Ambient temperature effect on mortality varies between places and populations, suggesting the existence of effect modifiers for this association. This study analyzes the influence of geographic, urban, and socioeconomic factors on the ambient temperature effect on non-accidental mortality in the general and older adults population of Brazilian metropolitan areas, and on that associated with circulatory, respiratory, and other mortality in older adults. Effects of this association were estimated for each group in 42 locations using a generalized additive model combined with the nonlinear distributed lag model. A meta-analysis was then performed to estimate the effects at the national and regional levels. Meta-regression determined the influence of effect modifiers. Estimated relative risks of the temperature-mortality association varied between locations in the Brazilian territory. Heat effects on non-accidental mortality at the national level were 1.09 (95%CI: 1.04-1.15) and 1.13 (95%CI: 1.07-1.20) for the General and Older Adult groups, respectively. Cold effects were 1.26 (95%CI: 1.21-1.32) and 1.30 (95%CI: 1.24-1.36) for the General and Older Adult groups, respectively. We observed a greater effect of cold than heat in both groups. For all causes of death, effects of heat and cold were greater in the Southeast and South Brazil. Amplitude of the mean temperature was the factor that best explained the heterogeneity between locations, followed by latitude, income and schooling. Hence, implementing adaptive measures to reduce the ambient temperature effects on mortality depends on the profile of each location.
Keywords:
Temperatures; Mortality; Epidemiologic Effect Modifier; Climate Effects
El efecto de la temperatura ambiente sobre la mortalidad varía entre sitios y poblaciones, lo que sugiere la presencia de modificadores del efecto de esta asociación. El objetivo de este estudio fue analizar la influencia de los factores geográficos, urbanos y socioeconómicos en el efecto de la temperatura ambiente sobre la mortalidad no accidental en la población general y anciana de las regiones metropolitanas brasileñas, y la influencia asociada con las causas de mortalidad circulatoria, respiratoria u otra en los ancianos. Los efectos de esta asociación se estimaron para cada grupo en 42 sitios mediante un modelo aditivo generalizado combinado con el modelo no lineal distribuido. A continuación, se realizó un metaanálisis para estimar los efectos a nivel Brasil y sus regiones. La influencia de los modificadores del efecto se determinó mediante metarregresión. Los riesgos relativos estimados de la asociación temperatura-mortalidad variaron entre las ubicaciones en el territorio brasileño. Los efectos del calor sobre la mortalidad no accidental a nivel nacional fueron de 1,09 (IC95%: 1,04-1,15) y de 1,13 (IC95%: 1,07-1,20) para el grupo General y Ancianos, respectivamente. Los efectos del frío fueron 1,26 (IC95%: 1,21-1,32) y 1,30 (IC95%: 1,24-1,36) para el grupo General y Anciano, respectivamente. Hay un mayor efecto del frío que del calor en ambos grupos. Para todas las causas de muerte, los efectos del calor y del frío fueron mayores en las regiones Sudeste y Sur de Brasil. El factor que explicó mejor la heterogeneidad entre los locales fue el rango de temperatura media, seguido de la latitud, los ingresos y el nivel de estudios. Por lo tanto, la implementación de medidas de adaptación para reducir los efectos de la temperatura ambiente sobre la mortalidad depende del perfil de cada lugar.
Palabras-clave:
Temperatura Ambiental; Mortalidad; Modificador del Efecto Epidemiológico; Efectos del Clima
Introduction
High and low ambient temperatures are related to increases in medical emergencies, hospitalization, and deaths 11. Song X, Wang S, Hu Y, Yue M, Zhang T, Liu Y, et al. Impact of ambient temperature on morbidity and mortality: an overview of reviews. Sci Total Environ 2017; 586:241-54.. Temperature-mortality association was observed for overall mortality 11. Song X, Wang S, Hu Y, Yue M, Zhang T, Liu Y, et al. Impact of ambient temperature on morbidity and mortality: an overview of reviews. Sci Total Environ 2017; 586:241-54.,22. Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet 2015; 386:369-75., cardiovascular disease 33. 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.,44. Alahmad B, Khraishah H, Royé D, Vicedo-Cabrera AM, Guo Y, Papatheodorou SI, et al. Associations between extreme temperatures and cardiovascular cause-specific mortality: results from 27 countries. Circulation 2023; 147:35-46., respiratory diseases 55. Bunker A, Wildenhain J, Vandenbergh A, Henschke N, Rocklöv J, Hajat S, et al. 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; 6:258-68., and cerebrovascular diseases 44. Alahmad B, Khraishah H, Royé D, Vicedo-Cabrera AM, Guo Y, Papatheodorou SI, et al. Associations between extreme temperatures and cardiovascular cause-specific mortality: results from 27 countries. Circulation 2023; 147:35-46.,66. Wen J, Zou L, Jiang Z, Li Y, Tao J, Liu Y, et al. Association between ambient temperature and risk of stroke morbidity and mortality: a systematic review and meta-analysis. Brain Behav 2023; 13:e3078., in locations across Europe 77. Marí-Dell'Olmo M, Tobías A, Gómez-Gutiérrez A, Rodríguez-Sanz M, García de Olalla P, Camprubí E, et al. Social inequalities in the association between temperature and mortality in a South European context. Int J Public Health 2019; 64:27-37., Asia 88. Ng CFS, Ueda K, Takeuchi A, Nitta H, Konishi S, Bagrowicz R, et al. Sociogeographic variation in the effects of heat and cold on daily mortality in Japan. J Epidemiol 2014; 24:15-24.,99. Luan G, Yin P, Wang L, Zhou M. The temperature - mortality relationship: an analysis from 31 Chinese provincial capital cities. Int J Environ Health Res 2018; 28:192-201., Africa 1010. Scovronick N, Sera F, Acquaotta F, Garzena D, Fratianni S, Wright CY, et al. The association between ambient temperature and mortality in South Africa: a time-series analysis. Environ Res 2018; 161:229-35., North America 1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72. and Latin America 1212. Kephart JL, Sánchez BN, Moore J, Schinasi LH, Bakhtsiyarava M, Ju Y, et al. City-level impact of extreme temperatures and mortality in Latin America. Nat Med 2022; 28:1700-5., thus highlighting a global health issue that can be amplified by climate change events 1313. Core Writing Team; Lee H, Romero J, editors. Climate change 2023: synthesis report. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva: International Panel on Climate Change; 2023..
