Effects of Extreme Temperatures on Years of Life Lost for Cardiovascular Deaths: A Time Series Study in Brisbane, Australia
Background—Extreme temperatures are associated with cardiovascular disease (CVD) deaths. Previous studies have investigated the relative CVD mortality risk of temperature, but this risk is heavily influenced by deaths in frail elderly people. To better estimate the burden of extreme temperatures, we estimated their effects on years of life lost due to CVD.
Methods and Results—The data were daily observations on weather and CVD mortality for Brisbane, Australia, between 1996 and 2004. We estimated the association between daily mean temperature and years of life lost due to CVD, after adjusting for trend, season, day of the week, and humidity. To examine the nonlinear and delayed effects of temperature, a distributed lag nonlinear model was used. The model’s residuals were examined to investigate whether there were any added effects due to cold spells and heat waves. The exposure–response curve between temperature and years of life lost was U-shaped, with the lowest years of life lost at 24°C. The curve had a sharper rise at extremes of heat than of cold. The effect of cold peaked 2 days after exposure, whereas the greatest effect of heat occurred on the day of exposure. There were significantly added effects of heat waves on years of life lost.
Conclusions—Increased years of life lost due to CVD are associated with both cold and hot temperatures. Research on specific interventions is needed to reduce temperature-related years of life lost from CVD deaths.
Climate change has created a growing interest in the relation between weather and health.1,2 Epidemiological studies have shown that ambient temperature has short-term effect on overall mortality, with a generally U-shaped relationship because of the increased risks in cold and hot weather.3,4 Extended periods of extreme temperatures, known as cold spells and heat waves, have also been associated with peaks in mortality.5,6 The effects of extreme temperatures may last for many days, especially for cold weather where the effects can be delayed by weeks.7,8
The harmful effects of cold and heat are strongly apparent in cardiovascular disease (CVD).9,10 In many countries, CVD is the leading cause of death and accounts for a large proportion of the total burden of disease.11–13 The well-known risk factors for the cause of CVD are tobacco smoking, risky alcohol consumption, poor diet, insufficient physical activity, high blood pressure and cholesterol, obesity, and diabetes mellitus.14 However, other environmental factors, such as ambient temperature and air pollution, also play a role.15,16 Changes in cholesterol levels and the response of autonomic nervous system have been reported to increase CVD events during temperature extremes, which is a particular concern among older adults with limited cardiovascular reserve.17–19
Previous studies have investigated the relative mortality risk of temperature on CVD.20–22 However, this risk is heavily influenced by deaths among elderly persons. If most temperature-related CVD deaths occur in people with a short life expectancy, then temperature exposure would be less of a public health concern.23 Conversely, if many temperature-related CVD deaths are among people with a longer life expectancy, then this would increase the current concern, especially in light of the impending higher temperatures because of climate change.24
Years of life lost is a measure of disease burden that uses the life expectancy at death, and so gives more weight to deaths among younger people compared with the traditional measure of relative mortality risk that weights all deaths equally.25 In this study we examined the effects of temperature on years of life lost due to CVD using data from Brisbane, Australia. We also investigated whether there were any added effects of extreme cold spells and heat waves.
WHAT IS KNOWN
Ambient temperature is associated with an increased risk of death, especially for cardiovascular diseases.
Previous studies have investigated the relative mortality risk of temperature, but this risk is heavily influenced by deaths in frail elderly persons.
WHAT THE STUDY ADDS
We used years of life lost to examine the effects of temperature on cardiovascular deaths in Brisbane, Australia.
The association between temperature and years of life lost was U-shaped, with increased years of life lost due to cold and hot temperatures.
There were significantly added effects of heat waves on years of life lost, whereas there was no added effect of cold spells.
Brisbane is the state capital of Queensland, Australia. It is Australia’s third-largest city, with a population of 896 649 as per records in 2001.26 It has a subtropical climate with hot, humid summers and mild, dry winters.
