for some studies (Murazaki and Hess 2006; Bell et al. 2007; Meleux et al. 2007; Huszar et al. 2011; Langner et al. 2012; Varotsos et al. 2013). For example, Meleux et al. (2007) projected an increase of up 25 per cent (16 – 18 ppb) in peak ozone and 10 to 16 per cent (7 –
10 ppb) in average ozone concentrations across Western Europe for 2071 to 2100, compared to 1961 to 1990, under the A2 scenario. Bell et al. (2007) projected an average increase of 4.8 and 4.4 ppb in daily 1-hour and 8-hour maximum ozone levels respectively in the 2050s, compared to the 1990s, across 50 cities in the
United States under the A2 scenario. Those cities with currently high levels of ozone are projected to have the largest increases in ozone concentrations in the future, such as the Northeast United States, and decreases are projected in areas currently with lower levels (Murazaki and Hess 2006; Bell et al. 2007; Schnell et al. 2016). Peak ozone concentrations have also been projected to increase in the
Greater Sydney Metropolitan region under climate change, as well as the number of exceedances of the current 1-hour and 4-hour national health standards (Cope
52 et al. 2008). For example, Cope et al. (2008) projected that the number of exceedance days of the 1-hour standard will increase by 27 per cent in 2021 to
2030, relative to 1996 to 2005, further increasing by 45 per cent in 2051 to 2060.
The impact of climate change on concentrations of particulate matter is less certain (Jacob and Winner 2009). Most areas across Europe are expected to
3 experience an increase of 1 day per year when the PM10 threshold of 50g/m is exceeded in the future under climate change, although decreases in the number of exceedances have been projected in some areas (Huszar et al. 2011). Shen et al.
3 (2017) projected an increase of 0.4 to 1.4g/m in annual mean levels of PM2.5 in eastern United States for 2050 to 2069, relative to 2000 to 2019, under RCP4.5, while a decrease of 0.3 to 1.2g/m3 was projected for the Intermountain West.
Seasonal differences were observed, with larger average increases projected during the summer in eastern United States of up to 2 to 3g/m3 (Shen et al.
2017). Wildfires, or bushfires, are an important source of particulate matter, and are expected to become more frequent and intense under climate change due to hotter and drier climatic conditions, particularly in south-east Australia (Hughes and Alexander 2016). Liu et al. (2016) projected an increase of 160 per cent in average levels of PM2.5 associated with wildfires for 2046 to 2051 under the A1B scenario in western United States.
Climate change is also expected to affect extreme levels of pollutants.
More frequent and longer extreme ozone episodes have been projected in Europe under the A2 and B2 scenarios (Meleux et al. 2007). Schnell et al. (2016) showed that such extreme air pollution episodes are likely to occur earlier over North
America, Europe and Asia, with higher ozone concentrations. Mickley et al.
(2004) projected an increase in the intensity and duration of extreme air pollution
53 episodes in the northeastern and midwestern United States for 2045 to 2052, estimating an increase of 5 to 10 per cent in pollutant levels during these episodes, and an increase from 2, to 3 to 4 days, relative to 1995 to 2002 under the A1B scenario.
These studies in assessing the impact of climate change on concentrations of ozone and particulate matter have assumed that levels of anthropogenic emissions will remain constant throughout this century. Other studies have sought to examine the effects of both future climate change and emissions reductions on air pollution levels (e.g. Tagaris et al. 2007; Wu et al. 2008; Lam et al. 2011;
Watson et al. 2016; Lacressonnière et al. 2017). These studies have projected that levels of ozone and particulate matter will decrease in the future due to a reduction in emissions, while climate change will lead to an increase in levels.
This in turn means that greater emission reduction targets will need to be implemented to account for this – Wu et al. (2008) described this as the ‘climate change penalty’. However, some of these studies have projected the increase in air pollution levels associated with climate change to be smaller than the decrease associated with the reduction in emissions, suggesting that climate change might have a negligible impact on air pollution levels, even at 2°C warming (e.g.
Watson et al. 2016). This however assumes that emission reduction targets are robustly implemented and met (Watson et al. 2016).
2.3.4 Linking Ambient Air Pollution and Health Outcomes
The London Smog event of December 1952 is an important example illustrating the adverse health impacts of air pollution, with an estimated 12,000 excess deaths occurring as a result of the event during December 1952 to
February 1953 (Bell et al. 2001).
54 The adverse effects of ozone exposure on health outcomes are well documented. A number of large, international and regional studies have found evidence of a short-term association between elevated concentrations of ozone and mortality and morbidity, including cause-specific outcomes, such as cardiovascular and respiratory conditions (Anderson et al. 2004; Bell et al. 2005;
Ji et al. 2011; Peng et al. 2013). Studies have also found some evidence of an association in Australia, despite the country’s comparatively low levels of air pollution. For example, Chen et al. (2016) observed significant, positive associations between ozone and hospitalisations for asthma among children and adults during the warm season in Adelaide. Jalaudin et al. (2008) found significant, positive associations between ozone and emergency presentations for asthma among those aged 1 to 4 and 5 to 9 years in Sydney, with the strongest associations observed during the warm season. Petroeschevsky et al. (2001) observed positive associations between ozone and hospitalisations for total respiratory diseases and asthma, but not total cardiovascular diseases, in Brisbane.
The adverse effects of particulate matter exposure on health outcomes are also well established. Several large international and regional studies have also found evidence of a short-term association between elevated levels of particulate matter (including PM10 and PM2.5) and mortality and morbidity, particularly cardiovascular conditions (Samet et al. 2000; Anderson et al. 2004; Dominici et al. 2006; Peng et al. 2008; Staffoggia et al. 2013). A number of studies have also found evidence of an association in Australia, with bushfires an important source of particulate matter and adversely affecting health outcomes (Barnett et al. 2006;
Chen et al. 2007; Tham et al. 2009; Johnston et al. 2011; Hansen et al. 2012). For example, Barnett et al. (2006) found significant, positive associations between
55 PM2.5 and PM10 and hospitalisations for cardiac diseases among the elderly for
Australian and New Zealand cities. Hansen et al. (2012) observed a 2.71 (95% CI:
0.16, 5.33) and 0.37 (95% CI: -0.51, 1.25) per cent increase in cardiovascular
3 hospital admissions per 10g/m increase in PM2.5 and PM10 in Adelaide, respectively. Tham et al. (2009) showed that increased concentrations of PM10 were associated with an increased risk of presenting to the emergency department with respiratory conditions during the 2002 to 2003 bushfire season in Victoria.
2.3.5 Ambient Air Pollution and Health Outcomes under Climate Change
A few studies have estimated the health impacts associated with exposure to increased levels of ambient ozone and particulate matter under climate change at the global level (e.g. West et al. 2007; Fang et al. 2013; Silva et al. 2017). Most of these studies have shown that the number of deaths associated with exposure to ozone and PM2.5 will increase under climate change, although the magnitude of this increase will differ across regions and time periods. For example, Silva et al.
(2017) estimated that an additional 3,340 ozone-related deaths would occur per annum under RCP8.5 in 2030, increasing to 43,600 deaths in 2100. Ozone-related deaths are expected to increase by the end of this century in most regions; however, the greatest increases are expected to occur in those areas currently with large populations and high levels of pollution, such as East Asia, India and North
America (Silva et al. 2017). Fang et al. (2013) projected an additional 6,300 premature ozone-related deaths per annum due to respiratory disease in those aged
30 years and above under the A1B scenario, with the largest increases projected to occur in South and East Asia and North America. The impacts arising from exposure to increased levels of PM2.5 are much greater, however. Silva et al.
(2017) projected an additional 55,600 deaths per annum associated with exposure
56 to PM2.5 in 2030, increasing to 215,000 in 2100. The greatest increases in 2100 were observed for India, the Middle East and East Asia, although both increases and decreases in the number of deaths were observed across regions (Silva et al.
2017). Fang et al. (2013) projected an additional 100,000 premature deaths associated with exposure to PM2.5 to occur per annum before the end of this century under the A1B scenario.
A number of regional studies have also assessed the impact of air pollutants on health outcomes under climate change (e.g. Knowlton et al. 2004;
Bell et al. 2007; Cheng et al. 2008; Chang et al. 2010; Sheffield et al. 2011; Orru et al. 2013). Most of these studies have considered the impact of increased ozone levels as a result of climate change on all-cause or cause-specific mortality and respiratory morbidity in Europe or North America. For example, Knowlton et al.
(2004) projected a 4.5 per cent increase in the number of ozone-related deaths during the summer season across 31 counties in the New York metropolitan region in the 2050s, compared to 1990s, under the A2 scenario. Further, Bell et al.
(2007) projected an increase of 0.11 to 0.27 per cent in total mortality due to elevated ozone levels under the A2 scenario across 50 cities in the United States, as well as an increase of 0.8 to 2.1 per cent in respiratory hospital admissions in those aged 65 years above. An increase in ozone-related deaths and respiratory hospital admissions has also been projected for most European countries. Orru et al. (2013) projected an additional 2,402 ozone-related premature deaths per annum and 3,135 respiratory hospitalisations in 2021 to 2050, compared to 1990 to 2009, under the A1B scenario in Europe. The largest increases are projected to occur in Belgium, France, Spain and Portugal, although decreases were observed in several Nordic and Baltic countries (Orru et al. 2013).
57 The regional differences highlighted above illustrate the critical importance of conducting localised studies. Few studies have considered how the health impacts of air pollution will differ under climate change in Australia (e.g.
Cope et al. 2008; Physick et al. 2014). Cope et al. (2008) projected an additional
350 and 750 ozone-related respiratory hospital admissions in 2021 to 2030 and
2051to 2060 respectively, representing an increase of 2.1 and 4.4 per cent from
2005. Physick et al. (2014) estimated that the number of ozone-related deaths due to climate change will increase by 55 to 65 in 2051 to 2060, compared to 1996 to
2005, representing an increase of between 2.3 to 27.3 per cent.
2.4 Ambient Air Pollution as a Confounder
This section briefly reviews the evidence regarding ambient air pollution as a confounder of the association between temperature or extreme heat and adverse health outcomes.
As shown in Section 2.3 of this Chapter, there is an established correlation between temperature and ambient air pollution, and an association between ambient air pollution and adverse health outcomes. As such, some studies have assessed, or controlled for, the potential confounding effects of ambient air pollution when examining the association between high temperatures or extreme heat and health outcomes (e.g. Basu et al. 2008; Bell et al. 2008; Anderson and
Bell 2009; Green et al. 2010; Anderson et al. 2013; Vaneckova and Bambrick
2013; Gronlund et al. 2014; Ogbomo et al. 2017). The effect of adjusting for air pollution, such as ozone and particulate matter, has not been consistent across studies. For example, some studies have reported that adjusting for ozone or
58 particulate matter had no substantial, or little effect on the health risk estimates for deaths and hospital admissions (Basu et al. 2008; Anderson et al. 2013; Gronlund et al. 2014). Other studies have observed a slight or moderate reduction in the health risk estimates after adjusting for ozone or particulate matter (O’Neill et al.
2005; Bell et al. 2008; Son et al. 2012).
It is noted however, that some researchers have questioned whether air pollution should be considered or characterised as a confounder of the association between temperature, or high temperatures, and health outcomes (e.g. Reid et al.
2012; Buckley et al. 2014). For example, Reid et al. (2012) used directed acyclic graphs (DAGs) to examine the role of ozone in the association between high temperature and mortality. The authors concluded that ozone should be regarded as a ‘causal intermediate’ of the association, as opposed to a confounder. Further, also using DAGs, Buckley et al. (2014) reasoned that there are few circumstances in which it is warranted to adjust for air pollution in studies examining the association between temperature and health outcomes, encouraging researchers to list clear reasons for doing so.
2.5 Ambient Air Pollution as an Effect Modifier
Despite the known independent effects of extreme heat and ambient air pollution on health outcomes, few studies have investigated the potential joint, or interactive effects of these two environmental exposures. This is of particular concern given that the joint effect of weather and air pollution on health outcomes is thought to be greater than the risk derived from the individual impacts of these two exposures (Zanobetti and Peters 2015). Indeed, as Schnell and Prather (2017)
59 note, the tendency for heat waves and extreme air pollution episodes to coincide is of considerable concern for human health as populations become exposed to both the hottest temperatures and highest levels of air pollution simultaneously
(Schnell and Prather 2017). There is some suggestion that an interactive effect between temperature and air pollution may be biologically plausible (Gordon
2003).
This section reviews the evidence regarding the potential joint, or interactive effects of temperature or extreme heat and ambient air pollution on adverse health outcomes. It considers the two pollutants that are the focus of this thesis, namely, ambient ozone and particulate matter.
2.5.1 Temperature, Ozone and Particulate Matter
A few studies have examined whether temperature modifies the association between ozone and all-cause or cardiovascular mortality (Ren et al.
2009; Pattenden et al. 2010; Jhun et al. 2014). These studies have generally observed stronger associations between ozone and mortality at higher temperatures, although evidence of statistical interaction has not been consistent and notable regional differences have been observed. For example, Pattenden et al. (2010) observed a statistically significant interaction between ozone and hot days on mortality in London and Cardiff, but found little evidence of an interaction in cities elsewhere in the United Kingdom. Ren et al. (2009) found that the association between ozone and cardiovascular mortality was generally stronger at higher temperatures in the northern regions of the United States, but little and inconsistent evidence of effect modification was found in the southern regions.
60 A few studies have also investigated whether ozone modifies the association between temperature and all-cause or cardiovascular mortality (Ren et al. 2007; Burkart et al. 2013; Breitner et al. 2014). Stronger temperature or heat effects at higher levels of ozone have been observed, although statistical evidence of effect modification has varied. For example, Breitner et al. (2014) found that the pooled association between high temperature and non-accidental mortality was significantly stronger at high compared to moderate-levels of ozone for three
German cities, but evidence of a statistical interaction was not found for cardiovascular mortality. Ren et al. (2007) reported that ozone modified the association between temperature and cardiovascular mortality in certain regions in the United States during the warm season, with stronger associations generally observed at higher levels of ozone.
A few studies from Europe and Asia have investigated whether temperature modifies the association between particulate matter (PM10) and total or cause-specific mortality, including cardiovascular and respiratory mortality
(Qian et al. 2008; Stafoggia et al. 2008; Li et al. 2011; Cheng and Kan 2012;
Meng et al. 2012; Burkart et al. 2013). Most have observed stronger associations between PM10 and mortality at high compared to moderate or low temperatures, although such evidence of effect modification has not been consistently statistically significant across studies and disease groups. For example, Stafoggia et al. (2008) observed that the pooled effects of PM10 on mortality across nine
Italian cities were strongest at high temperatures, but found little evidence of a statistical interaction between PM10 and temperature on mortality. Li et al. (2011) reported that the effects of PM10 on non-accidental and cause-specific mortality were stronger at high compared to low-level temperatures in Tianjin, China, with
61 evidence of statistically significant interactions found for cardiovascular and cardiopulmonary mortality, and ischemic heart disease. However, Cheng and Kan
(2012) found a statistically significant interaction between PM10 and low temperature on total and cardiovascular mortality in Shanghai, China, with the strongest associations generally observed at lower temperatures. Ren and Tong
(2006) found that maximum temperature modified the association between PM10 and all non-external cause and cardiovascular mortality in Brisbane, Australia at different lags, with stronger PM10 effects observed at high compared to low temperature levels.
Studies from Europe, North America, Asia and New Zealand have assessed whether PM10 modifies the association between temperature and mortality (Hales et al. 2000; Basu et al. 2008; Burkart et al. 2013; Breitner et al.
2014; Li et al. 2015). The results of these studies have largely been inconsistent, with some finding evidence of effect modification (e.g. Li et al. 2015) and others not (e.g. Hales et al. 2000; Basu et al. 2008). For example, Li et al. (2015) found that both cold and hot effects on mortality increased with increasing levels of
PM10 in Guangzhou, China, with the strongest cold and hot effects observed on days with the highest PM10 levels. In contrast, Basu et al. (2008) found little evidence that PM10 or PM2.5 modified the pooled association between apparent temperature and mortality during the warm season in nine Californian counties.
This finding is in general agreement with Hales et al. (2000), which found no evidence of an interactive effect between PM10 and temperature on mortality in
Christchurch, New Zealand. One Australian study however, reported that PM10 modified the association between temperature and all non-external cause and cardiovascular mortality in Brisbane at different lags (Ren et al. 2006).
62 2.5.2 Heat Waves, Ozone and Particulate Matter
Less is known about the potential interactive, or joint effects of heat waves and ambient air pollution on health outcomes. A few studies have estimated the number or proportion of excess deaths attributable to ozone or particulate matter exposure during specific heat wave events or summer seasons, such as the 2003
European heat wave and summer, and the 2004 Brisbane heat wave (Fischer et al.
2004; Stedman 2004; Tong et al. 2010a). For example, Stedman (2004) estimated that of the 2045 excess deaths that occurred during the 2003 European heat wave in England and Wales, 221 to 567 deaths could be associated with exposure to ozone and 202 could be associated with PM10 exposure. This estimation meant that 21 to 38 per cent of the total number of excess deaths that occurred during the heat wave could in fact be attributed to exposure to these two pollutants. Further,
Tong et al. (2010a) reported that ozone exposure contributed to the number of excess deaths during the 2004 Brisbane heat wave, but had a lesser impact than that of exposure to high temperatures. Dear et al. (2005) and Filleul et al. (2006) also examined the potential interactive effects of exposure to high temperatures and ozone during the 2003 heat wave and summer in France, respectively. Dear et al. (2005) found there was a statistically significant interaction between minimum temperature and ozone on mortality, but Filleul et al. (2006) reported that the addition of interaction terms between various temperature metrics and ozone did not improve the models, finding little evidence of a statistical interaction.
Time-series studies examining the potential joint, or interactive effects of heat waves and ambient air pollution are limited. One study investigated whether ozone and PM10 modifies the pooled association between heat waves and total and cause-specific mortality for nine European cities (Analitis et al. 2014). Stronger
63 heat wave effects were observed at high compared to low-levels ozone for total and cardiovascular mortality, although no evidence of a statistical interaction was found. Further, stronger heat wave effects were observed at high compared to low-levels of PM10 on total, cardiovascular and respiratory mortality, with effects found to be more pronounced in the North-Continental cities and among the elderly. Shaposhnikov et al. (2010) found evidence of an interactive effect between high temperature and PM10 on mortality during the 2010 Moscow heat wave and wildfires in Russia.
5.2.3 Heat Waves, Ozone, Particulate Matter and Morbidity
The studies discussed in Section 5.2.1 and 5.2.2 of this chapter considered total or cause-specific mortality as the health outcome. Little is known about the potential interactive effects of temperature, particularly heat waves, and ambient air pollution on morbidity.
A few studies have examined the potential joint effects of different meteorological variables (i.e. temperature, season, relative humidity) and PM10 on cardiovascular morbidity (Ren and Tong 2006; Ren et al. 2006; Qiu et al. 2013).
Ren et al. (2006) found evidence of a statistical interaction between temperature and PM10 on total cardiovascular hospital admissions at different lags in Brisbane,
Australia, but found no such evidence for total cardiovascular emergency presentations. Qiu et al. (2013) reported that the association between PM10 and emergency hospital admissions for ischemic heart disease was strongest in the cool season and at lower levels of relative humidity in Hong Kong, China.
Further, Kang et al. (2016) found no evidence of a significant interactive effect between heat waves and PM10 on out-of-hospital cardiac arrest in Korea.
64 Chapter Three: Methods
This Chapter describes the study setting, data and study design used in this thesis.
3.1 Study Setting
This section describes the location of the study, Greater Sydney, Australia.
3.1.1 Greater Sydney, Australia
Sydney, located on Australia’s southeast coast (33.87°S, 151.21°E), is the capital city of the state of New South Wales. It is Australia’s largest city, with an estimated residential population of 5.03 million at June 2016 (Australian Bureau of Statistics 2017a). Sydney’s population is projected to increase substantially in the future, with an additional 1.7 and 3.6 million people expected to live in city by
2036 and 2056 respectively (Greater Sydney Commission 2018). The median age of the city’s residents is 36, with 18.7 per cent of the population aged between 0 to 14 years and 13.9 per cent aged 65 years and over (Australian Bureau of
Statistics 2017b). Over 40 per cent of the population are born overseas (Australian
Bureau of Statistics 2017b).
Sydney has a temperate climate, with an average maximum temperature of
17.1°C during winter (June to August), and 25.7°C during summer (December to
February) (Bureau of Meteorology 2017b). The recent 2016/2017 summer season was the hottest on record for city, and more frequent and longer heat waves are projected for the city under climate change (Cowan et al. 2014; Bureau of
Meteorology 2017a). Sydney observed its second hottest day on record in the summer of 2017/2018, with daily maximum temperature reaching 47.3°C at
Penrith (Bureau of Meteorology 2018b). Future urban expansion in Sydney is
65 projected to have a considerable impact on the city’s overnight temperatures, with
Argüeso et al. (2014) showing that such development will further enhance the warming of minimum temperatures under climate change.
Levels of ozone and particulate matter (PM10) can exceed national quality standards in Sydney, particularly during the warm season (NSW Environment
Protection Authority 2013; NSW Office of Environment and Heritage 2014).
During 1994 to 2014, exceedances of the national 1-hour and 4-hour health standards for ozone occurred on up to 19 and 21 days per year respectively (NSW
Environment Protection Authority 2015). Exceedances of the national PM10 health standards are often associated with bushfires and dust storms (NSW
Environment Protection Authority 2015). Ozone concentrations are expected to increase under climate change, resulting in an increased ozone-related health burden for the city (Cope et al. 2008; Physick et al. 2014).
The Sydney Statistical Division (SSD) administrative unit, which is broadly consistent with the geographic boundaries of the Greater Sydney region, was selected as the predefined region for this study, similar to previous work
(Vaneckova and Bambrick 2013).
3.2 Data
This section provides an overview of each dataset used in this thesis including, where relevant, the processes and procedures undertaken to clean and sort each dataset.
66 3.2.1 Meteorological Data
The meteorological data was sourced from the Australian Government’s
Bureau of Meteorology. This agency operates an extensive, national network of weather monitoring stations that record daily and sub-daily values for a range of weather variables.
3.2.1.1 Selecting Weather Stations
The Bureau of Meteorology’s online Weather Station Directory (WSD) service was used to identify weather stations located within the SSD. The WSD required the completion of three search criteria, which are outlined in Table 3.1, along with the corresponding information that was entered.
Table 3.1 Search criteria used to identify weather stations located within the SSD.
Search Criteria Information Entered
Place Name Defined as Sydney (NSW) (33.87°S, 151.21°E)
Defined to a radius of 150km surrounding the geographical coordinates indicated above. This Search Within radius was selected to ensure that all potential weather stations within the SSD were captured during the search.
Defined as daily maximum and daily minimum Weather Element temperature
The search identified multiple weather stations located within the SSD.
The details and geographic coordinates of those stations that were identified as having temperature observations that covered, or partially covered, the period of
67 1993 to 2014 were recorded. A visual check was then conducted to confirm that the identified weather stations were indeed located within the SSD. This check involved mapping the location of each station in ArcGIS, and then inspecting whether each station fell within the Australian Bureaus of Statistics’ 2011 SSD boundary.
A request for weather data from each of the identified stations was then submitted to the Bureau of Meteorology via their online Climate Data Request
Form. Such data included daily and sub-daily observations for a range of weather variables including daily maximum temperature, daily minimum temperature, air temperature, dew-point temperature, relative humidity, wind speed and wind direction. It is noted that not all stations recorded observations for all of the variables requested, and that the coverage of observations for each of the variables differed between stations (i.e. not all stations had full coverage of the time period).
3.2.1.2 Quality Control Procedures and Checks
It is important for observational weather station data to undergo a series of quality control checks, particularly when using the data to identify extreme temperatures or events, such as heat waves (Alexander and Tebaldi 2012). This is because it is possible for incorrect data entries to be considered as real ‘extreme’ values and included in further analyses (Alexander and Tebaldi 2012). The
Bureau of Meteorology performs a series of quality control checks on its observational weather data (Bureau of Meteorology 2018c). It is possible, however, for errors to remain.
A series of quality control checks were performed on the observed daily maximum and minimum temperature values for each weather station, as well as
68 the sub-daily dew-point temperature values. Inhomogeneities were also tested for in each daily maximum and minimum temperature time-series to inspect the overall quality of the series. These checks were performed for those weather stations that had near complete data coverage for the period of 2001 to 2013. This amounted to seventeen weather stations in total. A detailed overview of these processes is outlined in the following paragraphs.
Maximum and Minimum Temperature Values
For each daily maximum and minimum temperature time series,
RClimDex (1.0) was used in the ‘R’ Statistical Computing Environment to identify any missing data, and to detect and correct any ‘unreasonable’ values including possible outliers (Zhang and Yang 2004). The program defines unreasonable values as those where daily minimum temperature is greater than or equal to daily maximum temperature (Zhang and Yang 2004). This check revealed that one station had a considerable amount of missing values (~8.3 per cent for daily maximum temperature and ~9 per cent for daily minimum temperature). This station was excluded from inclusion in any further analyses.
Those values that were detected as ‘unreasonable’ were removed and replaced with NA (missing). One potential outlier was identified for one station. To determine if it was a true outlier, this value was then compared against the observed value recorded for three neighbouring stations on the same day. On inspection and comparison, it was deemed to be a plausible value. It therefore remained within the time series for that station.
A quality flag description for the observed daily maximum and minimum temperature values was provided with the data (Bureau of Meteorology 2015).
There are six quality flag descriptions and these are outlined in Table 3.2. It is
69 noted that this information was missing for some observations for some weather stations.
Table 3.2 Description of quality control flags provided by the Bureau of
Meteorology.
Quality Flag Description
Y Quality controlled and acceptable N Not quality controlled W Quality controlled and considered wrong S Quality controlled and considered suspect I Quality controlled and inconsistent with other information X No quality information available
Those observed values that had a quality control flag of ‘W’, ‘S’ or ‘I’ were inspected. Values flagged as ‘W’ were removed and replaced with NA
(missing). Values flagged as ‘S’ or ‘I’ were compared against the observed value recorded on the same day for up to three neighbouring weather stations. If the value was considered plausible on comparison, then it remained within the series; or was removed and replaced with NA (missing), if not.
The metadata notes provided with the data by the Bureau of Meteorology indicated that some weather stations may not record observations over the weekend or on a public holiday (Bureau of Meteorology 2015). This means that a given daily value may be an ‘accumulated value’; that is, the value for daily maximum temperature may be the highest value observed over two to three days
(Bureau of Meteorology 2015). In this circumstance, the ‘days of accumulation’ variable indicates the number of days for which the given value represents
(Bureau of Meteorology 2015). Those daily maximum and minimum temperature observations that had a ‘days of accumulation’ value greater than one were
70 removed and replaced with NA (missing). The analyses in this thesis examine the association between extreme heat and hospital admissions at the daily level, and it is therefore important that each daily temperature value recorded is the true value for that given day.
All time series plots for daily maximum and minimum temperature generated as a result of the quality control checks using RClimDex (1.0) were inspected for any remaining suspect values. The overall patterns and trends for each time series plot were compared against the plots for neighbouring stations to assist in determining whether the values were suspect. A few suspect values were identified, but these values were removed during the quality control checks described above.
This quality controlled data was then tested for inhomogenities using
RHtests V4 in the ‘R’ Statistical Computing Environment. This test was performed to inspect the overall quality of each of the daily maximum and minimum temperature time series for each weather station. The tests were performed without using a reference series. This is because the length of the time series (13 years) was considered to short to derive any meaningful comparisons with a reference series. After a cautious inspection of the results due to the short length of the series, one station was deemed to be of poor quality and excluded from consideration in further analyses.
Dew-Point Temperature Values
Two of the seventeen stations were first excluded from further consideration due to their high missing value count. Values flagged as ‘W’ or had a ‘days of accumulation’ value of greater than one were removed and replaced with NA (missing). Values flagged as ‘S’ or ‘I’ were compared against the
71 observed value recorded on the same day for up to three neighbouring weather stations. If the value was considered plausible on comparison, then it remained within the series; or was removed and replaced with NA (missing), if not. Time series plots for each weather station were then created and manually inspected for any remaining implausible values.
This cleaned meteorological dataset was used in the analyses in Chapters
Four, Five and Six.
3.2.2 Ambient Air Pollution Data
The ambient air pollution dataset was sourced from the NSW Office of
Environment and Heritage. This state government agency maintains an extensive air quality monitoring network that covers several regions across NSW including
Greater Sydney, with monitoring sites located across metropolitan Sydney, the
Illawarra and Central Coast (NSW Office of Environment and Heritage 2018a).
The monitoring sites record daily concentrations of six ambient air pollutants including ozone, nitrogen dioxide, carbon monoxide, sulphur dioxide and particulate matter with an aerodynamic diameter of less than 10μm and 2.5μm
(PM10, PM2.5, respectively) (NSW Office of Environment and Heritage 2018a).
The ongoing monitoring of these six pollutants is essential to measure compliance with the national ambient air quality standards and goals outlined in the National
Environment Protection (Ambient Air Quality) Measure (NEPM) (NSW
Environment Protection Authority 2015). For each pollutant, the NEPM outlines the maximum allowable concentration for a specific averaging period, as well as the number of maximum allowable exceedances per annum (NEPM Schedule 2).
The standards and goals for nitrogen dioxide, carbon monoxide and sulphur dioxide are regularly met in the Greater Sydney region, but exceedances for ozone
72 and particulate matter can and still occur, particularly during the warmer months
(NSW Environment Protection Authority 2015).
Table 3.3 NEPM standards and goals for pollutants used in this thesis.
Maximum Averaging Maximum Pollutant Allowable Period Concentration Exceedancesa
1 hour 0.10ppm 1 day a year
Ozone
4 hours 0.08ppm 1 day a year
3 Particles as PM10 1 day 50μg/m 5 days a year
a Goal within 10 years Note: This table has been adapted from Schedule 2 of the NEPM
Daily observations of ambient air pollution concentrations for all six pollutants were requested and obtained from the NSW Office of Environment and
Heritage for all monitoring sites located within the SSD for the period of 1993 to
2014. The completeness of the data for this time period differed across each of the monitoring sites.
The NSW Office of Environment and Heritage follows several quality assurance procedures to ensure the data is precise, accurate, representative and comparable (NSW Office of Environment and Heritage 2015b). On inspection of the data, negative values were observed for some pollutants for some stations.
These values were assigned a value of 0, as advised by the NSW Office of
Environment and Heritage (pers. comm., 16 September 2015).
This dataset was used in the analyses in Chapters Five and Six.
73 3.2.3 School and Public Holiday Data
A list of the official public school holidays for NSW was obtained upon request from the NSW Department of Education for the period of 2001 to 2013.
This data was then transcribed into an excel file as a time series, with 1=yes
(school holiday day) and 0=no (not school holiday day). This data was used in the analyses in Chapter Five.