Variations of these mortality risks related to ambient temperature between locations 44. Alahmad B, Khraishah H, Royé D, Vicedo-Cabrera AM, Guo Y, Papatheodorou SI, et al. Associations between extreme temperatures and cardiovascular cause-specific mortality: results from 27 countries. Circulation 2023; 147:35-46.,99. Luan G, Yin P, Wang L, Zhou M. The temperature - mortality relationship: an analysis from 31 Chinese provincial capital cities. Int J Environ Health Res 2018; 28:192-201.,1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.,1212. Kephart JL, Sánchez BN, Moore J, Schinasi LH, Bakhtsiyarava M, Ju Y, et al. City-level impact of extreme temperatures and mortality in Latin America. Nat Med 2022; 28:1700-5.,1414. Anderson BG, Bell ML. Weather-related mortality: how heat, cold, and heat waves affect mortality in the United States. Epidemiology 2009; 20:205-13.,1515. Ma W, Wang L, Lin H, Liu T, Zhang Y, Rutherford S, et al. The temperature-mortality relationship in China: an analysis from 66 Chinese communities. Environ Res 2015; 137:72-7.,1616. Wang C, Zhang Z, Zhou M, Zhang L, Yin P, Ye W, et al. Nonlinear relationship between extreme temperature and mortality in different temperature zones: a systematic study of 122 communities across the mainland of China. Sci Total Environ 2017; 586:96-106. may be due to the variability of individual and community characteristics 1717. Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004., socioeconomic characteristics 77. Marí-Dell'Olmo M, Tobías A, Gómez-Gutiérrez A, Rodríguez-Sanz M, García de Olalla P, Camprubí E, et al. Social inequalities in the association between temperature and mortality in a South European context. Int J Public Health 2019; 64:27-37., geographic aspects 1616. Wang C, Zhang Z, Zhou M, Zhang L, Yin P, Ye W, et al. Nonlinear relationship between extreme temperature and mortality in different temperature zones: a systematic study of 122 communities across the mainland of China. Sci Total Environ 2017; 586:96-106. or different adaptation responses 22. Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet 2015; 386:369-75.. These effect modifiers of the temperature-mortality association were investigated in previous studies, especially in areas from North America, Asia, and Europe 1717. Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004.. However, these studies showed divergent results and few compared different geographic and climatic regions 1717. Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004.,1818. Sera F, Armstrong B, Tobias A, Vicedo-Cabrera AM, Åström C, Bell ML, et al. How urban characteristics affect vulnerability to heat and cold: a multi-country analysis. Int J Epidemiol 2019; 48:1101-12..
Brazil is also vulnerable to the effects of extreme temperatures, with temperature-related mortality ranging from 2.04% to 7.08% between Brazilian capitals 22. Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet 2015; 386:369-75.. Such heat stress conditions are associated with higher hospitalization rates 1919. Requia WJ, Vicedo-Cabrera AM, de Schrijver E, Amini H. Low ambient temperature and hospitalization for cardiorespiratory diseases in Brazil. Environ Res 2023; 231:116231.,2020. Requia WJ, Vicedo-Cabrera AM, de Schrijver E, Amini H, Gasparrini A. Association of high ambient temperature with daily hospitalization for cardiorespiratory diseases in Brazil: a national time-series study between 2008 and 2018. Environ Pollut 2023; 331:121851. and mortality due to cardiovascular and respiratory diseases 2121. Oliveira BFA, Jacobson LSV, Perez LP, Silveira IH, Junger WL, Hacon SS. Impacts of heat stress conditions on mortality from respiratory and cardiovascular diseases in Brazil. Sustainability in Debate 2020; 11:297-313..
Located in South America, Brazil has over 203 million inhabitants distributed across more than 8.5 million km2. As an emerging country, few locations in its territory have consistent active policies for climate change adaptation or mitigation 2222. Barbi F, Ferreira LC. Climate change in Brazilian cities: policy strategies and responses to global warming. Int J Environ Sci Dev 2013; 4:49-51.. With a large territorial extension, socio-environmental and sociodemographic diversity, Brazil is an appropriate place of study to expand research in this area.
Few national studies address the effect of extreme temperatures on mortality 2323. Requia WJ, Alahmad B, Schwartz JD, Koutrakis P. Association of low and high ambient temperature with mortality for cardiorespiratory diseases in Brazil. Environ Res 2023; 234:116532., especially on Brazilian older adults, an expanding group that is vulnerable to ambient temperature effects 1717. Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004.,2323. Requia WJ, Alahmad B, Schwartz JD, Koutrakis P. Association of low and high ambient temperature with mortality for cardiorespiratory diseases in Brazil. Environ Res 2023; 234:116532.. Additionally, these studies fail to address temperature effects on populations living in metropolitan areas. This area of great urbanization encompasses a set of contiguous municipalities socioeconomically integrated into a central city that share public services and infrastructure and concentrate a third of Brazil’s population, 60% of the national gross domestic product (GDP) and 70% of urban poverty 2424. Marguti BO, Tavares SR, editors. Política metropolitana: governança, instrumentos e planejamentometropolitanos - II Seminário e Oficina. Brasília: Instituto de Pesquisa Econômica Aplicada; 2019.. Studies in these areas enable analyzing data from a larger contingent of urban dwellers, a factor also related to greater vulnerability to extreme temperatures 1818. Sera F, Armstrong B, Tobias A, Vicedo-Cabrera AM, Åström C, Bell ML, et al. How urban characteristics affect vulnerability to heat and cold: a multi-country analysis. Int J Epidemiol 2019; 48:1101-12..
Understanding this association and its modifying factors enables planning appropriate interventions for subpopulations with different vulnerability statuses and providing reliable predictions of climate change effects 22. Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet 2015; 386:369-75.,1717. Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004..
Seeking to contribute to this issue, the present study estimated the effect of ambient temperature on non-accidental mortality among the general population and older adults across the Brazilian territory, and analyzed the influence of geographic, urban and socioeconomic diversity on this association.
Methods
Study design
A time series analysis of daily mortality and meteorological data was performed from January 1, 2000, to December 31, 2014, for 45 Brazilian metropolitan areas located in the Central-West (2), Northeast (16), North (7), Southeast (8) and South (12). Other 29 metropolitan areas were excluded for lacking a climate data collection station.
Mortality data
Mortality data were obtained from the Brazilian Mortality Information System (SIM, acronym in Portuguese), Brazilian Health Informatics Department (DATASUS, acronym in Portuguese). Non-accidental daily mortality (General group) is represented by the total count of deaths excluding external causes (International Classification of Diseases, 10th revision of the [ICD-10]: A00-R99). The Older Adult group used data on the deaths of individuals aged 60 years or older.
To verify the diversity of effects between causes of death, we analyzed the association between temperature mortality from circulatory, respiratory and other causes in the Older Adult group. Mortality data were stratified by cause of death, forming the subgroups: Circulatory (diseases of the circulatory system - ICD-10: I00-I99); Respiratory (diseases of the respiratory system - ICD-10: J00-J99), and Other Causes (ICD-10: A00-H95 and K00-R99).
