Skip main navigation

Associations Between Extreme Temperatures and Cardiovascular Cause-Specific Mortality: Results From 27 Countries

Originally publishedhttps://doi.org/10.1161/CIRCULATIONAHA.122.061832Circulation. 2023;147:35–46

Abstract

Background:

Cardiovascular disease is the leading cause of death worldwide. Existing studies on the association between temperatures and cardiovascular deaths have been limited in geographic zones and have generally considered associations with total cardiovascular deaths rather than cause-specific cardiovascular deaths.

Methods:

We used unified data collection protocols within the Multi-Country Multi-City Collaborative Network to assemble a database of daily counts of specific cardiovascular causes of death from 567 cities in 27 countries across 5 continents in overlapping periods ranging from 1979 to 2019. City-specific daily ambient temperatures were obtained from weather stations and climate reanalysis models. To investigate cardiovascular mortality associations with extreme hot and cold temperatures, we fit case-crossover models in each city and then used a mixed-effects meta-analytic framework to pool individual city estimates. Extreme temperature percentiles were compared with the minimum mortality temperature in each location. Excess deaths were calculated for a range of extreme temperature days.

Results:

The analyses included deaths from any cardiovascular cause (32 154  935), ischemic heart disease (11 745 880), stroke (9 351 312), heart failure (3 673 723), and arrhythmia (670 859). At extreme temperature percentiles, heat (99th percentile) and cold (1st percentile) were associated with higher risk of dying from any cardiovascular cause, ischemic heart disease, stroke, and heart failure as compared to the minimum mortality temperature, which is the temperature associated with least mortality. Across a range of extreme temperatures, hot days (above 97.5th percentile) and cold days (below 2.5th percentile) accounted for 2.2 (95% empirical CI [eCI], 2.1–2.3) and 9.1 (95% eCI, 8.9–9.2) excess deaths for every 1000 cardiovascular deaths, respectively. Heart failure was associated with the highest excess deaths proportion from extreme hot and cold days with 2.6 (95% eCI, 2.4–2.8) and 12.8 (95% eCI, 12.2–13.1) for every 1000 heart failure deaths, respectively.

Conclusions:

Across a large, multinational sample, exposure to extreme hot and cold temperatures was associated with a greater risk of mortality from multiple common cardiovascular conditions. The intersections between extreme temperatures and cardiovascular health need to be thoroughly characterized in the present day—and especially under a changing climate.

Clinical Perspective

What Is New?

  • This study provided evidence from what we believe is the largest multinational dataset ever assembled on cardiovascular outcomes and environmental exposures.

  • Extreme hot and cold temperatures were associated with increased risk of death from any cardiovascular cause, ischemic heart disease, stroke, and heart failure.

  • For every 1000 cardiovascular deaths, 2 and 9 excess deaths were attributed to extreme hot and cold days, respectively.

What Are the Clinical Implications?

  • Extreme temperatures from a warming planet may become emerging priorities for public health and preventative cardiology.

  • The findings of this study should prompt professional cardiology societies to commission scientific statements on the intersections of extreme temperature exposure and cardiovascular health.

The second half of the 20th century witnessed significant public health gains in cardiovascular disease (CVD) prevention, mainly by addressing CVD traditional risk factors, such as tobacco use, hypertension, hyperlipidemia, diet, and physical activity.1 Despite this, CVD remains the leading cause of death worldwide, claiming 18 million lives yearly.2 Expanding public health assessments towards nontraditional risk factors, such as environmental exposures, could bring new insights into CVD prevention.

Climate change has led to substantial swings in extreme ambient temperature and increases in mortality.3 The Global Burden of Disease Study has recently introduced nonoptimal temperatures among the leading risk factors of mortality worldwide.4 In addition, a recent global analysis showed that 9.4% of all deaths can be attributed to high and low nonoptimal temperatures.5 Both low and high ambient temperatures were shown to contribute to CVD mortality.6–10 However, most temperature-related mortality studies considered total death counts combining deaths from all-causes, nonaccidental causes, or all cardiovascular causes as one stratum. Previous studies also used different designs and methodological approaches and were confined to limited geographical regions and climate zones with little attention for potential confounding from other environmental factors. Analyzing associations between extreme temperatures and specific CVD outcomes can identify vulnerable subgroups to mobilize healthcare resources and inform recommendations that is specific to CVDs.

In this study, we assembled a large global database on CVD-specific causes of death across the world from the Multi-Country Multi-City (MCC) Collaborative Research Network to investigate associations between extreme temperatures and all CVD causes, ischemic heart disease, stroke, heart failure, and arrhythmia using a standardized analytic protocol.

Methods

Data Collection

Investigators from each country of the MCC network (http://mccstudy.lshtm.ac.uk/) were contacted to extract specific causes of CVD mortality from national and regional death registries based on InternationalClassification ofDiseases, Ninth (ICD-9) and Tenth Revisions (ICD-10) codes. We collected the statistical underlying cause of death, defined by the World Health Organization as the disease that initiated the series of events that directly led to death. The analyses to locations with available cause-specific CVD mortality data. We used daily death counts of all cardiovascular causes (ICD-9; 390-459, ICD-10; I00-I99), ischemic heart disease (ICD-9; 410-414, ICD-10; I20-I25), stroke (ICD-9; 430-438, ICD-10; I60-I69), heart failure (ICD-9; 428, ICD-10; I50), and arrhythmia (ICD-9; 427 [except for 427.5], ICD-10; I47-I49).