At the time when the studies were designed for this thesis, there was no official or authoritative source of public holiday data for the period of 2001 to
2013 for NSW or Greater Sydney. As a result, a list of official public holidays for this time period was developed by consulting a range of sources including previous official calendars and federal and state government websites. This data was then transcribed into an excel file as a time series, with 1=yes (public holiday day) and 0=no (not public holiday day). This data was used in the analyses in
Chapters Four, Five and Six.
3.2.4 Health Data
The NSW Ministry of Heath’s Admitted Patient Data Collection includes all health records for patients admitted to public and private hospitals in NSW
(Centre for Health Record Linkage 2018). De-identified hospital admissions records for all individuals admitted to both public and private hospitals located within the SSD were obtained upon request from this data collection for selected principal diagnoses for July 2001 to December 2013 (n=1,570,805). The specific diagnoses that were selected and used for analysis are outlined in Chapters Four,
Five and Six.
A flowchart of the process used to clean the hospital admissions dataset is outlined in Figure 3.1.
74
Individual level hospital admission records n=1570805
All exact duplicate records were removed n=64
Individual level hospital admission records n=1570741
Records with an admission date outside of 1 July 2001 to 30 June 2013 were removed n=71080
Individual level hospital
admission records n=1499661
Records not classified as ‘emergency’ were removed n=366924
Individual level hospital admission records n=1132737
Records with an implausible, unknown or missing value for the variables of age and sex were removed n=32
Individual level hospital admission records n=1132705
Figure 3.1 An overview of the process used to clean the hospital admissions dataset.
75 3.3 Methods
This section describes the study design used for the analyses in Chapters
Four, Five and Six.
3.3.1 Study Design
A time-stratified case-crossover design was used in Chapters Four, Five and Six (Maclure et al. 1991; Janes et al. 2005). The design is equivalent to a matched pair case-control design: it compares a case’s exposure on the day of an adverse health event (e.g. hospital admission) to their exposure on control days
(or referent times) that are selected before and/or after the event (Janes et al. 2005;
Bell et al. 2008; Barnett and Dobson 2010). A number of previous studies conducted in the SSD and other regions have used this design to estimate the association between extreme heat, including heat waves, and acute health outcomes (Vaneckova and Bambrick 2013; Wilson et al. 2013; Zhang et al. 2013;
Kent et al. 2014; Gronlund et al. 2014). The results obtained using a case- crossover design have been shown to be similar to the alternate time-series design
(Tong et al. 2012).
Control days are selected by generally using one of the following three approaches: a unidirectional, bidirectional, or time-stratified approach (Janes et al.
2005). A time-stratified approach was used to avoid any potential bias introduced by the other approaches, such as the potential for time trend bias using the unidirectional design (Janes et al. 2005). The selection and subsequent use of this approach is consistent with previous studies using the case-crossover design to estimate the association between extreme heat and adverse health outcomes (e.g.
Zhang et al. 2013; Grounlund et al. 2014). Cases and controls were matched on day of the week and within the same month. This matching technique allowed for
76 the confounding effects of season and long-term trends to be controlled for by study design.
3.3.2 Statistical Analysis
A detailed overview of the statistical analyses used for each study is provided in Chapters Four, Five and Six.
3.4 Ethics
This thesis research project was reviewed and approved by the University of New South Wales Human Research Low Risk Ethics Advisory Committee
Panel H: Reference Number: 08/2014/60.
77 Chapter Four: Aim 1
Aim 1: To examine and compare the impact of single and
consecutive days of extreme heat, including high minimum temperatures, on heat-related hospital admissions in Greater Sydney,
Australia, for a suite of extreme heat definitions.
4.1 Introduction
The patterns and nature of hot temperature extremes are changing. The number of hot days and warm nights has increased across most of Australia, as has the intensity, frequency and duration of heat waves (Alexander and Arblaster
2009; Perkins et al. 2013; Steffen et al. 2014; Bureau of Meteorology and CSIRO
2017). This warming trend is projected to continue in the future under climate change in Australia, with the magnitude and intensity of such changes projected to scale in accordance with future levels of GHG emissions (Cowan et al. 2014;
Alexander and Arblaster 2017).
At the global scale, observed and projected warming trends for minimum temperature extremes, such as warm nights, are generally larger than for maximum temperature extremes (Alexander et al. 2006; Perkins et al. 2012;
Donat et al. 2013; Sillmann et al. 2013). In Australia, minimum and maximum temperatures have warmed by 1.1 °C and 0.8 °C respectively since 1910 (Bureau of Meteorology and CSIRO 2014). Temperature records for high minimum temperatures have broken at a faster rate than records for high maximum
78 temperatures since the 1950s (Lewis and King 2015). By the end of this century, warming for certain hot minimum temperature extremes, such as warm and tropical nights, is projected to be more pronounced across Australia than for certain hot maximum temperature extremes, such as hot days (Alexander and
Arblaster 2017).
Extreme heat is known to have significant, adverse impacts on human health outcomes. An estimated 70,000 excess deaths occurred during the
European summer of 2003, with 40,000 excess deaths associated with the heat wave in August (Robine et al. 2008; García-Herrera et al. 2010). The 1995
Chicago heat wave led to 672 excess deaths and 1,072 excess hospital admissions, with most of these hospitalisations for direct heat-related conditions, such as dehydration (Semenza et al. 1999; Kaiser et al. 2007). The 2009 south-east
Australian heat wave led to 374 excess deaths in Victoria, and 32 excess deaths in
Adelaide, South Australia (Nitschke and Tucker 2009; Victorian Government
Department of Human Services 2009). A 14-fold increase in hospital admissions for direct heat-related conditions was also observed in Adelaide, and a 8-fold increase in emergency department presentations for these conditions was observed in Victoria (Nitschke and Tucker 2009; Victorian Government Department of
Human Services 2009).
A notable feature of each of these heat waves was the lack of night-time relief, with unusually high minimum temperatures recorded (Kunkel et al. 1996;
Poumadère et al. 2005; Victorian Government Department of Human Services
2009). Exposure to elevated overnight temperatures adversely impacts the quality and quantity of an individual’s sleep, leading to increased wakefulness, and decreased duration of rapid eye movement and slow wave sleep (Okatmoto-
79 Mizuno et al. 2010; Lan et al. 2014). This disrupted sleep, along with the need for the body to thermoregulate during the night, means that individuals are unable to naturally rest and recover from the day. Despite the known relevance and importance of such temperatures to human health, little work has quantified the association between extreme high minimum temperatures and health outcomes. A few studies have considered the impact of such temperatures on mortality (e.g.
Laaidi et al. 2012; Royé 2017), but less is known about their impact on hospital admissions, particularly heat-related conditions. Studies assessing the impact of single and consecutive days of extreme heat on heat-related hospital admissions have generally used metrics derived from maximum and mean temperature to define heat exposure (e.g. Vaneckova and Bambrick 2013; Bobb et al. 2014b;
Gronlund et al. 2016; Ogbomo et al. 2017).
The impact of heat waves on health outcomes can differ in accordance with their characteristics, such as their intensity, duration and timing within the summer season. Studies have shown that those heat waves that are more intense and longer in duration can have a greater impact on mortality risk (Anderson and
Bell 2009; Anderson and Bell 2011; D’Ippoliti et al. 2010; Tong et al. 2014).
Evidence for the impact of heat wave timing is somewhat less consistent. A few studies have found that those heat waves that occur first in the summer season have a greater impact on mortality risk than those that occur later in the season
(Anderson and Bell 2011; Son et al. 2012; Son et al. 2016). One study found little difference in the impact of the first heat wave compared to later heat waves on preterm births and non-accidental deaths in the United States (Kent et al. 2014).
Tong et al. (2014) reported that those heat waves that occurred later in the season to have a greater impact on mortality in Brisbane, Australia. Less is known about
80 the impact of heat wave timing on the association between heat waves and heat- related hospital admissions.
Some studies have suggested that the impact of heat waves on mortality risk can be captured more accurately by separating the effects of heat waves into two measures: the ‘main effect’ and the ‘added effect’ (Hajat et al. 2006;
Gasparrini and Armstrong 2011; Rocklöv et al. 2012). The main effect estimates the mortality risk associated with single days of high temperatures, while the added effect estimates the mortality risk associated with prolonged exposure to extreme heat over several consecutive days (Gasparrini and Armstrong 2011;
Zeng et al. 2014). Some evidence of an added heat wave effect on mortality risk has been found, although the size of this effect has differed across studies
(Gasparrini and Armstrong 2011; Barnett et al. 2012; Dong et al. 2016). Only a few studies have investigated whether there is evidence of an added heat wave effect in the association between heat waves and morbidity (e.g. Gronlund et al.
2014; Chen et al. 2017).
Sydney experienced its hottest summer season on record in 2016/2017, with new temperature records set for the number of hot days and warm nights occurring throughout the season (Bureau of Meteorology 2017a). The combined impact of climate change and urban expansion is expected to have a considerable impact on the city’s overnight temperatures in the future, with the number of nights that Sydney’s population will experience high heat-stress projected to triple
(Argüeso et al. 2015). Studies enhancing our understanding of how extreme heat, particularly extreme high minimum temperatures, affects health outcomes are critical for future health adaptation planning under climate change.
81 The aim of this study was to examine and compare the impact of single and consecutive days of extreme heat, including high minimum temperatures, on heat-related hospital admissions in Greater Sydney, Australia, for a suite of extreme heat definitions. We investigated the impact of heat wave characteristics, including intensity, duration and timing within the season, and whether there was evidence of an added heat wave effect.
4.2 Data and Methods
4.2.1 Meteorological Data
Daily weather data for all stations located in the SSD with near complete coverage of the period of 2001 to 2013 were obtained from the Australian Bureau of Meteorology (n=17). Before identifying extreme temperature events, such as summer heat waves, in a climate time series, it is important that the data undergo quality control checks (Alexander and Tebaldi 2012). This is because it is possible for incorrect data entries to be considered as real ‘extreme’ values and included in further analyses (Alexander and Tebaldi 2012). To ensure our observational weather data was of the highest possible quality, a series of quality control checks were performed on the observed daily maximum, minimum and dew-point temperature values for each weather station, and also tested for inhomogeneities in each daily maximum and minimum time series to inspect their overall quality
(see Chapter Three for an overview of this process). High quality stations (n=15) were then used to calculate the respective city-wide averages for each temperature metric if they had a total missing value count of ≤ 2.5% of the study period. The missing value threshold was set at ≤ 2.5% to maximise the number of stations
82 included the calculation of the average and subsequent spatial coverage of the
SSD, whilst ensuring that the quality of those stations included remained high.
The daily average mean temperature was calculated as the mean of the city-wide daily average maximum and minimum temperature values. For dew-point temperature, as the observations were recorded at 3-hour intervals over a 24-hour period, the city-wide average value for each time interval was first calculated with those stations where the missing value count was ≤ 2.5% of the study period, then the overall 24-hour daily average was calculated from these averaged time interval values.
The definition of an extreme hot day or warm night has two elements: the temperature metric (e.g. maximum temperature) and the temperature threshold
(i.e. absolute or relative). To define a heat wave, a third element is added, which is duration (e.g. ≥2 consecutive days). Multiple definitions for single and consecutive days (i.e. heat waves) of extreme heat were developed based on altering the combination of these elements (see Tables A2-A5, Appendix). Three measures were selected from the literature for the temperature metric (maximum temperature: Tmax, mean temperature: Tmean, minimum temperature: Tmin), five relative measures for the temperature threshold (90th, 95th, 97th, 98th, 99th percentile) (see Table A1 for calculations, Appendix), and for heat waves, three different lengths (≥2, ≥3 or ≥4 consecutive days). These definitions were used to identify all extreme heat events that occurred during the warm season (1
November to 31 March) for 2001 to 2013 (see Tables A2-A5, Appendix). The purpose of developing and using this suite of extreme heat definitions was not to determine which definition was ‘best’ or most appropriate to use in local heat
83 wave alerts, but rather to determine if, and how, the association differed across these definitions.
4.2.2 Health Data
Individual-level daily hospital admission records for all public and private hospitals located in the SSD were obtained from the NSW Ministry of Health,
Admitted Patient Data Collection, for 2001 to 2013 for a range of heat-related principal diagnoses (n=1570805). All exact duplicate records were extracted and removed (n=1570741, 64 records removed), as well those records with an admission date outside of 1 July 2001 – 30 June 2013 (n=1499661, 71080 records removed). Records that were classified as ‘emergency’ hospital admissions
(EHAs) were then selected for analysis to eliminate ‘pre-planned’ hospital admissions (n=1132737, records removed 366924) (Khalaj et al. 2010).
Remaining records with an implausible, unknown or missing entry for age
(ranged deemed plausible: 0 – 115 years) or sex (required entry: male or female) were extracted and removed (n=1132705, records removed 32). Those records with a principal diagnosis of acute renal failure (ICD-10-AM: N17), dehydration
(ICD-10-AM: E86), other disorders of fluid, electrolyte and acid-base balance
(fluid imbalance disorders, ICD-10-AM: E87), effects of heat and light (direct heat-related illnesses, ICD-10-AM: T67) and exposure to excessive natural heat
(ICD-10-AM: X31) were then selected and aggregated into daily total counts.
4.2.3 Study Design and Statistical Analysis
A time-stratified case-crossover study design was used (Maclure et al.
1991; Janes et al. 2005). This design has been used in previous studies to estimate the association between heat waves and hospital admissions (Zhang et al. 2013;
84 Gronlund et al. 2014), and has been shown to produce similar results to the alternate time-series design (Tong et al. 2012). The design is equivalent to a matched pair case-control design: it compares a case’s exposure on the day of an adverse health event (e.g. hospital admission) to their exposure on control days
(or referent times) that are selected before and/or after the event (Janes et al. 2005;
Bell et al. 2008; Barnett and Dobson 2010). Since each case acts as their own control, personal characteristics such as sex and smoking status are controlled for by matching (Barnett and Dobson 2010). A time-stratified approach was used to select control days to avoid potential bias introduced by other approaches, such as the unidirectional and bidirectional designs (Janes et al. 2005). Cases and controls were matched on day of the week and within the same month, and thus the confounding effects of season and long-term trends were controlled for by design.
Conditional logistic regression was used to estimate the association between single and consecutive days of extreme heat (exposure definitions: n=15 for single days and n=39 for heat waves) and hospital admissions for the selected diagnoses. The potential non-linear confounding effects of dew-point temperature were controlled for using a natural cubic spline with 3 degrees of freedom (knots at quantiles) (Davis et al. 2016; Kingsley et al. 2015), as well as public holidays
(binary variable). Ambient air pollution was not controlled for, as there is no evidence within the literature of an established association between this exposure and the selected diagnoses.
The statistical analyses were conducted in the ‘R’ Statistical Computing
Environment (Version 3.1.3) using the ‘season’ package. To examine the impact of extreme heat during the summer, the analyses were restricted to the warm season (1 November to 31 March) for 2001 to 2013. The effects are presented as
85 odds ratio with their corresponding 95% confidence intervals. Figures are presented on the log scale. A p-value of <0.05 was considered significant.
4.2.3.1 Heat Wave Timing
We examined the impact of the first heat wave of the season compared to later heat waves on our selected diagnoses. We stratified our heat waves into two groups: those heat waves that occurred first in each warm season and those heat waves that did not occur first in each warm season (Anderson and Bell 2011).
This was done only for those heat wave definitions with a temperature threshold equalling the ≥90th or ≥95th percentile, and a duration of ≥2 or ≥3 consecutive days, to ensure we had an adequate number of heat wave days to conduct the analysis (n=12 for total heat wave definitions).
4.2.3.2 Added Heat Wave Effect
We examined if there was an ‘added heat wave effect’ (Hajat et al. 2006) in the association between heat waves and our selected diagnoses. We controlled for average daily temperature using a natural cubic spline with 2 degrees of freedom (knots at quantiles) for the diagnoses of acute renal failure, dehydration and fluid imbalance disorders. For direct heat-related conditions, daily average temperature was controlled for as a linear variable (sensitivity tests modelling daily average temperature as a linear variable, natural cubic spline with two degrees of freedom and natural cubic spline with three degrees of freedom showed that modelling temperature as a linear variable was the most appropriate - data not shown). We controlled for the daily average of the respective temperature metric that was used in the heat wave definition.
86 4.3 Results
Table 4.1 presents descriptive statistics of the selected heat-related EHAs.
There was a total of 6,349 EHAs with a principal diagnosis of acute renal failure during the study period; 4,830 of dehydration; 6,079 of fluid imbalance disorders; and 433 of direct-heat related conditions. EHAs for acute renal failure and fluid imbalance disorders had the highest median, with EHAs for dehydration having the highest maximum, and all EHAs having a minimum of 0. The results for direct heat-related illnesses are presented in Appendix due to the small total EHA count for this diagnosis (Figure A1, Figure A2, Table A6). There were no EHAs with a principal diagnosis of X31.
Descriptive statistics for each definition of extreme heat used are outlined in Tables A1-A5 in the Appendix. A total of 15 single days of extreme heat and
39 heat wave definitions were calculated using the selected temperature metrics, temperature thresholds, and for heat waves, duration. Heat waves defined as ≥ 2
th consecutive days ≥ 90 percentile with Tmin had the highest number of total heat wave days, while heat waves defined as ≥ 2 consecutive days ≥ 90th percentile with Tmean had the highest number of total heat wave events.
Figure 4.1 shows the association between single days of extreme heat
th th th th th (defined by Tmax, Tmean, Tmin at thresholds of the ≥ 90 , 95 , 97 , 98 and 99 percentile) and EHAs with a principal diagnosis of acute renal failure, dehydration and fluid imbalance disorders in the SSD during the warm season for 2001 to
2013. Positive, significant associations were found between single days of extreme heat and EHAs for acute renal failure and dehydration for all 15
th definitions, with estimates ranging from 1.15 (95% CI 1.03, 1.27) (Tmin ≥90
th percentile) to 1.65 (95% CI 1.31, 2.07) (Tmin ≥99 percentile) and 1.42 (95% CI
87 th th 1.27, 1.59) (Tmin ≥90 percentile) to 2.87 (95% CI 2.32, 3.55) (Tmean ≥ 99 percentile) across the 15 definitions respectively. Positive associations were also found between single days of extreme heat and EHAs for fluid imbalance disorders, with statistically significant associations found for 14 of the 15
th definitions. Estimates ranged from 1.13 (95% CI 1.02, 1.25) (Tmin ≥90
th percentile) to 1.64 (95% CI 1.28, 2.09) (Tmax ≥99 percentile) across the definitions. For all three diagnoses, the weakest associations were found using the
th same definition (Tmin ≥ 90 percentile), while the strongest associations were found using different definitions (although the same temperature threshold, but
th different temperature metrics: acute renal failure, Tmin ≥ 99 percentile;
th th dehydration, Tmean ≥ 99 percentile; fluid imbalance disorders, Tmax ≥ 99 percentile). Positive, significant associations were also found between single days of extreme heat and EHAs for direct heat-related conditions for all 15 definitions
(Figure A1).
Figure 4.2 shows the association between heat wave days and EHAs with a principal diagnosis of acute renal failure, dehydration and fluid imbalance disorders in the SSD for 2001 to 2013 (warm season) for 39 heat wave definitions.
Positive associations were found between heat wave days and EHAs for all three diagnoses, for all 39 heat wave definitions, although not all associations were statistically significant. For acute renal failure, statistically significant positive associations were found for 18 of the 39 definitions, with estimates ranging from
th 1.09 (95% CI 0.83, 1.44) (Tmean ≥95 percentile for ≥3 consecutive days) to 1.53
th (95% CI: 1.11, 2.12) (Tmin ≥ 99 percentile for ≥2 consecutive days) (Figure 4.2a).
Statistically significant positive associations were found for 38 of the 39 definitions for dehydration, with estimates ranging from 1.50 (95% CI 1.31, 1.72)
88 th (Tmin ≥90 percentile for ≥2 consecutive days) to 4.74 (95% CI 3.29, 6.85)
th th (multiple definitions: Tmin ≥ 98 , 99 percentile for ≥3 consecutive days and Tmin
≥ 97th, 98th, 99th percentile for ≥4 consecutive days) (Figure 4.2b). For fluid
th imbalance disorders, estimates ranged from 1.02 (95% CI 0.76, 1.38) (Tmin ≥95
th percentile for ≥4 consecutive days) to 3.01 (95% CI 1.54, 5.88) (Tmax ≥99 percentile for ≥2 consecutive days) across the 39 definitions, with statistically significant positive associations found for 25 of the 39 definitions (Figure 4.2c).
The strongest associations between heat wave days and EHAs for acute renal failure and dehydration were found when heat waves were defined as consecutive
th days of extreme minimum temperatures (acute renal failure: Tmin ≥ 99 percentile
th th for ≥2 consecutive days; dehydration: multiple definitions: Tmin ≥ 98 , 99
th th th percentile for ≥3 consecutive days and Tmin ≥ 97 , 98 , 99 percentile for ≥4 consecutive days). For fluid imbalance disorders, the strongest associations were found when heat waves were defined as consecutive days of extreme maximum
th temperatures (Tmax ≥99 percentile for ≥2 consecutive days). Postive, significant associations between heat wave days and EHAs for direct heat-related conditions were found for 38 of the 39 definitions (Figure A1).
Tables 4.2, 4.3 and 4.4 show the impact of the first heat wave of the season on EHAs for acute renal failure, dehydration and fluid imbalance disorders compared to the impact of later heat waves of the season for 12 heat wave definitions. For each diagnosis, the comparative impact is shown to be sensitive to the choice of heat wave definition used, with some definitions showing that the first heat wave of the season has a greater impact, and others showing that it does not. The heat wave definitions that show that the first heat wave has a greater impact are generally not consistent all three diagnoses, with the exception of two
89 th definitions (Tmax ≥ 95 percentile for ≥ 2 and ≥3 consecutive days). Similar patterns were also observed for direct heat-related conditions in Table A6.
Figure 4.3 shows the association between heat wave days and EHAs for acute renal failure, dehydration and fluid imbalance disorders in the SSD for 2001 to 2013 (warm season) after controlling for average daily temperature (or the
‘added heat wave effect’). Positive associations were found between heat wave days and EHAs for acute renal failure for 20 of the 39 definitions, although these associations were not statistically significant (Figure 4.3a). The estimates of these
th positive associations ranged from 1.001 (95% CI 0.777, 1.291) (Tmax ≥95
th percentile for ≥3 consecutive days) to 1.25 (95% CI 0.95, 1.64) (Tmin ≥95 percentile for ≥4 consecutive days). Positive associations were found between heat wave days and EHAs for dehydration for 36 of the 39 definitions after controlling for daily average temperature, with statistically significant associations found for 30 of the 39 definitions (Figure 4.3b). Positive estimates
th were found to range from 1.08 (95% CI 0.92, 1.26) (Tmax ≥90 percentile for ≥2
th consecutive days) to 3.15 (95% CI 2.13, 4.65) (multiple definitions: Tmin ≥98 ,
th th th th 99 percentile for ≥3 consecutive days and Tmin ≥97 , 98 , 99 percentile for ≥4 consecutive days) across the 39 definitions. One statistically significant positive association was observed between heat wave days and EHAs fluid imbalance disorders after controlling for daily average temperature, with an odds ratio of
th 2.20 (95% CI 1.11, 4.36) (Tmax ≥99 percentile for ≥2 consecutive days) (Figure
4.3c). Positive, insignificant associations were found for 26 of the 39 definitions,
th with estimates ranging from 1.002 (95% CI: 0.687, 1.459) (Tmin ≥ 97 percentile for ≥3 consecutive days) to 1.31 (95% CI: 0.86, 2.00) (multiple definitions: Tmin
th th th th th ≥98 , 99 percentile for ≥3 consecutive days and Tmin ≥97 , 98 , 99 percentile
90 for ≥4 consecutive days). For each diagnosis, the added heat wave effect was generally found to be strongest when heat waves were defined as consecutive days of extreme minimum temperatures at high temperature thresholds, with the exception of fluid imbalance disorders, where the strongest association was found using Tmax. Positive associations were found between heatwave days and EHAs for direct heat-related conditions for 34 of the 39 definitions, with statistically significant associations found for 14 of the 39 definitions (Figure A2).
91 Table 4.1 Descriptive statistics for selected heat-related EHAs during the warm season in the SSD, 2001 to 2013.
ICD-10-AM Total Median Selected EHAs (Principal Maximum Minimum EHAs (IQR) Diagnosis) Heat-Related Condition Acute renal failure N17 6 349 3 (2-5) 15 0 Dehydration E86 4 830 2 (1-4) 25 0 Fluid imbalance disorders E87 6 079 3 (2-5) 14 0 Direct heat-related illnesses T67 433 0 (0-0) 16 0
92 Table 4.2 Effect of heat waves first in season compared to heat waves not first in season on EHAs for acute renal failure in the SSD during the warm season, 2001 to 2013.
Heat wave Definition Odds Ratio (95% CI)
Duration Metric Threshold (consecutive First in season Not first in season (°C) (percentile) days) th Maximum ≥90 ≥2 1.01(0.79, 1.27) 1.22(1.07, 1.39) Maximum ≥95th ≥2 1.35(1.10, 1.64) 0.93(0.69, 1.24) Maximum ≥90th ≥3 1.16(0.94, 1.42) 1.34(1.04, 1.73) Maximum ≥95th ≥3 1.27(0.99, 1.64) 1.21(0.61, 2.36) Mean ≥90th ≥2 1.07(0.86, 1.34) 1.25(1.11, 1.42) Mean ≥95th ≥2 0.99(0.75, 1.30) 1.40(1.13, 1.73) Mean ≥90th ≥3 1.04(0.82, 1.31) 1.35(1.10, 1.66) Mean ≥95th ≥3 0.69(0.40, 1.17) 1.34(0.97, 1.86) th Minimum ≥90 ≥2 1.12(0.90, 1.40) 1.15(0.99, 1.32) Minimum ≥95th ≥2 1.26(0.99, 1.60) 1.05(0.84, 1.31) Minimum ≥90th ≥3 1.14(0.94, 1.39) 1.08(0.87, 1.34) Minimum ≥95th ≥3 1.08(0.81, 1.45) 1.45(1.00, 2.10)
93 Table 4.3 Effect of heat waves first in season compared to heat waves not first in season on EHAs for dehydration in the SSD during the warm season, 2001 to 2013.
Heat wave Definition Odds Ratio (95% CI)
Duration Metric Threshold (consecutive First in season Not first in season (°C) (percentile) days) th Maximum ≥90 ≥2 1.48(1.17, 1.87) 1.75(1.53, 2.00) Maximum ≥95th ≥2 2.24(1.85, 2.71) 1.23(0.91, 1.66) Maximum ≥90th ≥3 1.50(1.20, 1.86) 2.50(1.99, 3.14) Maximum ≥95th ≥3 2.48(1.96, 3.14) 1.42(0.80, 2.50) Mean ≥90th ≥2 1.48(1.18, 1.86) 1.73(1.52, 1.97) Mean ≥95th ≥2 2.27(1.76, 2.94) 1.98(1.61, 2.44) Mean ≥90th ≥3 1.66(1.30, 2.11) 2.10(1.73, 2.54) Mean ≥95th ≥3 1.43(0.92, 2.25) 3.42(2.53, 4.63) th Minimum ≥90 ≥2 1.49(1.19, 1.85) 1.42(1.21, 1.66) Minimum ≥95th ≥2 1.35(1.03, 1.76) 2.36(1.88, 2.96) Minimum ≥90th ≥3 1.36(1.09, 1.69) 1.79(1.45, 2.22) Minimum ≥95th ≥3 1.79(1.33, 2.41) 4.74(3.29, 6.85)
94 Table 4.4 Effect of heat waves first in season compared to heat waves not first in season on EHAs for fluid imbalance disorders in the SSD during the warm season, 2001 to 2013.
Heat wave Definition Odds Ratio (95% CI)
Duration Metric Threshold (consecutive First in season Not first in season (°C) (percentile) days) th Maximum ≥90 ≥2 1.18(0.92, 1.50) 1.19(1.04, 1.36) Maximum ≥95th ≥2 1.37(1.12, 1.69) 1.27(0.95, 1.70) Maximum ≥90th ≥3 0.99(0.79, 1.25) 1.41(1.11, 1.80) Maximum ≥95th ≥3 1.44(1.11, 1.86) 1.21(0.71, 2.06) Mean ≥90th ≥2 1.15(0.92, 1.45) 1.23(1.08, 1.40) Mean ≥95th ≥2 1.32(1.03, 1.71) 1.34(1.08, 1.67) Mean ≥90th ≥3 1.20(0.94, 1.53) 1.21(0.98, 1.48) Mean ≥95th ≥3 1.20(0.82, 1.76) 1.70(1.21, 2.40) th Minimum ≥90 ≥2 1.14(0.91, 1.43) 1.15(1.00, 1.33) Minimum ≥95th ≥2 1.07(0.82, 1.40) 1.08(0.87, 1.35) Minimum ≥90th ≥3 1.02(0.83, 1.26) 1.16(0.95, 1.42) Minimum ≥95th ≥3 0.85(0.62, 1.16) 1.56(1.05, 2.34)
95
Figure 4.1 The association between single days of extreme heat and EHAs for acute renal failure (ARF), dehydration (DEH) and fluid imbalance disorders
(FIDs) in the SSD during the warm season, 2001 to 2013
96
a
b c
Figure 4.2 The association between heat wave days and EHAs for acute renal failure (Figure 4.2a), dehydration (Figure 4.2b) and fluid imbalance disorders
(Figure 4.2c) in the SSD during the warm season, 2001 to 2013.
97 a
b c
Figure 4.3 The association between heat wave days and EHAs for acute renal failure (Figure 4.3a), dehydration (Figure 4.3b) and fluid imbalance disorders
(Figure 4.3c) in the SSD during the warm season, 2001 to 2013, after controlling for daily average temperature (or the ‘added heat wave effect’).
98 4.4 Discussion
This study examined and compared the impact of single and consecutive days of extreme heat on heat-related hospital admissions during the warm season in Greater Sydney, Australia, using a suite of extreme heat definitions. Positive, statistically significant associations were generally found between single days of extreme heat and EHAs for acute renal failure, dehydration and fluid imbalance disorders (Figure 4.1). The strongest associations were found at the highest temperature threshold (i.e. 99th percentile), although using different temperature metrics. Positive associations were also found between consecutive days (i.e. heat wave days) of extreme heat and EHAs for acute renal failure, dehydration and fluid imbalance disorders (Figure 4.2). The strongest associations for acute renal failure and dehydration were found when heat waves were defined using minimum temperature (i.e. consecutive warm nights), and using maximum temperature (i.e. consecutive hot days) for fluid imbalance disorders. For these diagnoses, the strongest associations were found for heat waves defined using higher temperature thresholds, but not necessarily for the longest heat waves.
Previous studies have reported strong, positive associations between single and/or consecutive of days extreme heat and hospitalisations for acute renal failure and dehydration; but have not considered or compared the impact of extreme high minimum temperatures alone (e.g. Hansen et al. 2008b; Vaneckova and Bambrick 2013; Bobb et al. 2014b; Gronlund et al. 2016; Ogbomo et al.