Low number of deaths in the time series may lead to imprecision in the estimates 2525. Burkart KG, Brauer M, Aravkin AY, Godwin WW, Hay SI, He J, et al. Estimating the cause-specific relative risks of non-optimal temperature on daily mortality: a two-part modelling approach applied to the Global Burden of Disease Study. Lancet 2021; 398:685-97.. Presence of inaccurate data in the first statistical analyses can generate bias in the data estimated in the second phase 2626. Gasparrini A, Armstrong B. Reducing and meta-analyzing estimates of distributed lag non-linear models. BMC Med Res Methodol 2013; 13:1.. To minimize possible errors in the subsequent combined estimates, exclusion criteria were applied. First, we excluded metropolitan areas with a number of deaths per day lower than 1.5, remaining 43 metropolitan areas for the analyses of the General and Older Adult groups. Finally, one metropolitan area was excluded due to numerical inconsistencies (numerical problems in estimation). For the Older Adult subgroups, with lower death counts, metropolitan areas with a daily mean of deaths below 1 in all 3 subgroups were also excluded, thus totaling 38 metropolitan areas for the subgroup analysis.
Meteorological data
Mean daily temperature (ºC) was chosen as the exposure variable to be analyzed, as it represents exposure throughout the day and night, having the best performance in predicting temperature effects on mortality 99. Luan G, Yin P, Wang L, Zhou M. The temperature - mortality relationship: an analysis from 31 Chinese provincial capital cities. Int J Environ Health Res 2018; 28:192-201., and because different temperature measurements have similar predictive abilities 1010. Scovronick N, Sera F, Acquaotta F, Garzena D, Fratianni S, Wright CY, et al. The association between ambient temperature and mortality in South Africa: a time-series analysis. Environ Res 2018; 161:229-35.,2727. Barnett AG, Tong S, Clements ACA. What measure of temperature is the best predictor of mortality? Environ Res 2010; 110:604-11..
Average of the current and previous day was used as the average relative humidity indicator (%). It was included in the analysis as a confounding factor, as in previous studies 2323. Requia WJ, Alahmad B, Schwartz JD, Koutrakis P. Association of low and high ambient temperature with mortality for cardiorespiratory diseases in Brazil. Environ Res 2023; 234:116532.,2828. 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.. Humidity influences temperature by modulating thermal sensation 2929. Liu C, Yavar Z, Sun Q. Cardiovascular response to thermoregulatory challenges. Am J Physiol Heart Circ Physiol 2015; 309:H1793-812.,3030. Davis RE, McGregor GR, Enfield KB. Humidity: a review and primer on atmospheric moisture and human health. Environ Res 2016; 144:106-16.. It also influences the development of respiratory 3030. Davis RE, McGregor GR, Enfield KB. Humidity: a review and primer on atmospheric moisture and human health. Environ Res 2016; 144:106-16. and cardiovascular diseases by affecting heat stress, dehydration and proliferation of disease vectors 2929. Liu C, Yavar Z, Sun Q. Cardiovascular response to thermoregulatory challenges. Am J Physiol Heart Circ Physiol 2015; 309:H1793-812.,3030. Davis RE, McGregor GR, Enfield KB. Humidity: a review and primer on atmospheric moisture and human health. Environ Res 2016; 144:106-16..
Missing data on mean compensated temperature and mean relative humidity could be minimized for five locations that had two or more metropolitan stations. In these, imputation of missing values was performed using the mtsdi package 3131. Junger WL, Ponce de Leon A. Imputation of missing data in time series for air pollutants. Atmos Environ 2015; 102:96-104. of the R platform (http://www.r-project.org). A model with nonparametric cubic spline with 8 degrees of freedom (df) predicted the data to be imputed. Imputation occurred only for days in which there was at least one observation per weather station and there could be no loss of data for more than three days.
Previous studies have reported the influence of daily and yearly temperature variations 2828. 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.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306., locations, latitude 1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.,2828. 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.,3333. Bao J, Wang Z, Yu C, Li X. The influence of temperature on mortality and its lag effect: a study in four Chinese cities with different latitudes. BMC Public Health 2016; 16:375. and geographic region 1515. Ma W, Wang L, Lin H, Liu T, Zhang Y, Rutherford S, et al. The temperature-mortality relationship in China: an analysis from 66 Chinese communities. Environ Res 2015; 137:72-7.,2828. 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. on the temperature-mortality association. Thus, the annual average of mean daily temperatures, annual mean temperature range, daily amplitude of temperature (difference between maximum and minimum daily temperature), latitude and geographic region were included in the model as geographic factors modifying the temperature-mortality association.
Urban and socioeconomic data
Individual and community characteristics were identified as modifying factors of the temperature-mortality association 1717. Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004.. Among the factors at the individual level we have schooling 77. Marí-Dell'Olmo M, Tobías A, Gómez-Gutiérrez A, Rodríguez-Sanz M, García de Olalla P, Camprubí E, et al. Social inequalities in the association between temperature and mortality in a South European context. Int J Public Health 2019; 64:27-37.,1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306. and income 1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.,1818. Sera F, Armstrong B, Tobias A, Vicedo-Cabrera AM, Åström C, Bell ML, et al. How urban characteristics affect vulnerability to heat and cold: a multi-country analysis. Int J Epidemiol 2019; 48:1101-12.,3434. Ma W, Chen R, Kan H. Temperature-related mortality in 17 large Chinese cities: how heat and cold affect mortality in China. Environ Res 2014; 134:127-33.. Population density 88. Ng CFS, Ueda K, Takeuchi A, Nitta H, Konishi S, Bagrowicz R, et al. Sociogeographic variation in the effects of heat and cold on daily mortality in Japan. J Epidemiol 2014; 24:15-24.,1818. Sera F, Armstrong B, Tobias A, Vicedo-Cabrera AM, Åström C, Bell ML, et al. How urban characteristics affect vulnerability to heat and cold: a multi-country analysis. Int J Epidemiol 2019; 48:1101-12.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306.,3535. Medina-Ramón M, Schwartz J. Temperature, temperature extremes, and mortality: a study of acclimatisation and effect modification in 50 US cities. Occup Environ Med 2007; 64:827-33. and GDP per capita 1818. Sera F, Armstrong B, Tobias A, Vicedo-Cabrera AM, Åström C, Bell ML, et al. How urban characteristics affect vulnerability to heat and cold: a multi-country analysis. Int J Epidemiol 2019; 48:1101-12.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306. are community factors that characterize the level of urban development.
Hence, socioeconomic and urban development data were included in the model as modifying factors for the temperature-mortality association. These data were obtained from the 2010 Demographic Census (https://censo2010.ibge.gov.br/) by the Brazilian Institute of Geography and Statistics (IBGE, acronym in Portuguese) for each municipality, and then the respective metropolitan area averages were calculated.
Urban indicators used were demographic density (inhabitants/km2) and GDP per capita (BRL). Socioeconomic indicators used were income (percentage of individuals with no income or with an income of up to one minimum wage) and schooling (percentage of individuals over 10 years old with complete primary education). For each metropolitan area we estimated the annual average of each numerical variable and the average for the period studied, which was the value used in the analyses.
Data analysis
Two-step analysis investigated the temperature-mortality association separately for each group of mortality causes.
