doi: 10.1136/oem.2010.056168
published online October 29, 2010
Occup Environ Med
Pei-Chih Wu, Chuan-Yao Lin, Shih-Chun Lung, et al.
Taiwan
in
events: determinants of regional vulnerability
Cardiovascular mortality during heat and cold
http://oem.bmj.com/content/early/2010/10/28/oem.2010.056168.full.html
Updated information and services can be found at:
These include:
Data Supplement
http://oem.bmj.com/content/suppl/2010/10/14/oem.2010.056168.DC1.html
"Web Only Data"
References
http://oem.bmj.com/content/early/2010/10/28/oem.2010.056168.full.html#ref-list-1
This article cites 38 articles, 15 of which can be accessed free at:
P<P
Published online October 29, 2010 in advance of the print journal.
service
Email alerting
box at the top right corner of the online article.
Receive free email alerts when new articles cite this article. Sign up in the
Notes
articles must include the digital object identifier (DOIs) and date of initial publication.
priority; they are indexed by PubMed from initial publication. Citations to Advance online
prior to final publication). Advance online articles are citable and establish publication
yet appeared in the paper journal (edited, typeset versions may be posted when available
Advance online articles have been peer reviewed and accepted for publication but have not
http://group.bmj.com/group/rights-licensing/permissions
To request permissions go to:
http://journals.bmj.com/cgi/reprintform
To order reprints go to:
http://journals.bmj.com/cgi/ep
To subscribe to BMJ go to:
group.bmj.com on November 8, 2010 - Published by oem.bmj.com Downloaded fromCardiovascular mortality during heat and cold events:
determinants of regional vulnerability in Taiwan
Pei-Chih Wu,
1Chuan-Yao Lin,
2Shih-Chun Lung,
2How-Ran Guo,
3Chang-Hung Chou,
4Huey-Jen Su
3ABSTRACT
Objectives To identify the vulnerable regions with underlying susceptibility and poor adaptive capability in response to cold and heat events in Taiwan, and to characterise the determinants associated with such an increasing risk to design better adaptive strategies in view of predicted weather changes in the future. Methods The authors used spatial regression models to measure the relationships between the spatial characteristics of temperature, extracted factors from demographic and socio-economic parameters, and the mean cardiovascular mortality 2 weeks before and after cold or heat events from 1994 to 2003.
Results Metropolitan regions were found to have a substantially lower mortality than rural areas after cold and heat events. Events of cold, compared with heat, had greater impacts on the mortality ratio in most townships. A negative association was identified, using a spatial lag model, between the mortality after cold and heat events and urbanisation, and the availability of medical resources. A higher percentage of older people, vulnerable and aborigines might have contributed to the increasing vulnerability of townships during cold and heat events.
Conclusions These data, using an island-wide spatial analysis, suggest that urban areas have a greater adaptive capability than rural areas, plausibly because people in urban areas have a higher socio-economic status and more medical resources. Social inequality across urban and rural townships is apparent and developing customised adaptation programmes for vulnerable regions to cope with heat and cold event should be prioritised.
INTRODUCTION
Exposure to cold temperatures is a documented trigger for cardiac attacks. Further, the cardiovas-cular mortality is a major contributor to excess mortality in winter,1e3even in regions with rela-tively mild winters.4 5 Cold exposures may lead directly to cardiovascular stress, resulting in an increase in platelets, red blood cell counts, blood viscosity, blood pressure, vasoconstriction, and plasma cholesterol and fibrinogen.6e8 Haemo-concentration is believed to promote arterial thrombosis, and rapid death is caused by ischaemic heart diseases and cerebrovascular diseases. Several European countries, which are situated in the temperate zone, experienced the detrimental impacts of heatwaves in 2003.9e11Heat exposure stresses the thermoregulatory system, increasing blood viscosity and cholesterol levels.12People with
an underlying cardiovascular illness tend to fall in the category of a vulnerable subgroup in terms of adverse health outcomes.13
Previous studies have reported that higher excess deaths due to cold are generally found in less severe, milder winter climates, where people have less adaptive capacity to cold exposure14; older people
are also found to be more susceptible to cold and heat exposures.15In addition, the influence of other
factors, including genetic predisposition, environ-mental variability, healthcare access and different lifestyles, have been given sizeable considerations when studying the association between weather variability and impaired health and premature mortality.16 17 Yet, only a few studies have attempted a multivariate analysis of the relative contribution of the aforementioned factors on the excess mortality following exposure to extreme cold or heat. There is far less research on the major determinants of mortality in extreme cold or heat events in subtropical or tropical regions, particularly the socio-economic determinants of mortality.5 18 Taiwan is located in a humid,
subtropical region and has hot summers and mild winters compared with many countries at higher latitudes. Cardiovascular diseases, including ischaemic heart diseases, cerebrovascular diseases and hypertensive diseases, cause around 20% of the annual total deaths and are three of the 10 leading
<An additional figure (figure 1S) is published online only. To view this figure please visit the journal online (http://oem.bmj. com).
