Abstract
-
Objectives
Climate variability, particularly ambient air temperature, is an emerging environmental determinant of blood pressure (BP); however, evidence from tropical low- and middle-income countries (LMICs) remains limited. We examined associations between monthly average ambient temperature with systolic BP (SBP) and diastolic BP (DBP) among adults in Bogor City, Indonesia.
-
Methods
This longitudinal cohort analysis used secondary data collected between 2011 and 2018 from 1,648 participants with repeated BP measurements (19 assessments over 6 years). Average ambient temperature data were matched to the month of BP measurement. Generalized estimating equations models were used to assess the association between temperature and BP after adjustment for age, sex, education, and socioeconomic status.
-
Results
Mean ambient temperature ranged from 24.6 °C to 28.9 °C. SBP varied more than DBP and tended to increase as ambient temperature decreased. The overall main effect of temperature on BP was not statistically significant. However, several visit-specific temperature–time interactions were negative and significant, suggesting inverse patterns between lower temperatures and higher SBP or DBP. Marginal effects plots demonstrated time-varying associations, with predicted SBP and DBP differences of approximately −4 to +3 mmHg across follow-up visits at representative temperatures. Socioeconomic status and education were associated with SBP, whereas age and female sex were associated with DBP.
-
Conclusion
Temperature variations were associated with time-varying BP changes among adults with hypertension in Bogor City during 2011–2018. Although the main temperature effect was non-significant, temperature–time interactions suggested modest inverse patterns within clinically plausible ranges. These findings support the need for updated, multi-site studies in tropical LMIC settings.
-
Keywords: Blood pressure; Climate change; Hypertension; Temperature
Introduction
Non-communicable diseases (NCDs) are chronic conditions driven by genetic, physiological, environmental, and behavioral factors. Although they are often associated with aging, NCDs also affect younger populations, with 17 million people younger than 70 years dying annually from these diseases [
1]. Globally, NCDs represent a major health burden, and without innovative interventions, achieving the Sustainable Development Goal target of reducing premature NCD mortality by one-third by 2030 appears unlikely [
2].
According to the Global Burden of Disease Study, cardiovascular diseases were the leading cause of NCD-related death globally and in Indonesia from 1990 to 2019. Elevated systolic blood pressure (SBP) accounted for approximately 10.8 million deaths, representing 19.2% of all deaths in 2019 [
3].
Climate change is projected to exacerbate environmental hazards such as heat waves, heavy rainfall, and droughts, all of which substantially affect human health. These changes are associated with increased risks of cardiovascular, respiratory, and infectious diseases and may also threaten health infrastructure [
4].
Although numerous studies have examined temperature variability and cardiovascular mortality [
5–
11], fewer have investigated its specific association with blood pressure (BP). Cross-sectional studies have shown strong associations between ambient temperature and both SBP and diastolic BP (DBP) [
12–
15]. However, longitudinal evidence remains limited. A study in Scotland found that temperature-sensitive individuals experienced notable effects on mortality and SBP [
16], while research from Asian countries with 4 distinct seasons reported a 5.7 mmHg increase in SBP per 10 °C decrease in temperature [
17]. Most of these studies were conducted in high-income countries with temperate or subtropical climates, leaving an important gap in the evidence from tropical low- and middle-income countries (LMICs) such as Indonesia.
The human body can adapt to ambient temperatures, but extreme heat may induce physiological stress, increase heart rate, and raise hypertension risk via hormonal and nervous system activation [
18]. In Indonesia, the burden of NCDs—especially hypertension, heart disease, and stroke—continues to rise. Data from the Basic Health Research (Riskesdas) surveys show that the prevalence of hypertension among adults increased from 25.8% in 2013 to 34.1% in 2018, while coronary heart disease and stroke also displayed upward trends [
19].
Despite growing global attention to environmental temperature and cardiovascular health, research in Indonesia’s tropical setting remains limited. Local evidence from Banjarmasin showed a correlation between indoor temperature and hypertension [
20], and national data from 2015–2016 indicated higher numbers of outpatient visits for diabetes and cardiovascular diseases on days when temperatures exceeded 29.5 °C [
21].
Indonesia has developed national health sector guidelines to address climate-related health risks, reflecting growing recognition of the health impacts of climate variability [
22]. Nevertheless, the available studies remain inconsistent in methodology and outcomes, which limits their generalizability [
23]. Most of the existing evidence still originates from high-income countries with temperate climates, leaving a critical knowledge gap in tropical LMICs. Although the global literature on the health effects of climate change has expanded, evidence from LMICs has changed little over the past decade and remains concentrated in a small number of countries and climate-sensitive diseases [
24].