Firstly, a time series analysis was performed for each metropolitan area and group using the generalized additive model (GAM), assuming a quasi-Poisson model 3636. Bhaskaran K, Gasparrini A, Hajat S, Smeeth L, Armstrong B. Time series regression studies in environmental epidemiology. Int J Epidemiol 2013; 42:1187-95.. A cross-basis function of the distributed lag non-linear models (DLNM) 3737. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med 2010; 29:2224-34. was included to model the non-linear lagged effect of ambient temperature on mortality. This function was defined by cubic natural spline with three internal nodes placed on the 10th, 75th, and 90th percentiles of each site’s specific temperature distribution, and a cubic natural spline with nodes arranged at the intercept and three equally spaced internal nodes in the log scale of lag values. A 21-day analysis window (maximum lag up to 21 days) was used, allowing us to estimate the lagged relation between temperature and mortality, less effect of death anticipation, and to compare our results with previous studies.
Two thin plate splines were included on the regression model, one for adjusting the time and seasonal trend, and one for relative air humidity. Time trend and seasonality were adjusted using splines from 2 to 8 df and the choice was based on the Akaike information criterion (AIC) and analysis of residuals. Relative humidity adjustments tested 3 to 6 df, and the choice was made based on the lowest AIC. Finally, an indicator variable was included for each day of the week.
Based on this model, the minimum mortality percentile (MMP) and the respective minimum mortality temperature (MMT) were estimated for each metropolitan area and group of causes 3737. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med 2010; 29:2224-34.. The effect of temperature on mortality was estimated in relative risk (RR). The effect of cold was estimated by the RR of mortality between the 1st percentile and MMP, and the effect of heat was estimated from the RR of mortality between the 99th percentile and MMP. Confidence intervals (95%CI) were extracted from these values using a 95% confidence level.
Secondly, the degree of heterogeneity between locations was verified and the mean value of the temperature-mortality association was estimated for the entire country, for the geographic regions and for each group of causes. Using the same regression model described above, the entire temperature-mortality association accumulated in the lag period was reduced by extracting the vectors of the estimated coefficients and the respective matrix of estimated (co)variances for each location and group. This step reduces the number of parameters considered in the second-stage meta-analysis while preserving the complexity of the estimated dependency 2626. Gasparrini A, Armstrong B. Reducing and meta-analyzing estimates of distributed lag non-linear models. BMC Med Res Methodol 2013; 13:1.. Mean value of the MMP estimated in the individual analyses was chosen as the reference for the estimates.
A multivariate meta-analysis model 2626. Gasparrini A, Armstrong B. Reducing and meta-analyzing estimates of distributed lag non-linear models. BMC Med Res Methodol 2013; 13:1. defined the mean temperature-mortality association of the metropolitan areas at the national level and for each Brazilian region using the restricted maximum likelihood (REML) method. Quantification of heterogeneity in the exposure-response relations of the metropolitan areas used the Cochran Q test for (residual) heterogeneity and I2 statistics.
Univariable multivariate meta-regression models evaluated the modifying effect attributable to the following metapredictive variables: mean, daily and annual amplitude of mean temperature, latitude, demographic density, GDP per capita, schooling level and income. Meta-regression models, each including a single metapredictor, were specified and exposure-response associations were estimated for the 25th and 75th percentiles values of these metapredictor variables. Each model was tested for heterogeneity (Q-test and I2) and model fit (AIC). Wald test assessed the significance of the multivariate association between the outcome parameters and each predictor variable.
Sensitivity analysis was performed by testing different parameters for cross-basis. Two spline functions (natural cubic spline [ns] and quadratic B-spline [bs]) were tested, and two distributions for internal knots (10th, 75th and 90th percentiles and 25th, 50th and 75th percentiles) of the temperature distribution. Using the Q-AIC, an AIC modified for likelihood models 3838. 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., the model with the lowest value in the sum of Q-AICs of all metropolitan areas was chosen as the best fit.
All statistical analyses and graphs were performed on the R platform version 3.5.1 using dlnm, mgcv and mvmeta.
Results
A total of 6,483,270 deaths from non-accidental causes occurred between 2000 and 2014 in the 42 metropolitan areas analyzed, of which 4,290,322 were individuals over 60 years old. Table 1 summarizes mortality and climate data for each location. Metropolitan areas extend from latitude 2º82’ North (Metropolitan Area of Capital/Roraima State) to 30º5’ South (Metropolitan Area of Porto Alegre/Rio Grande do Sul State), with mean ambient temperature ranging from 14.87ºC (Metropolitan Area of Lages/Santa Catarina State) to 28.11ºC (Metropolitan Area of Capital/Roraima State).
Non-accidental temperature-mortality association accumulated in 21 days was estimated for the General and Older Adult groups. General group presented an estimated RR for the effect of heat of 1.09 (95%CI: 1.04-1.15) and for cold of 1.26 (95%CI: 1.21-1.32). In the Older Adult group, the estimated RR for the effect of heat was 1.13 (95%CI: 1.07-1.20) and for cold was 1.30 (95%CI: 1.24-1.36). We observed a greater effect of cold than heat in both groups. These values result from combining the estimated RR for each of the 42 Brazilian metropolitan areas.
Variability of the temperature-mortality association across the Brazilian territory
In observing the estimated values for each metropolitan area (Table 2), we note a significant increase of RR associated with extreme temperatures (both high and low) in non-accidental mortality for the General and Older Adult groups in several locations, especially in South and Southeast Brazil. Most metropolitan areas had higher RRs estimated for the effect of cold. RR values of the temperature-mortality association are higher in the Older Adult group than in the General group for most locations. Figure 1 illustrates the variability in RR results between metropolitan areas throughout the national territory, using the RR estimates of the Older Adult group as an example.
Geographic distribution of the non-accidental temperature-mortality association estimates in the Older Adult group.
We also note that the MMT also varies between metropolitan areas. MMT is the optimal 22. Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet 2015; 386:369-75., most comfortable or ideal temperature from which mortality increases 3838. 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.. Estimated MMT values ranged from 17.9 to 33.4ºC for the General group (mean: 26.1ºC, standard deviation - SD: 4.1) and from 16.7 to 29.9ºC for the Older Adult group (mean: 25.4ºC, SD: 3.1), with higher MMT values found in latitudes close to the Equator and lower in metropolises located further south.
Tables 3, 4, and 5 present the effects of extreme temperatures on circulatory, respiratory and other mortality causes in the Older Adult group by metropolitan area. Figure 2 shows the estimated effect (RR) of high and low temperatures on Older Adult subgroups’ mortality by geographic region. Differences in RR can be noted by geographic region and cause. Cold and heat effect was significant on the mortality of the three Older Adult subgroups in the South and Southeast regions. In the Central-West, only cold affected mortality. Northeast locations had only RR of circulatory mortality and other causes associated with extreme low temperatures. Northern metropolitan areas had no significant RRs.
Mean estimates of the cumulative effect of ambient temperature on the mortality over a 21-day period of the Older Adult subgroups for Brazil and each region.