2017). A few studies have examined the association between elevated temperature and renal morbidity using minimum temperatures as a continuous exposure metric, but little is known about the impact of elevated minimum temperatures on dehydration. Fletcher et al. (2012) observed a 6 per cent increase in the risk of
99 hospitalisation for acute renal failure per 2.78°C (5°F) for mean and minimum temperature at lag0 in New York State, United States; greater than the 4 per cent increase observed for maximum temperature. Williams et al. (2012) found an eleven per cent increase in risk of ED presentation for renal disease per 10°C increase in minimum temperature at lag0 in Perth, Australia.
We also investigated whether there was evidence of an ‘added heat wave effect’ for each condition. Some evidence of this effect was found, although such evidence was not statistically significant for acute renal failure, and not consistently statistically significant across all heat wave definitions for dehydration and fluid imbalance disorders (Figure 4.3). For acute renal failure and dehydration, the strongest associations were found when heat waves were defined using minimum temperature (i.e consecutive warm nights), and for fluid imbalance disorders, when heat waves were defined using maximum temperature
(i.e. consecutive hot days). Gronlund et al. (2014) found a small, insignificant added effect in the association between heat waves (defined when apparent temperature > 95th percentile for at least 2 and 4 days) and hospital admissions for renal diseases among the elderly in the United States (2.6%: 95% CI: -0.1, 6.4;
4.6%: 95% CI:-0.5, 9.9 respectively). Chen et al. (2017) found evidence of an added heat wave effect in the association heat waves and emergency department visits for acute renal failure at lag0, but such associations were only statistically significant for those heat waves defined with minimum temperature and maximum apparent temperature. Chen et al. (2017) also found the strongest associations between heat waves and emergency department visits for fluid and electrolyte imbalance when heat waves were defined using maximum temperature, which is consistent with our finding.
100 Exposure to high overnight temperatures can disrupt the body’s ability to naturally rest and recover from the day. Elevated overnight temperatures are known to adversely affect the quality and quantity of an individual’s sleep, resulting in increased wakefulness, and decreased rapid eye movement and slow wave sleep, as well as sleep efficiency (Okamoto-Mizuno et al. 2010; Lan et al.
2014). Libert et al. (1988) showed that sleep patterns and processes do not adapt to the heat, with sleep quality and quantity not improving after a period of continuous exposure to elevated day and night-time temperatures. Sleep is directly related to thermoregulation: it is theorised sleep occurs after a drop in core body temperature and that sleep time peaks when thermoneutrality is achieved
(Parmeggiani 1977; Gilbert et al. 2004). Such disrupted sleep patterns, along with the need to thermoregulate during the night, is likely to stress the body, where such stress may be additional to that already placed on the body from heat exposure during the day. There is some evidence to suggest that high minimum temperatures may have a greater impact on mortality than high maximum temperatures, particularly in urban areas. For example, Laaidi et al. (2012) found that exposure to consecutive nights of high minimum temperatures had a greater impact on mortality among the elderly in Paris during the 2003 European heatwave, than daily maximum temperatures.
As Fletcher et al. (2012) note, the exact biological pathways and mechanisms by which extreme heat exposure leads to acute renal failure are not clear. Some clinical studies have shown that acute renal failure can occur as a complication of a heat-related condition, such as heat stroke or dehydration, leading to kidney damage from rhabdomolysis and myoglobinurina (Salem 1968;
Tan et al. 1995; Vertel et al. 1967). Sweating is the most efficient means by which
101 our bodies shed heat, and when we release more fluids via sweating than we consume, our internal fluid balance can be disrupted and we can dehydrate
(Glazer 2005; Becker and Stewart 2011). This can lead to mild heat-related illnesses such as heat cramps and exhaustion, through to life-threating conditions such as heatstroke (internal body temperature ~≥40°C) (Becker and Stewart
2011). For most individuals, these conditions can be largely prevented through behavioural changes, such as maintaining adequate fluid intake, reducing heat exposure and reducing physical activity levels (Becker and Stewart 2011).
We also investigated how heat wave timing affects the risk of hospitalisation for each condition. We found that the impact of the first heat wave of the season compared to later heat waves was sensitive to the heat wave definition used to define heat exposure. For each diagnosis, the first heat wave of the season was found to have a greater impact than later heat waves for some definitions, but not others (Tables 4.2-4.4, Table A6). The first heat wave of the season has been found to have a greater impact on mortality risk than later heat waves in other regions (Anderson and Bell 2011; Son et al. 2012; Son et al. 2016).
Anderson and Bell (2011) showed that this finding was generally robust to changes in heat waves definitions, which is largely inconsistent with our results. It is thought that earlier heat waves have a greater impact on mortality because there is an increase number of susceptible individuals (or ‘pool’) at the beginning of the warm season, or because individuals can acclimatise to heat exposure throughout the warm season (Anderson and Bell 2011).
In contrast, Kent et al. (2014) reported that the impact of the first heat wave compared to later heat waves on preterm births and non-accidental deaths in
Alabama, United States was similar. In Australia, heat waves were found to have
102 stronger impacts on mortality and emergency hospital admissions in the later half of the warm season in Brisbane (Tong et al. 2014). These findings were attributed to the fact that longer and more intense heat waves occurred more frequently in the later part of the season (Tong et al. 2014). We found that the average and peak intensity, and average duration, of those heat waves first in the season differed across our 12 heat wave definitions (see Tables A7 and A8, Appendix). These differences might explain why our comparative estimates were inconsistent across our 12 heat wave definitions for each of our diagnoses.
A few limitations are noted for this study. Differences in daily maximum and minimum temperatures can be observed across Greater Sydney, with higher daily maximum temperatures often recorded in the western areas of the city
(Bureau of Meteorology 2018d; Bureau of Meteorology 2018e). As we used the daily city-wide average to estimate exposure, our study did not account for any climatic differences across the city. When examining the added heat wave effect, we used the temperature metric of the relevant heat wave definition to adjust for daily average temperature. This means that for those heat wave definitions defined using daily minimum temperature, we controlled for average daily minimum temperature only, and thus did not account for any potential confounding or accumulative effects of exposure to high temperatures during the day or on previous days. This limitation could also be extended to our assessment of the impact of single and consecutive nights of high minimum temperatures. It is likely that most of the population spends their evenings indoors, and therefore using observed minimum temperature values from weather stations across the city is unlikely to measure an individual’s true exposure to overnight temperatures. This is also true of exposure measurement for the other metrics (e.g. maximum
103 temperature), as individuals move between indoor and outdoor environments throughout the day. Future work could also consider the impact of various temperature metrics, especially minimum temperatures, on other heat-sensitive conditions including cardiovascular and respiratory conditions.
4.5 Chapter Conclusion
This study found strong, positive associations between extreme heat and heat-related hospital admissions in Greater Sydney, Australia, for a suite of extreme heat definitions. The strongest associations between heat wave days and
EHAs for acute renal failure and dehydration were observed when heat waves were defined using minimum temperature at the highest temperature thresholds.
The strongest added heat wave effects were also observed for these conditions when heat waves were defined in this way. These findings indicate that extreme high overnight temperatures have stronger impacts than extreme high daytime temperatures on these conditions. Such findings have important implications for population health in Greater Sydney, with minimum temperatures expected to increase in the future due to climate change and urban expansion.
104 Chapter Five: Aim 2
Aim 2: To examine whether ozone modifies the short-term
association between heat waves and hospital admissions for certain
respiratory diseases in Greater Sydney, Australia.
5.1 Introduction
Respiratory diseases were the sixth leading cause of the total burden of disease in Australia in 2011, contributing 8 per cent to the total burden (Australian
Institute of Health and Welfare 2016). Studies from several regions have shown that increased temperatures and heat waves are associated with an increased risk of hospitalisation for total and specific respiratory diseases including asthma, chronic obstructive pulmonary disease and respiratory infections, particularly among the elderly (Michelozzi et al. 2009; Anderson et al. 2013; Wilson et al.
2013; Qiu et al. 2016; Soneja et al. 2016). Other studies however, have found little evidence of an association between increased temperatures or heat waves and hospital admissions for respiratory diseases, or reported negative associations
(Nitschke et al. 2007; Ogbomo et al. 2017; Sherbakov et al. 2017). Nevertheless, work from the Northern Hemisphere has projected that the number of heat-related respiratory hospitalisations are likely to increase under climate change, resulting in increased healthcare costs to address this burden (Lin et al. 2012; Åström et al.
2013). To anticipate the future health impacts of climate change in Australia, and their associated increased costs, it is important to enhance and refine our understanding of the association between extreme heat and respiratory morbidity.
105 Elevated levels of tropospheric ozone are often observed during the summer season in urban areas, particularly on days of high temperatures and during heat waves (Sillman and Samson 1995; Fiala et al. 2003). This is generally because the chemical reactions between oxides of nitrogen and volatile organic compounds that create ozone intensify in the presence of sunlight and high ambient temperatures (US EPA 2017). Elevated levels of ozone have also been found to be associated with an increased risk of hospitalisation for respiratory diseases (Ji et al. 2011), and specific conditions including asthma (Chen et al.
2016; Goodman et al. 2017), and chronic obstructive pulmonary disease and pneumonia (Schwartz 1994; Anderson et al. 1997; Medina-Ramón et al. 2006).
As a consequence of these known associations and correlation between high temperatures and ozone, it has become common for studies to adjust for, or examine, the potential confounding effects of ozone when examining the relationship between high temperatures and hospital admissions for respiratory diseases (Michelozzi et al. 2009; Green et al. 2010; Anderson et al. 2013).
Few studies have investigated the potential joint, or interactive effects, between high temperatures, especially heat waves, and ozone on respiratory health outcomes. This is concerning given that the joint effect of weather and air pollution on health outcomes is thought to be greater than the risk derived from the individual impacts of these two exposures (Zanobetti and Peters 2015). There is also some suggestion that an interactive effect between temperature and air pollution may be biologically plausible (Gordon 2003). A few studies from
Europe and North America have examined whether temperature modifies the association between ozone and all-cause, respiratory or cardiovascular mortality
(Ren et al. 2009; Pattenden et al 2010; Burkart et al. 2013; Jhun et al. 2014).
106 These studies generally found stronger associations between ozone and mortality at higher temperatures, although important geographical differences were often observed in multi-city studies and evidence of effect modification was not consistently statistically significant. For example, Pattenden et al. (2010) observed a statistically significant interaction between ozone and hot days on mortality in
London and Cardiff, but found little evidence of an interaction in cities elsewhere in the United Kingdom. Further, Ren et al. (2006) found that the association between ozone and cardiovascular mortality was generally stronger at higher temperatures in the northern regions of the United States, but little and inconsistent evidence of effect modification was found in the southern regions.
Less is known about whether ozone modifies the association between temperature, especially heat waves, and respiratory health outcomes. Stronger associations between high temperatures and cardiovascular and non-accidental mortality have been observed at higher levels of ozone in the United States and
Europe respectively (Ren et al. 2007; Breitner et al. 2014). A few studies have estimated the number or proportion of excess deaths attributable to ozone exposure during specific heat wave events or summer seasons, such as the 2003
European heat wave and summer, and the 2004 Brisbane heat wave (Fischer et al.
2004; Stedman 2004; Tong et al. 2010a). Others have examined the potential interactive effects of exposure to high temperatures and ozone during the 2003 heat wave and summer in France, respectively (Dear et al. 2005; Filleul et al.
2006). Time-series studies examining whether ozone modifies the association between heat waves and health outcomes are limited. One study from Europe found that the pooled effects of heat waves on total and cardiovascular mortality
107 were stronger at high compared to low-levels of ozone across nine European cities, but not for respiratory mortality (Analitis et al. 2014).
Little is known about whether ozone modifies the association between heat waves and respiratory morbidity, particularly cause-specific respiratory morbidity. This study aimed to examine whether ozone modifies the short-term association between heat waves and hospital admissions for certain respiratory diseases in Greater Sydney, Australia. We also investigated the susceptibility of specific age groups, and tested the sensitivity of three heat wave definitions used.
5.2 Data and Methods
5.2.1 Meteorological Data
Daily weather data for all stations located in the SSD with near complete coverage of the period of 2001 to 2013 were obtained from the Australian
Government’s Bureau of Meteorology (n=17). Before identifying extreme temperature events in a climate time series, such as summer heat waves, it is important that the data undergo quality control checks (Alexander and Tebaldi
2012). This is because it is possible for incorrect data entries to be considered as real ‘extreme’ values and included in further analyses (Alexander and Tebaldi
2012). To ensure our observational weather data was of the highest possible quality, we performed a series of quality control checks on the observed daily maximum, minimum and dew-point temperature values for each weather station, and also tested for inhomogeneities in each daily maximum and minimum time series to inspect their overall quality (See Chapter Three, for an overview of this process). High quality stations (n=15) were then used to calculate the respective city-wide averages for each temperature metric if they had a total missing value
108 count of ≤ 2.5% of the study period. The missing value threshold was set at
≤2.5% to maximise the number of stations included the calculation of the average and subsequent spatial coverage of the SSD, while ensuring that the quality of those stations included remained high. The daily average mean temperature was calculated as the mean of the city-wide daily average maximum and minimum temperature values. For dew-point temperature, as the observations were recorded at 3-hour intervals over a 24-hour period, the city-wide average value for each time interval was first calculated with those stations where the missing value count was ≤ 2.5% of the study period, then the overall 24-hour daily average was calculated from these averaged time interval values.
In the absence of a standard heat wave definition, we selected and compared three heat wave definitions for this study. Previous studies have shown that the choice of heat wave definition can alter the magnitude and statistical significance of the association between heat wave and adverse heat outcomes.
(Tong et al. 2010; Kent et al. 2014). We defined a heat wave as two or more consecutive days where the temperature metric (three temperature metrics were selected and compared: maximum temperature (HWD01); mean temperature
(HWD02); and minimum temperature (HWD03)) is greater than or equal to the
90th percentile of the warm season (1 November to 31 March) during 2001 to
2013. We compared heat wave definitions with alternative temperature metrics, rather than temperature thresholds or durations, to ensure we kept an adequate number of heat wave days to conduct the analysis.
5.2.2 Ambient Air Pollution Data
Daily air pollution data for all stations located in the SSD were obtained from the NSW Office of Environment and Heritage for 2001 to 2013. Daily data
109 were obtained for the following air pollutants and used in this study: ozone (1hr average 24hr maximum value (pphm)), nitrogen dioxide (1hr average 24hr maximum value (pphm)) and particulate matter (particles with an aerodynamic diameter less than 10m, PM10) (1hr average 24hr average value). The NSW
Office of Environment and Heritage follows several quality assurance procedures to ensure the data is precise, accurate, representative and comparable (NSW
Office of Environment and Heritage 2015b). Negative daily values were assigned a value of 0, as advised by the NSW OEH (pers. comm., 16 September, 2015).
Stations that had a missing value count of ≤5% of the study period were used to calculate the daily city-wide average for each pollutant. Junger and Ponce de Leon
(2015) regarded a missing data level of 5% as the best case scenario in their application of time-series air pollution data. Similar to the threshold selected for our meteorological data, a threshold of 5% was optimal in allowing us to maximise the number of stations included the calculation of the average and subsequent spatial coverage of the SSD, while also ensuring that the quality of those stations included remained high.
5.2.3 Health Data
Individual-level daily hospital admission records with a principal diagnosis of J00-J99 (ICD-10-AM) for all public and private hospitals located in the SSD were obtained from the NSW Ministry of Health, Admitted Patient Data
Collection, for 2001 to 2013 as part of a larger dataset (n=1570805). All exact duplicate records were extracted and removed (n=1570741, 64 records removed), as well those records with an admission date outside of 1 July 2001 – 30 June
2013 (n=1499661, 71080 records removed). Records that were classified as
‘emergency’ hospital admissions (EHAs) were then selected for analysis to
110 eliminate ‘pre-planned’ hospital admissions (n=1132737, records removed
366924) (Khalaj et al. 2010). We then extracted and removed remaining records with an implausible, unknown or missing entry for age (ranged deemed plausible:
0 – 115 years) or sex (required entry: male or female) (n=1132705, records removed 32). Those records with a principal diagnosis of asthma (ICD-10-AM:
J45-46), chronic obstructive pulmonary disease and associated conditions (COPD,
ICD-10-AM: J40-J44, J47) and pneumonia (ICD-10-AM: J12-J18) were then selected, aggregated into daily counts. To test the susceptibility of specific age groups, we stratified the data in the following way: asthma: all ages, 0-14, 15-64,
65 years and over; COPD: all ages, 0-64, 65-74, 75 years and over; and pneumonia: all ages, 0-14, 15-64, 65-74, 75 years and over.
5.2.4 Study Design and Statistical Analysis
We used a time-stratified case-crossover study design (Maclure et al.
1991; Janes et al. 2005). This design has been used in previous studies to estimate the association between heat waves and hospital admissions (Zhang et al. 2013;
Gronlund et al. 2014), and has been shown to produce similar results to the alternate time-series design (Tong et al. 2012). The design is equivalent to a matched pair case-control design: it compares a case’s exposure on the day of an adverse health event (e.g. hospital admission) to their exposure on control days
(or referent times) that are selected before and/or after the event (Janes et al. 2005;
Bell et al. 2008; Barnett and Dobson 2010). Since each case acts as their own control, personal characteristics such as sex and smoking status are controlled for by matching (Barnett and Dobson 2010). We used the time-stratified approach to select control days to avoid potential bias introduced by other approaches, such as the unidirectional and bidirectional designs (Janes et al. 2005). We matched cases
111 and controls on day of the week and within the same month, and thus controlled for the confounding effects of season and long-term trends by design.
We used conditional logistic regression to estimate the association between heat wave days and EHAs for our three selected respiratory diseases. We first estimated the association with, and without, adjusting for daily average ozone at lag0, while also controlling for daily average dew-point temperature using a natural cubic spline (df=3, knots at quantiles) (Davis et al. 2016: Kingsley et al.
2016), daily average nitrogen dioxide (1hr average 24h maximum value (pphm)) and daily average particulate matter (PM10) (1hr average 24hr average value
(g/m3)). We also controlled for public and school holidays (school holidays were controlled for in the following age groups: asthma (all ages, 0-14 years); COPD
(all ages, 0-64 years) and pneumonia (all ages, 0-14 years)) to adjust for holiday effects.
To examine whether ozone modifies the association between heat waves and EHAs for our selected respiratory diagnoses, we estimated and compared heat wave effects on days with high and low levels of ozone. High and low level ozone days were defined as those where the daily average ozone value was ≥ 95th and <
95th percentile of the warm season during 2001 to 2013 respectively (Note: 95th percentile of the distribution was equal to 6.54 pphm; see Table A9 for further descriptive statistics, Appendix). We created an interaction term between high and low level ozone days (1=high, 0=low) and heat wave days (1=yes, 0=no). This term was added to the model, along with the respective individual variables, and potential confounding variables described in the previous paragraph. We selected the threshold of the 95th percentile for two main reasons: to ensure there was a reasonably equal distribution of high and low level ozone days across heat wave
112 days for a fair comparison; and to compare and estimate heat wave effects on the most ‘extreme’ of ozone days for Sydney. The city-wide daily average value calculated at the 95th percentile was below the current health standard threshold of
0.10 ppm (1 hour average) (equivalent to 0.0654 ppm) (NSW Office of
Environment and Heritage 2017).
The statistical analyses were conducted in the ‘R’ Statistical Computing
Environment (Version 3.2.1) using the ‘season’ and ‘dlnm’ packages. As we wanted to examine the impact of summer heat waves, we restricted our analyses to the warm season (1 November to 31 March) for 2001 to 2013. The effects are presented as odds ratio with their corresponding 95% confidence intervals. The figure is presented on the log scale. A p-value of <0.05 was considered statistically significant.
5.3 Results
Descriptive statistics for selected weather and ambient air pollutant variables are presented in Table 5.1. Table 5.2 presents the descriptive statistics for EHAs for the three selected respiratory diseases for all ages combined and specific age groups. The total count of EHAs during the study period for the three diseases was similar, with 37,931 for pneumonia, 35,811 for COPD and 35,374 for asthma. Differences were observed in the total number of counts for specific age groups. The 0-14 years age group had the highest total count for asthma with
24,719, and the 75 years and above age group had the highest total count for
COPD and pneumonia with 16,713 and 14,955 respectively. Table 5.3 summarises the heat wave characteristics for each heat wave definition used in the study. HWD03 had the highest total number of heat wave days during the study
113 period, and the longest average heat wave duration of 2.92 days. HWD02 had the most number of total heat wave events with 43.
Figure 5.1 shows the association between heat waves days and EHAs for three respiratory diseases with, and without, adjusting for daily average ozone at lag0. For all three diseases, and across the three heat wave definitions, controlling for daily average ozone generally had a small, but sometimes negligible, effect on the health risk estimates. After adjusting for daily average ozone, negative, statistically significant associations were found between heat waves days and
EHAs for asthma across the three heat wave definitions. Negative associations were also found between heat wave days and EHAs for COPD for HWD01 and
HWD02, with the association found to be statistically significant for HWD01. A small, positive association was found for HWD03 for EHAs for COPD, although this was not statistically significant. Small, negative associations were also found between heat wave days and EHAs for pneumonia across the three heat wave definitions, but these were not statistically significant.
Table 5.4 shows the association between heat wave days and EHAs for three respiratory diseases at two levels of ozone (high: ≥ 95th percentile; low <
th 95 percentile) for all ages at lag0 and lag1. The results for lag2 are presented in
Table A10 in the Appendix. Across all three heat wave definitions and lags, negative associations were found on both high and low-level ozone days for
EHAs for asthma, with some statistically significant associations found. Stronger heat wave effects on high-level ozone days were found on EHAs for COPD at lag1 and lag2 for HWD03. However, for HWD01 and HWD02, negative associations were generally found on both high and low-level ozone days for EHAs for COPD.
Stronger heat wave effects on high-level ozone days on EHAs for pneumonia
114 were found across most definitions and lags, with a positive, statistically significant interaction found for HWD02 at lag2.
Table 5.5 shows the association between heat wave days and EHAs for three respiratory diseases at two levels of ozone (high: ≥ 95th percentile; low <
th 95 percentile) for specific age groups at lag0 and lag1. The results for lag2 are presented in Table A10 in the Appendix. A positive, statistically significant interaction between heat wave and high-level ozone days on EHAs for COPD was found in the 0-64 years age group for HWD03 at lag0. Stronger heat wave effects on high-level ozone days were also found in this age group at lag0 for HWD02 and lag1 for HWD03, although the interaction term was not statistically significant. A positive, statistically significant interaction was also found between heat wave and high-level ozone days on EHAs for pneumonia in the 0-14 years age group at lag1 for HWD02, the 15-64 years age group at lag2 for HWD01, and the 75 years and above age group at lag2 for HWD02. Stronger heat wave effects on high-level ozone days were generally found across most heat wave definitions and lags in the 15-64 and 65-74 years age groups respectively, with one noted statistically significant interaction. Negative associations between heat waves and
EHAs for asthma were found in the 0-14 years and 15-64 years age groups across both high and low-level ozone days for most heat wave definitions and lags, with some associations found to be statistically significant in the 0-14 years age group.
Stronger heat wave effects were found on low-level ozone days in 75 years and above age group at lag0 for HWD01 and HWD02, and lag1 for HWD03, but the interaction terms were not statistically significant. A negative, statistically significant interaction between heat wave and high-level ozone days on EHAs for
115 COPD in the 75 years and above age group at lag0 for HWD01 was found, and on
EHAs for pneumonia for the same age group at lag1 for HWD03.
116 Table 5.1 Descriptive statistics for environmental variables in the SSD during the warm season, 2001 to 2013.
Environmental Variables Mean(SD) Maximum Minimum Value Value Value Weather (Degrees Celsius (°C)) Daily average maximum temperature 26.40(4.38) 43.99 14.41 Daily average mean temperature 21.34(3.05) 32.38 12.47 Daily average minimum temperature 16.27(2.70) 24.39 6.82 Daily average dew-point temperature 14.92(3.34) 22.10 -0.13
Ambient Air pollution Daily average ozone (pphm) 3.78(1.48) 11.52 1.04 3 Daily average PM10 (µg/m ) 20.43(11.48) 222.30 4.57 Daily average nitrogen dioxide (pphm) 1.44(0.59) 4.56 0.29
117 Table 5.2 Descriptive statistics for EHAs for three respiratory diseases in the SSD during the warm season, 2001 to 2013.
ICD Code Median(IQR) Maximum Minimum Total Count (ICD-10-AM) Daily Value Daily Value Daily Value Respiratory Disease Asthma J45-J46 All ages 35374 17(12-25) 66 1 0-14 24719 12(8-17) 59 0 15-64 8371 5(3-7) 22 0 65 years and over 2284 1(0-2) 6 0 COPD J40-J44, J47 All ages 35811 19(16-23) 47 6 0-64 8585 5(3-6) 17 0 65-74 10513 6(4-7) 19 0 75 years and above 16713 9(7-11) 23 1 Pneumonia J12-J18 All ages 37931 21(17-24) 44 6 0-14 6748 3(2-5) 16 0 15-64 11008 6(4-8) 16 0 65-74 5220 3(2-4) 12 0 75 years and over 14955 8(6-10) 22 0
118 Table 5.3 Summary of heat wave characteristics for each heat wave definition used.
Average Average Total number Total number Heat wave intensitya of duration of of heat wave of heat wave definition heat wave day heat wave days events (°C) (in days) HWD01 98 38 35.19 2.58 HWD02 113 43 27.31 2.63 HWD03 114 39 20.75 2.92 a The average intensity was calculated using the temperature metric used in each heat wave definition.
119
Figure 5.1 The association between heat wave days and EHAs for three respiratory diseases with, and without, controlling for daily average ozone at lag0 in the SSD during the warm season, 2001 to 2013.
120 Table 5.4 The effect of heat wave days on EHAs for three respiratory diseases on days with high levels of ozone compared to days with low levels of ozone in the SSD during the warm season, 2001 to 2013, for all ages. Effects are presented as odds ratios with their corresponding 95% confidence intervals.
HWD01 HWD02 HWD03 Lag0 Lag1 Lag0 Lag1 Lag0 Lag1 Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Low Ozone High Ozone Low Ozone High Ozone Low Ozone High Ozone Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Asthma 0.87 0.82 0.94 0.99 0.88 0.77 0.90 0.94 0.91 0.85 0.92 0.89 (0.80, 0.94) (0.73, 0.92) (0.87, 1.02) (0.88, 1.12) (0.82, 0.95) (0.69, 0.87) (0.84, 0.96) (0.84, 1.06) (0.86, 0.97) (0.75, 0.97) (0.87, 0.97) (0.71, 0.93) COPD 0.98 0.93 1.00 0.98 0.99 0.97 0.95 0.96 1.02 1.02 1.01 1.04 (0.91, 1.06) (0.83, 1.03) (0.93, 1.07) (0.88, 1.09) (0.92, 1.06) (0.87, 1.08) (0.89, 1.02) (0.86, 1.07) (0.96, 1.08) (0.90, 1.15) (0.95, 1.07) (0.92, 1.18) Pneumonia 1.01 1.06 1.00 1.05 1.01 1.07 0.94 0.98 1.01 1.07 0.95 0.83 (0.94, 1.09) (0.96, 1.18) (0.93, 1.07) (0.94, 1.17) (0.95, 1.07) (0.96, 1.19) (0.88, 1.00) (0.88, 1.09) (0.95, 1.07) (0.95, 1.21) (0.64, 1.41) (0.46, 1.51) *Denotes a statistically significant positive interaction term p-value (<0.05) ** Denotes a statistically significant negative interaction term p-value (<0.05)
121 Table 5.5 The effect of heat wave days on EHAs for three respiratory diseases on days with high levels of ozone compared to days with low levels of ozone in the SSD during the warm season, 2001 to 2013, for specific age groups. Effects are presented as odds ratios with their corresponding 95% confidence intervals.
HWD01 HWD02 HWD03 Lag0 Lag1 Lag0 Lag1 Lag0 Lag1 Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Low Ozone High Ozone Low Ozone High Ozone Low Ozone High Ozone Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Asthma 0.83 0.80 0.91 0.99 0.87 0.77 0.85 1.00 0.90 0.84 0.89 0.99 0-14 Years (0.75, 0.91) (0.69, 0.93) (0.83, 0.99) (0.86, 1.15) (0.79, 0.95) (0.67, 0.90) (0.78, 0.93) (0.86, 1.16) (0.84, 0.97) (0.72, 0.99) (0.83, 0.95) (0.84, 1.15) 0.94 0.86 1.06 1.04 0.88 0.83 1.06 0.88 0.90 0.94 0.97 0.84 15-64 Years (0.80, 1.11) (0.68, 1.08) (0.91, 1.25) (0.84, 1.30) (0.75, 1.03) (0.66, 1.04) (0.91, 1.23) (0.70, 1.10) (0.79, 1.03) (0.71, 1.23) (0.85, 1.11) (0.64, 1.10) 65 Years and 1.11 1.07 0.98 1.04 1.14 0.87 0.90 1.02 1.04 1.07 1.09 0.52 Over (0.84, 1.47) (0.71, 1.60) (0.75, 1.29) (0.69, 1.59) (0.88, 1.48) (0.58, 1.30) (0.69, 1.17) (0.68, 1.55) (0.82, 1.33) (0.66, 1.72) (0.87, 1.36) (0.30, 0.91) COPD 0.93 1.00 1.05 0.98 0.97 1.20 0.94 0.95 1.00 1.54* 0.93 1.17 0-64 Years (0.79, 1.08) (0.81, 1.24) (0.91, 1.22) (0.79, 1.22) (0.84, 1.12) (0.96, 1.49) (0.82, 1.08) (0.76, 1.18) (0.88, 1.13) (1.20, 1.97) (0.82, 1.05) (0.91, 1.49) 0.93 0.97 0.99 0.97 0.96 0.90 0.91 0.99 0.97 0.92 1.06 0.96 65-74 Years (0.81, 1.06) (0.80, 1.18) (0.86, 1.13) (0.80, 1.19) (0.85, 1.09) (0.74, 1.10) (0.80, 1.04) (0.75, 1.30) (0.86, 1.09) (0.72, 1.16) (0.95, 1.18) (0.76, 1.22) 75 Years and 1.08 0.88** 0.98 0.97 1.04 0.95 0.99 0.94 1.03 0.92 1.00 0.99 Over (0.97, 1.20) (0.75, 1.03) (0.88, 1.09) (0.83, 1.13) (0.94, 1.14) (0.81, 1.11) (0.90, 1.10) (0.81, 1.11) (0.94, 1.13) (0.76, 1.11) (0.92, 1.09) (0.82, 1.19) Pneumonia 1.02 1.00 0.89 1.04 0.97 0.98 0.77 1.14 * 0.97 0.97 0.96 0.97 0-14 Years (0.85, 1.21) (0.77, 1.30) (0.75, 1.06) (0.80, 1.35) (0.84, 1.12) (0.75, 1.29) (0.65, 0.91) (0.87, 1.49) (0.84, 1.11) (0.73, 1.29) (0.91, 1.02) (0.85, 1.10) 1.02 1.14 1.00 1.02 1.01 1.15 0.97 0.96 1.06 1.21 0.99 1.12 15-64 Years (0.89, 1.17) (0.94, 1.39) (0.87, 1.14) (0.84, 1.24) (0.91, 1.13) (0.94, 1.41) (0.86, 1.10) (0.79, 1.17) (0.96, 1.18) (0.97, 1.51) (0.86, 1.13) (0.83, 1.50) 1.08 1.20 1.18 0.94 1.08 1.25 0.95 0.83 0.97 1.16 1.01 1.06 64-75 Years (0.90, 1.31) (0.90, 1.60) (0.97, 1.43) (0.70, 1.26) (0.92, 1.26) (0.92, 1.69) (0.64, 1.41) (0.46, 1.50) (0.83, 1.13) (0.83, 1.60) (0.91, 1.13) (0.84, 1.32) 75 Years and 1.02 1.01 1.01 1.12 1.02 1.02 0.95 0.83 0.99 1.00 0.98 0.62** Over (0.91, 1.14) (0.85, 1.20) (0.90, 1.13) (0.94, 1.33) (0.93, 1.12) (0.86, 1.22) (0.64, 1.41) (0.46, 1.50) (0.90, 1.08) (0.82, 1.21) (0.84, 1.15) (0.42, 0.90) *Denotes a statistically significant positive interaction term p-value (<0.05) ** Denotes a statistically significant negative interaction term p-value (<0.05)
122 5.4 Discussion
This study examined whether ozone modifies the association between heat waves and EHAs for three respiratory diseases in Greater Sydney, Australia. We estimated and compared the effect of heat waves on high and low-level ozone days at lag0-lag2 for all ages and specific age groups, and tested the sensitivity of three heat wave definitions. We found some evidence that ozone modifies the association between heat waves and EHAs for specific respiratory diseases.