1Department of Occupational
Safety and Health, Chang Jung Christian University, Tainan, Taiwan, ROC
2Research Center for
Environmental Changes, Academia Sinica, Taipei, Taiwan, ROC
3Department of Environmental
and Occupational Health, College of Medicine, National Cheng Kung University, Tainan, Taiwan, ROC
4Graduate Institute of Ecology
and Evolutionary Biology, China Medical University, Taichung, Taiwan, ROC
Correspondence to Professor Huey-Jen Su, Department of Environmental and Occupational Health, College of Medicine, National Cheng Kung University, 138 Sheng-Li Road, Tainan 70428, Taiwan, ROC;
hjsu@mail.ncku.edu.tw
Accepted 13 September 2010
What this paper adds
1. Previous studies have reported that factors including climate zone, age, genetic predisposi-tion, health care access, and different lifestyles are potential determinants when studying the association between temperature and premature mortality.
2. This paper uses an island wide spatial analysis to identify the vulnerable regions with underlying susceptibility in response to cold and heat events in Taiwan with subtropical characteristics. 3. People in rural areas have less adaptive
capability to extreme temperature exposure than those of urban areas, potentially attribut-able to lower socioeconomic status and less availability of medical resources.
4. This is also the first study to provide scientific evidence that the aboriginal population is uniquely vulnerable to exposure of extreme temperatures.
causes of deaths in Taiwan.19 A U-shaped temper-atureemortality relationship, particularly for older people, was observed for coronary artery disease and cerebral infarction.13 Further, a cohort study in which blood coagulating factors were measured reported that there was a greater tendency for clot formation in the circulatory system in cold weather.20
Identifi-cation of vulnerable subpopulations is, therefore, of particular concern as they are often hit hardest when a disastrous situa-tion, such as extreme temperature, occurs. Therefore, their needs should be prioritised during preventive policy-setting. We conducted this study to assess the impact of cold and heat events on the cardiovascular mortality in Taiwan by using a high-spatial-resolution analysis not only to identify vulnerable regions but also to characterise factors determining the vulner-ability, integrated with susceptibility and adaptive capability under cold and heat events.
METHODS Data collection
Extreme temperature events Cold events
The East Asian cold surge is one of the most common features of the Asian winter monsoon. The arrival of a cold surge is generally characterised by a steep rise in surface pressure, a sharp drop in surface temperature and strengthening of northerly surface winds.21 22 According to the Central Weather Bureau (CWB) in Taiwan, a cold surge is characterised by the following: (1) the surface maximum temperature dropping at least 88C within 24 h; or (2) the minimum temperature in Taipei city dropping to below 108C. There were 24 cold events identified from 1994 to 2003, which was our study period.
Heat events
According to the weather records from 1994 to 2003, the daily maximum temperature in summer (JuneeSeptember) ranged from 20.68C to 38.88C in Taipei. The highest maximum temperature, 38.88C, occurred at noon on 9 August 2003. Our study defined the days with heat events using the following criteria, since no prior guidelines in this regard were promulgated by Taiwan’s CWB: (1) maximum summer temperatures $358C for more than 6 h per day; or (2) maximum temperatures$358C for more than 9 h within 3 days in Taipei. A total of 13 heat events were identified for data-analysis purposes.
Island-wide meteorological data
The weather monitoring stations in Taiwan are mostly clustered in urban areas, along the east coastline, and on the island. We, therefore, could not obtain a good estimation of temperature distribution around the whole island, especially in rural areas. Therefore, we collected daily data from approximately 80 routine monitoring stations that had complete temperature records. Data from approximately 300 monitoring stations of the CWB that had complete rainfall records from 1994 to 2003 were also obtained. Kriging estimation obtained using ArcGIS 8.1 was used to interpolate and extrapolate the available temperature data into raster data with a spatial resolution of 1000 m per pixel in order to generate a representative tempera-ture for each township, which was the basic, smallest geographical unit for diseases notification and also to reduce the probability of misclassification. Ordinary kriging was used to interpolate the temperature and rainfall data of a randomfield at an unobserved location from the values at nearby locations.23 Zonal estimation was used to aggregate the estimation of weather parameters falling into each cell of the target grid and to
calculate the mean temperature from the cells within the township polygon. This approach allowed us to obtain the relative representative temperature for each township during the cold or heat events.