Variability in the estimates of temperature-mortality associations between metropolitan areas was tested using the heterogeneity test (I2). Association between temperature and non-accidental mortality showed I2 values of 81% in the General group and 79% in the Older Adult group. Heterogeneity analyses found lower I2 values for the Older Adult subgroups. Including the geographic region factor into the model reduced the I2 values for all groups (Table 6).
Effect modifiers of the temperature-mortality association
This heterogeneity could be explained by different local factors that modify the association. We tested three groups of possible effect modifiers (geographic, urban and socioeconomic). Table 6 presents the heterogeneity analysis results.
By including geographic factors, heterogeneity of the temperature-non-accidental mortality association was partially explained mainly by the annual amplitude of mean temperature with a drop in the I2 value to 64.7% in the General group and 61.2% in the Older Adult group. Analyses conducted with data from the Older Adult subgroups also pointed to this importance, with lower I2 values in the three groups of causes. For all mortality groups tested, the model including the annual amplitude of mean temperature obtained the best model fit among the metapredictive variables considered, the lowest AIC and a significant Wald test. The model tested with the metapredictor variable latitude reached I2 values close to those obtained with the amplitude of mean temperature.
Regarding urbanization factors, including the metapredictive variables demographic density and GDP per capita (Table 7) into the model changed very little the I2 value. Both models had the highest AIC and nonsignificant Wald tests, except for GDP per capita for the circulatory group (Wald test, p = 0.03).
Socioeconomic factors (Table 7) reduced I2 values when included in the prediction models of non-accidental mortality groups and Older Adult subgroups. Models that included the income factor found lower I2 values when compared with the schooling factor.
Figure 3 shows the estimated temperature-mortality associations for the 25th and 75th percentiles values of the metapredictors included in each meta-regression model for the Older Adult circulatory mortality subgroup. We note steeper curves and higher relative risks associated with extreme temperatures for: locations at lower latitudes (towards the South), high annual amplitudes of mean temperature, lower average mean temperatures and lower income index value. Despite the proximity between the estimated temperature-mortality association curves for the variables daily amplitude of mean temperature and schooling, the results indicate that higher levels of these predictors lead to higher mortality risks. The other mortality groups studied had curve patterns similar to those of the cardiovascular subgroup, except for the metapredictor GDP per capita which was significant only in this group.
Cumulative 21-day lag temperature-circulatory mortality association curve of the Older Adult subgroups estimated for Brazil according to the 25th and 75th percentiles of each metapredictor.
Discussion
Using the DLNM method 2626. Gasparrini A, Armstrong B. Reducing and meta-analyzing estimates of distributed lag non-linear models. BMC Med Res Methodol 2013; 13:1.,3737. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med 2010; 29:2224-34. allowed us to capture the nonlinear and lagged relation dependent on the temperature-mortality association in Brazil. Our research advances in relation to previous studies 22. Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet 2015; 386:369-75.,2323. Requia WJ, Alahmad B, Schwartz JD, Koutrakis P. Association of low and high ambient temperature with mortality for cardiorespiratory diseases in Brazil. Environ Res 2023; 234:116532.,2828. 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.,3939. Ferreira LCM, Nogueira MC, Pereira RVB, de Farias WCM, Rodrigues MMDS, Teixeira MTB, et al. Ambient temperature and mortality due to acute myocardial infarction in Brazil: an ecological study of time-series analyses. Sci Rep 2019; 9:13790. by presenting the effect modifiers of extreme temperatures on non-accidental mortality in the general population and on four mortality causes among older adults, in addition to using a large number of metropolitan area distributed across the national territory.
Our results show the effects of extreme temperatures on the increased risk of mortality for non-accidental causes in the general population and for non-accidental, circulatory, respiratory and other causes in older adults in the metropolitan area, as well as in the Central-West, Northeast, Southeast and South regions. Effect shape and intensity and MMT/MMP values varied between the locations and causes studied. Geographic aspects, annual amplitude of the mean temperature and latitude were the effect modifier factors of the temperature-mortality association with the greatest impact, followed by income and, more discreetly, schooling. This modulating effect was found for the General and Older Adult groups, as well as for all causes of death subgroups.
The increased relative risk of non-accidental mortality associated with higher and lower temperature extremes found corroborates studies from China 3434. Ma W, Chen R, Kan H. Temperature-related mortality in 17 large Chinese cities: how heat and cold affect mortality in China. Environ Res 2014; 134:127-33., United States 1414. Anderson BG, Bell ML. Weather-related mortality: how heat, cold, and heat waves affect mortality in the United States. Epidemiology 2009; 20:205-13. and South Africa 1010. Scovronick N, Sera F, Acquaotta F, Garzena D, Fratianni S, Wright CY, et al. The association between ambient temperature and mortality in South Africa: a time-series analysis. Environ Res 2018; 161:229-35., which also included several locations. Such association variability between the analyzed locations has already been shown by studies in Brazil 2 and other countries 88. Ng CFS, Ueda K, Takeuchi A, Nitta H, Konishi S, Bagrowicz R, et al. Sociogeographic variation in the effects of heat and cold on daily mortality in Japan. J Epidemiol 2014; 24:15-24.,1616. Wang C, Zhang Z, Zhou M, Zhang L, Yin P, Ye W, et al. Nonlinear relationship between extreme temperature and mortality in different temperature zones: a systematic study of 122 communities across the mainland of China. Sci Total Environ 2017; 586:96-106.,4040. Zhang Y, Xiang Q, Yu Y, Zhan Z, Hu K, Ding Z. Socio-geographic disparity in cardiorespiratory mortality burden attributable to ambient temperature in the United States. Environ Sci Pollut Res 2019; 26:694-705..
Effects of extreme, high and low ambient temperatures on circulatory mortality in older adults were greater in locations southern and southeastern Brazil, as shown by previous studies in Brazil 2828. 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.,3939. Ferreira LCM, Nogueira MC, Pereira RVB, de Farias WCM, Rodrigues MMDS, Teixeira MTB, et al. Ambient temperature and mortality due to acute myocardial infarction in Brazil: an ecological study of time-series analyses. Sci Rep 2019; 9:13790. and other locations 33. 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.. Greatest impact of cold for this group was similar to previous studies with the Brazilian general population 2828. 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. and in other locations 1010. Scovronick N, Sera F, Acquaotta F, Garzena D, Fratianni S, Wright CY, et al. The association between ambient temperature and mortality in South Africa: a time-series analysis. Environ Res 2018; 161:229-35.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306.,4141. Wang X, Li G, Liu L, Westerdahl D, Jin X, Pan X. Effects of extreme temperatures on cause-specific cardiovascular mortality in China. Int J Environ Res Public Health 2015; 12:16136-56.. Hospitalization for cardiovascular disorders also suffers greater influence of cold 4242. Phung D, Thai PK, Guo Y, Morawska L, Rutherford S, Chu C. Ambient temperature and risk of cardiovascular hospitalization: an updated systematic review and meta-analysis. Sci Total Environ 2016; 550:1084-102.. Multiple physiological mechanisms are pointed out as promoters of cardiovascular responses induced by temperature changes, such as the high reactivity of the sympathetic nervous system and the renin-angiotensin system activated by cold, dehydration mediated by heat and cold, as well as systemic inflammatory response induced by heat stroke 2929. Liu C, Yavar Z, Sun Q. Cardiovascular response to thermoregulatory challenges. Am J Physiol Heart Circ Physiol 2015; 309:H1793-812.,4343. Stewart S, Keates AK, Redfern A, McMurray JJV. Seasonal variations in cardiovascular disease. Nat Rev Cardiol 2017; 14:654-64..