The MCC environmental database has been described previously.3,11,12 In brief, city-specific daily ambient temperatures (expressed in °C) were obtained either from meteorological stations in national and regional networks or climate reanalysis data (Table S1). The evidence suggests that, at a large global scale, reanalysis weather data is comparable to observed data obtained from monitoring station.13 In locations where 2 or more stations were available, the mean across all stations was computed for each day. Temperature and mortality data were available in overlapping periods ranging from as early as January 1, 1979 in Japan, to as late as December 31, 2019 in Ecuador and Paraguay. Wherever available, we also collected daily time series data of relative humidity (24-hour average, expressed as percentage) and air pollutants such as large particulate matter (ie, particulate matter with aerodynamic diameter <10 µm; 24-hour average, in µg/m3), fine particulate matter (particulate matter with aerodynamic diameter <2.5 µm; 24-hour average, in µg/m3), near surface ozone (maximum 8-hour average, in µg/m3), and nitrogen dioxide (24-hour average, in µg/m3). In general, city-specific air pollution concentrations were derived from daily measurements of 1 or more monitors in national or regional networks. The pollution data has been described in previous MCC studies.11,12,14 We collected values for country-level gross domestic product (GDP) per capita from the World Bank. Climate zones were assigned to each city using the Köppen-Geiger climate classification.15

Data was collected within the MCC Collaborative Research Network under a data sharing agreement and cannot be made publicly available. Institutional approvals were obtained by each MCC participant in their respective country. More information about institutional sources of data can be found in the supplemental material.

Statistical Analysis

The analyses follow a 2-stage approach: (1) at each city, we estimate the association between extreme temperatures and death from a cardiovascular cause; and (2) then we pool all the risk estimates from individual cities to get an overall estimate.

In a case-crossover design, conventionally, each individual case serves as their own control to estimate the acute associations between short-term temperature exposures and CVD events; effectively eliminating potential confounding from individual characteristics such as age, gender, diet, and smoking, among others.16 Since we had count data, we fit conditional quasi-Poisson models for each city, including 3-way interaction terms between year, month, and day of the week as a flexible time series alternative for a case-crossover design.17 The temperature-mortality association in each city was modeled as a continuous bidimensional function of temperature and lag using the distributed lag nonlinear models. Consistent with previous MCC studies, the exposure–response dimension was modeled with a quadratic B-splines with 3 internal knots placed at the 10th, 75th, and 90th percentiles of temperature for each city while the lag–response dimension was modeled with a natural spline with 3 internal knots equally spaced in the log scale.5,18 We considered lag periods up to 14 days to allow for delayed effects and the possibility of mortality displacement, in which deaths are advanced only a few days before when they would have occurred without the exposure.

For the second stage, to pool city-specific effect estimates, we applied a novel hierarchical extended mixed-effects framework for meta-analysis for nested groups.19 We fit city-specific mean summer temperature, mean winter temperature, and country-level GDP per capita as fixed meta-predictors in the meta-regression. These meta-predictors account for potential effect modification by climatologic and socioeconomic factors on the temperature–mortality relationships from different locations. In the same meta-regression model, we also fitted 2-level random effects where cities are nested within country-specific climate zones, allowing cities from the same country and climate zone to borrow information from each other. Each city has its own temperature–mortality relationships allowing us to account for differences across them. We examined the heterogeneity between cities first with intercept only and then with meta-predictors and random effects using extended versions of the Cochran Q test and I2 statistic.20 We predicted the pooled and country-level relative risks (RR) from the fixed meta-predictor model. The estimates of the temperature–mortality relationships at the city level were derived from the best linear unbiased predictions with risks modeled on continuous scale of temperature. For each cause of death, for each city and country, we empirically identified the Minimum Mortality Temperature (MMT), which is the temperature that is associated with the least mortality risk without imposing constraints on its location. The MMT is dependent on human adaptability to local climate reflecting the most comfortable, optimum temperature in a given location.21,22

There is evidence that people can adapt, at least partially, to the temperatures that they are used to experiencing.23–25 Therefore, the impact of given extreme temperature event on human health can depend on where and when it occurs. In some locations a 35°C day is a typical summer day, whereas in other locations it can cause incalculable damage. Accordingly, temperature extremes are defined by comparison to some local average (eg, the top 1% of warmest days recorded in a particular location) rather than to an absolute temperature (such as 35°C).25 The RR of death by CVD outcome is, therefore, reported for extreme heat as the location-specific 99th percentile temperature versus MMT, and extreme cold as the 1st percentile temperature versus the MMT. We then estimated the proportion of excess deaths that are attributable to extreme temperatures ranges. Ranges of extreme cold and hot days were also defined as location-specific days that were below the 2.5th or above the 97.5th temperature percentile, respectively.18 We computed empirical CIs (eCI) using Monte Carlo simulations assuming a multivariate normal distribution of the best linear unbiased predictions of reduced coefficients.

In additional analyses, we computed excess deaths associated with all hot and cold temperatures above and below the MMT. We also stratified the cities by quartiles of GDP per capita and ranked them by the relative risk for each CVD cause of death.

To assess robustness of findings, we modeled the exposure–response curves with 4 knots at the 10th, 50th, 75th, and 90th percentiles, 5 knots at the 5th, 25th, 50th, 75th, and 95th percentiles, and increased the lag period up to 21 days. We then adjusted for relative humidity, ozone, nitrogen dioxide, particulate matter with aerodynamic diameter <10 µm, and particulate matter with aerodynamic diameter <2.5 µm in cities where such data was available. We also adjusted for other temperature variables such as interday temperature variability and sustained durations of heatwaves. Last, we controlled for very slow decadal changes in mortality and/or temperature trends by including a natural spline function of time with 1 knot every 10 years. All analyses were carried out using R software (version 4.2.1) with gnm, dlnm, and mixmeta packages. Detailed steps of the statistical analyses are presented in the Supplemental Methods.