Mortality data
In Taiwan, all deaths are reported to the township and district household registry office; the National Death Registry database was obtained from the Department of Health (no personal information involved). The cardiovascular mortality (per 100 000) for 358 townships within 14 days before a cold or heat event was estimated using the number of deaths due to cardiovascular illness (ICD-9: 390e459) as the numerator and the total population in the corresponding region as the denominator. Cardiovascular mortality within 14 days after the event (including the event day) was also estimated.
Demographic and socio-economic factors
Selected socio-economic and demographic variables for each township, including medical resources (the number of clinics and doctors and demographic characteristics (percentage of aborigines, older persons, older persons living alone and disabled persons) occupation (residents engaged in service and those engaged in agriculture) and socio-economic conditions (house-hold ownership, uneducated population, unemployment rate and percentage of labourers working outside their county of residence) were estimated using the health statistics obtained from the Department of Health and also Taiwan Census 2000.
Data analysis
We used a geographic information system (GIS) to demonstrate the spatial patterns of cardiovascular mortality, and climatic and non-climatic factors of all 358 townships in Taiwan. The cardiovascular mortality 2 weeks before and after the event days, demographics and socio-economic factors were converted to polygon data for each township. Daily maximum, mean and minimum temperatures for each township were also converted using kriging and zonal estimations. Principal-component anal-ysis (PCA) was used to summarise the patterns of intercorrela-tion among 12 demographic and socio-economic factors for each township. Major factors accounting for most of the model variance were then extracted. We used spatial regression models implemented in GeoDa 0.9.5i to measure the relationships between the spatial characteristics of temperature, extracted factors from demographic and socio-economic parameters, and the mean cardiovascular mortality 2 weeks before and after the cold or heat events from 1994 to 2003. The mortality due to cardiovascular diseases for 358 townships was estimated using the number of deaths in the population of the year in the 2-week period before and after the cold and heat events.24In the spatial regression analysis, the mortality at 14 days after the events was a dependent variable (included the event day), and that at 14 days prior to the events was a covariate. Determining factors attributable to the geographical variation in the cardiovascular mortality were then identified under the threats of cold and heat events in Taiwan.
The relationship between the cardiovascular mortality before the cold or heat events, temperature on the day of the event and sociodemographic factors extracted using PCA were examined using spatial regression. Ordinary least square (OLS) was first used for these data, and multicollinearity condition, normality, and spatial dependence and heteroscedasticty were diagnosed by GeoDa 095i.25Spatial lag model estimation, which estimates the
maximum likelihood of a spatial regression model that includes
2 of 6 Wu P-C, Lin C-Y, Lung S-C, et al. Occup Environ Med (2010). doi:10.1136/oem.2010.056168
Original article
group.bmj.com
on November 8, 2010 - Published by
oem.bmj.com
a spatially lagged dependent variable, was then chosen for our analysis according to Lagrange multiplier tests.
RESULTS Disease mapping
The mortalities for cardiovascular diseases before and after cold events were significantly higher than that after heat events. Compared with the mean cardiovascular mortality before cold events (7.64 per 100 000 people), a significantly increase could be found after events (8.29 per 100 000 people). In contrast, cardiovascular mortality only increases slightly after heat events (table 1). Figure 1A,B illustrates the cardiovascular mortalities 14 days after cold or heat events in Taiwan between 1994 and 2003. The range of cardiovascular mortalities after cold events was from 2.4 to 27.0 per 100 000 people and from 1.1 to 21.0 per 100 000 people after heat events. In Taiwan, urban areas had spatial patterns of lower cardiovascular mortalities than rural and mountain areas, suggesting vulnerable regions and popula-tions for higher reported cardiovascular mortalities after exposure to extreme temperatures.