To address this gap, this study investigated the relationship between ambient temperature and BP variability using 6 years of cohort data from Bogor City collected between 2011 and 2018. Although the dataset provides rare longitudinal evidence from a tropical LMIC context, the period of data collection may limit its direct applicability to the current policy environment, a limitation discussed later in this paper. This study aimed to clarify how temperature fluctuations were associated with BP as an intermediate cardiovascular risk factor.
Materials and Methods
Study Design
This longitudinal study used data from the Non-Communicable Disease Risk Factor Cohort Study conducted in Bogor City by the Center for Research and Development of Public Health Efforts, Ministry of Health, Indonesia, between 2011 and 2018. These data were linked with ambient temperature data obtained from the Indonesian Agency of Meteorological, Climatological, and Geophysical (BMKG).
Study Participants
A total of 5,296 individuals were originally enrolled in the Bogor City cohort. BP measurements were obtained at least once during follow-up from 4,896 participants. Attendance varied across scheduled visits, which were conducted 3 times per year, resulting in intermittent missing measurements. A subset of 1,648 participants completed all 19 scheduled BP assessments over the 6-year follow-up period and was classified as having complete BP data.
Missing measurements primarily resulted from irregular attendance, as well as attrition due to illness, death, or relocation. The primary longitudinal analyses used a complete-case approach restricted to participants with BP measurements at all scheduled visits, thereby ensuring consistency in the repeated-measures analysis.
BP Measurement
BP was measured repeatedly over the 6-year follow-up period. Measurements were obtained every 4 months, resulting in 3 assessments per year and a total of 19 measurements per participant, including 1 baseline assessment in 2011 and 18 follow-up assessments.
Hypertension was defined according to the 2018 guidelines of the European Society of Cardiology/European Society of Hypertension, the Eighth Joint National Committee, and the Indonesian Society of Hypertension, all of which classify hypertension as SBP ≥140 mmHg and/or DBP ≥90 mmHg [
25].
BP was measured by trained paramedical staff from the Bogor City Health Office using a validated automated digital sphygmomanometer and a standardized protocol designed to minimize random error and systematic measurement bias. Participants were instructed to avoid smoking, eating, and physical activity for at least 30 minutes before measurement and to remain seated quietly for 5–15 minutes before assessment. Measurements were obtained in a calm indoor environment with participants seated, their feet flat on the floor, their legs uncrossed, and their right arm supported at heart level.
An appropriately sized cuff was placed on the right upper arm, with the lower edge positioned 1–2 cm above the antecubital fossa and the bladder centered over the brachial artery. Either 2 or 3 consecutive readings were obtained at 5-minute intervals, and their average was used for analysis. The right arm was used consistently for all participants to reduce inter-arm variability.
To reduce variability related to the timing of measurement, BP assessments were conducted during scheduled examination sessions using the same protocol and setting across all study visits. Although the exact clock time of measurement was not recorded, standardized premeasurement resting conditions and uniform measurement procedures were applied consistently throughout the study period to enhance comparability across follow-up visits [
26].
Temperature Measurement
Ambient air temperature data were obtained from the BMKG. Monthly average temperatures from the West Java Climatology Station, the only station providing complete and validated long-term records for Bogor City, were used for the 2011–2018 study period. This station serves as the official reference for local climate monitoring.
The monthly mean ambient temperature was selected to represent medium-term thermal exposure relevant to BP regulation. BP reflects an integrated physiological state influenced by sustained autonomic activity, vascular resistance, and volume regulation over days to weeks rather than acute responses to short-term temperature fluctuations. Because BP measurements were obtained at scheduled cohort visits approximately every 4 months, monthly averages provided a temporally aligned exposure metric and reduced random variability associated with day-to-day temperature changes.
Temperature exposure was matched to the calendar month of each BP measurement to ensure temporal correspondence between environmental conditions and clinical assessments. This approach assumes broadly similar ambient exposure across participants and does not capture neighborhood-level microclimatic variation or individual indoor–outdoor activity patterns. No missing values were reported in the BMKG temperature dataset during the study period.
Lagged temperature structures were not applied because the study was not designed to evaluate acute heat or cold effects within narrow physiological windows, such as daily lags. Instead, the analysis focused on medium-term exposure patterns and their time-varying associations with repeated BP measurements. Within the observed narrow temperature range, introducing multiple lag terms would have increased model complexity without clear physiological interpretability.
Covariates
Baseline covariates included age, sex, education level, and socioeconomic status (SES). Age was defined at cohort entry and categorized as 30–44 years, 45–59 years, and ≥60 years to capture baseline age-related differences in BP risk. Sex was classified as male or female. Education level reflected the highest level attained at baseline, was categorized as less than high school or high school and above, and was assumed to remain stable throughout follow-up.