Our findings confirmed the effect of extreme ambient temperature on respiratory mortality in older adults, corroborating other studies that also investigated this outcome 1010. Scovronick N, Sera F, Acquaotta F, Garzena D, Fratianni S, Wright CY, et al. The association between ambient temperature and mortality in South Africa: a time-series analysis. Environ Res 2018; 161:229-35.,1414. Anderson BG, Bell ML. Weather-related mortality: how heat, cold, and heat waves affect mortality in the United States. Epidemiology 2009; 20:205-13.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306.,4444. Su X, Cheng Y, Wang Y, Liu Y, Li N, Li Y, et al. Regional temperature-sensitive diseases and attributable fractions in China. Int J Environ Res Public Health 2020; 17:184.. Effects of heat and cold on other causes of death in the Older Adult group confirm the findings of previous studies that considered this group 1010. Scovronick N, Sera F, Acquaotta F, Garzena D, Fratianni S, Wright CY, et al. The association between ambient temperature and mortality in South Africa: a time-series analysis. Environ Res 2018; 161:229-35.. This population includes mortality from genitourinary, digestive and endocrine diseases that are sensitive to extreme ambient temperatures 4444. Su X, Cheng Y, Wang Y, Liu Y, Li N, Li Y, et al. Regional temperature-sensitive diseases and attributable fractions in China. Int J Environ Res Public Health 2020; 17:184..
The cold and heat effects (RRs) for all causes of mortality had different and increasing values in southern Brazil, with greater impact of extreme ambient temperature in the South and Southeast. These data corroborate previous studies that show response heterogeneity between regions of a given territory 99. Luan G, Yin P, Wang L, Zhou M. The temperature - mortality relationship: an analysis from 31 Chinese provincial capital cities. Int J Environ Health Res 2018; 28:192-201.,1515. Ma W, Wang L, Lin H, Liu T, Zhang Y, Rutherford S, et al. The temperature-mortality relationship in China: an analysis from 66 Chinese communities. Environ Res 2015; 137:72-7..
Of the effect modifiers of the temperature-mortality association tested, geographic factors had the greatest impact. Range of ambient temperature, latitude and mean ambient temperature helped to explain the heterogeneity between locations, the first showing the greatest effect. A previous study on cardiovascular mortality in Brazil identified the influence of mean temperature amplitude on the temperature-mortality association 2828. 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.. Significant effect of latitude on the heterogeneity of the associations between ambient temperature and mortality has been previously reported for non-accidental 1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306.,3333. Bao J, Wang Z, Yu C, Li X. The influence of temperature on mortality and its lag effect: a study in four Chinese cities with different latitudes. BMC Public Health 2016; 16:375.,3434. Ma W, Chen R, Kan H. Temperature-related mortality in 17 large Chinese cities: how heat and cold affect mortality in China. Environ Res 2014; 134:127-33. and cardiovascular 33. 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. mortality. The latitude indicator may represent the effect generated by the annual variation in mean temperature, since the amplitude of mean temperature is greater in the southernmost metropolitan areas.
Variations in the estimated RRs of the temperature-mortality association and the modulating effect of mean temperature amplitude may be related to different physiological adaptation (acclimatization) responses to different climatic situations 1515. Ma W, Wang L, Lin H, Liu T, Zhang Y, Rutherford S, et al. The temperature-mortality relationship in China: an analysis from 66 Chinese communities. Environ Res 2015; 137:72-7..
MMT data from our study point in the same direction. MMT is a characteristic aspect of the temperature-mortality association and how it can be influenced by many factors 4545. Krummenauer L, Prahl BF, Costa L, Holsten A, Walther C, Kropp JP. Global drivers of minimum mortality temperatures in cities. Sci Total Environ 2019; 695:133560.. MMT is the temperature with the least effect on the mortality rate 3838. 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., thus being a threshold and would be related to people’s ability to adapt to the local climate. Here, the estimated MMT values for the General group and the Older Adult group varied between locations, with higher values in places close to the Equator and decreasing along their distance, similar to previous studies 99. Luan G, Yin P, Wang L, Zhou M. The temperature - mortality relationship: an analysis from 31 Chinese provincial capital cities. Int J Environ Health Res 2018; 28:192-201.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306.,3535. Medina-Ramón M, Schwartz J. Temperature, temperature extremes, and mortality: a study of acclimatisation and effect modification in 50 US cities. Occup Environ Med 2007; 64:827-33.,3838. 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.. Locations with smaller ambient temperature ranges are close to the equator, thus its residents are consistently exposed to higher temperatures. Individuals routinely exposed to higher temperatures could develop acclimatization to this condition, with more efficient and less pronounced physiological responses to temperature extremes 4343. Stewart S, Keates AK, Redfern A, McMurray JJV. Seasonal variations in cardiovascular disease. Nat Rev Cardiol 2017; 14:654-64..
Additionally, there could be other behavioral adaptations (e.g., use of air conditioning/heater) of the populations to the local climate 3434. Ma W, Chen R, Kan H. Temperature-related mortality in 17 large Chinese cities: how heat and cold affect mortality in China. Environ Res 2014; 134:127-33. that could explain this heterogeneity.
Among the urban factors, population density had no effect on any of the groups, contrary to previous studies that reported its influence and showed that the effect of heat on mortality was greater in places with higher population density 88. Ng CFS, Ueda K, Takeuchi A, Nitta H, Konishi S, Bagrowicz R, et al. Sociogeographic variation in the effects of heat and cold on daily mortality in Japan. J Epidemiol 2014; 24:15-24.,3535. Medina-Ramón M, Schwartz J. Temperature, temperature extremes, and mortality: a study of acclimatisation and effect modification in 50 US cities. Occup Environ Med 2007; 64:827-33.. GDP per capita was a modulating factor of the temperature-mortality association only for the circulatory subgroup, but slightly reduced heterogeneity. Higher GDP per capita values had higher RR, in line with previous studies that showed the influence of this effect modifier 3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306.. A systematic review on the effect modifiers of the temperature-mortality association reported weak to limited evidence for the influence of community factors like population density, heating system, health facilities, proximity to water, housing quality, and level of air pollution, and limited or suggestive evidence for socioeconomic status, latitude, urban/rural, air conditioning, climatic condition, proportion of green areas or vegetation, and previous winter mortality 1717. Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004..