Positive, statistically significant interactions were found between heat wave and high-level ozone days on EHAs for COPD (0-64 years) and for pneumonia (all ages, 0-14 years, 15-64 years) at certain lags and for some heat wave definitions.
Stronger heat wave effects were also generally observed on high compared to low-level ozone days for EHAs for pneumonia across most definitions and lags in the all ages, 15-64 years and 64-75 years age groups.
Previous time series studies investigating the potential interactive effects of temperature or heat waves and ozone on human health have examined total or cause-specific mortality as the health outcome (e.g. Ren et al. 2007; Pattenden et al. 2010; Burkart et al. 2013; Analitis et al. 2014; Breitner et al. 2014). As such, it is difficult to make direct comparisons between our findings and the results of these studies. This previous work however, has generally found some evidence of effect modification, although regional differences have been observed in multi- city studies, and such evidence has not been consistently statistically significant.
For example, Analitis et al. (2014) observed stronger heat wave effects at high compared to low-levels of ozone on total and cardiovascular mortality for nine
European cities, although no evidence of a significant, statistical interaction for these pooled effects was found. Further, effect modification was generally found
123 to be more evident and pronounced in the North-Continental than Mediterranean cities (Analitis et al. 2014). Breitner et al. (2014) found that ozone modified the pooled association between high temperature and non-accidental mortality for three German cities, with significantly stronger heat effects observed at high compared to moderate-levels of ozone. One recent cohort study investigated whether air pollution modified the association between temperature and COPD morbidity (McCormack et al. 2016). Positive, statistically significant interactions were found between increasing indoor temperature and indoor concentrations of
PM2.5 and NO2 on COPD symptoms such as breathlessness and cough, and rescue inhaler use (McCormack et al. 2016). No evidence of an interaction was found between outdoor temperature and outdoor ambient air pollutants including PM2.5,
NO2 and ozone (McCormack et al. 2016). This later finding is somewhat consistent with ours, as we also found little evidence of an interactive effect between outdoor heat wave and ozone exposure on EHAs for COPD among older populations, although we did find some evidence among the younger population.
It is plausible that air pollutants and heat may interact on a biological level, although the exact casual pathways and mechanisms involved are not clear.
The activation of the body’s thermoregulatory system and mechanisms during heat stress can facilitate and increase the absorption and entry of toxins and air pollutants into the body, as well as alter the body’s response to such substances
(Gordon 2003). The strength of the toxicity of a chemical or toxin on a biological system can be exacerbated by increased body temperature (Gordon et al. 1988;
Gordon 2003). A few studies have examined the combined effects of exposure to ozone and high temperature on lung function in the context of exercise and athlete performance, with inconsistent findings (Folinsbee et al. 1977; Gibbons and
124 Adams 1984; Couto-Gomes et al. 2010). For example, Folinsbee et al. (1977) observed a general trend for a greater reduction in pulmonary function when exposure to ozone and high temperature was combined, while Couto-Gomes et al.
(2010) found little evidence that ozone further exacerbates the adverse effects of heat on athlete performance and lung function. However, a more recent study that examined the effect of ozone on lung function and vascular markers of coagulation and fibrinolysis at moderate (22°C) and high (32.5°C) temperature found evidence of a biological interaction between high temperature and ozone on markers of fibrinolysis, but not on lung function (Kahle et al. 2015). Nevertheless, independent exposure to ambient ozone (8hr metric) among COPD patients has been shown to reduce lung function (Li et al. 2018), as well as increase pulmonary airway inflammation and oxidative stress in former smokers with and without
COPD (Pirozzi et al. 2015). The independent biological effects of heat exposure on patients with COPD are less clear (Hansel et al. 2015), as are the independent biological effects of both heat and ambient ozone exposure on individuals with pneumonia. Although, pulmonary ventilation is known to increase in humans in response to heat stress, which is in turn accompanied by increases in alveolar ventilation, leading to changes in blood gas levels (Tsuji et al. 2016).
Stronger heat wave effects on high compared to low-level ozone days on
EHAs for COPD and pneumonia were generally observed among the younger age groups (e.g. COPD: 0-64 years; pneumonia: 0-14, 15-64, 65-74 years). This finding is in general disagreement with previous work examining the interactive effects of heat waves and ozone on mortality, which found effect modification to be more pronounced among the oldest age groups (i.e. 75-84 years and 85 years and above) (Analitis et al. 2014). The elderly are known to be particularly
125 susceptible to heat exposure due a decline in their ability to thermoregulate effectively (Kenney and Munce 2003), as well as to ozone exposure given the general decline in lung function with age, and higher prevalence of chronic respiratory diseases, such as COPD, among older populations (Simoni et al.
2015). Potential inaccuracies in exposure measurement may be a reason for finding little evidence of effect modification in older populations in our study.
Older patients with COPD (average age of 69 years) have been found to spend significantly more time in their homes and have air conditioning, than persons without COPD (Leech and Smith-Doiron 2006). Ambient temperature and air pollution observations from outdoor monitoring stations across the city are unlikely to capture an individual’s personal exposure level in their homes, particularly if they are using air conditioning. As noted above, McCormack et al.
(2016) found a significant interaction between indoor, but not outdoor temperature, and indoor air pollution levels on COPD morbidity. Another possible reason for our findings might be due to differences in the sample sizes across the age groups.
We observed positive, statistically significant interactions between heat wave and high-level ozone days at several lags across EHAs for COPD and pneumonia. Previous work examining the potential interactive effects of high temperature and ozone on mortality has also found some evidence of effect modification at a short lag (Ren et al. 2007; Burkart et al. 2013). For example,
Ren et al. (2007) found stronger associations between temperature and cardiovascular mortality at higher levels of ozone on the day of exposure, as well as at a lag of one day, in several regions in the United States. A short lag effect has also been observed when examining the independent effects of high
126 temperature and ozone on respiratory morbidity (Ji et al. 2011; Anderson et al.
2013). Positive, statistically significant interactions were found for some heat wave definitions only. The choice of heat wave definition has been shown to affect both the magnitude and statistical significance of the association between heat waves and health outcomes (Tong et al. 2010b; Kent et al. 2014). Each of the three heat wave definitions used in this study identified different days as
‘exposure’ days, and the total number of exposure days varied between our definitions (See, Table 5.3). It is likely that is affected our models, as well as the calculation of the interaction term between heat wave and high-level ozone days.
It is also possible that different temperature metrics (maximum, mean, minimum) may have different impacts on respiratory health outcomes, although the reasons for the differences in the findings of their interaction with ozone on COPD and pneumonia are not known. For example, we observed a positive, statistically significant interaction between heat wave and high-level ozone days on EHAs for
COPD (0-64 years) for HWD03 only. The disrupted sleep patterns associated with exposure to high minimum temperatures (Okamoto-Mizuno et al. 2010; Lan et al.
2014), along with the need for the body to thermoregulate during the night, means that individuals are unable to naturally rest and recover from the day. It is possible that this may place the respiratory systems of those with COPD under further stress, leading to an exacerbation of this condition.
We found negative associations between heat waves and EHAs for asthma in the all ages and 0-14 years age across all lags and definitions on both high and low-level ozone days, with some statistically significant associations found.
Previous studies conducted in Greater Sydney and surrounding regions have also found negative associations between extreme heat and EHAs with a principal
127 diagnosis of asthma (Khalaj et al. 2010; Vaneckova and Bambrick 2013; Wilson et al. 2013). Studies conducted elsewhere however, have reported positive associations between elevated temperature or heat waves and hospital admission for asthma (Lin et al. 2009; Isaksen et al. 2015; Soneja et al. 2016). Adverse, biological effects of exposure to heat and ozone among those individuals with asthma have also been observed (Kreit et al. 1989; Jörres et al. 1996; Hayes et al.
2013). Our findings might therefore be a result of how asthma exacerbations are managed or treated in the Greater Sydney region, or due to behavioural changes among young asthmatics during heat waves and air pollution episodes that reduce their exposure.
This study has several potential strengths. By examining and comparing specific respiratory health outcomes in various age groups, we have shown that some conditions may be more susceptible to the potential interactive effects of heat waves and ozone than others. We also tested the sensitivity of three heat wave definitions, showing that evidence of effect modification may also depend on the heat wave definition used to define heat exposure. We also analysed a relatively long period of time series data, and controlled for the potential confounding effects of other ambient air pollutants including particular matter
(PM10) and nitrogen dioxide.
This study also has some potential limitations. The analysis was performed for a single city, and therefore our results may not be generalisable given that ozone concentrations can vary geographically, as well as population acclimatisation to heat waves. As mentioned previously, we also estimated exposure to heat waves and ozone by calculating the daily city-wide average using monitoring stations, and not by measuring an individual’s personal exposure,
128 which may have resulted in some exposure misclassification. We also did not account for any potential spatial variation in the intensity of heat waves across
Greater Sydney, where daily maximum temperatures in the western areas of the city can be higher (Bureau of Meteorology 2018e).
5.5 Chapter Conclusion
This Chapter found some evidence that ozone modifies that association between heat waves and hospital admissions for certain respiratory diseases. The findings of this study showed inconsistencies and largely differed across disease, age group, lag and heat wave definition. In light of the differences found across the three diseases, this study highlights the need for future studies to consider, where possible, cause-specific outcomes when examining the potential interactive effects of heat waves and air ozone on morbidity. As heat waves and ozone concentrations are projected to increase under climate change, it is important to consider ozone as a potential effect modifier of the association between heat waves and hospital admissions in future work, where appropriate. As this Chapter has shown, this is true even for locations with comparatively low levels of ozone, such as Australia.
129 Chapter Six: Aim 3
Aim 3: To examine whether particulate matter modifies the short-
term association between heat waves and hospital admissions for
certain cardiovascular diseases in Greater Sydney, Australia.
6.1 Introduction
Cardiovascular disease is a major cause of death both worldwide and in
Australia (Nichols et al. 2014; World Health Organisation 2017). Some studies have shown that high temperatures and heat waves are associated with increased risk of hospitalisation for total cardiovascular diseases (Lin et al. 2009; Ostro et al. 2010; Ma et al. 2011), and specific cardiovascular diseases including ischemic heart disease and cardiac (or heart) dysrhythmias (Nitschke et al. 2007; Lin et al.
2009; Ostro et al. 2010). Elevated temperature and heat waves have also been shown to be associated with an increased risk of out-of-hospital cardiac arrest
(Kang et al. 2016). A short lag effect has been observed, with positive associations between high temperatures and hospitalisations for cardiovascular diseases reported on the same day of exposure (Ostro et al. 2010) and between 1-3 days after exposure (Lin et al. 2009). Other studies, however, including two meta- analyses, have reported null or negative associations between high temperatures and hospital admissions for cardiovascular diseases (Michelozzi et al. 2009;
Turner et al. 2012; Phung et al. 2016; Ogbomo et al. 2017); but Phung et al.
(2016)’s review reported a small, positive heat wave effect.
Ambient particulate matter with an aerodynamic diameter less than 10μm, known as PM10, is comprised of both solid particles and liquid droplets from
130 natural and anthropogenic sources (US EPA 2016). Levels and mixtures of PM10 can depend on season and temperature, with bushfire smoke and dust storms important sources during the warm season in Australia, and wood heaters an important source in the cool season (Keywood et al. 2016). Studies have shown that elevated levels of PM10 are associated with an increased risk of hospitalisation for all cardiovascular or cardiac diseases (Barnett et al. 2006; Colais et al. 2012;
Stafoggia et al. 2013), and specific diseases including ischemic heart disease
(Schwartz and Morris 1995; Xu et al. 2017), heart failure (Shah et al. 2013) and heart arrhythmias and conduction disorders (Colais et al. 2012), particularly among the elderly. Elevated levels of PM10 have also been shown to be associated with an increased risk of out-of-hospital cardiac arrest (Zhao et al. 2017). A few studies have assessed, or controlled for, the potential confounding effects of PM10 when estimating the association between extreme heat and hospitalisations for cardiovascular diseases (e.g. Vaneckova and Bambrick 2013; Wilson et al. 2013).
Little is known about the potential joint or interactive effects of high temperatures, particularly heat waves, and PM10 on cardiovascular heath outcomes. This is concerning given that the joint effect of weather and air pollution on health outcomes is thought to be greater than the risk derived from the individual impacts of these two exposures (Zanobetti and Peters 2015). There is also some suggestion that an interactive effect between air pollution and temperature may be biologically plausible (Gordon 2003). Some studies from
Europe and Asia have investigated whether temperature modifies the association between PM10 and all-cause and/or cardiovascular mortality (Qian et al. 2008;
Stafoggia et al. 2008; Li et al. 2011; Cheng and Kan 2012; Meng et al. 2012;
Burkart et al. 2013). Most of these studies have generally found stronger
131 associations at high compared to moderate or low level temperatures, although such evidence of effect modification has not been consistently statistically significant. However, Cheng and Kan (2012) found a statistically significant interaction between low, but not high, temperature and PM10 on total and cardiovascular mortality in Shanghai, China.
Few studies have investigated whether PM10 modifies the association between high temperatures, particularly heat waves, and cardiovascular health outcomes. Some have found stronger associations between high temperatures or heat waves and all-cause and/or cardiovascular mortality at higher levels of PM10, although not all have reported evidence of statistical significance (Burkart et al.
2013; Analitis et al. 2014; Breitner et al. 2014; Li et al. 2015). Other studies have found no evidence of an interaction between temperature and PM10 on mortality
(Hales et al. 2000; Basu et al. 2008). Little work, however, has examined whether
PM10 modifies the association between temperature or heat waves and cardiovascular morbidity, particularly cause-specific cardiovascular morbidity.
One Australian study found that PM10 modified the association between temperature and cardiovascular hospital admissions at different lags in Brisbane, but found little evidence of effect modification for cardiovascular emergency presentations (Ren et al. 2006). Further, a recent Korean study found no evidence of a significant interactive effect between heat waves and PM10 on out-of-hospital cardiac arrest (Kang et al. 2016).
The frequency, intensity and duration of heat waves is expected to increase in the future under climate change across most land areas globally, including
Australia (IPCC 2012; Cowan et al. 2014). It is therefore important to clarify and enhance our understanding of the association between heat waves and
132 cardiovascular morbidity to inform climate change adaptation planning in the health sector. This study aimed to examine whether PM10 modifies the short-term association between heat waves and hospital admissions for specific cardiovascular diseases in Greater Sydney, Australia. We investigated the susceptibility of both younger (0-64 years) and older populations (65 years and above), and tested the sensitivity of three heat wave definitions.
6.2 Data and Methods
6.2.1 Meteorological Data
Daily weather data for all stations located in the SSD with near complete coverage of the period of 2001 to 2013 were obtained from the Australian
Government’s Bureau of Meteorology (n=17). Before identifying extreme temperature events in a climate time series, such as summer heat waves, it is important that the data undergo quality control checks (Alexander and Tebaldi
2012). This is because it is possible for incorrect data entries to be considered as real ‘extreme’ values and included in further analyses (Alexander and Tebaldi
2012). To ensure our observational weather data was of the highest possible quality, we performed a series of quality control checks on the observed daily maximum, minimum and dew-point temperature values for each weather station, and also tested for inhomogeneities in each daily maximum and minimum time series to inspect their overall quality (See Chapter Three, for an overview of this process). High quality stations (n=15) were then used to calculate the respective city-wide averages for each temperature metric if they had a total missing value count of ≤2.5% of the study period. The missing value threshold was set at ≤2.5% to maximise the number of stations included the calculation of the average and
133 subsequent spatial coverage of the SSD, while also ensuring that the quality of those stations included remained high. The daily average mean temperature was calculated as the mean of the city-wide daily average maximum and minimum temperature values. For dew-point temperature, as the observations were recorded at 3-hour intervals over a 24-hour period, the city-wide average value for each time interval was first calculated with those stations where the missing value count was ≤2.5% of the study period, then the overall 24-hour daily average was calculated from these averaged time interval values.
In the absence of a standard heat wave definition, we selected and compared three heat wave definitions for this study. Previous studies have shown that the choice of heat wave definition can alter the magnitude and statistical significance of the association between heat waves and adverse health outcomes
(Tong et al. 2010; Kent et al. 2014). We defined a heat wave as two or more consecutive days where the temperature metric (three temperature metrics were selected and compared: maximum temperature (HWD01), mean temperature
(HWD02) and minimum temperature (HWD03)) is greater than or equal to the
90th percentile of the warm season (1 November to 31 March) during 2001 to
2013. We compared heat wave definitions with alternative temperature metrics, rather than temperature thresholds or durations, to ensure we kept an adequate number of heat wave days to conduct the analysis.
6.2.2 Ambient Air Pollution Data
Daily ambient air pollution data for all stations located in the SSD were obtained from the NSW Office of Environment and Heritage for 2001 to 2013.
Daily data were obtained for the following air pollutants and used in this study:
134 ozone (1hr average 24hr maximum value (pphm)); nitrogen dioxide (1hr average
24hr maximum value (pphm)) and particulate matter (particles with an aerodynamic diameter of less than 10μm, PM10) (1hr average 24hr average value).
The NSW Office of Environment and Heritage follows several quality assurance procedures to ensure the data are precise, accurate, representative and comparable
(NSW Office of Environment and Heritage 2015b). Negative daily values were assigned a value of 0, as advised by the NSW OEH (pers. comm., 16 September,
2015). Stations that had a missing value count of ≤5% of the study period were used to calculate the daily city-wide average for each pollutant. Junger and Ponce de Leon (2015) regarded a missing data level of 5% as the best case scenario in their application of time-series air pollution data. Similar to the threshold selection for our meteorological data, a threshold of 5% was optimal in allowing us to maximise the number of stations included the calculation of the average and subsequent spatial coverage of the SSD, while also ensuring that the quality of those stations included remained high.
6.2.3 Health Data
Individual-level daily hospital admission records with a principal diagnosis of I00-I99 (ICD-10-AM) for all public and private hospitals located in the SSD were obtained from the NSW Ministry of Health, Admitted Patient Data
Collection, for 2001 to 2013 as part of a larger dataset (n=1570805). All exact duplicate records were extracted and removed (n=1570741, 64 records removed), as well those records with an admission date outside of 1 July 2001 – 30 June
2013 (n=1499661, 71080 records removed). Records that were classified as
‘emergency’ hospital admissions (EHAs) were then selected for analysis to
135 eliminate ‘pre-planned’ hospital admissions (n=1132737, records removed
366924) (Khalaj et al. 2010). We then extracted and removed remaining records with an implausible, unknown or missing entry for age (ranged deemed plausible:
0 – 115 years) or sex (required entry: male or female) (n=1132705, records removed 32). Those records with a principal diagnosis of ischemic heart disease
(ICD-10-AM: I20-I25), heart failure (ICD-10-AM: I50), cardiac arrest (ICD-10-
AM: I46), heart arrhythmia (ICD-10-AM: 147-I49), conduction disorders (ICD-
10-AM: I44-I45) and hypertensive diseases (ICD-10-AM: I10-I15) were then selected and aggregated into daily counts. To investigate the susceptibility of both younger and older populations, we stratified the data into two age groups: 0-64 years and 65 years and above.
6.2.4 Study Design and Statistical Analysis
We used a time-stratified case-crossover study design (Maclure et al.
1991; Janes et al. 2005). This design has been used in previous studies to estimate the association between heat waves and hospital admissions (Zhang et al. 2013;
Gronlund et al. 2014), and has been shown to produce similar results to the alternate time-series design (Tong et al. 2012). The design is equivalent to a matched pair case-control design: it compares a case’s exposure on the day of an adverse health event (e.g. hospital admission) to their exposure on control days
(or referent times) that are selected before and/or after the event (Janes et al. 2005;
Bell et al. 2008; Barnett and Dobson 2010). Since each case acts as their own control, personal characteristics such as sex and smoking status are controlled for by matching (Barnett and Dobson 2010). We used the time-stratified approach to select control days to avoid potential bias introduced by other approaches, such as
136 the unidirectional and bidirectional designs (Janes et al. 2005). We matched cases and controls on day of the week and within the same month, and thus controlled for the confounding effects of season and long-term trends by design.
We used conditional logistic regression to estimate the association between heat waves and EHAs for our six selected cardiovascular diseases. We first estimated the association with, and without, adjusting for daily average PM10 at lag0, while also adjusting for daily average dew-point temperature (Davis et al.
2016) using a natural cubic spline (df=3, knots at quantiles), daily average nitrogen dioxide (1hr average 24h maximum value (pphm)), daily average ozone
(1hr maximum 24hr average value (pphm)) and public holidays. To determine the most appropriate way to model dew-point temperature, we conducted sensitivity tests modelling this variable as a natural cubic spline with 3 and 2 degrees of freedom, and as a linear variable at lag0. As the coefficients of the heat wave effect were largely similar across the three modelling approaches, we selected to model dew-point temperature as a natural cubic spline with 3 degrees of freedom to be consistent with previous work (Kingsley et al. 2016).
To examine whether PM10 modifies the association between heat waves and EHAs for our six selected cardiovascular diseases, we estimated and compared heat wave effects on days with high and low levels of PM10 at lag0-lag2.
High and low level PM10 days were defined as those where the daily average
th th PM10 value was ≥90 and <90 percentile of the warm season during 2001 to
2013 respectively (Note: 90th percentile of the distribution was equal to
30.52µg/m3; see Table A11 for further descriptive statistics, Appendix). We created an interaction term between high and low level PM10 days (1=high,
0=low) and heat wave days (1=yes, 0=no). This term was added to the model,
137 along with the respective individual variables and potential confounding variables described in the previous paragraph. We selected the threshold of the 90th percentile for two main reasons: to ensure there was a reasonably equal distribution of high and low level PM10 days across heat wave days for the three definitions for a fair comparison; and to compare and estimate heat wave effects on days with the more extreme values of PM10.
The statistical analyses were conducted in the ‘R’ Statistical Computing
Environment (Version 3.2.1) using the ‘season’ and ‘dlnm’ packages. As we wanted to examine the impact of summer heat waves, we restricted our analyses to the warm season (1 November to 31 March) for 2001 to 2013. The effects are presented as odds ratio with their corresponding 95% confidence intervals. The figure is presented on the log scale. A p-value of <0.05 was considered significant.
6.3 Results
Descriptive statistics for selected weather and ambient air pollution variables during the study period are presented in Table 6.1. The mean daily average maximum temperature was 26.0°C, and the mean daily average value of
3 PM10 was 20.43 µg/m . Table 6.2 shows descriptive statistics for selected EHAs for six cardiovascular diseases for all ages combined and two age groups: 0-64 years and 65 years and over. Ischemic heart disease had the highest number of total EHAs during the study period with 68,334, while cardiac arrest had the lowest with 1,861. For each cardiovascular disease, the older age group had a higher number of EHAs than the younger age group. A summary of the heat wave characteristics for each heat wave definition used is provided in Table 6.3.
138 HWD03 had the highest total number of heat wave days during the study period, and the longest average heat wave duration of 2.92 days. HWD02 had the highest number of total heat wave events with 43.
Figure 6.1 shows the association between heat wave days and EHAs for six cardiovascular diseases with, and without, controlling for daily average PM10 at lag0 for all ages. For all diseases, and across the three heat wave definitions, controlling for daily average PM10 had little effect on the health risk estimates.
Negative associations were found between heat wave days and EHAs for heart arrhythmia and hypertensive diseases for all three heat wave definitions, although these associations were not statistically significant. Negative associations were also found between heat wave days and EHAs for ischemic heart disease, heart failure and conduction disorders for HWD01 and HWD02, and small positive associations were found for HWD03. The negative associations found for EHAs for ischemic heart disease for HWD01 and HWD02 were statistically significant.
Small, positive associations were found between heat wave days and EHAs for cardiac arrest for HWD01 and HWD02, and negative associations were found for
HWD03.
Table 6.4 shows the association between heat wave days and EHAs for six
th cardiovascular diseases at two levels of PM10 (high: ≥90 percentile; and low:
th <90 percentile) for all ages at lag0 and lag1. The results for lag2 are presented in
Table A12 in the Appendix. A positive, statistically significant interaction was found between heat wave and high-level PM10 days on EHAs for hypertensive diseases at lag1 for HWD03. Heat wave effects were also stronger on high-level
PM10 days for hypertensive diseases for HWD03 at lag0 and lag2, but the p-value of the interaction term was not statistically significant. The impact of heat waves
139 on EHAs for cardiac arrest was generally found to be stronger on days with high levels of PM10 across most lags and definitions, although none of the interaction terms were statistically significant. A negative, statistically significant interaction was found between heat wave and high-level PM10 days on EHAs for ischemic heart disease at lag2 for HWD01, but not at lag0 or lag1.
Table 6.5 shows the association between heat wave days and EHAs for six
th cardiovascular diseases at two levels of PM10 (high: ≥90 percentile; and low:
th <90 percentile) for younger and older populations at lag0 and lag1. The results for lag2 are presented in Table A13 in the Appendix. A positive, statistically significant interaction was found between heat wave and high-level PM10 days on
EHAs for cardiac arrest in the older age group for HWD01 at lag1 and lag2, and for HWD02 at lag1. Heat waves effects were also found to be stronger on high- level PM10 days at lag0 for HWD02, and at lag0 and lag1 for HWD03 in the younger age group, but no evidence of a statistically significant interaction was found. The impact of heat waves on EHAs for conduction disorders was stronger on high-level PM10 days for all definitions and lags, and on EHAs for hypertensive diseases for HWD02 and HWD03 at all lags and lag1 for HWD01 in the younger population. Stronger heat wave effects on high compared to low-level
PM10 days were found for EHAs for heart failure at lag1 for HWD03 in the older age group. A negative, statistically significant interaction was found between heat wave and high-level PM10 days on EHAs for heart arrhythmia for HWD01 at lag1 in the younger age group.
140 Table 6.1 Descriptive statistics for environmental variables in the SSD during the warm season, 2001 to 2013.
Environmental Variables Mean(SD) Maximum Minimum Value Value Value Weather (Degrees Celsius (°C)) Daily average maximum temperature 26.40(4.38) 43.99 14.41 Daily average mean temperature 21.34(3.05) 32.38 12.47 Daily average minimum temperature 16.27(2.70) 24.39 6.82 Daily average dew-point temperature 14.92(3.34) 22.10 -0.13
Ambient Air pollution Daily average ozone (pphm) 3.78(1.48) 11.52 1.04 3 Daily average PM10 (µg/m ) 20.43(11.48) 222.30 4.57 Daily average nitrogen dioxide (pphm) 1.44(0.59) 4.56 0.29
141 Table 6.2 Descriptive statistics for EHAs for six cardiovascular diseases in the SSD during the warm season, 2001 to 2013.
ICD Code Median(IQR) Maximum Minimum Total Count (ICD-10-AM) Daily Value Daily Value Daily Value Cardiovascular Disease Ischemic Heart Disease I20-I25 All ages 68334 37(31-43) 70 14 0-64 years 28497 15(12-19) 34 3 65 years and over 39837 22(18-26) 46 5 Heart Failure I50 All ages 24721 13(10-17) 31 2 0-64 years 3470 2(1-3) 9 0 65 years and over 21251 11(9-14) 28 0 Cardiac Arrest I46 All ages 1861 1(0-2) 6 0 0-64 years 802 0(0-1) 4 0 65 years and over 1059 0(0-1) 4 0 Heart Arrhythmia I47-I49 All ages 32682 18(14-21) 36 5 0-64 years 12461 7(5-9) 19 0 65 years and over 20221 11(8-14) 25 1 Conduction Disorders I44-I45 All ages 3070 1(1-3) 7 0 0-64 years 641 0(0-1) 4 0 65 years and over 2429 1(0-2) 7 0 Hypertensive Diseases I10-I15 All ages 3859 2(1-3) 9 0 0-64 years 1571 1(0-1) 6 0 65 years and over 2288 1(0-2) 7 0
142 Table 6.3 Summary of heat wave characteristics for each heat wave definition used.
Average Average Total number Total number Heat wave intensitya of duration of of heat wave of heat wave definition heat wave day heat wave days events (°C) (in days) HWD01 98 38 35.19 2.58 HWD02 113 43 27.31 2.63 HWD03 114 39 20.75 2.92 a The average intensity was calculated using the temperature metric used in each heat wave definition.
143
Figure 6.1 The association between heat wave days and EHAs for six cardiovascular diseases with, and without, controlling for daily average PM10 at lag0 in the SSD during the warm season, 2001 to 2013. Note: Cond. Disorders is conduction disorders; Hyper. Diseases is hypertensive diseases; Isch. Heart
Disease is Ischemic Heart Disease.
144 Table 6.4 The effect of heat wave days on EHAs for six cardiovascular diseases on days with high levels of PM10 compared to days with low levels of PM10 in the SSD during the warm season, 2001 to 2013, for all ages. Effects are presented as odds ratios with their corresponding 95% confidence intervals.