Daily mortality data from January 1, 1996 to November 30, 2004 were requested from the Office of Economic and Statistical Research of the Queensland Treasury. All deaths recorded were of residents of Brisbane. The data were the latest available at the time of our request, which included date of death, sex, age, and cause of death. Causes of deaths were classified according to the International Classification of Diseases (ICD), 9th Revision for 1996 (ICD-9 codes: 390–459), and 10th Revision for 1997 to 2004 (ICD-10 codes: I00–I99). We used a fixed life expectancy to estimate the years of life lost for each CVD death by matching their age and sex with the Australian life table data from 1996 to 2004.27–30 According to these Australian life tables, for example, a man aged 50 years in 1996 had a life expectancy of 28.8 years,27 and a woman aged 80 years in 2004 had a life expectancy of 9.8 years.30 The daily total years of life lost was the sum of the years of life lost for all CVD deaths on the same day. To create the estimates on more comparable scales, we also standardized the years of life lost to a million people using the 2001 population data. We assumed a roughly stable population at risk as we examined all CVD deaths for the whole of Brisbane.
Daily weather data were provided by the Australia Bureau of Meteorology. We used daily values of maximum temperature, minimum temperature, and relative humidity from the Archerfield station located near the city center. Daily mean temperature was the average of the daily maximum and minimum temperature. When daily weather data were missing for Archerfield (<2%), data from Brisbane Airport were used. Daily air pollution data were obtained from the Queensland Environmental Protection Agency, which included the maximum 1-hour average concentrations of ozone (O3) and nitrogen dioxide (NO2), and the ambient 24-hour average concentration of particulate matter with diameters <10 μm (PM10).
Defining a Cold Spell and Heat Wave
There are no standard definitions of a cold spell or heat wave, but most are based on extreme cold and heat over a period of days.31,32 To define cold spells and heat waves, we used a combination of intensity and duration: (1) Intensity: the first percentile of daily mean temperature (11.7°C) as the cold threshold, and the 99th percentile of daily mean temperature (29.2°C) as the heat threshold; (2) Duration: a minimum of 2 to 4 consecutive days with temperatures below or above the thresholds.
We used regression models to investigate the association between temperature and years of life lost. The dependent variable was the total years of life lost due to CVD on each day.
Stage-1: Quantify the general effects of daily temperatures
A distributed lag nonlinear model was used to estimate the association between daily mean temperature and daily total years of life lost due to CVD.33 To capture the nonlinear effects of temperature, we used a natural spline with 4 degrees of freedom. Four degrees of freedom allow a nonlinear U-shaped association that is also asymmetrical with possibly stronger effects at temperature extremes. To capture the delayed effects of temperature, we assumed a maximum lag of 10 days between exposure and death, and this lag was modeled using a natural spline with 4 degrees of freedom. We adjusted for long-term trends in CVD mortality using a natural spline with 4 degrees of freedom. We adjusted for seasonal patterns (by month) and day of the week using categorical variables. We adjusted for daily humidity using a linear term with a maximum lag of 10 days. We plotted the mean and 95% confidence intervals (CIs) for the estimated years of life lost per day against daily mean temperature.
Stage-2: Examine the added effects of heat waves and cold spells
Excess mortality during prolonged extreme temperatures is often greater than the predicted using smoothed temperature-mortality association.34,35 This is because sustained period of extreme temperatures produce an extra effect beyond that predicted by daily temperatures. To investigate the potential added effects of heat waves and cold spells, we analyzed the residuals from our stage-1 model. Stage-2 model ensures that we estimate the extra effects of extreme events after removing the general effects of temperature.36 The heat wave and cold spell were defined as binary yes or no variables on each day. For example, if January 25 and 26 in 2004 were days with mean temperatures above the heat threshold, then January 26 would have a value of “yes” for heat wave. January 25 would not be a heat wave day because we defined a heat wave as a sustained period of extreme temperatures for ≥2 days. We assumed a maximum lag of 10 days for the delayed effects of heat waves and cold spells. We compared the years of life lost on extreme temperature days with nonextreme days.
Several sensitivity analyses were carried out. We stratified the analyses by sex, and adjusted for 3 air pollutants (O3, NO2, and PM10), assuming a maximum lag of 10 days between exposure and death. We used the Akaike information criterion as a measure of model fit to determine whether any pollutants should be adjusted for as confounders. Alternative definitions for cold spells and heat waves were also considered. We used the second percentile (12.3°C), third percentile (12.6°C), and fifth percentile (13.3°C) of daily mean temperature as a cold threshold, and the 98th percentile (28.5°C), 97th percentile (27.9°C), and 95th percentile (27.1°C) of daily mean temperature as a heat threshold.