SES was derived from baseline household income relative to the Bogor City municipal minimum wage (3,557,146 Indonesian Rupiah [IDR] in 2018), in accordance with official government standards. SES was categorized as below or above the municipal minimum wage and was treated as a baseline, time-invariant covariate. Annual changes in income or SES were not assessed because repeated measurements were not available in the cohort dataset. Accordingly, SES was modeled as an indicator of long-term socioeconomic position rather than short-term economic fluctuation.
Several established individual-level determinants of BP, including body mass index, smoking status, alcohol consumption, physical activity, dietary intake, comorbidities such as diabetes or kidney disease, and antihypertensive medication use, were unavailable in the dataset and therefore could not be included in the models. In addition, information on individual-level cooling access, such as air conditioning use, ventilation, housing characteristics, and indoor temperature, was unavailable. These variables were excluded because of data limitations.
Statistical Analysis
To estimate the longitudinal association between ambient temperature and BP, generalized estimating equation (GEE) models were used to account for repeated measurements within individuals. Follow-up visits, including baseline in 2011 and 18 subsequent assessments through 2018, were represented as indicator variables, with baseline as the reference category. Therefore, the visit-specific coefficients reflect contrasts relative to baseline rather than direct physiological changes. All models were adjusted for age, sex, education level, and SES.
An unstructured working correlation matrix was specified to allow maximum flexibility in modeling within-individual correlations across repeated BP measurements. This choice was motivated by the long follow-up period, irregular visit spacing inherent to cohort field implementation, and the absence of strong a priori assumptions regarding correlation decay over time, which could have made more parsimonious structures such as exchangeable or autoregressive (AR-1) overly restrictive. Alternative working correlation structures—independent, exchangeable, and AR-1—were formally compared using the quasi-likelihood under the independence model criterion (QIC) and its small-sample correction (QICC). Because QIC and QICC values were broadly comparable across structures, the unstructured correlation matrix was retained for the primary analyses (
Table S1). Additional details on the comparison of working correlation structures and sensitivity analyses are provided in
Supplementary Material 1.
Potential non-linear temperature effects were considered a priori, given previous evidence of U- or J-shaped associations between temperature and cardiovascular outcomes. However, the observed temperature range during the study period was relatively narrow (24.6 °C–28.9 °C), which limited statistical power to detect meaningful non-linearities or threshold effects. Exploratory analyses using alternative parameterizations, including temperature percentiles and spline-based terms, did not indicate substantial departures from linearity and yielded unstable estimates. Therefore, ambient temperature was modeled as a continuous linear term, with temporal heterogeneity captured through temperature–visit interaction terms.
Ethical Approval
The study protocol was approved by the Ethics Committee of the National Institute of Health Research and Development, Ministry of Health, Indonesia (approval numbers: KE.01.05/EC/394/2012 for baseline and LB.02.01/2/KE.076/2018 for follow-up assessments).
Results
During the 2011–2018 follow-up period in Bogor City, monthly mean ambient temperature varied within a relatively narrow range, from 24.6 °C to 28.9 °C. Within this context, 1,648 hypertensive adults aged ≥30 years completed all 19 scheduled BP assessments and were included in the analysis. Given the limited exposure contrast, temperature-related BP effects were expected to be modest and to primarily reflect temporal variability rather than responses to extreme temperatures.
The analytical sample was predominantly female (77.2%), and most participants were 45–59 years of age (46.4%). More than half had less than a high school education (58.9%), and most lived below the municipal minimum wage threshold (
Table 1).
As shown in
Figure 1, within-person variability was greater for SBP than for DBP, with mean within-person standard deviations of 10.0 mmHg and 6.4 mmHg, respectively, indicating greater temporal fluctuation in SBP. Simple Pearson correlation analyses showed near-zero associations between ambient temperature and BP (data not shown), reflecting between-individual correlations averaged across repeated measurements. In contrast, the GEE framework explicitly accounts for within-individual correlation over time and estimates population-averaged associations based on repeated measures, making it more sensitive to detecting time-dependent relationships driven by within-person changes across follow-up visits.
Regression coefficients from the GEE models represent the estimated change in BP per 1 °C difference in monthly mean ambient temperature at a given follow-up visit. To facilitate clinical interpretation, these coefficients were translated into predicted BP values at representative temperature levels corresponding to the 25th percentile (24.6 °C), median (26.8 °C), and 75th percentile (28.9 °C) of the observed temperature distribution.