Results

Descriptive Analyses

The analyses included 567 cities from 27 countries. We examined a total of 32 154 935 all-cause CVD deaths, 11 745 880 ischemic heart disease deaths, 9 351 312 stroke deaths, 3 673 723 heart failure deaths, and 670 859 arrhythmia deaths, covering largely overlapping years between 1979 and 2019 (average time series, 16 years per country; range, 3–36 years; Table 1). Overall temperature average varied by geographical and climate region (Figure 1). The variance of temperatures was also heterogeneous as evident by the distributions in each country (Figure S1). Countries that have temperate climates such as those in Caribbean and Central America and Southeastern Asia had a narrow range of temperatures, while the majority of other countries had a considerable variance with dual peaks of frequency for cold winters and hot summers. Across the cities, the range of temperatures varied from, for example, −30°C in Helsinki, Finland to 44°C in Kuwait City, Kuwait. Similarly, the MMT percentiles for all-cause CVD varied across countries but ranged from the 78th (Taiwan) to the 93rd (Estonia) percentiles (Table S2, Figure S1).

Table. Number of Cardiovascular Cause-Specific Deaths in Each Country

CountryYearsLocationsAll-Cause cardiovascular diseaseIschemic heart diseaseStrokeHeart failureArrhythmia
North America
 Canada1986–2015261 264  609692 655252 52464 93031 295
 United States1985–200620911 108  8245 894 9811 768 235487 512166 761
Caribbean and Central America
 Guatemala2009–2018115 137610833101636472
 Costa Rica2000–20171928832781910123
 Panama2013–20161974715931101
South America
 Uruguay2001–2018178 69219 41027 89059151667
 Ecuador2013–2019248 20217 79411 5271002619
 Paraguay2004–2019115 371423145081212435
 Brazil1997–2018121 176  298379 402358 91972 04221 077
South Africa
 South Africa1997–2013521 299  688203 319398 239245 9369995
North Europe
 Finland1987–2018190 99249 46223 30012011541
 Estonia1997–20189159 66472 41031 6731845764
 United Kingdom1990–2016702 258  2961 167 871568 58880 72936 372
Central Europe
 Switzerland1995–20168321 657130 15659 24530 1237485
 Moldova2001–2010133 08722 0629830
South Europe
 Portugal1990–20186531 859132 056241 55257 7869445
 Spain2000–20186356 60299 81090 29558 25314 346
 Italy2006–20156121 80540 76027 47952814524
 Cyprus2004–2017526 6298655541738211,151
Middle East Asia
 Iran2001–20172417 756120 78071 31255 09413 716
 Kuwait2000–2016135 28517 25166154620234
East Asia
 South Korea1997–201836701 638158 637350 137
 Japan1979–20154711 541  8972 348  9384 850  1752 449 671333 805
 Taiwan2008–20163107 21028 54732 70810 2274472
Southeast Asia
 Thailand1999–200855315 10587 701115 21928 2089579
 Philippines2006–2010487 40134 40630 1782618927
 Vietnam2010–2013122 196360794263872
Totals1979–201956732 154  93511 745  8809 351  3123 673 723670 859
Figure 1.

Figure 1. Map of all 567 locations. Each city is represented with average annual temperature in °C.

The overall missing rates for data for temperature, all-cause CVD, ischemic heart disease, stroke, heart failure, and arrhythmia time series were 0.71%, 0.08%, 0.38%, 0.26%, 0.76%, and 0.24%, respectively. Detailed descriptive statistics and information on missing data are summarized in Tables S1 to S3.

Regression Analyses

The pooled overall temperature and CVD mortality relationships were non-linear with increased mortality risk in both hot and cold temperature ranges (Figure 2). The RR of death increased gradually for the cold temperatures below the MMT, while the slope for hot temperatures was slightly steeper especially with heart failure where the RRs appear to escalate quickly. The pooled RRs of death associated with extreme heat (99th percentile vs MMT) from ischemic heart disease, stroke, and heart failure were 1.07 (95% CI, 1.04–1.10), 1.10 (95% CI, 1.06–1.15), and 1.12 (95% CI, 1.05–1.19), respectively (Figure 3A). Meanwhile, the pooled RRs of death associated with extreme cold (1st percentile vs MMT) from ischemic heart disease, stroke, and heart failure were 1.33 (95% CI, 1.26–1.41), 1.32 (95% CI, 1.26–1.38), and 1.37 (95% CI, 1.28–1.47), respectively (Figure 3B). Risk of dying from arrhythmias was associated with greater uncertainty and smaller effect size estimate for extreme heat (1.05 [95% CI, 0.98–1.12]) and cold (1.19 [95% CI, 1.07–1.33]). Exposure–response curves with temperature in the absolute scale (°C) are shown in Figures S2 and S3.

Figure 2.

Figure 2. Pooled exposure–response relationships. Relationships are described as temperature percentiles and relative risk (RR) of different cardiovascular causes of death: all-cause cardiovascular (567 cities), ischemic heart disease (567 cities), stroke (567 cities), heart failure (524 cities), and arrhythmia (441 cities). Dashed line indicates the minimum mortality temperature. Dotted line indicates the 1st percentile (extreme cold) and the 99th percentile (extreme heat). X-axis was transformed from absolute temperatures (°C) to percentiles to enable a comparative application of the association. CVD indicates cardiovascular disease.

Figure 3.

Figure 3. Relative risks of death from different cardiovascular-specific causes. A, Relative risks of death from extreme heat comparing the 99th percentile to the minimum mortality temperature in each country as well as the pooled estimate. B, Relative risks (RR) of death from extreme cold comparing the 1st percentile to the minimum mortality temperature in each country, as well as the pooled estimate.