Socioeconomic factors, schooling and income moderately influenced the temperature-mortality association between locations in all the groups studied. Schooling level was measured by the percentage of individuals with complete primary education, with higher RR values found for places with higher percentage values. Some studies show greater susceptibility of illiterate individuals with respect to the ambient temperature effect on non-accidental mortality 77. Marí-Dell'Olmo M, Tobías A, Gómez-Gutiérrez A, Rodríguez-Sanz M, García de Olalla P, Camprubí E, et al. Social inequalities in the association between temperature and mortality in a South European context. Int J Public Health 2019; 64:27-37.,3232. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306.,4646. Yang J, Ou CQ, Ding Y, Zhou YX, Chen PY. Daily temperature and mortality: a study of distributed lag non-linear effect and effect modification in Guangzhou. Environ Health 2012; 11:63., while other studies show that the percentage of individuals with a schooling level lower than the ninth grade (modifying factor) does not explain the heterogeneity between cities 1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.. In this case, the difference could be explained by the analysis approach and difference in schooling stratification. Studies on the cardiovascular temperature-mortality association that included the schooling factor presented divergent results 33. 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..
Regarding the income indicator, we note that places where a higher percentage of people have an income level equal to or less than one minimum wage or no income show a greater effect of ambient temperature on mortality. Previous studies that accounted for individuals’ poverty reported its influence only on heat effects 1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.,3434. Ma W, Chen R, Kan H. Temperature-related mortality in 17 large Chinese cities: how heat and cold affect mortality in China. Environ Res 2014; 134:127-33., while studies focusing on cardiovascular mortality presented divergent results in relation to this indicator 33. 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..
Study limitations includes those inherent to using secondary databases. Missing climate data throughout the time series in some metropolitan areas was minimized by using imputations for locations with more than one climate monitoring station. Another limitation was the lack of adjustments due to local air pollution levels given the lack of data in most locations studied. However, previous studies that analyzed the influence of air pollution on the temperature-mortality relation showed slight or no change in effects 1111. Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.,1414. Anderson BG, Bell ML. Weather-related mortality: how heat, cold, and heat waves affect mortality in the United States. Epidemiology 2009; 20:205-13.,4646. Yang J, Ou CQ, Ding Y, Zhou YX, Chen PY. Daily temperature and mortality: a study of distributed lag non-linear effect and effect modification in Guangzhou. Environ Health 2012; 11:63.. Thus, confounding in this case would be low 4646. Yang J, Ou CQ, Ding Y, Zhou YX, Chen PY. Daily temperature and mortality: a study of distributed lag non-linear effect and effect modification in Guangzhou. Environ Health 2012; 11:63..
We did not analyze the role of other climatic factors and events 1313. Core Writing Team; Lee H, Romero J, editors. Climate change 2023: synthesis report. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva: International Panel on Climate Change; 2023.,4747. Hartinger SM, Palmeiro-Silva YK, Llerena-Cayo C, Blanco-Villafuerte L, Escobar L E, Diaz A, et al. The 2023 Latin America report of the Lancet Countdown on health and climate change: the imperative for health-centred climate-resilient development. Lancet Reg Health Am 2024; 33:100746. such as precipitation, excessive rainfall, drought periods, heat and cold waves, which also occur in Brazil and could contribute to mortality events. Future studies should seek to verify the influences of these factors.
Research on factors modifying the effect of extreme ambient temperature on mortality is important to identify vulnerabilities that could amplify this effect and which can be minimized with appropriate mitigation proposals. Such adaptations, whether undertaken by a person or an institution, could reduce the impact of this climate factor 4848. Arbuthnott K, Hajat S, Heaviside C, Vardoulakis S. Changes in population susceptibility to heat and cold over time: assessing adaptation to climate change. Environ Health 2016; 15 Suppl 1:33. in population mortality, especially for those most susceptible such as older adults 1717. Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004..
Our study used a larger number and size of locations in Brazil to address the effect and its modifiers of the temperature-environment association on non-accidental, circulatory, respiratory and other causes mortality in older adults. Besides reinforcing the findings of previous studies, this work enables visualizing places and populations with more immediate needs for climate adaptation actions.
Daily meteorological indicators (mean, maximum, minimum temperature and relative humidity) and location (latitude) of the meteorological station were extracted from the Meteorological Database for Teaching and Research (BDMEP, acronym in Portuguese; https://bdmep.inmet.gov.br/), Brazilian National Institute of Meteorology (INMET, acronym in Portuguese).
References
- 1Song X, Wang S, Hu Y, Yue M, Zhang T, Liu Y, et al. Impact of ambient temperature on morbidity and mortality: an overview of reviews. Sci Total Environ 2017; 586:241-54.
- 2Gasparrini A, Guo Y, Hashizume M, Lavigne E, Zanobetti A, Schwartz J, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet 2015; 386:369-75.
- 3Moghadamnia 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.
- 4Alahmad B, Khraishah H, Royé D, Vicedo-Cabrera AM, Guo Y, Papatheodorou SI, et al. Associations between extreme temperatures and cardiovascular cause-specific mortality: results from 27 countries. Circulation 2023; 147:35-46.
- 5Bunker A, Wildenhain J, Vandenbergh A, Henschke N, Rocklöv J, Hajat S, et al. 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; 6:258-68.
- 6Wen J, Zou L, Jiang Z, Li Y, Tao J, Liu Y, et al. Association between ambient temperature and risk of stroke morbidity and mortality: a systematic review and meta-analysis. Brain Behav 2023; 13:e3078.
- 7Marí-Dell'Olmo M, Tobías A, Gómez-Gutiérrez A, Rodríguez-Sanz M, García de Olalla P, Camprubí E, et al. Social inequalities in the association between temperature and mortality in a South European context. Int J Public Health 2019; 64:27-37.
- 8Ng CFS, Ueda K, Takeuchi A, Nitta H, Konishi S, Bagrowicz R, et al. Sociogeographic variation in the effects of heat and cold on daily mortality in Japan. J Epidemiol 2014; 24:15-24.
- 9Luan G, Yin P, Wang L, Zhou M. The temperature - mortality relationship: an analysis from 31 Chinese provincial capital cities. Int J Environ Health Res 2018; 28:192-201.
- 10Scovronick N, Sera F, Acquaotta F, Garzena D, Fratianni S, Wright CY, et al. The association between ambient temperature and mortality in South Africa: a time-series analysis. Environ Res 2018; 161:229-35.
- 11Xiao J, Peng J, Zhang Y, Liu T, Rutherford S, Lin H, et al. How much does latitude modify temperature-mortality relationship in 13 Eastern US cities? Int J Biometeorol 2015; 59:365-72.
- 12Kephart JL, Sánchez BN, Moore J, Schinasi LH, Bakhtsiyarava M, Ju Y, et al. City-level impact of extreme temperatures and mortality in Latin America. Nat Med 2022; 28:1700-5.