All statistical analyses were performed using the R software (version 2.13.2) (R Development Core Team, Austria). The distributed lag nonlinear models were fitted using the R package dlnm (version 1.4.1).37
Characteristics of Daily Mortality, Years of Life Lost, and Ambient Temperature
Summary statistics for daily mortality and years of life lost in Brisbane are as given Table 1. Deaths due to CVD (7 deaths per day) accounted for more than 40% of total mortality (17 deaths per day), representing the most frequent cause of death. There were 65 years of life lost per day due to CVD and 219 years of life lost per day for total mortality (per 1 million people the years of life lost per day were 72 due to CVD and 243 for total mortality).
The average daily mean temperature in Brisbane was 20.5°C, with the minimum at 9.6°C and the maximum at 34.5°C (online-only Data Supplement Table I). Spearman correlations for the weather variables and air pollutants are in online-only Data Supplement Table II.
The effects of temporal trend and season on years of life lost for CVD deaths are shown in online-only Data Supplement Figures I and II. In Brisbane, daily total years of life lost due to CVD decreased over time during the study period, and a strong seasonal effect occurred in August.
Association of Temperature and Years of Life Lost
The exposure–response curve between daily mean temperature and years of life lost due to CVD is as given in Figure 1. The curve is U-shaped, with the lowest years of life lost at 24°C. For days with a mean temperature of 10°C, there were 31 years of life lost per day (95% CI, 11–52 years). There were 45 years of life lost per day for a mean temperature of 32°C (95% CI, 22–67 years). The exposure–response curve has a sharper rise at extremes of heat than of cold.
The residuals were checked to evaluate the adequacy of the model, by ensuring they were approximately normally distributed and independent over time (online-only Data Supplement Figure III). Several sensitivity analyses were also performed. We stratified our analyses by sex, but the exposure–response curves for men and women were similar (online-only Data Supplement Figure IV). We adjusted for daily air pollutants O3, NO2, and PM10 as potential confounders, but the effects of temperature on years of life lost did not change (online-only Data Supplement Figure V). The model with the lowest Akaike information criterion was the model without any of these air pollutants, we therefore did not include any pollutants in our final model.
To further explore whether the increased years of life lost were mainly due to the relatively large number of CVD deaths or a relatively large proportion of younger CVD deaths, we plotted the percentage of the daily number of CVD deaths by age group and daily temperature (online-only Data Supplement Figure VI). We also stratified our analyses by age groups (online-only Data Supplement Figure VII) and then calculated the annual temperature-related years of life lost for CVD deaths (online-only Data Supplement Table III). These results show that extreme temperatures result in more deaths in all age groups, so some of the increased years of life lost are due to deaths in younger people.
Delayed Effects of Temperature
Figure 2 shows the delayed effects of cold and hot temperatures on years of life lost due to CVD. We used the first percentile of temperature at 12°C to represent a cold day, and the 99th percentile of temperature at 29°C to represent a hot day. Cold effects lasted longer than heat effects. The greatest effect of heat occurred in the first day of exposure, then decreased rapidly, and returned to baseline levels within 5 days. Cold effect reached a peak 2 days after exposure, and then declined slightly with a delayed effect even after 10 days. A 3-dimensional plot of daily total years of life lost due to CVD by temperature and lag is as depicted in online-only Data Supplement Figure VIII.
Added Effects of Heat Waves and Cold Spells
Table 2 shows years of life lost due to the added effects of heat waves and cold spells in those days matching our definitions. Using a heat wave definition of 2 days’ temperature above the 99th percentile meant 85 years of life lost per day for CVD deaths (95% CI, 40–129 years). For a longer duration of 4 days, there were 264 years of life lost per day during heat waves (95% CI, 62–466 years). There was no significant increase in years of life lost for CVD deaths during cold spells. Sensitivity analyses using leave-one-out method and using different heat wave and cold spell definitions confirmed these findings (online-only Data Supplement Tables IV and V).
Mortality data are important for health risk assessments because they generally cover the whole population and include key characteristics of deaths. In this study, we considered deaths where CVD was the underlying cause. We used years of life lost to estimate the burden of temperature on CVD mortality. We found that the association between temperature and years of life lost due to CVD was U-shaped, with increased years of life lost attributable to cold and hot temperatures. We also observed significant added effects of heat waves, with between 85 and 264 years of life lost per day depending on the severity of heat waves.