For every 1000 all-cause CVD deaths, 2.2 (95% eCI, 2.1–2.3) excess deaths were attributed to extreme range of hot temperatures above the 97.5th percentile. The highest burden from extreme hot temperatures was seen for heart failure with 2.6 (95% eCI, 2.4–2.8) excess deaths for every 1000 heart failure deaths (Figure 4A). Excess mortality attributable to an extreme range of cold temperatures below the 2.5th percentile was consistently higher than heat and accounted for 9.1 (95% eCI, 8.9–9.2) for every 1000 all-cause CVD deaths and 12.8 (95% eCI, 12.2–13.1) for every 1000 heart failure deaths (Figure 4B). Country-specific risk estimates, excess deaths from extreme and all temperature ranges, and cities at highest risk are provided in the Tables S4 to S7.

Figure 4.

Figure 4. Excess deaths from different cardiovascular-specific causes. A, Proportion of excess deaths associated with a range of extreme hot temperatures above the 97.5th percentile, expressed as number of deaths for each 1000 cardiovascular-specific deaths in each country as well as the pooled estimate. B, Proportion of excess deaths associated with a range of extreme cold temperatures below the 2.5th percentile, expressed as number of deaths for each 1000 cardiovascular-specific deaths in each country as well as the pooled estimate.

Using simple models (with intercept only), the heterogeneity (I2 statistic) in second-stage meta-analyses for mortality from all-cause CVD, ischemic heart disease, stroke, heart failure, and arrhythmia were 55.0%, 36.9%, 23.7%, 21.9%, and 5.2%, respectively. After including the meta-predictors (GDP per capita, mean summer and winter temperatures) and random effects (country and climate zones), the corresponding heterogeneity was reduced to 50.0%, 33.5%, 17.6%, 17.9%, and 4.5%, respectively (Table S8). Exposure–response curves stratified by GDP per capita and mean summer and winter temperatures are shown in Figures S4 to S6. For stroke and heart failure, countries with low GDP per capita seem to have higher risks of mortality compared to countries with high GDP per capita, whereas for ischemic heart disease it is the opposite (Figure S4).

The results from the main model were robust to different modelling choices and adjustments for potential time-varying confounders (Table S9). Changing the number and location of knots in the temperature exposure–response curve did not substantially change the results. Increasing the lag period from 14 to 21 days was associated with slightly higher RR and excess deaths from cold temperatures. The results were not affected by adjustments for temperature variability, heatwaves, long-term trends, relative humidity, and air pollutants (eg, ozone, nitrogen dioxide, particulate matter with aerodynamic diameter <10 µm, and particulate matter with aerodynamic diameter <2.5 µm [included 1 at a time]).

Discussion

Our study provides evidence for an association between extreme temperatures and CVD cause-specific mortality from what we believe is the largest multinational dataset ever assembled. The data includes outcomes from countries in different climate zones, and with different socioeconomic and demographic characteristics. Previous studies reported associations for all-cause CVD mortality as an umbrella outcome but with substantial heterogeneity; this limits the ability to draw meaningful clinical conclusions about specific CVD causes of death to inform health care providers on vulnerable patients and direct public health planning. Using state-of-the-art unified analytical protocol, we examined specific causes and found considerably less heterogeneity and robust pooled-effect size estimates. Overall, extreme cold associations were found to be substantial for different causes of CVD deaths, with the largest excess deaths associated with heart failure. While hot temperatures showed steeper slopes and high RRs that increased with greater temperatures, these occurred during a small proportion of days; therefore, excess CVD deaths from sustained extreme cold were larger than those from quickly escalating heat risks.

Observed associations for temperature extremes and all-cause CVD in our study were in line with previous meta-analyses.8–10 Pooling across different definitions of cold and heat, 1 meta-analysis of 26 temperature–mortality studies found that cold exposure had a greater effect size on all-cause CVD mortality (pooled RR, 1.055; 95% CI, 1.050–1.060) compared to heat exposure (pooled RR, 1.013; 95% CI, 1.011–1.015).26 Another meta-analysis, restricted to the elderly population, found an increase in all-cause CVD mortality for every 1°C departure from the optimal temperature.27 The I2 statistic for heterogeneity in these meta-analyses ranged from 88.2% to 98.9%. In contrast to these studies, our findings had a markedly lower I2 of 50.0% after accounting for meta-predictors.

Ischemic heart disease is the commonest cause of CVD mortality, and it was the most frequent cause of death in our dataset accounting for 37% of CVD deaths. We estimated that about 1% of all ischemic heart disease deaths are attributed to extreme temperatures alone. Some have argued that future ischemic heart disease risk is expected to be more evident with increasing temperatures induced by climate change.28

Globally, stroke is estimated to be the third leading cause of death and disability combined in 2019.29 In this study, we estimate that for every 1000 stroke deaths, 1.6 and 9 excess deaths are attributed to extreme heat and cold days, respectively. While latitude, average temperatures, and sex were effect modifiers for previously examined temperature–stroke relationships,30 we found that GDP per capita and average summer and winter temperatures also partially explained heterogeneity. Previous studies support an association between temperature and stroke risk31; yet, the role of GDP per capita on temperature-related stroke deaths is not understood. The Global Burden of Disease Study showed that stroke mortality rates in low-income countries are 3× higher than high-income countries.29 Our effect modification analysis, albeit inconclusive, is suggestive of a potential link between GDP per capita and temperature–stroke relationship. High GDP per capita may reflect both increased ability to control temperature exposures and decreased outdoor work.

To our knowledge, little evidence exists on the risk of heart failure mortality from extreme temperatures. Some studies examined the relationship between heart failure admissions and seasonality or diurnal temperature range showing higher admissions during winter months and with greater diurnal temperature range.32,33 The exposure–response curves for heart failure outcomes and temperatures, however, were absent. We found heart failure mortality to have the highest burden of excess deaths for both heat and cold compared to other causes of CVD mortality. Additionally, the slope for hot temperatures beyond the MMT appears to accelerate sharply.