Sociodemographic factors
The descriptive statistics of 12 sociodemographic factors among 358 townships is shown in table 2. PCA was further applied to reduce the number of variables and to classify them into inte-grated factors that reflect their demographic or socio-economic characteristics. Four explanatory factors, including medical resources and urbanisation, percentage of susceptible population and aborigines, and lack of economic opportunity, were found to account for 71.97% variance of the total 12 variables, with each factor accounting for 24.43%, 24.57%, 15.05% and 10.92% variance, respectively (table 3). Factor 1 was a composite factor denoting ‘medical sources and level of urbanisation,’ where dominant variables included the number of clinics and doctors per 10 000 people (loading (L): 0.820, 0.683), home ownership rate (L:0.812), percentage of people engaging in service-related occupations (L: 0.784) and percentage of the population engaged in agricultural occupations (L:e0.484). The rate of older persons living alone (L: 0.902), rate of the older population (L: 0.902) and disability rate (L: 0.690) were the variables that together repre-sented factor 2, the factor of the percentage of susceptible population. The level of social disadvantage, factor 3, included the aborigine population rate (L: 0.873) and uneducated rate among persons older than 15 years (L: 0.770). Factor 4, lack of economic opportunity, was expressed by integrating the vari-ables of unemployment rate (L: 0.634) and percentage of labourers working outside the county of residence. The scores of these four explanatory factors served as independent variables in the following spatial regression models to avoid multi-collinearity. The geographical distribution of four extracted factors is also shown infigure 1s in the supplemental material.
Spatial analysis and regression
With an island-wide high-resolution analysis, the cardiovascular mortalities in each township after the cold events were not only well explained by the cardiovascular mortalities prior to the events but also significantly explained by sociodemographic factors with explanatory power of the full model (R2) of 0.767 for cold events and 0.569 for heat events (table 4). Except factor 4, lack of economic opportunity, three sociodemographic factors were significantly related with the mortalities after the cold and heat events. Independent of temperature changes, the higher score of factor 2, the ‘susceptible population,’ implied that a higher percentage of older people, older people living alone and the disabled was associated with rising mortalities at the township level, with both cold (
b
¼1.338) and heat (b
¼0.954) events. Elevated cardiovascular mortalities after the events were inversely associated with the score of factor 1,‘medical resources availability and the degree of urbanisation,’ with similar regression coefficients for both cold (b
¼e0.601) and heat (b
¼e0.456) exposures. To some extent, the explanatory power of these variables that was attributed to their in-township value, was really due to the neighbouring locations.To the best of our knowledge, our study provides the first scientific evidence, obtained via a nationwide data analysis, substantiating the critical contribution of medical resources availability in preventing deaths associated with exposure to extreme temperatures. Regarding factor 3, the higher percentage of the aborigine population was also found to be more vulner-able to extreme temperature exposures, particularly heat events (
b
¼0.760 vsb
¼0.308 in cold events). Moran’s I test statistic of the spatial residues wase0.022 for cold events and e0.041 for heat events, implying that including the spatially lagged dependent variable term in the model eliminated spatial dependence.DISCUSSION
Our results suggest that no independent effect could be assumed for the levels of temperature exposure on the higher cardiovas-cular mortalities during the cold and heat events because these rates appeared to be significantly modified by sociodemographic factors. The susceptible population (older people and disabled) and aborigine population in Taiwan are vulnerable subgroups whose needs should be prioritised when framing future adap-tation policies. Improving healthcare resources could be a critical determinant of the effectiveness of adaptation strategies. This is a tremendous public health challenge that will require some form of feasible and effective community-wide support and emergency response plans for persons with relatively fewer healthcare resources.
Our study reaffirms that older people are the most susceptible to temperature-related mortality, particularly exposure to extreme cold.26e28Ageing populations often have more under-lying diseases; they also have poorer thermoregulation. Furthermore, disabled persons are susceptible to dramatic changes in temperature and also not able to adapt to the changes because of self-care difficulties.