Figure 2 summarizes the time-varying temperature effects across all follow-up visits and highlights periods during which lower ambient temperatures were associated with higher BP. Several visits (e.g., visits 3–5 and 14–18) showed negative and statistically significant slopes, indicating higher SBP and DBP at relatively cooler temperatures during these periods. However, rather than emphasizing isolated visit-specific coefficients, the primary inference was based on global Wald tests, which confirmed statistically significant time effects and temperature–visit interactions for both SBP (Wald
χ²=272.9, degrees of freedom [df]=18,
p<0.001) and DBP (Wald
χ²=373.5, df=18,
p<0.001), indicating that temperature–BP associations varied across follow-up visits.
Marginal effects plots further illustrate these patterns by translating the coefficients into clinically interpretable predicted BP values at representative temperature contrasts (
Figures 3,
4). These predicted values correspond to the 25th percentile (24.6 °C), median (26.8 °C), and 75th percentile (28.9 °C) of the observed temperature distribution, facilitating interpretation of the time-varying effects. Across visits, predicted SBP differences associated with lower versus higher temperatures ranged from approximately −4 to +3 mmHg, whereas the corresponding DBP differences were smaller but followed similar time-varying patterns.
Sensitivity analyses excluding the second follow-up visit, which showed the largest visit-specific coefficient, yielded similar time-varying patterns, indicating that the findings were not driven by a single influential visit. Overall, no clear evidence of non-linear or threshold effects was observed within the narrow temperature range, and linear specifications provided stable and interpretable estimates.
Covariate effects were modest. Higher SES was associated with slightly higher SBP, whereas higher educational attainment was associated with lower SBP; corresponding patterns for DBP were less consistent. Model comparisons based on the QIC and QICC supported the use of an unstructured working correlation matrix for both SBP and DBP outcomes (
Table S1). Detailed visit-specific regression coefficients are presented in
Tables S2 and
S3, with additional model diagnostics and sensitivity analyses provided in
Tables S1–
S4.
Overall, when expressed in clinically interpretable terms, the observed associations corresponded to small BP differences on the order of a few mmHg across representative temperature contrasts. These modest effect sizes varied over time, and the primary inference was supported by joint tests of temporal heterogeneity rather than isolated visit-specific estimates.
Discussion
This longitudinal cohort study suggests that associations between ambient temperature and BP in a tropical urban setting varied over time rather than remaining uniform throughout follow-up. Although the main temperature effects were not statistically significant, significant time effects and temperature–time interactions indicate that BP responses to temperature differed across visits. These findings suggest that the association between ambient temperature and BP is context-dependent and shaped by temporal, physiological, and environmental factors rather than by a single stable exposure–response relationship.
Notably, the absence of a statistically significant overall temperature effect does not imply a lack of physiological relevance. Within the relatively narrow temperature range observed in this tropical environment, even modest variations in ambient temperature may affect BP regulation. Predicted differences in SBP across representative temperature contrasts were small, on the order of a few millimeters of mercury, but were physiologically plausible and consistent with gradual cardiovascular adaptation [
17,
27,
28]. Such modest effect sizes would be expected in tropical populations, in which long-term acclimatization and behavioral adaptation may attenuate the magnitude of temperature-related BP changes relative to temperate or cold climates [
29].
Several biological mechanisms may explain the inverse associations observed at specific follow-up visits between lower ambient temperature and higher BP. Even in tropical climates without extreme cold exposure, relatively cooler temperatures may trigger peripheral vasoconstriction, increased sympathetic nervous system activity, and elevated vascular resistance, leading to transient increases in BP [
27,
30]. Cold-related impairment of endothelial function, including reduced flow-mediated dilation, increased blood viscosity, and activation of the renin–angiotensin system, has been documented across diverse climatic settings and provides biologically plausible pathways linking lower temperatures to higher BP [
30–
32]. These mechanisms operate along a continuum and are not restricted to extreme cold conditions.
Temporal heterogeneity in the temperature–BP association may also reflect physiological acclimatization and changes in individual susceptibility over time. Repeated exposure to seasonal patterns may modify autonomic and vascular responses, whereas aging, disease progression, and changes in healthcare access may alter BP sensitivity to environmental stressors during follow-up [
17,
33]. In addition, tropical meteorological conditions vary across months and years in ways not fully captured by mean temperature alone, including changes in humidity, rainfall, cloud cover, and perceived thermal stress, all of which may modulate cardiovascular strain at specific time points [
34].
The use of monthly mean temperature captures cumulative thermal exposure relevant to BP regulation, which reflects integrated autonomic, vascular, and volume-related processes operating over days to weeks rather than acute responses to short-term temperature fluctuations [
17,
35]. Within this framework, the observed time-varying associations are best interpreted as reflecting relative thermal stress within a stable climatic range rather than responses to extreme heat or cold. Accordingly, modeling temperature as a linear exposure with time-varying interactions was both methodologically appropriate and biologically plausible in this cohort.