Few studies report on arrhythmia mortality and extreme temperatures, and findings have been inconsistent for occurrence of arrhythmia.34–37 This inconsistency may result from different arrhythmia end points and definitions across different studies. We did not find a significant association between extreme heat and death from arrhythmia, which may have resulted from misclassified recording of arrhythmia as a cause of death. For example, fatal ventricular arrhythmias may have occurred but often is the end result of ischemia, cardiomyopathy, or other CVD causes.

There are some postulated underlying mechanisms that could explain the CVD mortality risk associated with extreme temperatures. Both cold and heat exposures evoke a series of synchronized autonomic and cardiovascular responses that ensures core body thermal homeostasis. Those responses are often altered and exaggerated with aging and in individuals with preexisting cardiovascular risk factors, leading to CVD events.38,39 Heat exposure leads to increased skin blood flow which dissipates heat, as well as increased sweating which evaporates heat. This leads to potential volume depletion and sympathetic system activation, resulting in increased heart rate and inotropy to maintain stroke volume despite decreased preload.40 Rising core body temperature also increases metabolic state and oxygen consumption. In susceptible individuals, these cascades may lead to demand ischemia or plaque rupture.41 Volume depletion and insensible body water losses leads to hemoconcentration and hypercoagulable states further increasing the risk of thrombosis and myocardial ischemia.42 Fluid shifts can disturb electrolyte balance (specifically, potassium, magnesium), increasing risk of arrhythmias in susceptible individuals.42 In patients with heart failure who receive diuretics, heat may result in severe volume depletion and potentially, shock.43 On the other hand, cold exposure leads to increased sympathetic activity that vasoconstricts skin and increases skeletal muscle tone to generate and conserve heat. This leads to catecholamine-driven rise in blood pressure (mainly through an increase in peripheral vascular resistant), and hence, increased cardiac oxygen demand to overcome this increase in afterload.38 Cold also leads to increased cholesterol crystals deposition in atherosclerotic plaques, rendering susceptible individuals to plaque rupture and myocardial infarction.39 Last, hypothermia is known to induce hypercoagulability because of increased viscosity and hemoconcentration from fluid shifting into extravascular space and clotting factor abnormalities.39

Climate change produces both hotter summers and colder winters, rendering populations not accustomed to these unusual weather conditions vulnerable, especially in low-income areas where there may be less adaptability to changing conditions. From a clinical perspective, the extent to which healthcare providers who treat CVD patients understand and act to minimize the risks of extreme temperatures is not fully understood. A search of all American Heart Association and European Society of Cardiology journals finds no official statements related to guidance for providers on preventing adverse outcomes from extreme temperature exposures. From a public health standpoint, evidence suggests that targeted interventions, such as warning systems for vulnerable subpopulations, may be needed to effectively prevent CVD deaths.41

This study has several limitations. Readers must be cautioned to interpret our findings as global estimates since some regions (especially low-income countries) were underrepresented in our data such as South Asia, the Middle East, and Africa. Different populations have markedly different characteristics, baseline rates of CVD, and are exposed to a wide range of climates. Even within the studied countries, differences in population characteristics, urbanization, health care infrastructure and housing characteristics, warrant further investigation. More specifically, socioeconomic determinants are particularly important to understand the effects that extreme temperatures might have on socially and economically vulnerable subpopulations. While we account for GDP per capita across countries, it will not be enough to provide any information on the variations of socioeconomic resources that may drive these CVD outcomes. Potential effect measure modifiers at the individual level (eg, age, sex, and education) were not examined. However, in terms of internal validity, we note that our results were robust to multiple adjustments of potential time-varying confounders. On the other hand, exposure misclassification may result from assigning measured ambient temperature averaged from monitoring stations to all individuals at a study site. Exposure misclassification will be lower for variables that have less within-city heterogeneity compared to other variables with more subscale variance (eg, temperature compared to particulate matter with aerodynamic diameter <2.5 µm). Additionally, causes of CVD deaths may be coded differently across sites leading to misclassification bias. We could not examine subtypes of heart failure such as those with preserved or reduced ejection fraction nor disentangle types of arrhythmias. Using mortality from arrhythmia as an outcome may not be the best approach, since it may be difficult to pinpoint arrhythmic death as the underlying cause. Alternatively, studying the associations between extreme temperatures and the incidence of nonfatal cardiovascular diseases (including arrhythmias) may provide us with more answers.

Conclusions

This large, multicountry, multicity investigation shows increased risk and burden of all-cause CVD, ischemic heart disease, stroke, and heart failure mortality from extreme hot and cold temperatures. The results illustrate the relevance of environmental exposures to specific CVD causes of mortality and have direct relevance to prompt optimal medical and public health responses to temperature extremes in present day and under a changing climate.

Article Information

Acknowledgments

B.Alahmad, A.Z., and AG set up the collaborative network. B.Alahmad and H. Khraishah conceptualized the idea. B.Alahmad, A.V.C., and A.G. designed the study. B.Alahmad, J.S., D.R., A.V.C., and A.G. developed the statistical methods. B.Alahmad and H.Khraishah drafted the manuscript and interpreted the results. A.Z., A.V.C., D.R., Y.G., S.I.P., S.A., F.A., B.Armstrong, M.B., B.Y.C., T.N.D., D.V.D., A.E., Y.L.L.G., M.H., Y.H., E.I., C.I., J.J.K., J.S., H.Kim, E.L., W.L., S.L., J.M., F.M., H.O., A.O., M.S.R., N.R., P.H.S., N.S., X.S., F.S., S.P.S., F.D., M.S., and A.T. provided the data, and contributed to the interpretation of the results and reviewed the manuscript. P.K., J.S., A.S.B., E.G., and A.G. supervised the research project and contributed to the interpretation of the results. B.Alahmad and F.S. accessed and verified the data. All authors had the final responsibility for the decision to submit for publication.