The majority of previous studies have examined the relative risk of temperature-related mortality.3,10 However, mortality risk does not account for deaths at different ages. Years of life lost is a useful measure for assessing premature mortality, which represents the preventable loss of life years.25,38 Understanding how temperature exposure impacts on years of life lost will be helpful for ranking the health risks against other exposures.
Ambient temperature is an important determinant of human health. Exposure to extreme temperatures can act as a trigger for CVD events due to changes in blood pressure, blood viscosity, blood cholesterol, and heart rate.9,18,19 Cold weather is well known to be associated with a variety of autonomic responses in humans, including peripheral vasoconstriction, shivering, and increased blood pressure and heart rate. In patients suffering from ischemic heart disease, exposure to cold may cause a decrease in coronary blood flow, which could contribute to coronary spasm, chest pain, and even myocardial infarction.39–41 Heat has also been found to induce profound physiologic changes such as an increase in blood viscosity and cardiac output leading to dehydration, hypotension, and even endothelial cell damage. When the temperature rises, the heat balance of the body is generally restored by increased blood flow to the skin and by sweating. These responses may overload the heart and cause hemoconcentration, which could lead to coronary and cerebral thrombosis.17,42
Understanding the lag time between temperature exposure and years of life lost is important for health care providers to develop response plans for extreme temperature events.21,24 This study showed that the greatest effect of cold on years of life lost was not acute, but occurred 2 days later. Conversely, the greatest effect of heat occurred on the day of exposure, and the effect was limited to 5 days after exposure. Hence, health care providers should expect an immediate increase in ambulance call-outs and hospital admissions during hot weather.
A significant added effect of heat waves implies that an extra risk arises when the exposure to extreme heat is sustained for ≥2 days (Table 2). However, there was no added effect of cold spells. It might be that people took better protective actions during prolonged cold weather. Such evidence may shed new light on future changes in years of life lost related to global climate change. It is likely that climate change will produce more frequent, more intense, and longer lasting heat waves,43 and consequently put additional stress on vulnerable people such as the elderly and those with preexisting CVD. Furthermore, with the trends in increasing rates of obesity and related conditions including diabetes mellitus, CVD will continue to have a major impact on population health in terms of prevalence, morbidity, and mortality.14,44 The growing size of this vulnerable population could mean that the future disease burden of temperature will increase.
Temperature extremes can aggravate existing health conditions and ultimately accelerate the death of patients.1,31 It is important to create greater awareness of the dangers of extreme temperatures, particularly heat waves, to inform the public about how to minimize their risks. Advice on avoiding and managing temperature-related deaths could be distributed via the media. Further individual level studies collecting clinic, behavioral, and demographic data may help elucidate which subgroups are likely to be the most vulnerable, and to clarify the role of adaptive measures such as air conditioning.45 Doctors would then be able to give evidence-based advice to their CVD patients.
Climate change will create more stress on health systems through increased frequency and intensity of heat waves.2,4,32 Health systems can be challenged when they strive to achieve maximum coverage of the population. Therefore, health service delivery needs to be assured during extreme heat events. Hospitals and emergency departments may adapt their procedures to meet the added demands, such as reducing elective services to free-up staff and beds, and treating patients in nontraditional locations.46 However, in the long run, reducing climate change-related years of life lost for CVD deaths has to be achieved through improved responses that come from integrating the emergency preparedness and disaster risk reduction throughout society.47,48
Strengths and Limitations
To the best of our knowledge, this is the first study to examine the effects of temperature on years of life lost for CVD deaths. Our 2-stage approach also allows a more accurate estimation of temperature effects by making a distinction between effects from independent daily temperatures and from the duration of prolonged extreme temperatures. Several sensitivity analyses confirmed that our main conclusions were robust to changes in the model’s specification.
The study has some limitations. First, the data were limited to one city, which makes our results hard to generalize to other communities. The associations between temperature and years of life lost due to CVD are likely to be different for other geographic locations as a result of population acclimatization and sociodemographic differences. Second, temperature measures collected from 1 monitoring station may not represent individual exposures, creating a potential misclassification of exposure for those persons who died far from the station. Third, we have only considered the deaths where CVD was the underlying cause. However, to see whether people with CVD are more susceptible to extreme temperatures, we would need to separately investigate people with and without preexisting CVD.