We believe that this is the first study to provide scientific evidence that the aboriginal population is uniquely vulnerable to exposure to extreme temperatures, particularly heat. The underlying risk factors present in the aboriginal population could possibly justify, to a great extent, why the observations are plausible. Researchers have offered the several biological and genetic differences in the thermoregulatory system of aborigines, along with a higher prevalence of chronic illnesses in this Table 1 Mean cardiovascular mortality 14 days before/after events
across all townships (N¼358)
Cardiovascular mortality (1/100 000)* Cold events (24 events) Heat events (13 events) p Valuey 14 days before events 7.64 (3.63) 5.87 (3.06) <0.001 14 days after events 8.29 (3.76) 5.91 (3.26) <0.001 p Valuez <0.001 0.782
*Mean (SD).
yp Value for independent-sample t test. zp Value for paired-sample t test.
population, as explanatory rationales.29We have quantitatively demonstrated that newly identified socio-economic and lifestyle risk factors may be far more critical than the variables studied before. In Taiwanese society, the aborigines in general receive less formal education and earn less. A higher percentage of aborigines also engage in unhealthy lifestyles, such as heavy alcohol drinking and tobacco smoking.30 Health statistics shows that
the overall mortality in aborigine townships is about 70% higher than in the general population in Taiwan.31 This underlying
inequality in health status appears to have affected the aborig-ines, making them even more vulnerable to extreme weather
events. Therefore, a health promotion policy should start by improving the socio-economic status of, and education oppor-tunities for, this disadvantaged population. Moreover, traditional practices of biomass burning to keep warm might also contribute to severe indoor pollutants that have been known to be involved in the biological mechanism that triggers cardio-vascular attacks.32Thisfinding suggests that it is essential that
local administrative agencies immediately implement an early warning and emergency response programme for aborigine communities, which includes air-conditioning of public build-ings for cooling during heat events and an education programme about proper activities for keeping warm such as proper fuel selection and the use of an overcoat, gloves, scarf and hat during winter.
Table 2 Descriptive statistics of socio-economic factors (N¼358)
Minimum Maximum Mean SD No of clinics per 10 000 population 0.010 0.396 0.0616 0.042 No of doctors per 10 000 population 0.000 1.734 0.0944 0.141 Home ownership rate (percentage of
residence household)
67.497 98.935 90.902 5.208
Service occupation rate (percentage of working population)
22.542 77.314 46.300 12.401
Agriculture occupation rate (percentage of working population)
0.084 0.588 0.157 0.142
Rate of older population (percentage of persons over 65)
0.038 0.230 0.108 0.037
Rate of older persons living alone (percentage of persons over 65 living alone)
0.004 0.048 0.017 0.008
*Disability rate (percentage of any disability item among total population)
0.818 11.610 2.011 0.872
Aborigine population rate (percentage of total population)
0.024 97.588 9.320 24.084
Uneducated rate (percentage of uneducated among persons over 15)
0.000 14.286 1.203 1.381
yUnemployment rate (percentage of unemployment among persons over 15)
10.007 47.489 24.348 5.016
Rate of labourers working outside the county of residence (percentage of working population)
6.667 66.652 36.974 10.320
*Disability: percentage of any disability item (self-care difficulty, independent living difficulty and ambulatory difficulty) among the total population.
yUnemployment rate was estimated by the sum of subjects not studying at school and also unemployed among persons over 15.
Table 3 Factors loading and percentage of variance explained by social and demographic factors
Components
Percentage
variance explained Loading Factor 1: Medical resources and
urbanisation
24.43
No of clinics per 10 000 population 0.82 No of doctors per 10 000 population 0.68
Home ownership rate 0.81
Service occupation rate 0.78 Agriculture occupation rate 0.48 Factor 2: Susceptible population 21.57
Rate of older persons living alone 0.90 Rate of older population 0.90
Disability rate 0.69
Factor 3: Aborigine population 15.05
Aborigine population rate 0.87
Uneducated rate 0.77
Factor 4: Lack of economic opportunity 10.92
Unemployment rate 0.63
Rate of labourers working outside the county of residence
0.75
Sum of variance explained 71.97
figure 1 Mean cardiovascular mortality 14 days after (A) cold and (B) heat events.
4 of 6 Wu P-C, Lin C-Y, Lung S-C, et al. Occup Environ Med (2010). doi:10.1136/oem.2010.056168
Original article
group.bmj.com
on November 8, 2010 - Published by
oem.bmj.com
Previous studies have indicated that the regional variation in cardiovascular mortalities could be attributed to treatment delays because of the distance from the emergency hospital or healthcare services.33 We also identified that adequate medical
resources, in terms of facilities or professionals, could have a significantly protective effect on the reported cardiovascular mortalities after cold or heat events. One can easily conclude from the established relationship that when formulating a related preventive public health policy, important consider-ations are improving the availability of prehospital care such as thrombolytic therapy for myocardial infarctions and enhancing ambulance services for remote communities or rural areas.