Supplemental Material

Supplemental Methods

Tables S1 and S2

Figures S1–S5

References 44–51

Nonstandard Abbreviations and Acronyms

CVD

cardiovascular disease

eCI

empirical confidence interval

GDP

gross domestic product

ICD-9

International Classification of Diseases, Ninth Revision

ICD-10

International Classification of Diseases, Tenth Revision

MCC

Multi-Country Multi-City Collaborative Research Network

MMT

minimum mortality temperature

Disclosures None.

Footnotes

Supplemental Material is available at https://www.ahajournals.org/doi/suppl/10.1161/CIRCULATIONAHA.122.061832.

Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.

Data were collected within the Multi-Country Multi-City Collaborative Research Network under a data sharing agreement and cannot be made publicly available. Researchers can refer to Multi-Country Multi-City participants, who are listed as coauthors of this article, for information on accessing the data for each country.

For Sources of Funding and Disclosures, see page 45.

Circulation is available at www.ahajournals.org/journal/circ

Correspondence to: Barrak Alahmad, MD, MPH, PhD, Department of Environmental Health, Harvard T.H. Chan School of Public Health, 401 Park Dr, Landmark Center Room 412J, Boston, MA 02115. Email

References

  • 1. Mensah GA, Wei GS, Sorlie PD, Fine LJ, Rosenberg Y, Kaufmann PG, Mussolino ME, Hsu LL, Addou E, Engelgau MM, et al. Decline in cardiovascular mortality: possible causes and implications.Circ Res. 2017; 120:366–380. doi: 10.1161/CIRCRESAHA.116.309115LinkGoogle Scholar
  • 2. Roth GA, Johnson C, Abajobir A, Abd-Allah F, Abera SF, Abyu G, Ahmed M, Aksut B, Alam T, Alam K, et al. Global, regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015.J Am Coll Cardiol. 2017; 70:1–25. doi: 10.1016/j.jacc.2017.04.052CrossrefMedlineGoogle Scholar
  • 3. Vicedo-Cabrera AM, Scovronick N, Sera F, Royé D, Schneider R, Tobias A, Astrom C, Guo Y, Honda Y, Hondula DM, et al. The burden of heat-related mortality attributable to recent human-induced climate change.Nat Clim Chang. 2021; 11:492–500. doi: 10.1038/s41558-021-01058-xCrossrefMedlineGoogle Scholar
  • 4. GBD 2019 Risk Factors Collaborators.Global burden of 87 risk factors in 204 countries and territories, 1990-2019: a systematic analysis for the global burden of disease study 2019.Lancet. 2020; 396:1223–1249. doi: 10.1016/S0140-6736(20)30752-2CrossrefMedlineGoogle Scholar
  • 5. Zhao Q, Guo Y, Ye T, Gasparrini A, Tong S, Overcenco A, Urban A, Schneider A, Entezari A, Vicedo-Cabrera AM, et al. Global, regional, and national burden of mortality associated with non-optimal ambient temperatures from 2000 to 2019: a three-stage modelling study.Lancet Planet Health. 2021; 5:e415–e425. doi: 10.1016/S2542-5196(21)00081-4CrossrefMedlineGoogle Scholar
  • 6. Braga ALF, Zanobetti A, Schwartz J. The effect of weather on respiratory and cardiovascular deaths in 12 U.S. cities.Environ Health Perspect. 2002; 110:859–863. doi: 10.1289/ehp.02110859CrossrefMedlineGoogle Scholar
  • 7. Baccini M, Biggeri A, Accetta G, Kosatsky T, Katsouyanni K, Analitis A, Anderson HR, Bisanti L, D’Ippoliti D, Danova J, et al. Heat effects on mortality in 15 European cities.Epidemiology. 2008; 19:711–719. doi: 10.1097/EDE.0b013e318176bfcdCrossrefMedlineGoogle Scholar
  • 8. Yang J, Yin P, Zhou M, Ou C-Q, Guo Y, Gasparrini A, Liu Y, Yue Y, Gu S, Sang S, et al. Cardiovascular mortality risk attributable to ambient temperature in China.Heart. 2015; 101:1966–1972. doi: 10.1136/heartjnl-2015-308062CrossrefMedlineGoogle Scholar
  • 9. Scovronick N, Sera F, Acquaotta F, Garzena D, Fratianni S, Wright CY, Gasparrini A. The association between ambient temperature and mortality in South Africa: a time-series analysis.Environ Res. 2018; 161:229–235. doi: 10.1016/j.envres.2017.11.001CrossrefMedlineGoogle Scholar
  • 10. Silveira IH, Oliveira BFA, Cortes TR, Junger WL. The effect of ambient temperature on cardiovascular mortality in 27 Brazilian cities.Sci Total Environ. 2019; 691:996–1004. doi: 10.1016/j.scitotenv.2019.06.493CrossrefMedlineGoogle Scholar
  • 11. Vicedo-Cabrera AM, Sera F, Liu C, Armstrong B, Milojevic A, Guo Y, Tong S, Lavigne E, Kyselý J, Urban A, et al. Short term association between ozone and mortality: global two stage time series study in 406 locations in 20 countries.BMJ. 2020; 368:m108. doi: 10.1136/bmj.m108CrossrefMedlineGoogle Scholar
  • 12. Liu C, Chen R, Sera F, Vicedo-Cabrera AM, Guo Y, Tong S, Coelho MSZS, Saldiva PHN, Lavigne E, Matus P, et al. Ambient particulate air pollution and daily mortality in 652 cities.N Engl J Med. 2019; 381:705–715.doi: 10.1056/NEJMoa1817364CrossrefMedlineGoogle Scholar