HWD01 HWD02 HWD03
Lag0 Lag1 Lag0 Lag1 Lag0 Lag1 Heat Effecta Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Cardiovascular Disease Ischemic Heart 0.92 0.92 0.97 0.94 0.98 0.92 0.98 0.92 1.03 0.95 1.04 1.01 Disease (0.87, 0.98) (0.86, 0.98) (0.92, 1.02) (0.87, 1.01) (0.93, 1.03) (0.86, 0.97) (0.94, 1.03) (0.85, 0.99) (0.98, 1.07) (0.87, 1.03) (1.00, 1.08) (0.93, 1.09) Heart Failure 0.94 1.00 0.83 0.94 0.99 0.97 0.89 0.94 1.01 1.01 0.95 0.93 (0.85, 1.04) (0.90, 1.11) (0.76, 0.90) (0.83, 1.06) (0.90, 1.08) (0.87, 1.08) (0.83, 0.95) (0.83, 1.06) (0.93, 1.09) (0.88, 1.16) (0.89, 1.01) (0.80, 1.07) Cardiac Arrest 1.06 0.99 1.05 1.30 0.88 1.22 1.08 1.41 0.93 1.13 1.21 1.24 (0.73, 1.55) (0.69, 1.41) (0.77, 1.41) (0.86, 1.97) (0.62, 1.23) (0.86, 1.74) (0.84, 1.40) (0.93, 2.14) (0.68, 1.26) (0.70, 1.83) (0.95, 1.55) (0.75, 2.07) Heart 0.95 1.01 0.99 0.94 0.98 1.02 0.99 0.97 1.00 0.96 1.06 0.96 Arrhythmia (0.86, 1.04) (0.92, 1.10) (0.92, 1.06) (0.84, 1.04) (0.92, 1.06) (0.93, 1.12) (0.93, 1.05) (0.87, 1.08) (0.94, 1.07) (0.85, 1.08) (1.00, 1.12) (0.85, 1.09) Conduction 0.84 1.04 0.91 0.92 0.90 0.96 0.94 0.87 1.11 0.89 0.89 0.86 Disorders (0.63, 1.12) (0.77, 1.41) (0.73, 1.13) (0.64, 1.33) (0.71, 1.14) (0.71, 1.30) (0.77, 1.13) (0.60, 1.25) (0.91, 1.37) (0.60, 1.31) (0.74, 1.07) (0.57, 1.30) Hypertensive 0.86 0.87 0.91 0.92 0.88 0.82 0.91 0.91 0.90 1.10 0.86 1.30* Diseases (0.65, 1.13) (0.66, 1.16) (0.83, 1.25) (0.65, 1.30) (0.70, 1.12) (0.61, 1.09) (0.76, 1.09) (0.64, 1.29) (0.74, 1.10) (0.75, 1.60) (0.73, 1.02) (0.90, 1.89) *Denotes a statistically significant positive interaction term p-value (<0.05) ** Denotes a statistically significant negative interaction term p-value (<0.05)
145 Table 6.5 The effect of heat wave days on EHAs for six cardiovascular diseases on days with high levels of PM10 compared to days with low levels of PM10 in the SSD during the warm season, 2001 to 2013, for those aged 0-64 years and 65 years and over. Effects are presented as odds ratios with their corresponding 95% confidence intervals.
HWD01 HWD02 HWD03
Lag0 Lag1 Lag0 Lag1 Lag0 Lag1 Heat Effecta Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Cardiovascular Disease Ischemic Heart Disease 0-64 years 0.97 0.94 1.03 1.00 1.02 0.92 1.00 0.94 1.07 1.01 1.08 1.09 (0.88, 1.07) (0.85, 1.03) (0.96, 1.11) (0.89, 1.11) (0.94, 1.11) (0.84, 1.02) (0.94, 1.07) (0.84, 1.05) (1.00, 1.15) (0.89, 1.14) (1.02, 1.15) (0.96, 1.23) 65 years and 0.89 0.90 0.93 0.89 0.95 0.91 0.97 0.90 0.99 0.90 1.00 0.95 over (0.82, 0.96) (0.83, 0.98) (0.87, 0.99) (0.81, 0.98) (0.88, 1.01) (0.84, 0.99) (0.92, 1.03) (0.82, 0.99) (0.93, 1.05) (0.81, 1.01) (0.95, 1.06) (0.85, 1.06) Heart Failure 0-64 years 1.05 0.88 0.74 0.90 1.17 0.91 0.85 0.93 1.22 1.32 1.03 0.97 (0.80, 1.38) (0.66, 1.17) (0.59, 0.92) (0.64, 1.27) (0.93, 1.47) (0.68, 1.21) (0.71, 1.02) (0.67, 1.31) (1.001, 1.49) (0.92, 1.89) (0.87, 1.23) (0.66, 1.43) 65 years and 0.92 1.03 0.84 0.96 0.96 0.98 0.89 0.94 0.97 0.97 0.93 0.92 over (0.83, 1.03) (0.92, 1.15) (0.77, 0.92) (0.83, 1.08) (0.87, 1.06) (0.88, 1.10) (0.83, 0.97) (0.82, 1.08) (0.90, 1.06) (0.83, 1.13) (0.87, 1.01) (0.78, 1.07) Cardiac Arrest 0-64 years 0.90 0.83 1.31 0.78 0.99 1.14 1.30 1.08 0.96 1.35 1.25 1.35 (0.52, 1.58) (0.47, 1.45) (0.84, 2.04) (0.40, 1.52) (0.60, 1.62) (0.67, 1.97) (0.89, 1.89) (0.56, 2.08) (0.62, 1.49) (0.68, 2.67) (0.88, 1.77) (0.63, 2.90) 65 years and 1.19 1.11 0.91 1.85* 0.78 1.28 0.93 1.65* 0.90 0.96 1.18 1.16 over (0.71, 2.01) (0.70, 1.78) (0.62, 1.34) (1.08, 3.16) (0.48, 1.25) (0.80, 2.05) (0.65, 1.32) (0.96, 2.84) (0.58, 1.38) (0.48, 1.89) (0.84, 1.67) (0.59, 2.29) Heart Arrhythmia 0-64 years 0.91 0.99 1.13 0.91** 0.96 1.07 1.07 0.91 0.96 0.95 1.03 0.84 (0.78, 1.06) (0.86, 1.14) (1.01, 1.27) (0.77, 1.09) (0.85, 1.05) (0.93, 1.24) (0.97, 1.18) (0.76, 1.08) (0.86, 1.06) (0.79, 1.15) (0.94, 1.13) (0.69, 1.03) 65 years and 0.97 1.02 0.91 0.95 0.99 0.99 0.94 1.01 1.03 0.97 1.07 1.05 over (0.86, 1.09) (0.91, 1.15) (0.83, 0.998) (0.83, 1.09) (0.90, 1.09) (0.88, 1.11) (0.87, 1.01) (0.89, 1.16) (0.95, 1.12) (0.83, 1.13) (1.00, 1.15) (0.90, 1.23) Conduction Disorders 0-64 years 0.80 1.03 1.26 1.80 0.91 1.21 1.24 1.36 1.07 1.22 1.06 1.20 (0.41, 1.54) (0.52, 2.02) (0.77, 2.06) (0.83, 4.01) (0.54, 1.53) (0.63, 2.35) (0.81, 1.91) (0.61, 3.04) (0.67, 1.71) (0.53, 2.78) (0.72, 1.57) (0.52, 2.80) 65 years and 0.84 1.04 0.83 0.77 0.89 0.89 0.87 0.76 1.12 0.81 0.84 0.77 over (0.61, 1.16) (0.74, 1.47) (0.65, 1.07) (0.51, 1.16) (0.68, 1.16) (0.63, 1.27) (0.70, 1.08) (0.50, 1.16) (0.89, 1.41) (0.52, 1.26) (0.68, 1.04) (0.48, 1.24) Hypertensive Diseases 0-64 years 0.89 0.96 0.89 0.93 0.91 1.16 1.06 1.22 1.11 1.88 1.08 1.82 (0.58, 1.36) (0.62, 1.48) (0.65, 1.24) (0.55, 1.56) (0.63, 1.31) (0.75, 1.78) (0.80, 1.39) (0.73, 2.03) (0.81, 1.52) (1.07, 3.28) (0.83, 1.40) (1.03, 3.20) 65 years and 0.84 0.82 1.11 0.92 0.87 0.62 0.82 0.74 0.79 0.74 0.74 1.02 over (0.59, 1.21) (0.56, 1.19) (0.86, 1.44) (0.58, 1.44) (0.65, 1.18) (0.42, 0.93) (0.65, 1.05) (0.46, 1.18) (0.61, 1.02) (0.43, 1.25) (0.59, 0.92) (0.62, 1.68) *Denotes a statistically significant positive interaction term p-value (<0.05) **Denotes a statistically significant negative interaction term p-value (<0.05)
146 6.4 Discussion
This study examined whether PM10 modifies the association between heat waves and EHAs for six cardiovascular diseases in Greater Sydney, Australia. We estimated and compared the effect of heat waves on high and low-level PM10 days at lag0-lag2 for three age groups: all ages combined, 0-64 years and 65 years and above, and tested the sensitivity of three heat wave definitions. We found some evidence that PM10 modifies the association between heat waves and EHAs for certain cardiovascular diseases. Stronger heat wave effects were observed on high compared to low-level PM10 days for EHAs for cardiac arrest for all three age groups; conduction disorders for 0-64 years; and hypertensive diseases for all ages combined and 0-64 years. These findings however, were generally not consistent across all heat wave definitions and lags. Positive, statistically significant interactions were found only for EHAs for hypertensive diseases (all ages combined) and cardiac arrest (65 years and above).
It is difficult to directly compare our findings to previous studies, as most of the work to date examining the potential interactive effects of temperature or heat waves and PM10 on cardiovascular health outcomes has considered cardiovascular mortality (e.g. Qian et al. 2008; Stafoggia et al. 2008; Meng et al.
2012; Analitis et al. 2014; Breitner et al. 2014; Li et al. 2015). Few studies have considered cardiovascular morbidity as the health outcome, particularly cause- specific cardiovascular morbidity (Ren and Tong, 2006; Ren et al. 2006; Qiu et al.
2013). Much like our findings, the results of the studies considering cardiovascular morbidity have been broadly inconsistent, although different exposure variables have been considered (i.e. temperature, season and relative humidity). For example, Ren et al. (2006) found evidence of a statistical
147 interaction between temperature and total cardiovascular hospital admissions at different lags in Brisbane, Australia, but found no such evidence for total cardiovascular emergency presentations. Qiu et al. (2013) reported that the association between PM10 and emergency hospital admissions for ischemic heart disease was strongest in the cool season and at lower levels of relative humidity in
Hong Kong, China. Further, Kang et al. (2016) found no evidence of a significant interactive effect between heat waves and PM10 on out-of-hospital cardiac arrest in Korea, which is in general disagreement with our findings regarding EHAs for cardiac arrest. The level and source composition of PM10 differs across regions and cities (Karagulian et al. 2015; World Health Organization 2016), as does population acclimatisation to temperature changes and heat waves (Guo et al.
2014; Guo et al. 2017). It is therefore important to conduct further localised studies to account for these differences, and clarify our understanding of any potential interactive effects of these environmental exposures on cardiovascular morbidity.
It is plausible that air pollution and heat exposure may interact on a biological level, although the exact causal pathways and mechanisms involved are not known. The activation of the body’s thermoregulatory system and mechanisms during heat stress can facilitate and increase the absorption and entry of toxins and air pollutants into the body, as well as alter the body’s response to such substances (Gordon 2003). The strength of the toxicity of a chemical or toxin on a biological system can be exacerbated by increased body temperature (Gordon et al. 1988; Gordon 2003). Passive heat exposure can stress the cardiovascular system, where increased skin blood flow during thermoregulation results in increased cardiac output, which in turn is mediated by increases in heart rate
148 (Crandall and González-Alonso 2010). Madaniyazi et al. (2016) observed a ‘V’ shaped relationship between mean temperature and heart rate and blood pressure
(systolic and diastolic) in Chinese adults, finding heat effects above certain thresholds. Others have, however, observed a decrease in systolic blood pressure with an increase in ambient temperature (Barnett et al. 2007). Ren et al. (2011) found that increased ambient temperature is associated with decreased heart rate variability (HRV) during the warm season, but found no evidence of an interactive effect between ambient temperature and PM2.5 on HRV. Particulate matter may also adversely affect the cardiovascular system by directly entering into the systemic circulation (smaller particles: PM2.5 or PM1.0), or indirectly by affecting the autonomic nervous system or inducing an inflammatory response
(Nelin et al. 2012). Stafoggia et al. (2008) noted that their findings of stronger
PM10 effects on mortality during the warm season might be a result of increased exposure to this pollutant, with individuals more likely to open their windows and spend time outdoors during the summer months.
We observed positive, statistically significant interactions between heat wave and high-level PM10 days on EHAs for cardiac arrest among the elderly.
Previous studies examining the susceptibility of specific age groups to the potential interactive effects of high temperatures or heat waves and PM10 on cardiovascular mortality have generally found effect modification to be more pronounced among the elderly (Li et al. 2011; Analitis et al. 2014; Li et al. 2015).
The elderly are particularly susceptible to extreme heat exposure due to their decreased capacity to effectively thermoregulate, with sweat gland output, blood flow to the skin and cardiac output reduced (Kenney and Munce 2003). Given the general decline of the body’s physiological processes with age, and the higher
149 prevalence of cardiovascular diseases among older age groups, the elderly are also susceptible to the adverse effects of particulate matter (Sacks et al. 2011). We also found some evidence of effect modification in the younger age group for certain diseases. The reasons for this are unclear, although it may be because younger populations are generally more physically active than older populations (NSW
Health 2017), which may result in more time spent outdoors, subsequently increasing their exposure levels.
We found positive, statistically significant interactions at lag1 and lag2 for certain cardiovascular diseases, but not at lag0. Evidence of an interactive effect between high temperature and high-levels of PM10 on cardiovascular health outcomes has also been found at certain lags (Ren et al. 2006; Qian et al. 2008).
For example, Qian et al. (2008) observed stronger PM10 effects on cardiovascular mortality at high compared to normal level temperatures at lag0-1 in Wuhan,
China. Short lag effects have also been observed when examining the independent effects of high temperatures and PM10 on cardiovascular morbidity (Lin et al.
2009; Colais et al. 2012). Positive, statistically significant interactions were also found for some heat wave definitions only. The choice of heat wave definition has been shown to affect both the magnitude and statistical significance of the association between heat waves and health outcomes (Tong et al. 2010b). Each of the three heat wave definitions used in this study identified different days as
‘exposure’ days, and the total number of exposure days varied between our definitions (See, Table 6.3). It is likely that is affected our models, as well as the calculation of the interaction term between heat wave and high-level PM10 days. It is also possible that different temperature metrics (maximum, mean, minimum) may have different impacts on cardiovascular health outcomes, although
150 differences in their interaction with PM10 is unclear. For example, Kang et al.
(2016) found that the risk of out-of-hospital cardiac arrest during heat waves was highest in the afternoon (3pm to 5pm), which coincided with the peak of daily outdoor temperature.
A few negative, statistically significant interactions were found, and negative associations were observed across both high and low-level PM10 days and in Figure 1 for certain cardiovascular diseases. Several previous studies have also found null or negative associations between increased temperature or extreme heat and hospital admissions for cardiovascular diseases (Michelozzi et al. 2009;
Turner et al. 2012; Ogbomo et al. 2017). Such findings are in contrast to the positive associations often observed between high temperature or heat waves and cardiovascular mortality across several regions, particularly among the elderly
(Anderson and Bell 2009; D’Ippoliti et al. 2010). The exact reasons for the differences found between these cardiovascular health outcomes are not known.
One possible explanation is that individuals may die quickly from cardiovascular disease during high temperatures before they are able to seek medical attention or be admitted to hospital (Kovats et al. 2004).
This study has some potential strengths. To our knowledge, this is the first study to examine the potential interactive effects of heat waves and PM10 on cause-specific cardiovascular hospital admissions in an Australian city. By examining and comparing six specific cardiovascular diseases, we have shown that some conditions may be more susceptible to the potential interactive effects of heat waves and PM10 than others (e.g. cardiac arrest). We also analysed a relatively long period of time series data (12 years), and controlled for other ambient air pollutants including ozone and nitrogen dioxide.
151 This study has some potential limitations. The analysis was performed for a single city, and therefore our results may not be generalisable given that PM10 levels and mixtures can vary geographically, as well as population acclimatisation to heat waves. The samples sizes for some of the cardiovascular diseases were relatively small when stratified by age group (e.g. cardiac arrest, conductions disorders), and caution should therefore be given to the significance of these results. We also estimated exposure to heat waves and PM10 by calculating the daily city-wide average using monitoring stations, and not by measuring an individual’s personal exposure level, which may have resulted in some exposure misclassification.
6.5 Chapter Conclusion
This study found some evidence that PM10 modifies the association between heat waves and hospital admissions for certain cardiovascular diseases.
Our findings however, showed inconsistencies and largely differed across age group, disease, lag and heat wave definition. Given the differences found across diseases, our study highlights the need for future studies to consider, where possible, cause-specific outcomes when examining the potential interactive effects of heat waves and ambient air pollution. With both heat waves and levels of ambient particulate matter expected to increase under climate change, it is important to consider potential effect modification by air pollution when examining the impacts of heat waves on cardiovascular morbidity. As our study has shown, this is true even for locations with comparatively low levels of particulate matter, such as Australia.
152 Acknowledgement:
In accordance with the Inclusion of Publications Statement, work from this chapter (i.e. Chapter Six) has been published in a peer-reviewed journal.
Citation: Parry M, Green D, Zhang Y, Hayen A. 2019. Does Particular Matter
Modify the Short-Term Association between Heat Waves and Hospital
Admissions for Cardiovascular Diseases in Greater Sydney, Australia? Int J
Environ Res Public Health 16:3270; https://doi.org/10.3390/ijerph16183270.
153 Chapter Seven: Conclusion
This chapter provides a summary of the key findings of this thesis, outlines its potential strengths, limitations and implications; and provides direction for future research.
7.1 Summary of the Key Findings
This section briefly summarises the main findings of Chapters Four, Five and Six of this thesis.
7.1.1 Aim 1: To examine and compare the impact of single and consecutive days of hot temperature extremes, including high minimum temperatures, on heat-related hospital admissions in Greater Sydney, Australia, for a suite of definitions.
A time-stratified case-crossover design was used in Chapter Four to estimate the association between extreme heat and EHAs for four heat-related conditions including: acute renal failure; dehydration; fluid imbalance disorders; and direct heat-related conditions, during the warm season for 2001 to 2013. A range of definitions was used to define extreme heat exposure. These were developed using three different ambient temperature metrics (maximum, mean and minimum temperature); five different relative temperature thresholds (≥90th,
95th, 97th, 98th, 99th); and four different durations (≥1, 2, 3 and 4 consecutive days). Chapter Four also examined the impact of heat wave timing in the season
154 on each of the four diagnoses, and investigated whether there was evidence of an added heat wave effect.
Chapter Four found positive associations between single and consecutive days (i.e. heat waves) of extreme heat and EHAs for each diagnosis across all definitions, but the strength and statistical significance of these associations differed across definitions. The strongest associations between heat waves and
EHAs for acute renal failure and dehydration were observed when heat waves were defined using minimum temperature (i.e. consecutive warm nights), and for fluid imbalance disorders when heat waves were defined using maximum temperature (i.e. consecutive hot days). Some evidence of an added heat wave effect was found for each condition, with the strongest effects for EHAs for acute renal failure and dehydration observed again when heat waves were defined using minimum temperature. The impact of the first heat wave of the season compared to later heat waves for each diagnosis was sensitive to the heat wave definition used to define heat exposure, producing inconsistent results.
7.1.2 Aim 2: To examine whether ozone modifies the short-term association between heat waves and hospital admissions for certain respiratory diseases in Greater Sydney, Australia.
A time-stratified case-crossover design was used in Chapter Five to estimate the association between heat waves and EHAs for three respiratory conditions including: asthma; COPD; and pneumonia, during the warm season for
2001 to 2013. The association was first estimated with, and without, adjusting for ozone, while also adjusting for relevant confounders. To examine potential effect modification by ozone, heat wave effects were estimated and compared on days
155 with high (≥ 95th percentile) and low (<95th percentile) levels of ozone at different lags (lag0-lag2). This Chapter also investigated the susceptibility of specific age groups and tested the sensitivity of three heat wave definitions.
Chapter Five found that adjusting for ozone generally had a small, but sometimes negligible effect on the association between heat waves and each of the respiratory conditions. Positive, statistically significant interactions were found between heat wave and high level ozone days on EHAs for COPD (0-64 years) and for pneumonia (all ages, 0-14 years, 15-64 years) for some heat wave definitions at certain lags. Stronger heat wave effects were also generally observed on high compared to low level ozone days for EHAs for pneumonia across most definitions and lags in the all ages, 15-64 years and 65-74 years age groups. Negative associations between heat waves and EHAs for asthma were observed on both high and low level ozone days across all lags and heat wave definitions in the all ages and 0-14 years age group, with some statistically significant associations found. Overall, this Chapter found some evidence that ozone modifies the association between heat waves and hospital admissions for certain respiratory diseases, although the findings largely differed across disease, age group, lag and heat wave definition.
7.1.3 Aim 3: To examine whether particulate matter modifies the short-term association between heat waves and hospital admissions for cardiovascular diseases in Greater Sydney, Australia.
A time-stratified case-crossover design was used in Chapter Six to examine the association between heat waves and EHAs for six cardiovascular
156 diseases including: ischemic heart disease; heart failure; cardiac arrest; heart arrhythmia; conduction disorders; and hypertensive diseases, during the warm season for 2001 to 2013. The association was first estimated with, and without, adjusting for particulate matter (PM10), while also adjusting for relevant confounders. To examine potential effect modification by PM10, heat wave effects were estimated and compared on days with high (≥ 90th percentile) and low (<90th percentile) levels of PM10 at different lags (lag0-lag2). This Chapter also investigated the susceptibility of younger (0-64 years) and older (65 years and above) populations, and tested the sensitivity of three heat wave definitions.
Chapter Six found that adjusting for PM10 had little effect on the association between heat waves and the selected cardiovascular diseases. Stronger heat wave effects were observed on high compared to low level PM10 days for
EHAs for cardiac arrest for all ages combined, 0-64 years and 65 years and above; conduction disorders for 0-64 years; and hypertensive diseases for all ages combined and 0-64 years. These findings however, were generally not consistent across all heat wave definitions and lags. Positive, statistically significant interactions were found only for EHAs for cardiac arrest (65 years and above) and hypertensive diseases (all ages combined). Overall, Chapter Six found some evidence that PM10 modifies the association between heat waves and hospital admissions for certain cardiovascular diseases, although our findings largely differed across disease, age group, lag and heat wave definition.
7.2 Potential Strengths and Limitations
The two main environmental datasets used in this thesis (i.e. meteorological and air pollution) were generally of high quality. The respective 157 government agencies that collect and manage the data use a range of quality control procedures and processes to ensure that is of the highest possible quality
(Bureau of Meteorology 2018c; NSW Office of Environment and Heritage
2015b). As outlined in Section 3.2.1 in Chapter Three, these two datasets underwent further quality control checks to detect and correct any further remaining errors. A number of monitoring stations from various locations across the city were used to calculate the respective daily average temperature and air pollution values used in the analyses. This ensured that the calculated values provided a good representation of the daily average across the city. Further, a relatively long time series of meteorological data was analysed (twelve warm seasons), which allowed for a number of extreme heat events (both single and consecutive days) to be identified and analysed.
Multiple definitions of extreme heat, including heat waves, were constructed and used to estimate the respective associations in Chapters Four,
Five and Six. The purpose of doing so was not to establish or determine which definition is ‘best’ or most applicable to use in public health/heat wave alerts, but rather to examine if, and how, the association differed across the definitions. This is because previous work has shown that the choice of definition can affect both the magnitude and statistical significance of the association (Tong et al. 2010b;
Kent et al. 2014). The findings in Chapters Five and Six extend this understanding and show that heat wave definition choice can also affect whether air pollution can be considered as an effect modifier of the association between heat waves and hospital admissions. This is because the comparative estimates differed across heat wave definitions for some conditions. This suggests that future work should
158 also test the sensitivity of heat wave definitions when considering air pollution as a potential effect modifier of the association.
Chapters Five and Six examined cause-specific health outcomes. This allowed us to show that specific conditions (e.g. COPD, cardiac arrest) may be more susceptible to the interactive effects of heat waves and ozone or particulate matter (PM10). The interpretation and implications of these findings are subject, however, to certain limitations. By examining specific conditions, and stratifying these conditions by age group, the sample size was reduced. This can in turn affect the precision of the effect estimates. Further, the NSW Ministry of Health’s
Admitted Patient Data Collection is a routinely collected administrative dataset, where the information is not collected for the purposes of clinical or population health research. Therefore, it is possible that errors exist with the coding of the principal diagnosis, which may adversely impact the accuracy of analyses at the disease level. Although, such information is coded by professional coders according to strict criteria.
Differences in daily maximum and minimum temperatures are often observed across Greater Sydney during the warm season, with higher extreme maximum temperatures recorded in the western areas of the city (Bureau of
Meteorology 2018d; Bureau of Meteorology 2018e). Further, under certain synoptic conditions, elevated levels of ozone can be observed in the western and south-western areas of the city, with the sea breeze transporting precursor pollutants from other areas (Jiang et al. 2017). The analyses in this thesis used the daily city-wide average to estimate exposure to both extreme heat and air pollution. Therefore, any potential spatial variation in exposure to these
159 environmental variables, and its impact on the associations estimated in Chapters
Four, Five and Six, was not accounted for.
Daily and sub-daily observational values from government monitoring stations across Greater Sydney were used to estimate daily exposure to extreme heat and air pollution. This method is unlikely to have accurately estimated an individual’s ‘true’ personal exposure, and it is therefore possible that some exposure misclassification occurred. This is because individuals are also likely to spend time indoors throughout the day and evenings, which may be air- conditioned. Most studies conducted in the Greater Sydney region to date have estimated temperature or air pollution exposure using stationary monitoring stations, with very few estimating an individual’s own personal exposure (e.g.
Chertok et al. 2004).
The results of this thesis may not be generalisable to other areas and populations. This is because the magnitude of the association between extreme heat and adverse health outcomes differs across cities, countries and regions due to differences in population demographics; population acclimatisation to temperature changes; and a range socioeconomic, technological, cultural and behavioural factors (Hajat and Kosatky 2010). The concentration and mixture of air pollutants, particularly particulate matter, also differs across cities, countries and regions. Despite this, the results presented in this thesis provide important information for the local health sector and agencies, especially in regard to climate change adaptation planning. Further, the study design and statistical methods used in this thesis can be applied in other locations and used to evaluate similar research questions.
160 This thesis used a time-stratified case-crossover study design to estimate the association between extreme heat and selected hospital admissions. Results obtained using the time-series and case-crossover designs have shown to be similar (Tong et al. 2012). However, unlike the time-series design, the case- crossover design does not allow for possible overdispersion to be considered in modelling or design (Lu and Zeger 2007). Further, a study has also shown that the time-series design was more suitable in accounting for autocorrelation than the case-crossover design (Guo et al. 2010).
7.3 Potential Implications
Chapter Four found some evidence to suggest that extreme high minimum temperatures may have a greater impact than extreme high maximum temperatures on certain heat-related conditions. Chapters Five and Chapter Six found some evidence of effect modification, although the results showed inconsistencies and differed across age group, lag and heat wave definition. This section outlines the potential implications of these main findings for a range of stakeholders.
As noted in Chapters One and Two, there is no universal or standard definition of extreme heat. It can be argued that the general public may perhaps consider a single day or period to be extremely hot when daily maximum temperatures are particularly high (e.g. 35°C, 40°C). As such, they may not be aware of the potential impacts that extreme high minimum (or overnight) temperatures can have on their health, and the subsequent need to reduce their exposure to such temperatures. As the evidence base for the impact of these
161 temperatures on health outcomes develops, it may be appropriate for health authorities to develop and implement a public awareness campaign to equip individuals with knowledge regarding the health impacts of these temperatures, and strategies they can implement to reduce exposure during the evenings.
Currently, NSW Health’s ‘Beat the Heat’ campaign makes reference to ‘hot weather’ and ‘extreme heat’, but the information available to the public on their website does not appear differentiate between the impacts of extreme high minimum and maximum temperatures (NSW Health 2016). It may also be appropriate for clinicians and health professionals to communicate with their patients and clients about the importance and relevance of extreme high minimum temperatures to their health, especially if certain population groups are identified in future work as being particularly susceptible or vulnerable, such as those with certain pre-existing medical conditions.
As noted in Chapter One, future urban expansion and land use change is expected to enhance the warming of minimum temperatures under anthropogenic climate change across the city. There are, however, strategies that can be adopted to mitigate the impact of urban expansion and climate change on these temperatures, and in turn reduce the associated health impacts. One such option is the implementation of green infrastructure, such as green roofs and walls, and urban forests. There appears to be some momentum for the implementation of such infrastructure in Sydney and NSW more broadly, with the City of Sydney
Council developing a Green Roofs and Walls Policy, and the Government
Architect of NSW releasing a draft framework for the implementation of green infrastructure across urban areas in NSW in late 2017, entitled Greener Places
(City of Sydney Council 2014; NSW Government Architect 2017). This thesis
162 provides some empirical evidence of the adverse impacts that extreme high minimum temperatures can have on heat-related conditions in Greater Sydney. It is therefore recommended that public health considerations should be a key focus in the planning and development of urban and green infrastructure across the city, particularly in western Sydney where urban and housing development is rapidly expanding.
Different national and state government agencies detect and monitor the occurrence of heat waves and extreme air pollution episodes in Greater Sydney.
The Bureau of Meteorology uses the Excess Heat Factor to define and predict heat waves across Australia (Bureau of Meteorology 2018f). When a heat wave is forecast for Greater Sydney, NSW Health issues an appropriate health alert for the region. The NSW Office of Environment and Heritage calculates an air quality index for Greater Sydney, and issues a forecast of the index at 4pm each day for the following day (NSW Office of Environment and Heritage 2018b; NSW Health
2018). If the value is forecast to be above a particular threshold (i.e. 100), then an air pollution health alert is issued (NSW Health 2018). Media reports have highlighted that air pollution health alerts have been issued during recent heat waves and days of extreme high temperatures in the region (Hunt 2017; McInnes
2018). However, both of these alerts are issued independently and do not take into consideration that populations may be more vulnerable to the health impacts of heat waves during high air pollution episodes (or vice-versa). As the evidence regarding the synergistic health effects of these two exposures becomes clearer and more certain with future work, it may be appropriate to develop a more integrated approach when issuing these alerts to account for the combined health impacts of these two exposures. This approach may also include educating the
163 public about the potential health risks associated with exposure to both heat waves and high levels of air pollution simultaneously, and the strategies and behaviours they can adopt to reduce their exposure during these events.
Most previous work examining whether ambient air pollution modifies the association between temperature or heat waves and human health has considered mortality as the health outcome, with few considering morbidity, particularly cause-specific morbidity. The findings of Chapters Five and Six highlight the importance of considering ozone and particulate matter as potential effect modifiers in future work when examining the association between heat waves and morbidity. This is particularly relevant for other regions that have higher average and peak concentrations of ozone and particulate matter, such as Asia, Europe and
North America.
Chapters Five and Six found some evidence of effect modification for certain diseases only. As the evidence base develops in relation to the interactive effects of extreme heat and air pollution on specific conditions, it may become relevant for clinicians to inform their patients of the potential health risks associated with exposure to heat and air pollution simultaneously, as well as strategies they can adopt to minimise and respond to these risks.