We have not seen in our study the influence of ‘economic opportunities,’ a factor integrating the unemployment rate and working opportunities in the township, on the reported cardiovascular mortalities after cold or heat events. This result might indicate that this factor may not properly explain the ‘poverty’ status and housing conditions that might play key roles in preventing exposure to extreme cold or heat.
Studies, using different geographical levels as the unit of analysis, including cross country, nationwide and inner city levels, have yielded inconsistent conclusions on how socio-economic factors, such as income poverty, social deprivation levels and fuel poverty, may be associated with diseases outcomes linked to environmental changes.4 5 17 34e37Future investigations should aim to clarify whether the differences found across regions with respect to the observed disease outcomes can be linked to ethnics or the related biological mechanisms.
Overall, cold exposures, rather than heat events, appear to have a higher impact on the reported cardiovascular mortalities of populations in Taiwan, which has warm summers and mild winters. Although the Taiwanese are expected to have developed physiological and behavioural adaptations to warmer tempera-tures, most buildings in Taiwan have air-conditioning, especially those in urban areas. We also found that the vulnerable popula-tions were located in areas with relatively poor housing insu-lation and therefore were less protected from cold exposures compared with subjects of studies conducted in southern Europe.14 17 More comprehensive comparative investigations should be conducted to evaluate how elevated temperatures during cold events in winter due to climate change may affect the burdens of mortality or morbidities. Moreover, research suggests that the effects of extreme heat on mortality have a shorter lag (0e1 day) than the effects of cold and might cause heat-related short-term mortality displacement.38 This preliminary study,
which used data from a 2-week window during cold and heat events, might not be able to detect heat-related increases in cardiovascular mortality. This might also explain why more substantial findings were obtained during cold events. High humidity is a well-known crucial factor in evaporative heat loss during heat exposure.39In the context of spatial analysis, spatial
differences in relative humidity might also partially explain our model. In this study, we could not obtain geographic represen-tative data for humidity because the monitoring stations lacked proper spatial distribution. The heat index in combination with temperature and humidity in certain vulnerable regions could be further used to examine the combined effects of these parameters in the future. Studies to detect transient cold- or heat-related increases in cardiovascular mortality should be conducted in regions with different vulnerabilities in Taiwan to develop response plans.
This spatial analysis used the highest quality of spatial data to identify the vulnerable regions in terms of climate-change-related health risks and to identify the variables that determine the vulnerability integrated with susceptibility and adaptive capability. Our study supports the hypothesis that cold events in winter may cause more deaths than heat in summer, even in many subtropical regions.40 The percentage of the susceptible population comprising older people, disabled and aborigines, and the availability of medical resources are considered key deter-minants of the reported cardiovascular mortalities. Given our findings, a national intervention programme that prioritises the needs of these high-risk subgroups should be developed.
When considering our conclusions regarding cold and heat events, it is important to bear in mind that these results depend greatly on the definition of cold and heat events used in our region. We attempted to explore the geographical differences in vulnerability under the threat of heat and cold exposure, using cardiovascular mortality before and after the events in our spatial model. Instead of concerning ourselves with over-smoothing using a over-smoothing technique in displaying the map of mortality, we used mortality from 14 days after the event as the dependent variable and the mortality from 14 days prior to the event as a covariate in the full model. However, a potential bias might have resulted from estimating mortality, particularly from a small population at risk, and not using smoothing to adjust the data.41In the context of this study, a confounder is a factor that is statistically associated with the exposure and also an independent risk factor for diseases. In this study, many non-climatic factors such as change in urbanisation status, personal protection and housing quality are important potential Table 4 Examining the spatial relationships among cardiovascular mortality after the cold and heat
events, baseline of mortality, event temperature, socio-economic and demographic factors by using the spatial lag model.
Cold events (24 events), R2[0.77
Heat events (13 events), R2[0.57
Coefficients 95% CI Coefficients 95% CI Cardiovascular mortality before cold or heat events 0.54y 0.44 to 0.64 0.40y 0.29 to 0.51 Mean temperature of cold or heat events <0.01 0.08 to 0.08 0.03 0.07 to 0.12 Factor 1: Medical resources and urbanisation 0.60y 0.83 to 0.37 0.46y 0.71 to 0.21 Factor 2: Susceptible population 1.34y 1.01 to 1.66 0.95y 0.61 to 1.29 Factor 3: Aborigine population 0.31y 0.10 to 0.51 0.76y 0.50 to 1.02 Factor 4: Lack of economic opportunity 0.08 0.27 to 0.12 0.21* 0.44 to 0.02 Constants 4.19y 2.66 to 5.71 2.30* 0.44 to 5.04 Rho <0.01 0.09 to 0.08 0.10* 0.01 to 0.21 Moran’s I value residues 0.022 0.04
*p<0.1. yp<0.05.
confounders. Moreover, we cannot test the modifying effects of air pollution on cardiovascular mortality in this study due to the lack of high-spatial-resolution data on air pollution across Taiwan. Another limitation of this ecological analysis for making causal inferences is ecological bias, which is the failure of an expected ecological effect estimate to reflect the biological effect at the individual level.