Socioeconomic and housing-related factors may also modify temperature–BP relationships in tropical urban populations. Variations in housing quality, ventilation, and behavioral responses to cooler periods may influence individual exposure and vulnerability over time [
35]. Because information on indoor environments and adaptive behaviors was unavailable in this study, the observed associations should be interpreted as average population-level effects and may underestimate temperature sensitivity among individuals with limited adaptive capacity. Future studies incorporating indoor environmental data, higher-resolution meteorological exposure measures, and individual adaptive behaviors are needed to better characterize vulnerability to temperature-related BP changes in tropical settings.
In this study, participants with higher educational attainment tended to have slightly lower SBP, whereas those with higher SES had marginally higher SBP. These associations should be interpreted cautiously and should not be construed as evidence of a direct causal relationship. Importantly, individual-level behavioral factors such as dietary patterns, physical activity, smoking, alcohol consumption, and healthcare utilization were not measured and therefore could not be directly evaluated.
The observed pattern may reflect broader social and environmental gradients associated with socioeconomic position rather than the intrinsic effects of income or education alone. In several LMIC settings undergoing epidemiological and nutritional transition, higher SES has been associated with a higher prevalence of hypertension. For example, studies from Nepal have reported higher BP and hypertension prevalence among employed and higher-income groups, potentially reflecting shifts toward more sedentary occupations and increased consumption of processed or energy-dense foods [
36–
38].
Alternative explanations may include differences in occupational stress, work-related demands, healthcare access, or other unmeasured lifestyle factors associated with socioeconomic position. In contrast, higher educational attainment may confer modest protective effects through improved health literacy, greater awareness of cardiovascular risk, and healthier long-term behaviors, even in contexts in which income-related exposures increase. Given the absence of detailed behavioral and clinical covariates, these findings should be interpreted as descriptive associations that highlight the social patterning of BP rather than causal pathways.
DBP was 3.5–3.7 mmHg higher among participants aged 45–59 years and ≥60 years than among those aged 30–44 years, and it was 1.2 mmHg higher in women than in men. These findings are consistent with research from China reporting a 6.4 mmHg increase in SBP per decade of age [
39]. Age is a well-established determinant of BP, with both baseline age and aging positively associated with higher values [
40]. However, Khajavi et al. [
40] found stronger aging effects in men, whereas other studies have shown that postmenopausal women experience steeper increases in SBP that eventually exceed those in men [
41].
The associations observed in this study reflect short- to medium-term physiological responses to ambient temperature variability during 2011–2018 rather than long-term climate change trends. The analysis captured routine fluctuations in monthly mean temperature and did not assess gradual warming trajectories or future climate scenarios. Accordingly, these findings should not be interpreted as direct evidence of climate change effects but rather as evidence that temperature variability may influence BP regulation in a tropical urban setting [
34,
35].
Nevertheless, these findings remain relevant to public health. Increasing climate variability and the growing frequency of extreme weather events are expected to intensify population exposure to fluctuating thermal conditions, especially in rapidly urbanizing tropical regions [
34]. Understanding short- to medium-term physiological responses to temperature may therefore help generate hypotheses regarding population vulnerability and adaptive capacity under changing climatic conditions, even in the absence of extreme heat or cold exposure [
35].
From a public health perspective, the modest magnitude of the observed BP changes suggests that these results should be interpreted as hypothesis-generating rather than immediately actionable. Future studies using updated cohorts, higher-resolution climate data, and information on indoor environments and adaptive behaviors will be needed to better understand how increasing climate variability may interact with hypertension risk in tropical populations [
34,
35].
Strengths and Limitations
This 6-year cohort study (2011–2018) from Bogor City provides one of the few longitudinal analyses of ambient temperature and BP in a tropical LMIC setting. By leveraging repeated BP measurements and population-based cohort data, the study contributes valuable historical evidence on time-varying climate–health relationships in an understudied context.
However, several limitations should be acknowledged. First, the dataset did not include several individual-level factors, such as body mass index, smoking, alcohol consumption, physical activity, dietary patterns, comorbidities (e.g., diabetes or kidney disease), or antihypertensive medication use, which may have contributed to residual confounding. These factors are well-established determinants of BP and may also modify physiological responses to ambient temperature; failure to account for them may therefore have attenuated or distorted the estimated associations. In addition, SES and education were treated as baseline characteristics; therefore, potential changes in income or social position during follow-up could not be evaluated, which may have introduced additional residual confounding. Second, the primary analyses were restricted to participants with complete BP measurements at all follow-up visits. If participants with incomplete follow-up differed systematically from those retained in the analytical sample, this complete-case approach may have introduced selection bias. Third, because the cohort was drawn from an urban population in Bogor City, the findings may have limited generalizability to rural populations or to other regions with different climatic, socioeconomic, or health-system contexts.