7.4 Directions for Future Research
Future urban expansion and land use change is expected to enhance the warming of minimum temperatures under anthropogenic climate change in
Greater Sydney. Future research could examine how projected increases in minimum temperatures arising from future urban expansion and land use will
164 affect human health outcomes in the region, particularly future mortality and morbidity. This could also include, if possible, a comparison of the relative contributions of anthropogenic climate change and urban expansion to these estimated health impacts. Such information would be useful in developing strategies to mitigate the effects of climate change and urban expansion on temperature changes, and their associated health impacts, in the region.
The intensity of extreme heat and concentrations of ozone and particulate matter can vary spatially across the Greater Sydney, with more intense heat and higher concentrations of air pollutions, particularly ozone, observed in the western regions at times. Future research could conduct a spatial analysis to investigate if the potential interactive effects of heat waves and air pollution on health outcomes differ across Greater Sydney. This has the potential to identify areas where populations may be more susceptible or vulnerable to the interactive effects of these two exposures. Such knowledge could then be used to develop more localised and targeted interventions and responses to protect population health across the city.
Humidity can influence the body’s physiological responses to extreme heat exposure, decreasing the rate at which an individual can shed excess heat
(Davis et al. 2016; Coffel et al. 2018). The three studies in thesis controlled for the potential confounding effects of air moisture by using dew-point temperature, but did not consider or assess any potential interactive effects between extreme heat (including heat waves) and humidity and hospital admissions. Future research work could explore this potential interaction, including the influence of other meteorological variables such as wind speed.
165 Climate change is expected to increase the intensity, frequency and duration of heat waves, and affect average and peak concentrations of ozone and particular matter. Future research could examine the potential interactive effects of these two environmental exposures on various health outcomes under different climate change scenarios. The general approach to date has been for studies to examine how climate change will affect the respective independent associations between these two exposures and health outcomes, with little consideration of how it will affect their potential joint impacts on health outcomes. This work has the potential to inform climate change adaptation planning in the health sector.
7.5 Conclusion
This thesis increased our understanding of the association between extreme heat and hospital admissions in Greater Sydney, Australia; although given some of the inconsistencies and differences in the findings, uncertainties still remain. It is vital that future research work continues to enhance and refine our understanding of this association across the globe, given that the number of hot days and warm nights is expected to increase due to climate change, as well as the intensity, frequency and duration of heat waves. Such work is critically important to inform climate change adaptation planning in the health sector to ensure populations are well prepared for a future of unprecedented extreme heat.
166 References
Alexander LV, Zhang X, Peterson TC, Caeser J, Gleason B, Klein Tank AMG,
Haylock M, Collins D, Trewin B, Rahumzadeh F, et al. 2006. Global
observed changes in daily extremes of temperature and precipitation. J
Geophys Res 111:DO5109; doi: 10.1029/2005JD006290.
Alexander LV, Arblaster JM. 2009. Assessing trends in observed and modelled
climate extremes over Australia in relation to future projections. Int J
Climatol 29:417-435.
Alexander L, Tebaldi C. 2012. Climate and weather extremes: Observations,
modelling, and projections. In: The future of the world’s climate.
(Henderson-Sellers A, McGuffie K, eds). New York: Elsevier, 253-288.
Alexander LV, Arblaster JM. 2017. Historical and projected trends in temperature
and precipitation extremes in Australia in observations and CMIP5.
Weather Clim Extrem 15:34-56.
Analitis A, Michelozzi P, D’Ippoliti D, de’Donato F, Menne B, Matthies F,
Atkinson RW, Iñiguez C, Basagaña X, Schneider A, et al. 2014. Effects of
heat waves on mortality: Effect modification and confounding by air
pollutants. Epidemiology 25:15-22.
Anderson GB, Bell ML. 2009. Weather-related mortality: How heat, cold and heat
waves affect mortality in the United States. Epidemiology 20:205-213.
Anderson GB, Bell ML. 2011. Heat waves in the United States: Mortality risk
during heat waves and effect modification by heat wave characteristics in
43 U.S. communities. Environ Health Perspect 119:210-218.
167 Anderson GB, Dominici F, Wang Y, McCormack MC, Bell ML, Peng RD. 2013.
Heat-related emergency hospitalizations for respiratory diseases in the
Medicare population. Am J Respir Crit Care Med 187:1098-1103.
Anderson HR, Atkinson RW. Peacock JL, Marston L, Konstantinou K. 2004.
Meta-analysis of time-series studies and panel studies of particulate matter
(PM) and ozone (O3). World Health Organization Regional Office for
Europe. Copenhagen, Demark.
Anderson HR, Spix C, Medina S, Schouten JP, Castellsague J, Rossi G, Zmirou
D, Touloumi G, Wojtyniak B, Ponka A, et al. 1997. Air pollution and
daily admissions for chronic obstructive pulmonary disease in 6 European
cities: results from the APHEA project. Eur Respir J 10:1064-1071.
Arbuthnott K, Hajat S, Heaviside C, Vardoulakis S. 2016. Changes in population
susceptibility to heat and cold over time: Assessing adaptation to climate
change. Environ Health 15(Suppl 1):33; doi:10.1186/s12940-016-0102-7.
Argüeso D, Evans JP, Fita L, Bormann KJ. 2014. Temperature response to future
urbanization and climate change. Clim Dyn 42:2183-2199.
Argüeso D, Evans JP, Pitman AJ, Di Luca A. 2015. Effects of city expansion on
heat stress under climate change conditions. PloS One 10: e0117066;
doi:10.1371/journal.pone.0117066.
Åström C, Orru H, Rocklöv J, Strandberg G, Ebi KL, Forsberg B. 2013. Heat-
related respiratory hospital admissions in Europe in a changing climate: a
health impact assessment. BMJ Open 3:e001842; doi:10.1136/bmjopen-
2012- 001842.
168 Australian Bureau of Statistics. 2017a. Regional Population Growth, Australia,
2016. Available: http://www.abs.gov.au/ausstats/[email protected]/mf/3218.0
[Last accessed 20 October 2017].
Australian Bureau of Statistics. 2017b. 2016 Census Quick Stats: Greater Sydney.
Available:
http://www.censusdata.abs.gov.au/census_services/getproduct/census/201
6/quickstat/1GSYD?opendocument [Last accessed 23 April 2018].
Australian Government and CSIRO. 2018. Australian Climate Trends. Available:
https://www.climatechangeinaustralia.gov.au/en/climate-
campus/australian-climate-change/australian-trends/ [Last accessed 15
April 2018].
Australian Institute of Health and Welfare. 2016. Australian burden of disease
study: Impact and causes of illness and death in Australia. Australian
Burden of Disease Study series no.3. BOD 4. Canberra, Australian Capital
Teritory, Australia.
Bao J, Wang Z, Yu C, Li X. 2016. The influence of temperature on mortality and
its Lag effect: a study in four Chinese cities with different latitudes. BMC
Public Health 16:375; doi:10.1186/s12889-016-3031-z.
Barnett AG, Williams GM, Schwartz J, Best TL, Neller AH, Petroeschevsky AL,
Simpson RW. 2006. The effects of air pollution on hospitalizations for
cardiovascular disease in elderly people in Australian and New Zealand
cities. Environ Health Perspect 114:1018-1023.
Barnett AG. 2007. Temperature and cardiovascular deaths in the US elderly.
Epidemiology 18(3):369-372.
169 Barnett AG, Sans S, Salomaa V, Kuulasmaa K, Dobson AJ, WHO MONICA
Project. 2007. The effect of temperature on systolic blood pressure. Blood
Press Monit 12:195-203.
Barnett AG, Dobson AJ. 2010. Analysing seasonal health data. Statistics for
Biology and Health. Springer, Germany.
Barnett AG, Hajat S, Gasparrini A, Rocklöv J. 2012. Cold and heat waves in the
United States. Environ Res 112:218-224.
Barriopedro D, Fischer EM, Luterbacher J, Trigo RM, García-Herrera R. 2011.
The hot summer of 2010: Redrawing the temperature record map of
Europe. Science 332:220-224.
Basu R, Feng WY, Ostro B. 2008. Characterizing temperature and mortality in
Nine California Counties. Epidemiology 19:138-145.
Becker JA, Stewart LK. 2011. Heat-related illness. Am Fam Physician 83:1325-
1330.
Bell ML, Davis DL. 2001. Reassessment of the lethal London fog of 1952: Novel
indicators of acute and chronic consequences of acute exposure to air
pollution. Environ Health Perspect 109:389-294. Supp. 3.
Bell ML, Dominic F, Samet JM. 2005. A meta-analysis of time-series studies of
ozone and mortality with comparison to the National Morbidity, Mortality,
and Air Pollution Study. Epidemiology 164(4):436-445.
Bell ML, Goldberg R, Hogrefe C, Kinney PL, Knowlton K, Lynn B, Rosenthal J,
Rosenzwieg C, Patz JA. 2007. Climate change, ambient ozone and health
in 50 US cities. Clim Change 82:61-76.
170 Bell ML, O’Neill MS, Ranjit N, Borja-Aburto VH, Cifuentes LA, Gouveia NC.
2008. Vulnerability to heat-related mortality in Latin America: a case-
crossover study in São Paulo, Brazil, Santiago, Chile and Mexico City,
Mexico. Int J Epidemiol 37:796-804.
Bhaskaran K, Gasparrini A, Hajat S, Smeeth L, Armstrong B. 2013. Time series
regression studies in environmental epidemiology. Int J Epidemiol 1-9;
doi:10.1093/ije/dyt092.
Bobb JF, Peng RD, Bell ML, Dominici F. 2014a. Heat-related mortality and
adaptation to heat in the United States. Environ Health Perspect
122(8):811-816.
Bobb JF, Obermeyer Z, Wang Y, Dominici F. 2014b. Cause-specific risk of
hospital admission related to extreme heat in older adults. JAMA
312(24):2659-2667.
Breitner S, Wolf K, Devlin RB, Diaz-Sanchez D, Peters A, Schneider A. 2014.
Short-term effects of air temperature on mortality and effect modification
by air pollution in three cities of Bavaria, Germany: A time-series
analysis. Sci Total Environ 485-486:49-61.
Buckley JP, Samet JM, Richardson DB. 2014. Does air pollution confound studies
of temperature? Epidemiology. 25(2):242-245.
Bureau of Meteorology. 2014. Annual Climate Statement 2013. Available:
http://www.bom.gov.au/climate/current/annual/aus/2013/ [Last accessed
14 April 2018].
Bureau of Meteorology. 2015. Metadata Notes. Australian Government.
[Explanatory Notes provided with the data].
171 Bureau of Meteorology. 2017a. Sydney in summer 2016-17: Warmest summer on
record. Available:
http://www.bom.gov.au/climate/current/season/nsw/archive/201702.sydne
y.shtml [Last accessed 14 April 2018].
Bureau of Meteorology. 2017b. Climate Statistics for Australian Locations.
Available: http://www.bom.gov.au/climate/averages/tables/cw_066062.sht
ml [Last accessed 4 February 2018].
Bureau of Meteorology. 2018a. Annual climate statement 2017. Available:
http://www.bom.gov.au/climate/current/annual/aus/ [Last accessed 15
April 2018].
Bureau of Meteorology. 2018b. Greater Sydney in summer 2017-18: Another hot
and dry summer. Available:
http://www.bom.gov.au/climate/current/season/nsw/sydney.shtml [Last
accessed 23 April 2018].
Bureau of Meteorology. 2018c. Quality control of climate data. Available:
http://www.bom.gov.au/climate/data-services/about-data-
observations.shtml#tabs=Quality-control [Last accessed 23 April 2018].
Bureau of Meteorology. 2018d. Climate statistics for Australian locations: Sydney
(Observatory Hill). Available:
http://www.bom.gov.au/climate/averages/tables/cw_066062_All.shtml
[Last accessed 23 January 2018].
Bureau of Meteorology. 2018e. Climate statistics for Australian locations: Penrith
Lakes AWS. Available:
http://www.bom.gov.au/climate/averages/tables/cw_067113_All.shtml
[Last accessed at 23 January 2018].
172 Bureau of Meteorology. 2018f. Heatwave service for Australia. Available:
http://www.bom.gov.au/australia/heatwave/. [Last accessed 25 April
2018].
Bureau of Meteorology, CSIRO. 2014. State of the Climate 2014. Available:
http://www.bom.gov.au/state-of-the-climate/2014/ [Last accessed 14
April 2018].
Bureau of Meteorology, CSIRO. 2017. State of the Climate 2016. Available:
http://www.bom.gov.au/state-of-the-climate/index.shtml [Last accessed 14
April 2018].
Burkart K, Canário P, Breitner S, Schneider A, Scherber K, Andrade H,
Alcoforado MJ, Endlicher W. 2013. Interactive short-term effects of
equivalent temperature and air pollution on human mortality in Berlin and
Lisbon. Environ Pollut 183:54-63.
Camalier L, Cox W, Dolwick P. 2007. The effects of meteorology on ozone in
urban areas and their use in assessing ozone trends. Atmos Environ
41:7127-7137.
Centre for Health Record Linkage. 2018. Data dictionaries. Available:
http://www.cherel.org.au/data-dictionaries#section1[Last accessed 26
April 2018].
Chang HH, Zhou J, Fuentes M. 2010. Impact of climate change on ambient ozone
level and mortality in Southeastern United States. Int J Environ Res Public
Health. 7:2866-2880.
Chen K, Glonek G, Hansen A, Williams S, Tuke J, Salter A, Bi P. 2016. The
effects of air pollution on asthma hospital admissions in Adelaide, South
173 Australia, 2003-2013: time-series and case-crossover analyses. Clin Exp
Allergy 0:1-15; doi: 10.1111/cea.12795.
Chen L, Mengersen K, Tong S. 2007. Spatiotemporal relationship between
particle air pollution and respiratory emergency hospital admissions in
Brisbane, Australia. Sci Total Environ 373:57-67.
Chen T, Sarnat SE, Grundstein AJ, Winquist A, Chang HH. 2017. Time-series
analysis of heat waves and emergency department visits in Atlanta, 1993
to 2012. Environ Health Persp 125(5):057009; doi: 10.1289/EHP44.
Cheng CS, Campbell M, Li Q, Li G, Auld H, Day N, Pengelly D, Gingrich S,
Klaassen J, MacLver D, et al. 2008. Differential and combined impacts of
extreme temperatures and air pollution on human mortality in south-
central Canada. Part II: future estimates. Air Qual Atmos Health 1:223-
235.
Cheng Y, Kan H. 2012. Effect of the interaction between outdoor air pollution
and extreme temperature on daily mortality in Shanghai, China. J
Epidemiol 22:28-36.
Chertok M, Voukelatos A, Sheppeard V, Rissel C. 2004. Comparison of personal
exposures to air pollutants by commuting mode in Sydney BTEX & NO2.
New South Wales Department of Health, Sydney, Australia.
Chung Y, Noh H, Honda Y, Hashizume M, Bell ML, Guo YYL, Kim H. 2017.
Temporal changes in mortality related to extreme temperatures for 15
cities in Northeast Asia: Adaptation to heat and maladaptation to cold. Am
J Epidemiol 185(10):907-913.
City of Sydney Council. 2014. Green Roofs and Walls Policy. Available at:
http://www.cityofsydney.nsw.gov.au/vision/towards-
174 2030/sustainability/greening-the-city/green-roofs-and-walls [Last accessed
25 April 2018].
Coates L, Haynes K, O’Brien J, McAneney J, Dimer de Oliveira F. 2014.
Exploring 167 years of vulnerability: An examination of extreme heat
events in Australia 1844-2010. Environ Sci Policy 42:33-44.
Coffel ED, Horton RM, de Sherbinin A. 2018. Temperature and humidity based
projections of a rapid rise in global heat stress exposure during the 21st
century. Environ Res Lett 13: 014001.
Colais P, Faustini A, Stafoggia M, Berti G, Bisanti L, Cadum E, Cernigliaro A,
Mallone S, Pacelli B, Serinelli M, et al. 2012. Particulate air pollution and
hospital admissions for cardiac diseases in potentially sensitive subgroups.
Epidemiology 23:473-481.
Collins, M., R. Knutti, J. Arblaster, J.-L. Dufresne, T. Fichefet, P. Friedlingstein,
X. Gao, W.J. Gutowski, T. Johns, G. Krinner, M. Shongwe, C. Tebaldi,
A.J. Weaver and M. Wehner, 2013: Long-term Climate Change:
Projections, Com- mitments and Irreversibility. In: Climate Change 2013:
The Physical Science Basis. Contribution of Working Group I to the Fifth
Assessment Report of the Intergovernmental Panel on Climate Change
[Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung,
A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.)]. Cambridge
University Press, Cambridge, United Kingdom and New York, NY, USA.
Cope M, Lee S, Physick B, Abbs D, Nguyen KC, McGregor J. 2008. A
methodology for determining the impact of climate change on ozone
levels in an urban area: Final Report. CSIRO. Canberra, Australian Capital
Territory, Australia.
175 Couto-Gomes E, Stone V, Florida-James G. 2010. Investigating performance and
lung function in a hot, humid and ozone-polluted environment. Eur J Appl
Physiol 110:199-205.
Cowan T, Purich A, Perkins S, Pezza A, Boschat G, Sadler K. 2014. More
frequent, longer and hotter heat waves for Australia in the Twenty-First
Century. J Clim 27: 5851-5871.
Crandall CG, González-Alonso J. 2010. Cardiovascular function in the heat-
stressed human. Acta Physiol 199:407-423.
CSIRO (Commonwealth Scientific and Industrial Research Organisation) and
Bureau of Meteorology. 2015. Climate change in Australia information
for Australia’s natural resource management regions: Technical report.
CSIRO and Bureau of Meteorology. Australia.
Dahlman L. 2017. Climate change: Global temperature. Accessed at:
https://www.climate.gov/news-features/understanding-climate/climate-
change-global-temperature. [Last accessed 15 April 2018].
Davis RE, McGregor GR, Enfield KB. 2016. Humidity: A review and primer on
atmospheric moisture and human health. Environ Res 144:106-116.
Dawson JP, Adams PJ, Pandis SN. 2007a. Sensitivity of ozone to summertime
climate in eastern USA. A modelling case study. Atmos Environ 41:1494-
1511.
Dawson JP, Adams PJ, Pandis SN. 2007b. Sensitivity of PM2.5 to climate in the
Eastern US: a modelling case study. Atmos Environ 7:4295-4309.
Dear K, Ranmuthugala G, Kjellström T, Skinner C, Hanigan I. 2003. Effects of
temperature and ozone on daily mortality during the August 2003 heat
wave in France. Arch Environ Occup Health 60:205- 212.
176 den Elzen M, Fransen T, Kuramochi T. 2017. The Emissions Gap Report: A UN
Environment Synthesis Report. Chapter 2: Pre-2020 action: trends and
progress. United Nations Environment Programme. Nairobi, Kenya.
Department of Environment and Climate Change NSW. 2007. Technical Report
No.2 Air Emissions Inventory for the Greater Metropolitan Region in New
South Wales. Anthropogenic Ozone Precursors and Particle Emissions in
the Greater Metropolitan and Sydney Regions: Results. State Government
of NSW. Sydney, New South Wales Australia.
Department of Health. 2014. The health impacts of the January 2014 heatwave in
Victoria. State Government of Victoria. Melbourne, Victoria, Australia.
D’Ippoliti D, Michelozzi P, Marino C, de’Donato F, Menne B, Katsouyanni K, et
al. 2010. The impact of heat waves on mortality in 9 European cities:
results from the EuroHEAT project. Environ Health 9:37;
https://doi.org/10.1186/1476-069X-9-37.
Dominici F, Peng RD, Bell ML, Pham L, McDermott A, Zeger S, Samet JM.
2006. Fine particulate air pollution and hospital admission for
cardiovascular and respiratory diseases. JAMA. 295(10):1127-1134.
Donat MG, Alexander LV, Yang H, Durre I, Vose R, Dunn RJH, Willett KM,
Aguliar E, Brunet M, Caeser J, et al. 2013. Updated analyses of
temperature and precipitation extremes indices since the beginning of the
twentieth century: The HadEX2 dataset. J Geophys Res Atmos 118:1-16;
doi: 10.1002/jgrd.50150.
Dong W, Zeng Q, Ma Y, Li G, Pan X. 2016. Impact of heat wave definitions on
the added effect of heat waves on cardiovascular mortality in Beijing,
177 China. Int J Environ Res Public Health. 13:933;
doi:10.3390/ijerph13090933.
Fang Y, Mauzerall DL, Liu J, Fiore AM, Horowitz LW. 2013. Impacts of 21st
century climate change on global air pollution-related premature mortality.
Clim Change 121:239-253.
Fiala J, Cernikovsky L, de Leeuw F, Kurfuerst, P. 2003. Overview of exceedances
of EC ozone threshold values during the summer season April-August
2003 and comparisons with previous years. European Environment
Agency. Copenhagen, Denmark.
Filleul L, Cassadou S, Médina S, Fabres P, Lefranc A, Eilstein D, et al. 2006. The
relation between temperature, ozone, and mortality in nine French cities
during the heat wave of 2003. Environ Health Perspect 114:1334-1347.
Fiore MA, Naik V, Spracklen DV, Steiner A, Unger N, Prather M, Bergmann D,
Camerson-Smith PJ, Cionni I, Collins WJ, Dalsøren S, et al. 2012. Global
air quality and climate. Chem Soc Rev 41:6663-6683.
Fischer PH, Brunekreef B, Lebret E. 2004. Air pollution related deaths during the
2003 heat wave in the Netherlands. Atmos Environ 38:1083-1085.
Fletcher BA, Lin S, Fitzgerald E, Hwang SA. 2012. Association of summer
temperatures with hospital admisisons for renal diseases in New York
State: A case-crossover study. Am J Epidemiol 175(9):907-916.
Folinsbee LJ, Horvath SM, Raven PB, Bedi JF, Morton AR, Drinkwater BL,
Bolduan NW, Gliner JA. 1977. Influence of exercise and heat stress on
pulmonary function during ozone exposure. J Appl Physiol Respir Environ
Exerc Physiol 43(3):409-413.
178 Fouillet A, Rey G, Laurent F, Pavillon G, Bellec S, Guihenneuc-Jouyaux C,
Clavel J, Jougla E, Hémon D. 2006. Excess mortality related to the August
2003 heat wave in France. Int Arch Occup Environ Health 80:16-24.
Fouillet A, Rey G, Wagner V, Laaidi K, Empereur-Bissonnet P, Le Tertre A,
Frayssinet P, Bessemoulin P, Laurent F, De Crouy-Chanel P. 2008. Has
the impact of heat waves on mortality changed in France since the
European heat wave of summer 2003? A study of the 2006 heat wave. Int
J Epidemiol 37:309-317.
Galindo N, Varea M, Gil-Moltó J, Yubero E, Nicolás J. 2011. The influence of
meteorology on particulate matter concentrations at an urban
Mediterranean location. Water Air Soil Pollut 215:365-372.
García-Herrera R, Díaz J, Trigo RM, Luterbacjer J, Fisher EM. 2010. A review of
the European summer heat wave of 2003. Crit Rev Environ Sci Technol
40:267-306.
Gasparrini A, Armstrong B. 2011. The impact of heat waves on mortality.
Epidemiology 22:68-73.
Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, Tobias
A, Tong S, Rocklöv J, Forsberg B, et al. 2015a. Mortality risk attributable
to high and low ambient temperature: a multicounty observational study.
The Lancet 386:369-375.
Gasparrini A, Guo Y, Hashizume M, Kinney PL, Petkova EP, Laving E,
Zanobetti A, Schwartz JD, Tobias A, Leone M. 2015b. Temporal
Variation in Heat-Mortality Associations: A multicountry study. Environ
Health Perspect 123(11):1200-1207.
179 Gasparrini A, Guo Y, Hashizume M, Lavigne E, Tobias A, Zanobetti A, Schwartz
JD, Leone M, Michelozzi P, Kan H. 2016. Changes in susceptibility to
heat during the summer: A multicountry analysis. Am J Epidemiol
183(11):1027-1036.
Gasparrini A, Guo Y, Sera F, Vicedo-Cabrera AM, Huber V, Tong S, de Sousa
Zanotti Stagliorio Coelho M, Hilario Nascimento Saldiva P, Lavigne E,
Matus Correa P. 2017. Projections of temperature-related excess mortality
under climate change scenarios. Lancet Planet Health. 1:e360–67.
Gibbons SI, Adams WC. 1984. Combined effects of ozone exposure and ambient
heat on exercising females. J Appl Physiol Respir Environ Exerc Physiol
57(2):450-456.
Gilbert SS, van den Heuvel CJ, Ferguson SA, Dawson D. 2004. Thermoregulation
as a sleep signalling system. Sleep Med Rev 8:81-93.
Glazer JL. 2005. Management of heatstroke and heat exhaustion Am Fam
Physician 71(11):2133-2140.
Goodman JE, Zu K, Loftus CT, Tao G, Liu X, Lange S. Ambient ozone and
asthma hospital admissions in Texas: a time-series analysis. Asthma Res
Pract 3:6; doi:10.1186/s40733-017-0034-1.
Gordon CJ, Mohler FS, Watkinson WP, Rezvani AH. 1988. Temperature
regulation in laboratory mammals following acute toxic insult. Toxicology
53:161-178.
Gordon C. 2003. Role of environmental stress in the physiological response to
chemical toxicants. Environ Res 92:1-7.
180 Gosling SN, McGregor GR, Páldy A. 2007. Climate change and heat-related
mortality in six cities Part 1: model construction and validation. Int J
Biometeorol 51:525-540.
Greater Sydney Comission. 2018. Greater Sydney Region Plan. A Metropolis of
Three Cities – Connecting people. NSW Government. Sydney, New South
Wales, Australia.
Green SR, Basu R, Malig B, Broadwin R, Kim JJ, Ostro B. 2010. The effect of
temperature on hospital admissions in nine California counties. Int J
Public Health 55:113-121.
Gronlund CJ, Zanobetti A, Schwartz JD, Wellenius GA, O’Neill MS. 2014. Heat,
heat waves, and hospital admissions among the elderly in the United
States, 1992-2006. Environ Health Perspect 122:1187-1192.
Gronlund CJ, Zanobetti A, Wellenius GA, Schwartz JD, O’Neill MS. 2016.
Vulnerability to renal, heat and respiratory hospitalizations during extreme
heat among U.S elderly. Clim Change 136:631-645.
Gryparis A, Forsberg B, Katsouyanni K, Analitis A, Touloumi G, Schwartz J,
Samoli E, Medina S, Anderson HR, Niciu EM, et al. 2004. Acute effects
of ozone on mortality from the “Air pollution and health: A European
approach” project. Am J Respir Crit Care Med 170:1080–1087.
Guo Y, Barnett AG, Zhang Y, Tong S, Yu W, Pan X. 2010. The short-term effect
of air pollution on cardiovascular mortality in Tianjin, China: Comparison
of time series and case-crossover analyses. Sci Total Environ 409:300-
306.
Guo Y, Gasparrini A, Armstrong B, Li S, Tawatsupa B, Tobias A, Lavigne E, de
Sousa Zanotti Stagliorio Coelho M, Tobias A, Lavigne E, et al. 2014.
181 Global variation in the effects of ambient temperature on mortality: A
systematic evaluation. Epidemiology 25:781-789.
Guo Y, Li S, Liu L, Chen D, Williams G, Tong S. 2016. Projecting future
temperature-related mortality in three largest Australian cities. Environ
Pollut 208(Pt A):66-73.
Guo Y, Gasparrini A, Armstrong BG, Tawatsupa B, Tobias A, Lavigne E, de
Sousa Zanotti Stagliorio Coelho M, Pan X, Kim H, Hashizume M. 2017.
Heat wave and mortality: A multicountry, multicommunity study. Environ
Health Perspect 125(8); doi: 10.1289/EHP1026.
Hajat S, Armstrong B, Baccini M, Biggeri A, Bisanti L, Russo A, Páldy A, Menne
B, Kosatsky T, et al. 2006. Impact of high temperatures on mortality: Is
there an added heat wave effect? Epidemiology 17:632-638.
Hajat S, Kosatky T. 2010. Heat-related mortality: a review and exploration of
heterogeneity. J Epidemiol Community Health 64:753-760.
Hales S, Salmond C, Town GI, Kjellstron T, Woodward A. 2000. Daily mortality
in relation to weather and air pollution in Christchurch, New Zealand.
Aust N Z J Public Health 24(1):89-91.
Hanna EG, Tait PW. 2015. Limitations to thermoregulation and acclimatization
challenge human adaptation to global warming. Int J Environ Res Public
Health 12: 8034-8074.
Hansel NN, McCormack MC, Kim V. 2016. The effects of air pollution and
temperature on COPD. COPD. 13:372-379.
Hansen A, Bi P, Nitschke M, Ryan P, Pisaniello D, Tucker G. 2008a. The effect
of heat waves on mental health in temperate Australian city. Environ
Health Perspect 116:1369-1375.
182 Hansen AL, Bi P, Ryan P, Nitschke M, Pisaniello D, Tucker G. 2008b. The effect
of heat waves on hospital admissions for renal disease in a temperate city
of Australia. Int J Epidemiol 37: 1359-1365.
Hansen A, Bi P, Nitschke M, Pisaniello D, Ryan P, Sullivan T, Barnett AG. 2012.
Particulate air pollution and cardiorespiratory hospital admissions in a
temperate Australian city: A case-crossover analysis. Sci Total Environ
416:48-52.
Hansen A, Bi P, Saniotis A, Nitschke M, Benson J, Tan Y, Smyth V, Wilson L
and Han G-S. 2013. Extreme heat and climate change: Adaptation in
culturally and linguistically diverse (CALD) communities. National
Climate Change Adaptation Research Facility. Gold Coast, Queensland,
Australia.
Hayes Jr D, Collins PB, Khosravi M, Lin RL, Lee LY. 2012. Bronchoconstriction
triggered by breathing hot humid air in patients with asthma. Am J Respir
Crit Care Med 185:1190-1196.
Heaton MJ, Sain SR, Greasby TA, Uejio CK, Hayden MH, Monaghan AJ,
Boehnert J, Sampson K, Banerjee D, Nepal V, et al. 2014. Characterising
urban vulnerability to heat stress using a spatially varying coefficient
model. Spat Spatio-temporal Epidemiol 8:23-33.
Hime N, Cowie C, Marks G. 2015. Review of the health impacts of emission
sources, types and levels of particulate matter air pollution in ambient air in
NSW. Centre for Air Quality and Health Research and Evaluation (CAR)
and Woolcock Institute of Medical Research. Sydney, New South Wales,
Australia.
183 Hou P, Wu S. 2016. Long-term changes in extreme air pollution meteorology and
the implications for air quality. Sci Rep 6:23792; doi: 10.1038/srep23792.
Hughes L, Alexander DJ. 2016. Climate Change and the NSW/ACT bushfire
threat: Update 2016. Climate Council of Australia.