Acknowledgements The authors thank the Department of Health for providing access to the computerised database of daily death registrations. We are also indebted to our colleagues who participated in data key-in and management. Funding Taiwan National Science Council (NSC95-2625-Z006-018) grants have, in part, provided critical financial support to this study.
Competing interests None. Patient consent Obtained.
Provenance and peer review Not commissioned; externally peer reviewed.
REFERENCES
1. Mercer JB. Cold: an underrated risk factor for health: Combined impact of exercise and temperature stress on the physiological response to toxic agent. Environ Res 2003;92:8e13.
2. Curriero FC, Heiner KS, Samet JM, et al. Temperature and mortality in 11 cities of the eastern United States [see comment]. Am J Epidemiol 2002;155:80e7. 3. Carder M, McNamee R, Beverland I, et al. The lagged effect of cold temperature
and wind chill on cardiorespiratory mortality in Scotland. Clin Occup Environ Med 2005;62:702e10.
4. El-Zein A, Tewtel-Salem M. On the association between high temperature and mortality in warm climates. Sci Total Environ 2005;343:273e5.
5. Gouveia N, Hajat S, Armstrong B. Socioeconomic differentials in the temperatureemortality relationship in Sa˜o Paulo, Brazil [see comment]. Int J Epidemiol 2003;32:390e7.
6. Keatinge WR, Coleshaw SR, Cotter F, et al. Increases in platelet and red cell counts, blood viscosity, and arterial pressure during mild surface cooling: factors in mortality from coronary and cerebral thrombosis in winter. BMJ 1984;289:1405e8. 7. Neild PJ, Syndercombe-Court D, Keatinge WR, et al. Cold-induced increases in
erythrocyte count, plasma cholesterol and plasma fibrinogen of elderly people without a comparable rise in protein C or factor X. Clin Sci (Lond) 1994;86:43e8. 8. Imai Y, Munakata M, Tsuji I, et al. Seasonal variation in blood pressure in
normotensive women studied by home measurements. Clin Sci 1996;90:55e60. 9. Kovats RS, Johnson H, Griffith C. Mortality in southern England during the 2003 heat
wave by place of death. Health Stat Q 2006;29:6e8.
10. Rozzini R, Zanetti E, Trabucchi M. Elevated temperature and nursing home mortality during 2003 European heat wave. J Am Med Dir Assoc 2004;5:138e9. 11. Kovats RS, Kristie LE. Heatwaves and public health in Europe [see comment]. Eur J
Public Health 2006;16:592e9.
12. Keatinge W. Increased platelet and red cell counts, blood viscosity, and plasma cholesterol level during heat stress, and mortality from coronary and cerebral thrombosis. Am J Med 1986;81:795e800.
13. Pan WH, Li LA, Tsai MJ. Temperature extremes and mortality from coronary heart disease and cerebral infarction in elderly Chinese. [see comment]. Lancet 1995;345:353e5.
14. The Eurowriter group. Cold exposure and winter mortality from ischaemic heart disease, cerebrovascular disease, respiratory disease, and all causes in warm and cold regions of Europe. The Eurowinter Group. [see comment]. Lancet 1997;349:1341e6.
15. Hajat S, Kovats RS, Lachowycz K. Heat-related and cold-related deaths in England and Wales: who is at risk? Occup Environ Med 2007;64:93e100.
16. Wilkinson P, Pattenden S, Armstrong B, et al. Vulnerability to winter mortality in elderly people in Britain: population based study [see comment]. BMJ 2004;329:647.
17. Healy JD. Excess winter mortality in Europe: a cross country analysis identifying key risk factors. J Epidemiol Community Health 2003;57:784e9.