  • 13. Mistry MN, Schneider R, Masselot P, Royé D, Armstrong B, Kyselý J, Orru H, Sera F, Tong S, Lavigne É, et al. Comparison of weather station and climate reanalysis data for modelling temperature-related mortality.Sci Rep. 2022; 12:5178.doi: 10.1038/s41598-022-09049-4.CrossrefMedlineGoogle Scholar
  • 14. Meng X, Liu C, Chen R, Sera F, Vicedo-Cabrera AM, Milojevic A, Guo Y, Tong S, Coelho MD, Saldiva PH, et al. Short term associations of ambient nitrogen dioxide with daily total, cardiovascular, and respiratory mortality: multilocation analysis in 398 cities.BMJ. 2021; 372:n534. doi: 10.1136/bmj.n534CrossrefMedlineGoogle Scholar
  • 15. Kottek M, Grieser J, Beck C, Rudolf B, Rubel F. World map of the Köppen-Geiger climate classification updated.Meteorol Z. 2006; 15:259–263. doi: 10.1127/0941-2948/2006/0130CrossrefGoogle Scholar
  • 16. Jaakkola JJK. Case-crossover design in air pollution epidemiology.Eur Respir J Suppl. 2003; 40:81s–85s. doi: 10.1183/09031936.03.00402703CrossrefMedlineGoogle Scholar
  • 17. Armstrong BG, Gasparrini A, Tobias A. Conditional Poisson models: a flexible alternative to conditional logistic case cross-over analysis.BMC Med Res Methodol. 2014; 14:122. doi: 10.1186/1471-2288-14-122CrossrefMedlineGoogle Scholar
  • 18. Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, Tobias A, Tong S, Rocklöv J, Forsberg B, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study.Lancet. 2015; 386:369–375. doi: 10.1016/S0140-6736(14)62114-0CrossrefMedlineGoogle Scholar
  • 19. Sera F, Armstrong B, Blangiardo M, Gasparrini A. An extended mixed-effects framework for meta-analysis.Stat Med. 2019; 38:5429–5444. doi: 10.1002/sim.8362CrossrefMedlineGoogle Scholar
  • 20. Gasparrini A, Armstrong B, Kenward MG. Multivariate meta-analysis for non-linear and other multi-parameter associations.Stat Med. 2012; 31:3821–3839. doi: 10.1002/sim.5471CrossrefMedlineGoogle Scholar
  • 21. Yin Q, Wang J, Ren Z, Li J, Guo Y. Mapping the increased minimum mortality temperatures in the context of global climate change.Nat Commun. 2019; 10:4640. doi: 10.1038/s41467-019-12663-y.CrossrefMedlineGoogle Scholar
  • 22. Tobías A, Hashizume M, Honda Y, Sera F, Ng CFS, Kim Y, Roye D, Chung Y, Dang TN, Kim H, et al. Geographical variations of the minimum mortality temperature at a global scale.Environ Epidemiol. 2021; 5:e169.doi: 10.1097/EE9.0000000000000169CrossrefMedlineGoogle Scholar
  • 23. Bobb JF, Peng RD, Bell ML, Dominici F. Heat-related mortality and adaptation to heat in the United States.Environ Health Perspect. 2014; 122:811–816. doi: 10.1289/ehp.1307392CrossrefMedlineGoogle Scholar
  • 24. Guo Y, Barnett AG, Tong S. High temperatures-related elderly mortality varied greatly from year to year: important information for heat-warning systems.Sci Rep. 2012; 2:830. doi: 10.1038/srep00830CrossrefMedlineGoogle Scholar
  • 25. Sarofim MC, Saha S, Hawkins MD, Mills DM, Hess J, Horton R, Kinney P, Schwartz J, St Juliana A. Temperature-related death and illness.Crimmins A, Balbus J, Gamble JL, Beard CB, Bell JE, Dodgen D, Eisen RJ, Fann N, Hawkins MD, Herring SC, et al, eds. In: The Impacts of Climate Change on Human Health in the United States: A Scientific Assessment. USGlobal Change Research Program; 2016.Google Scholar
  • 26. Moghadamnia MT, Ardalan A, Mesdaghinia A, Keshtkar A, Naddafi K, Yekaninejad MS. Ambient temperature and cardiovascular mortality: a systematic review and meta-analysis.PeerJ. 2017; 5:e3574. doi: 10.7717/peerj.3574CrossrefMedlineGoogle Scholar
  • 27. Bunker A, Wildenhain J, Vandenbergh A, Henschke N, Rocklöv J, Hajat S, Sauerborn R. Effects of air temperature on climate-sensitive mortality and morbidity outcomes in the elderly; a systematic review and meta-analysis of epidemiological evidence.EBioMedicine. 2016; 6:258–268. doi: 10.1016/j.ebiom.2016.02.034CrossrefMedlineGoogle Scholar
  • 28. Chen K, Breitner S, Wolf K, Hampel R, Meisinger C, Heier M, von Scheidt W, Kuch B, Peters A, Schneider A, et al. Temporal variations in the triggering of myocardial infarction by air temperature in Augsburg, Germany, 1987-2014.Eur Heart J. 2019; 40:1600–1608. doi: 10.1093/eurheartj/ehz116CrossrefMedlineGoogle Scholar
  • 29. GBD. 2019 Stroke Collaborators.Global, regional, and national burden of stroke and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019.Lancet Neurol. 2021; 20:795–820.doi: 10.1016/S1474-4422(21)00252-0CrossrefMedlineGoogle Scholar
  • 30. Zorrilla-Vaca A, Healy RJ, Silva-Medina MM. Revealing the association between cerebrovascular accidents and ambient temperature: a meta-analysis.Int J Biometeorol. 2017; 61:821–832. doi: 10.1007/s00484-016-1260-6CrossrefMedlineGoogle Scholar