Hunt E. 2017. Sydney air pollution alert issued as temperature heads to 38C.
Available at: https://www.theguardian.com/weather/2017/jan/10/sydney-air-
pollution-alert-issued-due-to-high-ozone-level [Last accessed 25 April
2018].
Huszar P, Rezler-Juda K, Halenka T, Chervenkov H, Syrakov D, Krüger BC,
Zanis P, Melas, D, Katragkou E, Reixer M. 2011. Effects of climate change
on ozone and particulate matter over Central and Eastern Europe. Clim Res
50:51-68.
IPCC. 2012. Managing the Risks of Extreme Events and Disasters to Advance
Climate Change Adaptation. A Special Report of Working Groups I and II
of the Intergovernmental Panel on Climate Change [Field, C.B., V. Barros,
T.F. Stocker, D. Qin, D.J. Dokken, K.L. Ebi, M.D. Mastrandrea, K.J. Mach,
G.-K. Plattner, S.K. Allen, M. Tignor, and P.M. Midgley (eds.)]. Cambridge
University Press, Cambridge, UK, and New York, NY, USA, 582 pp.
IPCC. 2014. Climate Change 2014. Synthesis report. Contribution of Working
Groups I, II, III to the Fifth Assessment Report of the Intergovernmental
Panel on Climate Change [Core Writing Team, R.K. Pachauri and L.A.
Meyer (eds)]. IPCC, Geneva, Switzerland, 151 pp.
Isaksen TB, Yost MG, Hom EK, Ren Y, Lyons H, Fenske RA. 2015. Increased
hospital admissions associated with extreme-heat exposure in King
County, Washington, 1990-2010. Rev Environ Health 30(1):51-64.
184 Jacob DJ, Winner DA. 2009. Effect of climate change on air quality. Atmos
Environ 43:51-63.
Jalaludin B, Khalaj B, Sheppeard V, Morgan G. 2008. Air pollution and ED visits
for asthma in Australian children: a case-crossover analysis. Int Arch
Occup Environ Health. 81:967-974.
Janes H, Sheppard L, Lumley T. 2005. Case-crossover analyses of air pollution
exposure data: Referent selection strategies and their implications for bias.
Epidemiology 16:717–726.
Jegasothy E, McGuire R, Nairn J, Fawcett R, Scalley B. 2017. Extreme climatic
conditions and health service utilisation across rural and metropolitan New
South Wales. Int J Biometerol 61(8):1359-1370.
Jhun I, Fann N, Zanobetti A, Hubbell B. 2014. Effect modification of ozone-
related mortality risks by temperature in 97 US cities. Environ Int 73:128-
134.
Ji M, Cohan DS, Bell ML. 2011. Meta-analysis of the association between short-
term exposure to ambient ozone and respiratory hospital admissions.
Environ Res Lett 6:024006; doi:10.1088/1748-9326/6/2/024006.
Jiang N, Scorgie Y, Hart M, Riley ML, Crawford J, Beggs PJ, Edwards GC,
Chang L, Salter D, Di Virgilio G. 2017. Visualizing the relationship
between synoptic circulation type and air quality in Sydney, a subtropical
coastal-basin environment. Int J Climatol 37:1211-1228.
Johnston F, Hanigan I, Henderson S, Morgan G, Bownan D. 2011. Extreme air
pollution events from bushfires and dust storms and their association with
mortality in Sydney, Australia 1994-2007. Environ Res 11:811-816.
185 Jörres R, Nowak D, Magnussen H. 1996. The effect of ozone exposure on
allergen responsiveness in subjects with asthma or rhinitis. Am J Respir
Crit Care Med 153:56-64.
Junger WL, Ponce de Leon A. 2015. Imputation of missing data in time series for
air pollutants Atmos Environ 102:96-104.
Kahle JJ, Neas LM, Devlin RB, Case MW, Schmitt MT, Madden MC, Diaz-
Sanchez D. 2015. Interaction effects of temperature and ozone on lung
function and markers of systemic inflammation, coagulation, and
fibrinolysis: A caseover study of healthy young volunteers. Environ
Health Persp 123(4):310-316.
Kaiser R, Le Tertre A, Schwartz J, Gotway W, Daley R, Rubin CH. 2007. The
effect of the 1995 heat wave in Chicago on all-cause and cause-specific
mortality. Am J Public Health 97:S158-S162.
Kang SH, Oh IY, Heo J, Lee H, Kim J, Lim WH, Cho Y, Choi EK, Yi SM, Shin
SD, et al. 2016. Heat, heat waves, and out-of-hospital cardiac arrest. Int J
Cardiol 221:232-237.
Karagulian F, Belis CA, Dora CFC, Prüss-Ustün AM, Bonjour S, Adair-Rohani
H, Amann M. 2015. Contributions to cities’ ambient particulate matter
(PM): A systematic review of local source contributions at the global
level. Atmos Environ 120:475-483.
Kenney WL, Munce TA. 2003. Aging and human temperature regulation. J Appl
Physiol 95:2598-2603.
Kent ST, McClure LA, Zaitchik BF, Smith TT, Gohlke JM. 2014. Heat waves and
health outcomes in Alabama (USA): The importance of heat wave
definition. Environ Health Perspect 122:151-158.
186 Keywood MD, Emmerson KM, Hibberd MF. 2016. Ambient air quality: Coarse
particulate matter (PM10). In: Australia state of the environment 2016.
Available: https://soe.environment.gov.au/theme/ambient-air-
quality/topic/2016/coarse-particulate-matter-pm10 [Last accessed 17 April
2018].
Khalaj B, Lloyd G, Sheppeard V, Dear K. 2010. The health impacts of heat waves
in five regions of New South Wales, Australia: a case-only analysis. Int
Arch Occup Environ Health 83:833-842.
Kingsley SL, Eliot MN, Gold J, Vanderslice RR, Wellenius GA. 2016. Current
and projected heat-related morbidity and mortality in Rhode Island.
Environ Health Perspect 124:460-467.
Kirtman B, SB. Power, J.A. Adedoyin, G.J. Boer, R. Bojariu, I. Camilloni, F.J.
Doblas-Reyes, A.M. Fiore, M. Kimoto, G.A. Meehl, M. Prather, A. Sarr,
C. Schär, R. Sutton, G.J. van Oldenborgh, G. Vecchi and H.J. Wang,
2013: Near-term Climate Change: Projections and Predictability. In:
Climate Change 2013: The Physical Science Basis. Contribution of
Working Group I to the Fifth Assessment Report of the Intergovernmental
Panel on Climate Change [Stocker, T.F., D. Qin, G.-K. Plattner, M.
Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M.
Midgley (eds.)]. Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA.
Knowlton K, Rosenthal JE, Hogrefe C, Lynn B, Gaffin S, Goldberg R,
Rosenzweig C, Civerolo K, Ku JY, Kinney PL. 2004. Assessing ozone-
related health impacts under a changing climate. Environ Health Perspect
112:1557-1563.
187 Kovats S, Hajat S, Wilkinson P. 2004. Contrasting patterns of mortality and
hospital admissions during hot weather and heat waves in Greater London,
UK. Occup Environ Med 61:893-898.
Kreit JW, Gross KB, Moore TB, Lorenzen TJ, D’Arcy J, Eschenbacher WL.
1989. Ozone-induced changes in pulmonary function and bronchial
responsiveness in asthmatics. J Appl Physiol 66:217-222.
Kunkel KE, Changnon SA, Reinke BC, Arritt RW. 1996. The July 1995 heat
wave in the Midwest: A climatic perspective and critical weather factors.
Bull Am Meteorol Soc 77(7): 1507-1518.
Laaidi K, Abdelkrim Z, Doussett B, Bretin P, Vandentorren S, Giraudet E,
Beaudeau P. 2012. The impact of heat islands on mortality in Paris during
the August 2003 heat wave. Environ Health Perspect 120:254-259.
Lacressonnière G, Watson L, Gauss M, Engardt M, Andersson C, Beeknamm M,
Colette A, Foret G, Josse B, Marcècal V, et al. 2017. Particulate matter air
pollution in Europe in a +2 °C warming world. Atmos Environ 154:129-
140.
Lan L, Pan L, Lian Z, Huang H, Lin Y. 2014. Experimental study on thermal
comfort of sleep people at different air temperatures. Build Environ 73:24-
31.
Langner J, Engardt M, Baklanov A, Christensen JH, Gauss M, Geels C,
Hedegaard GB, Nuterman R, Simpson J, Soares J, Sofiev M, et al. 2012. A
multi-model study of impacts of climate change on surface ozone in
Europe. Atmos Chem Phys 12:10423-10440.
188 Lam YF, Fu JS, Wu S, Mickley LJ. 2011. Impacts of future climate change and
effects of biogenic emissions on surface ozone and particulate matter
concentrations in the United States. Atmos Chem Phys 11:4789-4806.
Leech JA and Smith-Doiron M. 2006. Exposure time and place: Do COPD
patients differ from the general population? J Expo Sci Environ Epidemiol
16: 238-241.
Lewis SC, Karoly DJ. 2013. Anthropogenic contributions to Australia’s record
summer temperatures of 2013. Geophys Res Lett 40:3705-3709.
Lewis SC, Karoly DJ. 2014. The role of anthropogenic forcing in the record 2013
Australia-wide annual and spring temperatures. [in “Explaining Extremes
of 2013 from a Climate Perspective”]. Bull Amer Meteor Soc 95(9): S31-
S34.
Lewis SC, King AD. 2015. Dramatically increased rate of observed hot record
breaking in recent Australian temperatures. Geophys Res Lett 42:7776-
7784.
Lewis SC, King AD, Mitchell DM. 2017. Australia’s unprecedented future
temperature extremes under Paris limits to warming. Geophys Res Lett
44:9947-9956.
Li G, Zhou M, Cai Y, Zhang Y, Pan X. 2011. Does temperature enhance acute
mortality effects of ambient particle pollution in Tianjin City, China. Sci
Total Environ 409:1811-1817.
Li H, Wu S, Pan L, Xu J, Shan J, Yang X, Dong W, Deng F, Chen Y, Shima M,
Guo X. 2018. Short-term effects of various ozone metrics on
cardiopulmonary function in chronic obstructive pulmonary disease
189 patients: Results from a panel study in Beijing, China. Environ Pollut
232:358-366.
Li L, Yang J, Guo C, Chen PY, Ou CQ, Guo Y. 2015. Particulate matter modifies
the magnitude and time course of the non-linear temperature-mortality
association. Environ Pollut 196:423-430.
Libert JP, Di Nisi J, Fukuda H, Muzet A, Ehrhart J and Amoros C. 1988. Effect
of continuous heat exposure on sleep stages in humans. Sleep 11(2):
195-209.
Lin S, Luo M, Walker RJ, Liu X, Hwang S, Chinery R. 2009. Extreme high
temperatures and hospital admissions for respiratory and cardiovascular
diseases. Epidemiology 20:738-746.
Lin S, Hsu WH, Van Zutphen AR, Saha S, Luber G, Hwang S. 2012. Excessive
heat and respiratory hospitalizations in New York State: Estimating
current and future public health burden related to climate change. Environ
Health Perspect 120:1571-1577.
Linfoot S, Johnson M, Young M, Angri L, Quigley S, Spencer J, Duc H, Trieu T,
Xu C. 2010. State of Knowledge: Ozone. Department of Environment,
Climate Change and Water. Sydney, New South Wales, Australia.
Liu JC, Mickley LJ, Sulprizio MP, Dominici F, Yue X, Ebisu K, Anderson GB,
Khan RFA, Bravo MA, Bell ML. 2016. Particulate air pollution from
wildfires in the Western US under climate change. Clim Change 138:655-
666.
Lu Y and Zeger SL. 2007. On the equivalence of case-crossover and time series
methods in environmental epidemiology. Biostatistics 8:337-334.
190 Ma W, Xu X, Peng L, Kan H. 2011. Impact of extreme temperature on hospital
admission in Shanghai, China. Sci Total Environ 409:3634-3637.
Maclure M. 1991. The case-crossover design: a method for studying transient e
ffects on the risk of acute events. Am J Epidemiol 133:144-153.
Madaniyazi L, Zhou Y, Li S, Williams G, Jaakkola J J.K., Liang X, Liu Y, Wu S,
Guo Y. 2016. Outdoor temperature, heart rate, and blood pressure in
Chinese adults: Effect modification by individual characteristics. Sci Rep
6:21003; doi: doi:10.1038/srep21003.
McCormarck MC, Belli AJ, Waugh D, Mastsui EC, Peng RD, Williams DL,
Paulin L, Saha A, Aloe CM, Diette GB, et al. Respiratory effects of indoor
heat and the interaction with air pollution in chronic obstructive
pulmonary disease. Annals ATS 13:2125-2131.
McInnes W. 2018. Health warnings issued as Sydney faces heat wave. Available:
https://www.smh.com.au/national/nsw/health-warnings-issued-as-sydney-
faces-heatwave-20180106-h0ecky.html [Last accessed 16 April 2018].
Medina- Ramón M, Zanobetti A, Schwartz J. 2006. The effect of ozone and PM10
on hospital admissions for pneumonia and chronic obstructive pulmonary
disease: A national multicity study. Am J Epidemiol 163:579-588.
Meleux F, Solmon F, Giorgi F. 2007. Increase in summer European ozone
amounts due to climate change. Atmos Environ 41:7577-7587.
Meng X, Zhang Y, Zhao Z, Duan X, Xu X, Kan H. 2012. Temperature modifies
the acute effect of particulate air pollution on mortality in eight Chinese
cities. Sci Total Environ 435-436:215-221.
191 Michelozzi P, Accetta G, De Sario M, D’Ippoliti D, Marino C, Baccini M, Biggeri
A, Anderson HR, Katsouyanni K, Ballester F, Cadum E, et al. 2009. High
temperature and hospitalizations for cardiovascular and respiratory causes
in 12 European cites. Am J Respir Crit Care Med 179:383-389.
Mickley LJ, Jacob DJ, Field BD, Rind D. 2004. Effects of future climate change
on regional air pollution episodes in the United States. Geophys Res Lett
31:L24103; doi: 10.1029/2004GL021216.
Middleton N. 2003. Going to extremes: mud, sweat and frozen tears. Pan Books.
United Kingdom.
Murazaki K, Hess P. 2006. How does climate change contribute to surface ozone
change over the United States? Geophys Res Lett 111:D05301; doi:
10.1029/2005JD005873.
Nelin TD, Joseph AM, Gorr MW, Wold LE. 2012. Direct and indirect effects of
particulate matter on the cardiovascular system. Toxico Lett 208(3):293-
299.
NEPM (National Environment Protection (Ambient Air Quality) Measure).
Available: https://www.legislation.gov.au/Details/C2004H03935 [Last
accessed 23 April 2018].
Nichols M, Peterson K, Alston L, Allender S. 2014. Australian heart disease
statistics 2014. National Heart Foundation of Australia.
Nitschke M, Tucker GR, Bi P. 2007. Morbidity and mortality during heatwaves in
metropolitan Adelaide. Med J Aust 187:662-665.
Nitschke M, Tucker G. 2009. The unfolding story of heat waves in metropolitan
Adelaide. SA Health, Government of South Australia. Australia.
192 Nitschke M, Tucker GR, Hansen AL, Williams S, Zhang Y, Bi P. 2011. Impact of
two recent extreme heat episodes on morbidity and mortality in Adelaide,
South Australia: a case-series analysis. Environ Health 10:42;
https://doi.org/10.1186/1476-069X-10-42.
NOAA (National Oceanic and Atmospheric Administration). 2017. State of the
Climate: Global Climate Report for Annual 2016. Available:
https://www.ncdc.noaa.gov/sotc/global/201613 [Last accessed 14 April
2018].
NOAA (National Oceanic and Atmospheric Administration). 2018a. Climate at a
glance: Global time series.
Available:
https://www.ncdc.noaa.gov/cag/global/timeseries/globe/land/ytd/12/1880-
2016?trend=true&trend_base=10&firsttrendyear=1880&lasttrendyear=201
7 [Last accessed 15 April 2018].
NOAA (National Oceanic and Atmospheric Administration). 2018b. Climate at a
glance: Global time series. Available:
https://www.ncdc.noaa.gov/cag/global/time-
series/globe/ocean/ytd/12/1880-
2016?trend=true&trend_base=10&firsttrendyear=1880&lasttrendyear=201
7 [Last accessed 29 April 2018].
NSW Environment Protection Authority. 2013. Managing particles and
improving air quality in NSW. NSW Environment Protection
Authority. Sydney, New South Wales, Australia.
193 NSW Environment Protection Authority. 2015. New South Wales State of the
Environment. NSW Environment Protection Authority. Sydney, New
South Wales, Australia.
NSW Government Architect. 2017. Greener Places. NSW Government. Sydney,
New South Wales, Australia.
NSW Health. 2016. Beat the Heat. Available:
http://www.health.nsw.gov.au/environment/beattheheat/Pages/default.aspx
[Last accessed 25 April 2018].
NSW Health. 2017. Physical activity in adults. HealthStats NSW. Available:
http://www.healthstats.nsw.gov.au/Indicator/beh_phys_age/beh_phys_age
_snap [Last accessed 2 February 2018].
NSW Health. 2018. Air quality index: AQI. Available:
http://www.health.nsw.gov.au/environment/air/Pages/aqi.aspx [Last
accessed at 25 April 2018].
NSW Office of Environment and Heritage. 2014. TP02: Air Quality Trends in
Sydney. NSW Government. Sydney, New South Wales, Australia.
NSW Office of Environment and Heritage. 2015a. Metropolitan Sydney: Climate
change snapshot. NSW Government. Sydney, New South Wales,
Australia.
NSW Office of Environment and Heritage. 2015b. Quality assurance for the air
quality monitoring network.
Available: http://www.environment.nsw.gov.au/topics/air/understanding-
air-quality-data/data-validation [Last accessed 19 April 2018].
194 NSW Office of Environment and Heritage. 2017. Standards and goals for
measuring air pollution.
Available: http://www.environment.nsw.gov.au/topics/air/understanding-
air-quality-data/standards-and-goals [Last accessed 20 October 2017].
NSW Office of Environment and Heritage. 2018a. Air quality monitoring
network.
Available: http://www.environment.nsw.gov.au/topics/air/monitoring-air-
quality [Last accessed 16 April 2018].
NSW Office of Environment and Heritage. 2018b. Current and forcast air quality.
Available: http://www.environment.nsw.gov.au/aqms/aqi.htm [Last
accessed at 25 April 2018].
Ogbomo AS, Gronlund CJ, O’Neill MS, Konen T, Cameron L, Wahl R. 2017.
Vulnerability to extreme-heat-associated hospitalization in three counties
in Michigan, USA, 2000-2009. Int J Biometeorol 61:833-843.
Okamoto-Mizuno K, Tsuzuki K, Mizuno K. 2010. Effects of mild heat exposure
on sleep stages and body temperature in older men. Int J Biometeorol
49:32-36.
Olivier JGJ, Schure KM, Peters JAHW. 2017. Trends in global CO2 and total
greenhouse gas emissions: 2017 Report. PBL Netherlands Environmental
Assessment Agency. The Hague, Netherlands.
O’Neill MS, Hajat S, Zanobetti A, Ramirez-Aguilar M, Schwartz J. 2005. Impact
of control for air pollution and respiratory epidemics on the estimated
associations of temperature and daily mortality. Int J Biometeorol 50:121-
129.
195 Orlowsky B, Seneviratne SI. 2012. Global changes in extreme events: regional
and seasonal dimension. Clim Change 110:669-696.
Orru H, Andersson C, Ebi KL, Langner J, Åström C, Forsberg B. 2013. Impact of
climate change on ozone-related mortality and morbidity in Europe. Eur
Respir J 41:285-294.
Ostro B, Rauch S, Green R, Malig B, Basu R. 2010. The effects of temperature
and use of air conditioning on hospitalizations. Am J Epidemiol 172:1053-
1061.
Otero N, Sillmann J, Schnell JL, Rust HW, Butler T. 2016. Synoptic and
meteorological drivers of extreme ozone concentrations over Europe.
Environ Res Lett 11:024005; doi:10.1088/1748-9326/11/2/024005.
Paris Agreement, opened for signature 22 April 2016, UN DOC
FCCC/CP/2015/10 (entered into force 4 November 2016). Available:
https://unfccc.int/process/the-paris-agreement/what-is-the-paris-agreement
[Last accessed 14 April 2018].
Parmeggiani PL. 1987. Interaction between sleep and thermoregulation: An aspect
of the control of behavioral states. Sleep 10(5):426-435.
Pattenden S, Armstrong B, Milojevic A, Heal MR, Chalabi Z, Doherty R, Barratt
B, Kovats RS, Wilkinson P. 2010. Ozone, heat and mortality: acute effects
in 15 British conurbations. Occup Environ Med 67:699-707.
Pearce JL, Beringer J, Nicholls N, Hyndman RJ, Tapper NJ. 2011. Quantifying
the influence of local meteorology on air quality using generalised
additive models. Atmos Environ 45:1328-1336.
196 Peng RD, Chang HH, Bell ML, McDermott A, Zeger SL, Samet JM, Dominici F.
2008. Coarse particulate matter air pollution and hospital admissions for
cardiovascular disease and respiratory diseases among Medicare patients.
JAMA 299(18):2172-2179.
Peng RD, Bobb JF, Telbaldi C, McDaniel L, Bell ML, Dominici F. 2011. Toward
a quantitative estimate of future heat wave mortality under global climate
change. Environ Health Perspect 119(5):701-706.
Peng RD, Samoli E, Pham L, Dominici F, Touloumi G, Ramsay T, Burnett RT,
Krewski D, Le Tertre A, Cohen A, et al. 2013. Acute effects of ambient
ozone on mortality in Europe and North America: results from the
APHENA study Air Qual Atmos Health 6:445-453.
Perkins SE, Alexander LV, Nairn JR. 2012. Increasing frequency, intensity and
duration of observed global heatwaves and warm spells. Geophys Res Lett
39: L20714; doi:10.1029/2012GL053361.
Perkins SE, Alexander LV. 2013. On the measurement of heat waves. J Clim
26:4500-4516.
Perkins SE, Lewis SC, King AD, Alexander LV. 2014. Increased simulated risk
of the hot Australian summer of 2012/13 due to anthropogenic activity as
measured by heat wave frequency and intensity. [in “Explaining Extremes
of 2013 from a Climate Perspective”]. Bull Amer Meteor Soc 95(9): S34-
S37.
Petroeschevsky A, Simpson RW, Thalib L, Rutherford S. 2001. Associations
between outdoor air pollution and hospital admissions in Brisbane,
Australia. Archives of Environment Heath: An International Journal.
56(1): 37-52.
197 Phung D, Thai PK, Guo Y, Morawska L, Rutherford S, Chu C. 2016. Ambient
temperature and risk of cardiovascular hospitalization: An updated
systematic review and meta-analysis. Sci Total Environ 550:1084-1102.
Physick W, Cope M, Lee S. 2014. The impact of climate change on ozone-related
mortality in Sydney. Int J Environ Res Public Health 11:1034-1048; doi:
doi:10.3390/ijerph110101034.
Pirozzi C, Sturrock A, Weng HY, Greene T, Scholand MB, Kanner R, Paine III,
R. 2015. Effect of naturally occurring ozone air pollution episodes on
pulmonary oxidative stress and inflammation. Int J Environ Res Public
Health 12:5061-5075.
Poumadére M, Mays C, Le Mer S, Blong R. 2005. The 2003 heat wave in France:
Dangerous climate change here and now. Risk Anal 25(6):1483-1494.
Pricewaterhouse Coopers (PwC). 2011. Protecting human health and safety during
severe and extreme heat events. A national framework. Department of
Climate Change and Energy and Efficiency. Commonwealth Government,
Australia.
Qian Z, He Q, Lin HM, Kong L, Bentley CM, Liu W, Zhou D. 2008. High
temperatures enhanced acute mortality effects of ambient particle pollution
in the ‘oven’ city of Wuhan, China. Environ Health Perspect 116:1172-
1178.
Qiu H, Sun S, Tang R, Chan KP, Tian L. 2016. Pneumonia hospitalization risk in
the elderly attributable to cold and hot temperatures in Hong Kong, China.
Am J Epidemiol 184(8):570-578.
198 Qiu H, Yu IT, Wang X, Tian L, Ah Tse L, Wai Wong T. 2013. Cool and dry
weather enhances the effects of air pollution on emergency IHD hospital
admissions. Int J Cardiol 168:500-505.
Queensland Floods Commission of Inquiry. 2012. Final Report. Queensland
Government. Brisbane, Queensland, Australia.
Raftery AE, Zimmer A, Frierson DMW, Startz R, Liu P. 2017. Less than 2°C
warming by 2100 unlikely. Nat Clim Change 7:637-641.
Reid CE, Snowden JM, Kontgis C, Tager IB. 2012. The role of ambient ozone in
epidemiologic studies of heat-related mortality. Environ Health Perspect
120(12): 1627-1630.
Ren C, Tong S. 2006. Temperature modifies the health effects of particulate
matter in Brisbane, Australia. Int J Biometerol 51:87-96.
Ren C, Williams GM, Tong S. 2006. Does particulate matter modify the
association between temperature and cardiorespiratory diseases? Environ
Health Perspect 114:1690-1696.
Ren C, Williams GM, Morawska L, Mengersen K, Tong S. 2007. Ozone modifies
associations between temperature and cardiovascular mortality: analysis of
the NMMAPS data. Occup Environ Med 65:225-260.
Ren C, Williams GM, Mengersen K, Morawska L. 2009. Temperature enhanced
effects of ozone on cardiovascular mortality in 95 large US communities,
1987-2000: Assessment using the NMMAPS data. Arch Environ Occup
Health 64:177-184.
Ren C, O’Neill MS, Park SK, Sparrow D, Vokonas P, Schwartz J. 2011. Ambient
temperature, air pollution, and heart rate variability in an aging population.
Am J Epidemiol 173:1013-1021.
199 Robine JM, Cheung SLK, Le Roy S, Van Oyen H, Griffiths C, Michel JP,
Herrmann FR. 2003. C.R. Biologies. 331:171-178.
Rocklöv J, Barnett AG, Woodward A. 2012. On the estimation of heat-intensity
and heat-duration effects in time series models of temperature-related
mortality in Stockholm, Sweden. Environ Health 11:23;
https://doi.org/10.1186/1476-069X-11-23.
Royé D. 2017. The effects of hot nights on mortality in Barcelona, Spain. Int J
Biometerol 61:2127-2140.
Sacks JD, Stanek LW, Luben TJ, O. Johns D, Buckley BJ, Brown JS, Ross M.
2011. Particulate matter induced health effects: Who is susceptible?
Environ Health Perspect 119:446-454.
Salem SN. 1968. Heat-stress and renal failure. Lancet 872.
Samet JM, Dominici F, Curriero FC, Coursac I, Zeger SL. 2000. Fine particulate
air pollution and mortality in 20 U.S. cities, 1987-1994. N Engl J Med
343:1742-1749.
Sánchez-Lugo A, Morice C, Berrisford P, Argüez A. 2017. Global surface
temperatures [in ‘State of the Climate in 2016’]. Bull Amer Meteor Soc
98(8):S11-S13.
Sanford T, Frumhoff PC, Luers A, Gulledge J. 2014. The climate policy narrative
for a dangerously warming world. Nat Clim Change 4:164-166.
Schnell JL, Prather MJ, Josse B, Naik V, Horowitz LW, Zeng G, Shindell DT,
Faluvegi G. 2016. Effect of climate change in surface ozone over North
America, Europe and East Asia. Geophys Res Lett 43:3509-3518.
200 Schnell JL, Prather MJ. 2017. Co-occurrence of extremes in surface ozone,
particulate matter, and temperature over eastern North America. PNAS
114(11):2854-2859.
Schwartz J. 1994. Air pollution and hospital admissions for the elderly in Detroit,
Michigan. Am J Respir Crit Care Med 150:648-655.
Schwartz J, Morris R. 1995. Air pollution and hospital admissions for
cardiovascular disease in Detroit, Michigan. Am J Epidemiol 142:23-35.
Semenza JC, McCullough JE, Dana Flanders W, McGeehin MA, Lumpkin J.R.
1999. Excess hospital admissions during the July 1995 heat wave in
Chicago. Am J Prev Med 16(4):269-277.
Seneviratne SI, Donat MG, Pitman AJ, Knutti R, Wilby RL. 2016. Allowable CO2
emissions based on regional and impact-related climate targets. Nature
529:477-483.
Shah ASV, Langrish JP, Nair H, McAllister DA, Hunter AL, Donaldson K,
Newby DE, Mills NL. 2013. Global association of air pollution and heart
failure: a systematic review and meta-analysis. The Lancet 382:1039-
1048.
Shaposhnikov D, Revich B, Bellander T, Bero Bedada G, Bottai M, Kharkova T,
Kvasha E, Lezina E, Lind T, Semutnikova E. 2014. Epidemiology 25:359-
364.
Sheffield PE, Knowlton K, Carr JL, Kinney PL. 2011. Modeling of regional
climate change effects on ground-level ozone and childhood asthma. Am J
Prev Med 41(3):251-257.
201 Shen L, Mickley LJ, Murray LT. 2017. Influence of 2000-2050 climate change on
particulate matter in the United States: results from a new statistical
model. Atmos Chem Phys. 17:4355-4367.
Sherbakov T, Malig B, Guirguis K, Gershunov A, Basu R. 2017. Ambient
temperature and added heat wave effects on hospitalizations in California
form 1999-2009. Environ Res 160: 83-90.
Sherwood SC, Huber M. 2010. An adaptability limit to climate change due to heat
stress. PNAS 107(21):9552-9555.
Sillman S, Samson PJ. 1995. Impact of temperature on oxidant photochemistry in
urban, polluted rural and remote environments. J Geophys Res 100:11497-
11507.
Sillmann J, Kharin VV, Zwiers FW, Zhang X, Bronaugh D. 2013. Climate
extremes indices in the CMIP5 multimodel ensemble: Part 2. Future
climate projections. Geophys Res Lett 118:2473-2493.
Silva RA, West JJ, Lamarque JF, Shindell DT, Collins W, Faluvegi G, Folberth
GA, Horowitz LW, Nagashima T, Naik V, et al. 2017. Future global
mortality from changes in air pollution attributable to climate change. Nat
Clim Change 7:647-651.
Simoni M, Baldacci S, Maio S, Cerrai S, Sarno G, Viegi G. 2015. Adverse effects
on outdoor air pollution in the elderly. J Thorac Dis 7:34-45.
Solberg S, Hov O, Søvde A, Isaken ISA, Coddeville P, De Backer H, Forster C,
Orsolini Y, Uhse K. 2008. European surface ozone in the extreme summer
2003. J Geophys Res 113:D07307; doi:10.1029/2007JD009098.
202 Son JY, Lee JT, Anderson GB, Bell ML. 2012. The impact of heat waves on
mortality in severn major cities in Korea. Environ Health Perspect
120:566-571.
Son JY, Gouveia N, Bravo MA, de Freitas CU, Bell ML. 2016. The impact of
temperature on mortality in a sub-tropical city: effects of cold, heat, and
heat waves in São Paulo, Brazil. Int J Biometeorol 60:113-121.