Fourth, temperature exposure was derived from monthly averages obtained from the official BMKG Data Online system. For Bogor City, only 1 meteorological station, the West Java Climatology Station, provides complete long-term records. Although this reflects national meteorological infrastructure rather than a weakness of the study design, reliance on a single station may have resulted in nondifferential exposure misclassification because it could not capture microclimatic variation, occupational exposure, or indoor–outdoor temperature differences. Such nondifferential misclassification would most likely bias the estimated associations toward the null, potentially leading to underestimation of the true temperature–BP association. Fifth, humidity and other indicators of seasonal variation were unavailable, limiting the ability to disentangle temperature-specific effects from broader seasonal influences. Although BP measurement procedures were standardized, information on short-term behavioral factors such as caffeine intake or smoking immediately before measurement was unavailable, which may have introduced nondifferential measurement variability across study visits.
Sixth, time in the GEE models was specified as indicator variables for each follow-up visit, resulting in multiple temperature–time interaction terms and increased model complexity. Although this approach allowed a flexible representation of temporal heterogeneity, it also introduced a potential risk of overfitting. To mitigate this concern, the interpretation emphasized global Wald tests and marginal effects summaries rather than isolated visit-specific coefficients, and the overall interaction pattern remained consistent across sensitivity analyses. Seventh, the observed ambient temperature range (24.6 °C–28.9 °C) was relatively narrow, and the study was not designed to evaluate extreme heat or cold thresholds. Because daily temperature data and short-term lag structures were not examined, potential non-linear or acute BP responses to short-term temperature extremes could not be assessed. The findings should therefore be interpreted as reflecting medium-term exposure patterns rather than immediate physiological effects within the context of gradual temperature variability typical of tropical urban environments. Eighth, stratified subgroup analyses (e.g., by sex or age) were not conducted because the study was not powered for such comparisons.
Finally, the study analyzed data collected between 2011 and 2018 and should therefore be interpreted as historical evidence. Subsequent changes in climate patterns, lifestyle behaviors, and hypertension management in Indonesia, including those during the coronavirus disease 2019 period, may influence the current magnitude or direction of temperature–BP associations.
Despite these limitations, the study strengthens understanding of temporal variability in BP in a tropical urban cohort and provides hypothesis-generating evidence to inform future multi-site longitudinal investigations incorporating richer individual-level covariates and higher-resolution environmental data. Overall, this study offers historical cohort evidence indicating that, although the main temperature effect on BP was not statistically significant, temperature–time interactions revealed modest, time-varying associations within clinically plausible ranges.
Conclusion
This 6-year cohort study (2011–2018) in Bogor City provides one of the few longitudinal analyses examining the relationship between ambient temperature and BP in a tropical LMIC. Although the main temperature effects on SBP and DBP were not statistically significant, several temperature–time interactions were observed, indicating that the associations varied across follow-up visits. Lower ambient temperatures were generally associated with higher BP, and these fluctuations, although modest (approximately −4 to +3 mmHg for SBP), fell within clinically plausible ranges.
SES and educational attainment were associated with SBP, whereas age and sex were associated with DBP. These findings highlight that both environmental and sociodemographic factors shape BP variability in tropical urban populations. Although temperature exposure itself is not modifiable, its interaction with population vulnerability and adaptation capacity should be considered when developing strategies for hypertension prevention and control.
Future studies should incorporate individual-level information on cooling access and indoor thermal environments to better characterize vulnerability to temperature-related BP changes.
This study contributes valuable historical evidence from Indonesia’s tropical setting and underscores the need for updated multi-site studies with richer individual- and environmental-level data. Such research will be critical to understanding how increasing climate variability and urban heat exposure interact with chronic disease risk, particularly in resource-limited settings.
HIGHLIGHTS
• Ambient temperature showed no overall effect on blood pressure but varied across time.
• Lower temperatures were associated with modest increases in systolic blood pressure.
• Effect sizes were small but biologically plausible in tropical populations.
• Associations were time-dependent rather than uniform across follow-up visits.
• Results emphasize temporal variability in climate–health relationships.
Article information
Ethics Approval
This study was approved by the Ethics Committee of the National Institute of Health Research and Development, Ministry of Health, Indonesia (approval numbers KE.01.05/EC/394/2012 for baseline and LB.02.01/2/KE.076/2018 for follow-up assessments).
Conflicts of Interest
The authors have no conflicts of interest to declare.
Availability of Data
The datasets are not publicly available but are available from the corresponding author upon reasonable request.