  • 31. Lavados PM, Olavarría VV, Hoffmeister L. Ambient temperature and stroke risk: evidence supporting a short-term effect at a population level from acute environmental exposures.Stroke. 2018; 49:255–261. doi: 10.1161/STROKEAHA.117.017838LinkGoogle Scholar
  • 32. Inglis SC, Clark RA, Shakib S, Wong DT, Molaee P, Wilkinson D, Stewart S. Hot summers and heart failure: seasonal variations in morbidity and mortality in Australian heart failure patients (1994-2005).Eur J Heart Fail. 2008; 10:540–549. doi: 10.1016/j.ejheart.2008.03.008CrossrefMedlineGoogle Scholar
  • 33. Qiu H, Yu IT, Tse LA, Tian L, Wang X, Wong TW. Is greater temperature change within a day associated with increased emergency hospital admissions for heart failure?.Circ Heart Fail. 2013; 6:930–935. doi: 10.1161/CIRCHEARTFAILURE.113.000360LinkGoogle Scholar
  • 34. McGuinn L, Hajat S, Wilkinson P, Armstrong B, Anderson HR, Monk V, Harrison R. Ambient temperature and activation of implantable cardioverter defibrillators.Int J Biometeorol. 2013; 57:655–662. doi: 10.1007/s00484-012-0591-1CrossrefMedlineGoogle Scholar
  • 35. Zanobetti A, Coull BA, Kloog I, Sparrow D, Vokonas PS, Gold DR, Schwartz J. Fine-scale spatial and temporal variation in temperature and arrhythmia episodes in the VA Normative Aging Study.J Air Waste Manag Assoc. 2017; 67:96–104. doi: 10.1080/10962247.2016.1252808CrossrefMedlineGoogle Scholar
  • 36. Fries RP, Heisel AG, Jung JK, Schieffer HJ. Circannual variation of malignant ventricular tachyarrhythmias in patients with implantable cardioverter-defibrillators and either coronary artery disease or idiopathic dilated cardiomyopathy.Am J Cardiol. 1997; 79:1194–1197. doi: 10.1016/s0002-9149(97)00081-7CrossrefMedlineGoogle Scholar
  • 37. Pimentel M, Grüdtner L, Zimerman LI. Seasonal variation of ventricular tachycardia registered in 24-hour Holter monitoring.Arq Bras Cardiol. 2006; 87:403–406. doi: 10.1590/s0066-782x2006001700002CrossrefMedlineGoogle Scholar
  • 38. Greaney JL, Kenney WL, Alexander LM. Sympathetic regulation during thermal stress in human aging and disease.Auton Neurosci. 2016; 196:81–90. doi: 10.1016/j.autneu.2015.11.00CrossrefMedlineGoogle Scholar
  • 39. Stewart S, Keates AK, Redfern A, McMurray JJV. Seasonal variations in cardiovascular disease.Nat Rev Cardiol. 2017; 14:654–664. doi: 10.1038/nrcardio.2017.76CrossrefMedlineGoogle Scholar
  • 40. Crandall CG, Wilson TE. Human cardiovascular responses to passive heat stress.Compr Physiol. 2015; 5:17–43. doi: 10.1002/cphy.c140015CrossrefMedlineGoogle Scholar
  • 41. Peters A, Schneider A. Cardiovascular risks of climate change.Nat Rev Cardiol. 2021; 18:1–2. doi: 10.1038/s41569-020-00473-5CrossrefMedlineGoogle Scholar
  • 42. Liu C, Yavar Z, Sun Q. Cardiovascular response to thermoregulatory challenges.Am J Physiol Heart Circ Physiol. 2015; 309:H1793–H1812.doi: 10.1152/ajpheart.00199.2015CrossrefMedlineGoogle Scholar
  • 43. Schulte F, Röösli M, Ragettli MS. Heat-related cardiovascular morbidity and mortality in Switzerland: a clinical perspective.Swiss Med Wkly. 2021; 151:w30013. doi: 10.4414/SMW.2021.w30013CrossrefMedlineGoogle Scholar
  • 44. Turner H, Firth D. Generalized nonlinear models in R: An overview of the gnm package. 2007.Google Scholar
  • 45. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models.Stat Med. 2010; 29:2224–2234. doi: 10.1002/sim.3940CrossrefMedlineGoogle Scholar
  • 46. Barrett JR. Increased minimum mortality temperature in France: data suggest humans are adapting to climate change.Environ Health Perspect. 2015; 123:A184. doi: 10.1289/ehp.123-A184CrossrefMedlineGoogle Scholar
  • 47. Gasparrini A, Armstrong B. Reducing and meta-analysing estimates from distributed lag non-linear models.BMC Med Res Methodol. 2013; 13:1. doi: 10.1186/1471-2288-13-1CrossrefMedlineGoogle Scholar
  • 48. Anderson BG, Bell ML. Weather-related mortality: how heat, cold, and heat waves affect mortality in the United States.Epidemiology. 2009; 20:205–213. doi: 10.1097/EDE.0b013e318190ee08CrossrefMedlineGoogle Scholar
  • 49. Peng RD, Bobb JF, Tebaldi C, McDaniel L, Bell ML, Dominici F. Toward a quantitative estimate of future heat wave mortality under global climate change.Environ Health Perspect. 2011; 119:701–706. doi: 10.1289/ehp.1002430CrossrefMedlineGoogle Scholar
  • 50. Gasparrini A, Leone M. Attributable risk from distributed lag models.BMC Med Res Methodol. 2014; 14:55. doi: 10.1186/1471-2288-14-55CrossrefMedlineGoogle Scholar
  • 51. Vicedo-Cabrera AM, Forsberg B, Tobias A, Zanobetti A, Schwartz J, Armstrong B, Gasparrini A. Associations of inter- and intraday temperature change with mortality.Am J Epidemiol. 2016; 183:286–293. doi: 10.1093/aje/kwv205CrossrefMedlineGoogle Scholar

eLetters(0)

eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.