- 13Core Writing Team; Lee H, Romero J, editors. Climate change 2023: synthesis report. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva: International Panel on Climate Change; 2023.
- 14Anderson BG, Bell ML. Weather-related mortality: how heat, cold, and heat waves affect mortality in the United States. Epidemiology 2009; 20:205-13.
- 15Ma W, Wang L, Lin H, Liu T, Zhang Y, Rutherford S, et al. The temperature-mortality relationship in China: an analysis from 66 Chinese communities. Environ Res 2015; 137:72-7.
- 16Wang C, Zhang Z, Zhou M, Zhang L, Yin P, Ye W, et al. Nonlinear relationship between extreme temperature and mortality in different temperature zones: a systematic study of 122 communities across the mainland of China. Sci Total Environ 2017; 586:96-106.
- 17Son JY, Liu JC, Bell ML. Temperature-related mortality: a systematic review and investigation of effect modifiers. Environ Res Lett 2019; 14:073004.
- 18Sera F, Armstrong B, Tobias A, Vicedo-Cabrera AM, Åström C, Bell ML, et al. How urban characteristics affect vulnerability to heat and cold: a multi-country analysis. Int J Epidemiol 2019; 48:1101-12.
- 19Requia WJ, Vicedo-Cabrera AM, de Schrijver E, Amini H. Low ambient temperature and hospitalization for cardiorespiratory diseases in Brazil. Environ Res 2023; 231:116231.
- 20Requia WJ, Vicedo-Cabrera AM, de Schrijver E, Amini H, Gasparrini A. Association of high ambient temperature with daily hospitalization for cardiorespiratory diseases in Brazil: a national time-series study between 2008 and 2018. Environ Pollut 2023; 331:121851.
- 21Oliveira BFA, Jacobson LSV, Perez LP, Silveira IH, Junger WL, Hacon SS. Impacts of heat stress conditions on mortality from respiratory and cardiovascular diseases in Brazil. Sustainability in Debate 2020; 11:297-313.
- 22Barbi F, Ferreira LC. Climate change in Brazilian cities: policy strategies and responses to global warming. Int J Environ Sci Dev 2013; 4:49-51.
- 23Requia WJ, Alahmad B, Schwartz JD, Koutrakis P. Association of low and high ambient temperature with mortality for cardiorespiratory diseases in Brazil. Environ Res 2023; 234:116532.
- 24Marguti BO, Tavares SR, editors. Política metropolitana: governança, instrumentos e planejamentometropolitanos - II Seminário e Oficina. Brasília: Instituto de Pesquisa Econômica Aplicada; 2019.
- 25Burkart KG, Brauer M, Aravkin AY, Godwin WW, Hay SI, He J, et al. Estimating the cause-specific relative risks of non-optimal temperature on daily mortality: a two-part modelling approach applied to the Global Burden of Disease Study. Lancet 2021; 398:685-97.
- 26Gasparrini A, Armstrong B. Reducing and meta-analyzing estimates of distributed lag non-linear models. BMC Med Res Methodol 2013; 13:1.
- 27Barnett AG, Tong S, Clements ACA. What measure of temperature is the best predictor of mortality? Environ Res 2010; 110:604-11.
- 28Silveira 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.
- 29Liu C, Yavar Z, Sun Q. Cardiovascular response to thermoregulatory challenges. Am J Physiol Heart Circ Physiol 2015; 309:H1793-812.
- 30Davis RE, McGregor GR, Enfield KB. Humidity: a review and primer on atmospheric moisture and human health. Environ Res 2016; 144:106-16.
- 31Junger WL, Ponce de Leon A. Imputation of missing data in time series for air pollutants. Atmos Environ 2015; 102:96-104.
- 32Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, et al. Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities. BMJ 2018; 363:k4306.
- 33Bao J, Wang Z, Yu C, Li X. The influence of temperature on mortality and its lag effect: a study in four Chinese cities with different latitudes. BMC Public Health 2016; 16:375.
- 34Ma W, Chen R, Kan H. Temperature-related mortality in 17 large Chinese cities: how heat and cold affect mortality in China. Environ Res 2014; 134:127-33.
- 35Medina-Ramón M, Schwartz J. Temperature, temperature extremes, and mortality: a study of acclimatisation and effect modification in 50 US cities. Occup Environ Med 2007; 64:827-33.
- 36Bhaskaran K, Gasparrini A, Hajat S, Smeeth L, Armstrong B. Time series regression studies in environmental epidemiology. Int J Epidemiol 2013; 42:1187-95.
- 37Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med 2010; 29:2224-34.
- 38Yin 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.
- 39Ferreira LCM, Nogueira MC, Pereira RVB, de Farias WCM, Rodrigues MMDS, Teixeira MTB, et al. Ambient temperature and mortality due to acute myocardial infarction in Brazil: an ecological study of time-series analyses. Sci Rep 2019; 9:13790.
- 40Zhang Y, Xiang Q, Yu Y, Zhan Z, Hu K, Ding Z. Socio-geographic disparity in cardiorespiratory mortality burden attributable to ambient temperature in the United States. Environ Sci Pollut Res 2019; 26:694-705.
- 41Wang X, Li G, Liu L, Westerdahl D, Jin X, Pan X. Effects of extreme temperatures on cause-specific cardiovascular mortality in China. Int J Environ Res Public Health 2015; 12:16136-56.
- 42Phung D, Thai PK, Guo Y, Morawska L, Rutherford S, Chu C. Ambient temperature and risk of cardiovascular hospitalization: an updated systematic review and meta-analysis. Sci Total Environ 2016; 550:1084-102.
- 43Stewart S, Keates AK, Redfern A, McMurray JJV. Seasonal variations in cardiovascular disease. Nat Rev Cardiol 2017; 14:654-64.
- 44Su X, Cheng Y, Wang Y, Liu Y, Li N, Li Y, et al. Regional temperature-sensitive diseases and attributable fractions in China. Int J Environ Res Public Health 2020; 17:184.
- 45Krummenauer L, Prahl BF, Costa L, Holsten A, Walther C, Kropp JP. Global drivers of minimum mortality temperatures in cities. Sci Total Environ 2019; 695:133560.
- 46Yang J, Ou CQ, Ding Y, Zhou YX, Chen PY. Daily temperature and mortality: a study of distributed lag non-linear effect and effect modification in Guangzhou. Environ Health 2012; 11:63.
- 47Hartinger SM, Palmeiro-Silva YK, Llerena-Cayo C, Blanco-Villafuerte L, Escobar L E, Diaz A, et al. The 2023 Latin America report of the Lancet Countdown on health and climate change: the imperative for health-centred climate-resilient development. Lancet Reg Health Am 2024; 33:100746.
- 48Arbuthnott K, Hajat S, Heaviside C, Vardoulakis S. Changes in population susceptibility to heat and cold over time: assessing adaptation to climate change. Environ Health 2016; 15 Suppl 1:33.