18. Bell ML, Neill MSO, Ranjit N, et al. Vulnerability to heat-related mortality in Latin America: a case-crossover study in Sa˜o Paulo, Brazil, Santiago, Chile and Mexico City, Mexico. Int J Epidemiol 2008;37:796e804.
19. Department of Health T. Statistics of Causes of Death. Taipei, Taiwan, ROC: Department of Health, 2007.
20. Yeh CJ, Chan P, Pan WH. Values of blood coagulating factors vary with ambient temperature: the Cardiovascular Disease Risk Factor Two-Township Study in Taiwan. Chin J Physiol 1996;39:111e16.
21. Chen TC, Huang WR, Yoon Jh. Interannual variation of the East Asian cold surge activity. J Clim 2004;17:401e13.
22. Lin CY, Lung SC, Guo HR, et al. Climate variability of cold surge and its impacton the air quality of Taiwan. Clim Change 2009;94:457e71.
23. Baily TC, Gatrell AC. Interactive Spatial Data Analysis. Harlow, UK: Longman Group Limited, 1995.
24. Yanga TC, Wu PC, Chen VYJ, et al. Cold surge: A sudden and spatially varying threat to health? Sci Total Environ 2009;407:3421e4.
25. Anselin L. Exploring Spatial Data with GeoDa: A Workbook. Urbana-Champaign. Department of Geography, University of Illinois. IL: Spatial Analysis Laboratory, 2005. 26. Woodall AA. Winter mortality in elderly people in Britain: outdoor exposure and
effect of windchill should be taken into consideration [comment]. BMJ 2004;329:976; author reply 977.
27. Keatinge WR, Donaldson GC. Winter mortality in elderly people in Britain: action on outdoor cold stress is needed to reduce winter mortality [comment]. BMJ 2004;329:976; author reply 977.
28. Koken PJM, Piver WT, Ye F, et al. Temperature, air pollution, and hospitalization for cardiovascular diseases among elderly people in Denver. Environ Health Perspect 2003;111:1312e17.
29. Taylor NAS. Ethnic differences in thermoregulation: Genotypic versus phenotypic heat adaptation J Therm Biol 2006;31:90e104.
30. Ho CS, Tsai AC. Prevalence of overweight and obesity and its associated factors in aboriginal Taiwanese: findings from the 2001 National Health Interview Survey in Taiwan. Asia Pac J Clin Nutr 2007;16:572e9.
31. Wen CP, Shan PT, Shih YT, et al. Bridging the gap in life expectancy of the aborigines in Taiwan. Int J Epidemiol 2004;33:320e7.
32. Fullerton DG, Bruce N, Gordon SB. Indoor air pollution from biomass fuel smoke is a major health concern in the developing world. Trans R Soc Trop Med Hyg 2008;102:843e51.
33. Goldberg RJ, Steg PG, Sadiq I, et al. Extent of, and factors associated with, delay to hospital presentation in patients with Acute Coronary disease (the GRACE registry). Am J Cardiol 2002;89:791e6.
34. Gyllerup S, Lanke J, Lindholm LH, et al. Socioeconomic factors in the community fail to explain the high coronary mortality in cold parts of Sweden. Eur Heart J 1992;13:878e81.
35. Shah S, Peacock J. Deprivation and excess winter mortality [see comment]. J Epidemiol Community Health 1999;53:499e502.
36. Seretakis D, Lagiou P, Lipworth L, et al. Changing seasonality of mortality from coronary heart disease. JAMA 1997;278:1012e14.
37. O’Neill MS, Zanobetti A, Schwartz J. Modifiers of the temperature and mortality association in seven US cities. Am J Epidemiol 2003;157:1074e82.
38. Hajat S, Armstrong BG, Gouveia N, et al. Mortality displacement of heat-related deaths: a comparison of Delhi, Sa˜o Paulo, and London. Epidemiology
2005;16:613e20.
39. Kim H, Ha JS, Park J. High temperature, heat index, and mortality in 6 major cities in South Korea. Arch Environ Occup Health 2006;61:265e70.
40. Keatinge WR, Donaldson GC. The impact of global warming on health and mortality. South Med J 2004;97:1093e9.
41. Marshall RJ. Mapping disease and mortality rates using empirical Bayes estimators. Appl Stat 1991;40:283e94.
6 of 6 Wu P-C, Lin C-Y, Lung S-C, et al. Occup Environ Med (2010). doi:10.1136/oem.2010.056168
Original article
group.bmj.com
on November 8, 2010 - Published by
oem.bmj.com