This study shows evidence that ambient temperature is associated with years of life lost due to CVD in a subtropical city in Australia. We also found significant added effects from prolonged extreme heat events. To reduce the temperature-related years of life lost for CVD deaths, research on specific interventions are needed.
Sources of Funding
C. Huang was supported by Queensland University of Technology Postgraduate Research Award and Commonwealth Scientific and Industrial Research Organisation Climate Adaptation Flagship Collaboration Fund. Dr Tong was supported by a National Health & Medical Research Council Research Fellowship (#553043).
The online-only Data Supplement is available at http://circoutcomes.ahajournals.org/lookup/suppl/doi:10.1161/CIRCOUTCOMES.112.965707/-/DC1.
- Received December 27, 2011.
- Accepted August 10, 2012.
- © 2012 American Heart Association, Inc.
- Hajat S,
- Kosatky T
- Revich B,
- Shaposhnikov D
- Rocklöv J,
- Forsberg B
- Bhaskaran K,
- Hajat S,
- Haines A,
- Herrett E,
- Wilkinson P,
- Smeeth L
- Go AS,
- Magid DJ,
- Wells B,
- Sung SH,
- Cassidy-Bushrow AE,
- Greenlee RT,
- Langer RD,
- Lieu TA,
- Margolis KL,
- Masoudi FA,
- McNeal CJ,
- Murata GH,
- Newton KM,
- Novotny R,
- Reynolds K,
- Roblin DW,
- Smith DH,
- Vupputuri S,
- White RE,
- Olson J,
- Rumsfeld JS,
- Gurwitz JH
- Moran A,
- Gu D,
- Zhao D,
- Coxson P,
- Wang YC,
- Chen CS,
- Liu J,
- Cheng J,
- Bibbins-Domingo K,
- Shen YM,
- He J,
- Goldman L
- Stewart S,
- Ekman I,
- Ekman T,
- Odén A,
- Rosengren A
- 14.↵WHO. Prevention of cardiovascular disease: guideline for assessment and management of cardiovascular risk. Geneva, Switzerland: World Health Organization; 2007.
- von Klot S,
- Peters A,
- Aalto P,
- Bellander T,
- Berglind N,
- D’Ippoliti D,
- Elosua R,
- Hörmann A,
- Kulmala M,
- Lanki T,
- Löwel H,
- Pekkanen J,
- Picciotto S,
- Sunyer J,
- Forastiere F
- Ren C,
- O’Neill MS,
- Park SK,
- Sparrow D,
- Vokonas P,
- Schwartz J
- Basu R,
- Ostro BD
- Lopez AD,
- Mathers CD,
- Ezzati M,
- Jamison DT,
- Murray CJL
- 26.↵Australian Bureau of Statistics. 3218.0 Population Estimates by Local Government Area, 2001 to 2011. Available at: http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3218.02010-11?OpenDocument. Accessed March 12, 2012.
- 27.↵Australian Bureau of Statistics. 3302.0 - Deaths, Australia, 1998. Available at: http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3302.01998?OpenDocument. Accessed February 10, 2012.
- 28.↵Australian Bureau of Statistics. 3302.0 - Deaths, Australia, 2000. Available at: http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3302.02000?OpenDocument. Accessed February 10, 2012.
- 29.↵Australian Bureau of Statistics. 3302.0 - Deaths, Australia, 2002. Available at: http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3302.02002?OpenDocument. Accessed February 10, 2012.
- 30.↵Australian Bureau of Statistics. 3302.0 - Deaths, Australia, 2004. Available at: http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3302.02004?OpenDocument. Accessed February 10, 2012.
- 31.↵WHO. Improving public health responses to extreme weather/heat-waves – EuroHEAT. Copenhagen Ø, Denmark: WHO Regional Office for Europe; 2009.
- Gasparrini A,
- Armstrong B
- Wolf K,
- Schneider A,
- Breitner S,
- von Klot S,
- Meisinger C,
- Cyrys J,
- Hymer H,
- Wichmann HE,
- Peters A
- 43.↵IPCC. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, UK: Cambridge University Press; 2007.
- Roger VL,
- O’Donnell CJ
- Bhaskaran K,
- Hajat S,
- Haines A,
- Herrett E,
- Wilkinson P,
- Smeeth L