Authors’ Contributions
Conceptualization: PSH, WR; Data curation: PSH, WR, ID; Formal analysis: PSH, WR, ID, PWD; Investigation: NEWS, NS; Methodology: WR, PWD, YT; Supervision: WR, PWD, YT; Validation: PWD, YT; Visualization: NEWS, NS; Writing–original draft: PSH, WR, NEWS, NS; Writing–review & editing: all authors. All authors read and approved the final manuscript.
Acknowledgements
The authors express their sincere appreciation to the Research and Development Center for Public Health Efforts, Ministry of Health of the Republic of Indonesia, and to the Meteorology, Climatology, and Geophysics Agency (BMKG) for providing access to the datasets used in this study. The authors also extend their gratitude to the Non-Communicable Disease Risk Factor Cohort research team for their dedication and hard work in conducting the cohort study, which made it possible to use these valuable data for this analysis. Constructive comments from the anonymous reviewers are also gratefully acknowledged for improving the quality and clarity of this manuscript.
Supplementary Material
Table S2.
Estimated effects of ambient temperature and covariates on systolic blood pressure (SBP) from generalized estimating equation (GEE) models, Bogor City cohort, 2011–2018.
Table S3.
Estimated effects of ambient temperature and covariates on diastolic blood pressure (DBP) from generalized estimating equation (GEE) models, Bogor City cohort, 2011–2018.
Figure 1.Temporal changes in systolic and diastolic blood pressure (BP) across follow-up visits, Bogor City cohort, 2011–2018. Mean systolic BP (blue) and diastolic BP (red) across 18 follow-up visits. Each point represents the average value across all participants at each visit, illustrating between-person mean trends and temporal fluctuations in BP over the 6-year study period.
Figure 2.Time-varying temperature effects on systolic and diastolic blood pressure across follow-up visits, Bogor City cohort, 2011–2018. Panel A shows the estimated effect of ambient temperature on systolic blood pressure (SBP), and panel B shows the corresponding effect on diastolic blood pressure (DBP). Each point represents the temperature coefficient (°C) from the generalized estimating equation model for each follow-up visit, with positive values indicating higher blood pressure at warmer temperatures. The dashed horizontal line at 0 represents the null effect and illustrates how the direction and magnitude of temperature–blood pressure associations varied across the 19 observation points.
Figure 3.Predicted systolic blood pressure (SBP) from the generalized estimating equation models across 19 follow-up visits (Bogor City, Indonesia, 2011–2018) at 3 representative ambient temperature levels (24.6 °C, 26.8 °C, and 28.9 °C), with 95% confidence intervals. Each line represents the estimated mean SBP at low, median, and high ambient temperature levels derived from the fitted model. Shaded ribbons indicate the 95% confidence intervals. The x-axis corresponds to the sequence of follow-up visits (baseline plus 18 subsequent visits), and the y-axis shows model-predicted SBP (mmHg).
Figure 4.Predicted diastolic blood pressure (DBP) from the generalized estimating equation models across 19 follow-up visits (Bogor City, Indonesia, 2011–2018) at 3 representative ambient temperature levels (24.6 °C, 26.8 °C, and 28.9 °C), with 95% confidence intervals. Each line represents the estimated mean DBP at low, median, and high ambient temperature levels derived from the fitted model. Shaded ribbons indicate the 95% confidence intervals. The x-axis corresponds to the sequence of follow-up visits (baseline plus 18 subsequent visits), and the y-axis shows model-predicted DBP (mmHg).
Table 1.Descriptive characteristics of participants in the Non-Communicable Disease Risk Factor Cohort, Bogor City, 2011–2018 (n=1,648)
Table 1.
| Characteristic |
Frequency (n, %) |
| Age group (y) |
|
| 30–44 |
603 (36.6) |
| 45–59 |
764 (46.4) |
| ≥60 |
281 (17.1) |
| Sex |
|
| Male |
376 (22.8) |
| Female |
1,272 (77.2) |
| Education level |
|
| <High school |
970 (58.9) |
| ≥High school |
678 (41.1) |
| Economic status |
|
| Low (<UMK) |
1,501 (91.1) |
| High (≥UMK) |
147 (8.9) |
References
- 1. World Health Organization (WHO) (CH). Noncommunicable diseases [Internet]. WHO; 2022 [cited 2023 Sep 3]. Available from: https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases.
- 2. Nugent R, Fottrell E. Non-communicable diseases and climate change: linked global emergencies. Lancet 2019;394:622−3.
- 3. 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;396:1223−49.
- 4. World Health Organization (WHO), United Nations Framework Convention on Climate Change (CH). Climate and health country profile: 2015 Indonesia demographic estimates. WHO; 2016.