Soneja S, Jiang C, Fisher J, Upperman CR, Mitchell C, Sapkota A. 2016.
Exposure to extreme heat and precipitation events associated with
increased risk of hospitalization for asthma in Maryland, U.S.A. Environ
Health 15:57;doi:10.1186/s12940-016-0142-z.
Stafoggia M, Schwartz J, Forastiere F, Perucci CA, the SISTI Group. 2008. Does
temperature modify the association between air pollution and mortality? A
multicity case-crossover analysis in Italy. Am J Epidemiol 167:1476-1485.
Stafoggia M, Samoli E, Alessandrini E, Cadum E, Ostro B, Berti G, Faustini A,
Jacquemin B, Linares C, Pascal M, et al. 2013. Short-term associations
between fine and coarse particulate matter and hospitalizations in Southern
Europe: Results from the MED-PARTICLES Project. Environ Health
Perspect 121:1026-1033.
Stedman J. 2004. The predicted number of air pollution related deaths in the UK
during the August 2003 heatwave. Atmos Environ 38:1087-1090.
Steffen W, Hughes L, Perkins S. 2014. Heatwaves: Hotter, Longer, More Often.
The Climate Council of Australia. Australia.
203 Tagaris E, Manomaiphiboon K, Liao KJ, Leung R, Woo JH, He S, Amar P,
Russell AG. 2007. Impacts of global climate change and emissions on
regional ozone and fine particulate matter concentrations over the United
States. J Geophys Res 112:D14312; doi:10.1029/2006JD008262.
Tai APK, Mickley LJ, Jacob DJ. 2010. Correlations between fine particulate
matter (PM2.5) and meteorological variables in the United States:
Implications for the sensitivity of PM2.5 to climate change. Atmos Environ
44:3976-3984.
Tan W, Herzlich BC, Funaro R, Konstantinos K, Pagala M, Amaladevi B, et al.
1995. Rhabdomyolysis and Myoglobinuric acute renal failure associated
with classic heat stroke. South Med J 88:1065-1068.
Tham R, Erbas B, Akram M, Dennekamp M, Abramson MJ. 2009. The impact of
smoke on respiratory hospital outcomes during the 2002-2003 bushfire
season, Victoria, Australia. Respirology 14:69-75.
Tong S, Ren C, Becker N. 2010a. Excess deaths during the 2004 heat wave in
Brisbane, Australia. Int J Biometeorol 54:393-400.
Tong S, Wang XY, Barnett AG. 2010b. Assessment of heat-related health impacts
in Brisbane, Australia: Comparison of different heat wave definitions.
PLoS One 5: e12155; doi:10.1371/journal.pone.0012155.
Tong S, Wang XY, Guo Y. 2012. Assessing the short-term effects of heatwaves
on mortality and morbidity in Brisbane, Australia: Comparison of case-
crossover and time series analyses. PLoS One 7:e37500; doi:
10.1371/journal.pone.0037500.
204 Tong S, Wang XY, FitzGerald G, McRae D, Neville G, Tippett V, Aitken P,
Verrall K. 2014. Development of health risk-based metrics for defining a
heat wave: a time series study in Brisbane, Australia. BMC Public Health
14:435; doi:10.1186/1471-2458-14-435.
Tsuji B, Hayashi K, Kondo N, Nishiyasu T. 2016. Characteristics of
hyperthermia-induced hyperventilation in humans. Temperature 3:146-
160.
Turner LR, Barnett AG, Connell D, Tong S. 2012. Ambient temperature and
cardiorespiratory morbidity: A systematic review and meta-analysis.
Epidemiology 23:594-606.
US EPA (United States Environment Protection Authority). 2016. Particulate
Matter (PM) Pollution. Available: https://www.epa.gov/pm-
pollution/particulate-matter-pm-basics#PM. [Last accessed 6 September
2017].
US EPA (United States Environment Protection Authority). 2017. Ozone
Pollution. Available: https://www.epa.gov/ozone-pollution [Last accessed
26 February 2018].
Vaneckova P, Bambrick H. 2013. Cause-specific hospital admissions on hot days
in Sydney, Australia. PLoS One 8:e55459; doi:10.1371/
journal.pone.0055459.
Vardoulakis S, Dear K, Hajat S, Eggen B, McMichael AJ. 2014. Comparative
assessment of the effects of climate change on heat-and cold-related
mortality in the United Kingdom and Australia. Enviorn Health Persp
122(12): 1285-1292.
205 Varotsos KV, Tombrou M, Giannakopoulous C. 2013. Statistical estimations of
the number of future ozone exceedances due to climate change in Europe.
J Geophys Res Atmos. 118:6080-6099.
Vertel RM, Knochel JP. 1967. Acute renal failure due to heat injury. Am J Med
43:435-451.
Victorian Bushfires Royal Commission. 2010. Final report summary. Parliament
of Victoria. Melbourne, Victoria, Australia.
Victorian Government Department of Human Services. 2009. January 2009
heatwave in Victoria: An assessment of health impacts. State Government
of Victoria, Melbourne, Victoria, Australia.
Watson L, Lacressonniè G, Gauss M, Engardt M, Andersson C, Josse B, Marécal
V, Nyiri A, Sobolowski S, Siour G, et al. 2016. Impact of emissions and
+2 °C climate change upon future ozone and nitrogen dioxide over
Europe. Atmos Environ 142:271:285.
Weinberger KR, Haykin L, Eliot MN, Schwartz JD, Gasparrini A, Wellenius GA.
2017. Projected temperature-related deaths in ten large U.S metropolitan
areas under different climate change scenarios. Environ Int 107:196-204.
West JJ, Szopa S, Hauglustaine DA. 2007. Human mortality effects of future
concentrations of tropospheric ozone. C R Geosci 339:775-783.
Williams S, Nitschke M, Weinstein P, Pisaniello DL, Parton KA, Bi P. 2012. The
impact of summer temperatures and heatwaves on mortality and morbidty
in Perth, Australia 1994-2008. Environ Int 40:33-38.
206 Wilson LA, Morgan GG, Hanigan IV, Johnston FH, Hisham AR, Broome R,
Gaskin C, Jalaludin B. 2013. The impact of heat on mortality and
morbidity in the Greater Metropolitan Sydney Region: a case crossover
analysis. Environ Health 12:98; doi: 10.1186/1476-069X-12-98.
World Health Organization. 2016. Concentration of particulate matter with an
aerodynamic diameter of 10 m or less (PM10) in nearly 3000 urban
areas, 2008-2015.
Available:
http://gamapserver.who.int/mapLibrary/Files/Maps/Global_pm10_cities.p
ng?ua=1 [Last accessed 2 December 2017].
World Health Organization. 2017. Cardiovascular diseases (CVDs). Available:
http://www.who.int/mediacentre/factsheets/fs317/en/ [Last accessed 2
February 2018].
World Meteorological Organization. 2017. WMO Greenhouse Gas Bulletin. The
State of Greenhouse Gases in the Atmosphere Based on Global
Observations through 2016. No. 13. Available:
https://public.wmo.int/en/resources/library/wmo-greenhouse-gas-bulletin
[Last accessed 14 April 2018].
Wu J, Zhou Y, Gao Y, Fu JS, Johnson BA, Huang C, Kim YM, Liu Y. 2014.
Estimation and uncertainty analysis of impacts of future heat waves on
mortality in the Eastern United States. Environ Health Perspect 122(1):10-
16.
Wu S, Mickley LJ, Leibensperger EM, Jacob DJ, Rind D, Streets DG. 2008.
Effects of 2000-2050 global change on ozone air quality in the United
States. J Geophys Res 113:D06302; doi:10.1029/2007JD008917.
207 Xiang J, Bi P, Pisaniello D, Hansen A. 2014. The impact of heatwaves on
workers’ health and safety in Adelaide, South Australia. Environ Res
133:90-95.
Xiao J, Spicer T, Jian L, Yun GY, Shao C, Nairn J, Fawcett RJB, Robertson A,
Weeramanthri TS. 2017. Variation in population vulnerability to heat
wave in Western Australia. Front Public Health 5:64; doi:
10.3389/fpubh.2017.00064.
Xu A, Mu Z, Jiang B, Wang W, Yu H, Zhang L, Li J. 2017. Acute effects of
particulate air pollution in ischemic heart disease hospitalizations in
Shanghai, China. Int J Environ Res Public Health 14:168;
doi:10.3390/ijerph14020168.
Xu Z, Huang C, Hu W, Turner LR, Su H, Tong S. 2013. Extreme temperatures
and emergency department admissions for childhood asthma in Brisbane,
Australia. Occup Environ Med 70:730-735.
Xu Z, FitzGerald G, Guo Y, Jalaudin B, Tong S. 2016. Impact of heatwave on
mortality under different heatwave definitions: A systematic review and
meta-anlysis. Environ Int 89-90:193-203.
Yu W, Mengersen K, Hu W, Guo Y, Pan X, Tong S. 2011. Assessing the
relationship between global warming and mortality: Lag effects of
temperature fluctuations by age and mortality categories. Environ Pollut
159:1789-1793.
Zanobetti A, Peters A. 2015. Disentangling interactions between atmospheric
pollution and weather. J Epidemiol Community Health 69:613-615.
208 Zeng W, Lao X, Rutherford S, Xu Y, Xu X, Lin H, Liu T, Luo Y, Xiao J, Hu M,
Chu C, Ma X. 2014. The effect of heat waves on mortality and effect
modifiers in four communities of Guangdong Province, China. Sci Total
Environ 482-483:214-221.
Zhang X, Yang F. 2004. RClimDex(1.0) User Manual. Climate Research Branch,
Environment Canada. Downsview, Ontario, Canada.
Zhang Y, Nitschke M, Bi P. 2013. Risk factors for direct heat-related
hospitalization during the 2009 Adelaide heat wave: A case-crossover
study. Sci Total Environ 442:1-5.
Zhang Y, Nitschke M, Krackowizer A, Dear K, Pisaniello D, Weinstein P, Tucker
G, Shakib S, Bi P. 2017. Risk factors for deaths during the 2009 heat wave
in Adelaide, Australia: a matched case-control study. Int J Biometeorol
61:35-47.
Zhao R, Chen S, Wang W, Huang J, Wang K, Liu L, Wei S. 2017. The impact of
short-term exposure to air pollutants on the onset of out-of-hospital
cardiac arrest: A systematic review and meta-analysis. Int J Cardiol
226:110-117.
209 Appendix
Table A1 Temperature thresholds calculated for the warm season in the SSD,
2001 to 2013.
City-wide daily City-wide daily City-wide daily Threshold average maximum average mean average minimum (percentile) temperature temperature temperature (°C) (°C) (°C) 90th 32.13 25.26 19.55 95th 34.42 26.53 20.28 97th 35.39 27.41 20.75 98th 36.35 28.18 21.20 99th 38.13 29.27 21.87
210 Table A2 Fifteen definitions used to define single days of extreme heat.
Variable Definition Extremes Characteristics
Average number of Average intensity Threshold Definition Number Metric Total number of days extreme days per of extreme days (percentile) season (°C) 1 Maximum ≥90th 182 15.17 34.86 2 Maximum ≥95th 91 7.58 36.52 3 Maximum ≥97th 55 4.58 37.56 4 Maximum ≥98th 37 3.08 38.42 5 Maximum ≥99th 19 1.58 39.72 6 Mean ≥90th 182 15.17 26.96 7 Mean ≥95th 91 7.58 28.10 8 Mean ≥97th 55 4.58 28.88 9 Mean ≥98th 37 3.08 29.47 10 Mean ≥99th 19 1.58 30.28 11 Minimum ≥90th 182 15.17 20.59 12 Minimum ≥95th 91 7.58 21.26 13 Minimum ≥97th 55 4.58 21.74 14 Minimum ≥98th 37 3.08 22.10 15 Minimum ≥99th 19 1.58 22.70
211 Table A3 Definitions used for heat waves with maximum temperature as the metric.
Variable Heat Wave Definition Heat Wave Characteristics
Average Average Average Duration Total number Total number number of intensity of Definition Threshold duration of Metric (consecutive of heat wave of heat wave heat wave heat wave number (percentile) heat wave days) days events days per days (days) season (°C) 1 Maximum ≥90th ≥2 98 38 8.17 2.58 35.19 2 Maximum ≥95th ≥2 41 16 3.42 2.56 36.45 3 Maximum ≥97th ≥2 15 7 1.25 2.14 37.62 4 Maximum ≥98th ≥2 10 5 0.83 2.00 38.06 5 Maximum ≥99th ≥2 2 1 0.17 2.00 38.72 6 Maximum ≥90th ≥3 44 11 3.67 4.00 35.45 7 Maximum ≥95th ≥3 19 5 1.58 3.80 36.71 8 Maximum ≥97th ≥3 3 1 0.25 3.00 36.44 N/A Maximum ≥98th ≥3 0 0 0 0.00 N/A N/A Maximum ≥99th ≥3 0 0 0 0.00 N/A 9 Maximum ≥90th ≥4 26 5 2.17 5.20 35.38 10 Maximum ≥95th ≥4 10 2 0.83 5.00 36.66 N/A Maximum ≥97th ≥4 0 0 0 0.00 N/A N/A Maximum ≥98th ≥4 0 0 0 0.00 N/A N/A Maximum ≥99th ≥4 0 0 N/A N/A N/A a The average intensity was calculated using the temperature metric used in each heat wave definition.
212 Table A4 Definitions used for heat waves with mean temperature as the metric.
Variable Heat Wave Definition Heat Wave Characteristics
Average Average Average Duration Total number Total number number of intensity of Definition Threshold duration of Metric (consecutive of heat wave of heat wave heat wave heat wave number (percentile) heat wave days) days events days per days (days) season (°C) 1 Mean ≥90th ≥2 113 43 9.42 2.63 27.31 2 Mean ≥95th ≥2 42 17 3.50 2.47 28.39 3 Mean ≥97th ≥2 21 7 1.75 3.00 29.21 4 Mean ≥98th ≥2 12 4 1.00 3.00 29.48 5 Mean ≥99th ≥2 3 1 0.25 3.00 29.82 6 Mean ≥90th ≥3 51 12 4.25 4.25 27.64 7 Mean ≥95th ≥3 14 3 1.17 4.67 28.61 8 Mean ≥97th ≥3 11 2 0.92 5.50 29.01 9 Mean ≥98th ≥3 6 1 0.50 6.00 29.63 10 Mean ≥99th ≥3 3 1 0.25 3.00 29.82 11 Mean ≥90th ≥4 30 5 2.50 6.00 27.83 12 Mean ≥95th ≥4 11 2 0.92 5.50 29.01 13 Mean ≥97th ≥4 11 2 0.92 5.50 29.01 14 Mean ≥98th ≥4 6 1 0.50 6.00 29.63 N/A Mean ≥99th ≥4 0 0 N/A N/A N/A a The average intensity was calculated using the temperature metric used in each heat wave definition.
213 Table A5 Definitions used for heat waves with minimum temperature as the metric.
Variable Heat Wave Definition Heat Wave Characteristics
Average Average Average Duration Total number Total number number of intensitya of Definition Threshold duration of Metric (consecutive of heat wave of heat wave heat wave heat wave number (percentile) heat wave days) days events days per days (days) season (°C) 1 Minimum ≥90th ≥2 114 39 9.50 2.93 20.75 2 Minimum ≥95th ≥2 46 18 3.83 2.56 21.44 3 Minimum ≥97th ≥2 20 8 1.67 2.50 22.12 4 Minimum ≥98th ≥2 9 3 0.75 3.00 22.90 5 Minimum ≥99th ≥2 7 2 0.58 3.50 23.28 6 Minimum ≥90th ≥3 64 14 5.33 4.57 21.00 7 Minimum ≥95th ≥3 20 5 1.67 4.00 21.82 8 Minimum ≥97th ≥3 8 2 0.67 4.00 22.78 9 Minimum ≥98th ≥3 5 1 0.42 5.00 23.58 10 Minimum ≥99th ≥3 5 1 0.42 5.00 23.58 11 Minimum ≥90th ≥4 49 9 4.08 5.44 21.05 12 Minimum ≥95th ≥4 14 3 1.17 4.67 22.21 13 Minimum ≥97th ≥4 5 1 0.42 5.00 23.58 14 Minimum ≥98th ≥4 5 1 0.42 5.00 23.58 15 Minimum ≥99th ≥4 5 1 0.42 5.00 23.58 a The average intensity was calculated using the temperature metric used in each heat wave definition.
214
Figure A1 The association between single days of extreme heat and EHAs for direct heat-related conditions in the SSD during the warm season, 2001 to 2013.
215 Table A6 Effect of heat waves first in season compared to heat waves not first in season on EHAs for direct heat-related conditions in the SSD during the warm season, 2001 to 2013.
Heat wave Definition Odds Ratio (95% CI) Duration Metric Threshold (consecutive First in season Not first in season (°C) (percentile) days) th Maximum ≥90 ≥2 4.01(2.05, 7.83) 7.31(5.15, 10.37) Maximum ≥95th ≥2 14.99(8.46, 26.57) 2.48(1.29, 4.78) Maximum ≥90th ≥3 4.97(3.09, 7.99) 29.34(8.91, 96.62) Maximum ≥95th ≥3 16.27(7.50, 35.29) 6.68(1.21, 36.82) Mean ≥90th ≥2 4.46(2.46, 8.06) 7.52(5.14, 11.00) Mean ≥95th ≥2 10.89(5.91, 20.08) 7.83(4.15, 14.76) Mean ≥90th ≥3 5.43(3.12, 9.48) 10.17(5.33, 19.42) Mean ≥95th ≥3 5.60(1.61, 19.52) 60.51(8.15, 449.27) th Minimum ≥90 ≥2 5.20(2.94, 9.19) 3.93(2.39, 6.45) Minimum ≥95th ≥2 4.40(2.05, 9.45) 7.30(3.62, 14.68) Minimum ≥90th ≥3 3.12(1.76, 5.52) 6.29(3.04, 13.00) Minimum ≥95th ≥3 4.20(1.87, 9.44) 47.51(6.32, 356.96)
216
Figure A2 The association between heat wave days and EHAs for direct heat- related conditions in the SSD during the warm season, 2001 to 2013, after controlling for daily average temperature (the ‘added heat wave effect’).
217 Table A7 The average and peak intensity of heat wave days comprising of the first heat wave of the season and those heat wave days that do not during the warm season in the SSD, 2001 to 2013. Note: The average intensity was calculated using the temperature metric used in each heat wave definition.
Average Peak Average Peak Average Mean Peak Mean Heat wave definition Maximum Maximum Minimum Minimum (°C) (°C) (°C) (°C) (°C) (°C) Duration Metric Threshold Not Not Not Not Not Not (consecutive First First First First First First (°C) (percentile) firsta first first first first first days) Maximum ≥90th ≥2 34.48 35.42 38.55 42.69 26.33 27.36 29.79 32.38 18.18 19.29 22.29 24.39 Maximum ≥95th ≥2 36.78 35.95 42.69 38.89 28.41 27.68 32.38 29.64 20.04 19.40 24.39 23.05 Maximum ≥90th ≥3 35.07 36.10 42.69 40.61 27.04 28.21 32.38 31.37 19.00 20.31 22.29 24.39 Maximum ≥95th ≥3 36.86 36.11 42.69 37.81 28.59 28.28 32.38 29.55 20.32 20.46 24.39 21.29 Mean ≥90th ≥2 34.41 34.79 39.66 43.99 26.99 27.40 29.79 32.38 19.56 20.01 22.29 24.39 Mean ≥95th ≥2 36.04 35.82 42.69 40.61 28.04 28.63 32.38 31.37 20.04 21.44 22.09 24.39 Mean ≥90th ≥3 34.54 35.54 42.69 40.61 27.29 28.00 32.38 31.37 20.05 20.46 22.09 24.39 Mean ≥95th ≥3 34.93 36.36 37.81 39.61 27.80 29.42 29.55 31.3 20.68 22.47 22.09 24.39 Minimum ≥90th ≥2 31.94 30.35 42.69 39.61 26.33 25.56 32.38 31.3 20.72 20.77 22.89 24.39 Minimum ≥95th ≥2 32.17 31.99 41.08 39.61 26.67 26.80 31.03 31.3 21.18 21.61 22.95 24.39 Minimum ≥90th ≥3 30.84 31.60 41.08 39.61 25.87 26.35 31.03 31.3 20.90 21.10 22.95 24.39 Minimum ≥95th ≥3 32.97 35.76 39.17 39.61 27.10 29.67 31.03 31.3 21.23 23.58 22.89 24.39 a Consists of heat wave days belonging to the second, third heat wave of the season
218 Table A8 The average duration of heat waves first in the warm season and those heat waves not first in the SSD during the warm season, 2001 to
2013.
Total Average duration Heat wave definition number of days (days) (days) Duration Metric Threshold (consecutive First Not first First Not first (°C) (percentile) days) Maximum ≥90th ≥2 24 74 2.4 2.6 Maximum ≥95th ≥2 25 16 2.8 2.3 Maximum ≥90th ≥3 28 16 3.5 5.3 Maximum ≥95th ≥3 15 4 3.8 4 Mean ≥90th ≥2 26 87 2.4 2.7 Mean ≥95th ≥2 17 25 2.1 2.8 Mean ≥90th ≥3 26 25 4.3 4.2 Mean ≥95th ≥3 7 7 3.5 7 Minimum ≥90th ≥2 27 87 2.3 3.2 Minimum ≥95th ≥2 18 28 2.3 2.8 Minimum ≥90th ≥3 33 31 4.7 4.4 Minimum ≥95th ≥3 15 5 3.8 5
219 Table A9 Characterisation of heat wave days as high and low level ozone days.
Number of heat wave Number of heat wave Average heat wave Average heat wave Heat wave days identified as high days identified as low intensitya of high intensitya of low ozone Definition ozone days ozone days ozone day (°C) day (°C) HWD01 46 52 35.38 35.03
HWD02 49 64 27.38 27.25
HWD03 21 93 20.79 20.62 a The average intensity was calculated using the temperature metric used in each heat wave definition.
220 Table A10 The effect of heat wave days on EHAs for three respiratory diseases on days with high levels of ozone compared to days with low
levels of ozone in the SSD during the warm season, 2001 to 2013, for all ages and specific age groups at lag2.
HWD01 HWD02 HWD03 Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Low Ozone High Ozone Low Ozone High Ozone Low Ozone High Ozone Asthma All Ages 0.94 0.97 0.96 0.91 0.91 0.91 (0.87, 1.02) (0.86, 1.08) (0.90, 1.03) (0.81, 1.02) (0.86, 0.96) (0.74, 0.95) 0-14 Years 0.86 0.94 0.92 0.89 0.87 0.93 (0.78, 0.94) (0.82, 1.08) (0.85, 1.00) (0.78, 1.03) (0.82, 0.93) (0.80, 1.09) 15-64 Years 1.24 1.05 1.15 0.98 0.99 0.86 (1.06, 1.45) (0.84, 1.31) (1.00, 1.32) (0.79, 1.23) (0.88, 1.12) (0.66, 1.13) 65 Years and 0.96 1.02 0.87 0.97 1.04 1.06 Over (0.73, 1.26) (0.66, 1.59) (0.67, 1.13) (0.63, 1.50) (0.83, 1.30) (0.63, 1.77) COPD All Ages 1.05 0.98 0.96 0.99 1.02 1.10 (0.98, 1.13) (0.88, 1.09) (0.90, 1.03) (0.89, 1.10) (0.97, 1.08) (0.97, 1.25) 0-64 Years 1.18 1.11 0.96 1.01 0.98 1.00 (1.02, 1.37) (0.90, 1.39) (0.83, 1.10) (0.82, 1.26) (0.87, 1.10) (0.77, 1.30) 65-74 Years 1.01 0.91 0.99 1.00 1.04 1.29 (0.88, 1.16) (0.74, 1.10) (0.87, 1.12) (0.82, 1.21) (0.93, 1.15) (1.03, 1.62) 75 Years and 1.02 0.95 0.96 1.00 1.03 1.04 Over (0.92, 1.14) (0.82, 1.12) (0.87, 1.05) (0.85, 1.17) (0.95, 1.12) (0.86, 1.25) Pneumonia All Ages 1.02 1.13 0.95 1.11* 0.97 1.08 (0.95, 1.09) (1.01, 1.25) (0.89, 1.01) (0.99, 1.23) (0.92, 1.03) (0.81, 1.04) 0-14 Years 0.91 0.85 0.77 0.98 0.94 1.10 (0.77, 1.08) (0.66, 1.10) (0.65, 0.92) (0.75, 1.26) (0.82, 1.08) (0.82, 1.47) 15-64 Years 0.97 1.26* 1.00 1.10 1.02 1.26 (0.85, 1.11) (1.04, 1.54) (0.89, 1.13) (0.90, 1.34) (0.92, 1.13) (1.01, 1.58) 64-75 Years 1.13 1.00 1.05 1.10 0.97 0.96 (0.93, 1.37) (0.75, 1.33) (0.87, 1.25) (0.82, 1.46) (0.83, 1.13) (0.67, 1.36) 75 Years and 1.08 1.23 0.95 1.19* 0.97 0.96 Over (0.96, 1.20) (1.04, 1.45) (0.86, 1.06) (1.004, 1.412) (0.83, 1.13) (0.67, 1.36)
*Denotes a statistically significant positive interaction term p-value (<0.05) ** Denotes a statistically significant negative interaction term p-value (<0.05)
221 Table A11 Characterisation of heat wave days as high and low level PM10 days.
Number of heat wave Number of heat wave Average Average a a Heat wave definition days identified as high days identified as low heat wave intensity of heat wave intensity of PM10 days PM10 days high PM10 day (°C) low PM10 day (°C)
HWD01 56 42 35.76 34.43 HWD02 52 61 27.88 26.81
HWD03 24 90 20.79 20.74
a The average intensity was calculated using the temperature metric used in each heat wave definition.
222 Table A12 The effect of heat wave days on EHAs for six cardiovascular diseases on days with high levels of PM10 compared to days with low levels of PM10 in the SSD during the warm season, 2001 to 2013, for all ages at lag2.
HWD01 HWD02 HWD03 Disease Heat Effecta Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Ischemic Heart 1.02 0.93** 1.00 0.95 0.99 0.98 Disease (0.98, 1.07) (0.85, 1.02) (0.96, 1.04) (0.87, 1.03) (0.96, 1.03) (0.89, 1.07) Heart Failure 0.87 0.89 0.92 0.89 0.94 0.91 (0.81, 0.94) (0.76, 1.03) (0.86, 0.99) (0.76, 1.03) (0.88, 1.002) (0.77, 1.07) Cardiac Arrest 1.13 1.26 1.13 1.37 1.13 0.73 (0.87, 1.47) (0.75, 2.12) (0.89, 1.45) (0.89, 2.09) (0.89, 1.43) (0.40, 1.35) Heart 0.98 1.00 1.00 1.04 1.02 1.08 Arrhythmia (0.92, 1.05) (0.87, 1.14) (0.95, 1.06) (0.92, 1.19) (0.97, 1.08) (0.94, 1.24) Conduction 1.01 0.81 0.92 0.88 0.85 0.90 Disorders (0.82, 1.23) (0.51, 1.28) (0.76, 1.11) (0.57, 1.35) (0.71, 1.01) (0.58, 1.39) Hypertensive 1.02 0.94 1.02 0.91 1.00 1.11 Disease (0.85, 1.23) (0.62, 1.41) (0.86, 1.20) (0.61, 1.35) (0.85, 1.17) (0.73, 1.68) *Denotes a statistically significant postive interaction term p-value (<0.05) **Denotes a statistically significant negative interaction term p-value (<0.05)
223 Table A13 The effect of heat wave days on EHAs for cardiovascular diseases on days with high levels of PM10 compared to days with low levels of PM10 in the
SSD during the warm season, 2001 to 2013, for those aged 0-64 years and 65 years and over at lag2.
HWD01 HWD02 HWD03 Disease Heat Effecta Heat Effect Heat Effect Heat Effect Heat Effect Heat Effect Low PM10 High PM10 Low PM10 High PM10 Low PM10 High PM10 Ischemic Heart Disease 0-64 years 1.07 0.96 1.01 0.95 1.00 1.04 (1.00, 1.14) (0.84, 1.11) (0.95, 1.07) (0.84, 1.09) (0.94, 1.06) (0.86, 1.14) 65 years and over 0.99 0.91 0.99 0.94 0.99 0.93 (0.94, 1.05) (0.80, 1.02) (0.94, 1.05) (0.84, 1.06) (0.94, 1.04) (0.82, 1.06) Heart Failure 0-64 years 0.84 0.78 0.90 1.02 0.99 0.92 (0.69, 1.02) (0.51, 1.19) (0.75, 1.07) (0.69, 1.49) (0.84, 1.18) (0.60, 1.40) 65 years and over 0.90 0.90 0.92 0.86 0.93 0.91 (0.83, 0.98) (0.75, 1.05) (0.86, 0.99) (0.73, 1.02) (0.87, 0.997) (0.76, 1.08) Cardiac Arrest 0-64 years 1.51 0.76 1.33 0.77 1.17 0.61 (1.01, 2.26) (0.31, 1.85) (0.93, 1.91) (0.34, 1.74) (0.83, 1.65) (0.24, 1.57) 65 years and over 0.92 1.71* 0.98 1.68 1.10 0.83 (0.65, 1.31) (0.89, 3.28) (0.70, 1.38) (0.89, 3.18) (0.80, 1.51) (0.36, 1.90) Heart Arrhythmia 0-64 years 1.05 1.04 1.03 1.00 0.98 0.96 (0.95, 1.16) (0.84, 1.29) (0.94, 1.13) (0.81, 1.22) (0.89, 1.07) (0.76, 1.20) 65 years and over 0.95 0.97 0.99 1.08 1.05 1.16 (0.87, 1.03) (0.82, 1.15) (0.92, 1.07) (0.91, 1.27) (0.98, 1.12) (0.97, 1.39) Conduction Disorders 0-64 years 1.09 2.07 1.08 1.80 0.81 1.58 (0.70, 1.72) (0.79, 5.42) (0.72, 1.62) (0.70, 4.66) (0.55, 1.18) (0.59, 4.22) 65 years and over 0.99 0.62 0.88 0.73 0.86 0.78 (0.79, 1.24) (0.36, 1.07) (0.72, 1.09) (0.45, 1.19) (0.71, 1.06) (0.47, 1.27) Hypertensive Disease 0-64 years 0.83 1.08 1.06 1.21 1.00 1.64 (0.62, 1.11) (0.60, 1.94) (0.82, 1.38) (0.69, 2.14) (0.78, 1.28) (0.91, 2.98) 65 years and over 1.18 0.77 0.98 0.69 0.99 0.78 (0.93, 1.50) (0.43, 1.40) (0.79, 1.22) (0.39, 1.21) (0.81, 1.22) (0.43, 1.42) *Denotes a statistically significant positive interaction term p-value (<0.05) ** Denotes a statistically significant negative interaction term p-value (<0.05)
224