- 5. Li Y, Ma Z, Zheng C, et al. Ambient temperature enhanced acute cardiovascular-respiratory mortality effects of PM2.5 in Beijing, China. Int J Biometeorol 2015;59:1761−70.
- 6. Phung D, Thai PK, Guo Y, et al. Ambient temperature and risk of cardiovascular hospitalization: an updated systematic review and meta-analysis. Sci Total Environ 2016;550:1084−102.
- 7. Moghadamnia MT, Ardalan A, Mesdaghinia A, et al. Ambient temperature and cardiovascular mortality: a systematic review and meta-analysis. PeerJ 2017;5:e3574.
- 8. Cui L, Geng X, Ding T, et al. Impact of ambient temperature on hospital admissions for cardiovascular disease in Hefei City, China. Int J Biometeorol 2019;63:723−34.
- 9. Lu P, Zhao Q, Xia G, et al. Temporal trends of the association between ambient temperature and cardiovascular mortality: a 17-year case-crossover study. Environ Res Lett 2021;16:045004.
- 10. Alahmad B, Khraishah H, Roye D, et al. Associations between extreme temperatures and cardiovascular cause-specific mortality: results from 27 countries. Circulation 2023;147:35−46.
- 11. Jin J, Meng X, Wang D, et al. Association between ambient temperature and cardiovascular diseases related hospital admissions in Lanzhou, China. Heliyon 2023;9:e12997.
- 12. Chakrabarty M, Bora B. The measurement of blood pressure in winter season and its co-relation with blood pressure after cold exposure in summer season: a cross-sectional study in Gauhati medical college and hospital, Guwahati, Assam, India. Int J Res Med Sci 2017;5:3111−3.
- 13. Umishio W, Ikaga T, Kario K, et al. Cross-sectional analysis of the relationship between home blood pressure and indoor temperature in winter: a nationwide smart wellness housing survey in Japan. Hypertension 2019;74:756−66.
- 14. Li M, Meng R, Luo J, et al. The association of temperature variability with blood pressure in Southern China: a series of cross-sectional studies in Guangdong Province, China. Res Sq [Preprint] 2021 Sep 7 https://doi.org/10.21203/rs.3.rs-847370/v1.
- 15. Sun Y, Zhang M, Chen S, et al. Potential impact of ambient temperature on maternal blood pressure and hypertensive disorders of pregnancy: a nationwide multicenter study based on the China birth cohort. Environ Res 2023;227:115733.
- 16. Aubiniere-Robb L, Jeemon P, Hastie CE, et al. Blood pressure response to patterns of weather fluctuations and effect on mortality. Hypertension 2013;62:190−6.
- 17. Park S, Kario K, Chia YC, et al. The influence of the ambient temperature on blood pressure and how it will affect the epidemiology of hypertension in Asia. J Clin Hypertens (Greenwich) 2020;22:438−44.
- 18. Cheng BJ, Li H, Meng K, et al. Short-term effects of heatwaves on clinical and subclinical cardiovascular indicators in Chinese adults: a distributed lag analysis. Environ Int 2024;183:108358.
- 19. National Institute of Health Research and Development, Ministry of Health (ID). National report of Basic Health Research (Riskesdas) 2018. Ministry of Health; 2018.
- 20. Gibran MH, Heriyani F, Djallalluddin D. Hubungan suhu rumah dengan kejadian hipertensi. Homeostasis 2020;3:441−6. Indonesian.
- 21. Fritz M. Temperature and non-communicable diseases: evidence from Indonesia’s primary health care system. Health Econ 2022;31:2445−64.
- 22. Ministry of Health, Republic of Indonesia (ID). Guidelines for identifying health risk factors related to climate change (Peraturan Menteri Kesehatan Republik Indonesia No. 035 Tahun 2012). Ministry of Health; 2012. Indonesian.
- 23. Wang C, Yuan Y, Zheng M, et al. Association of age of onset of hypertension with cardiovascular diseases and mortality. J Am Coll Cardiol 2020;75:2921−30.
- 24. Bianco G, Espinoza-Chavez RM, Ashigbie PG, et al. Projected impact of climate change on human health in low- and middle-income countries: a systematic review. BMJ Glob Health 2024;8(Suppl 3). e015550.
- 25. James PA, Oparil S, Carter BL, et al. 2014 evidence-based guideline for the management of high blood pressure in adults: report from the panel members appointed to the Eighth Joint National Committee (JNC 8). JAMA 2014;311:507−20.
- 26. Riyadina W. Track record of cohort study of risk factors of non-communicable diseases in Bogor City. UI Publishing; 2020. Indonesian.
- 27. Brook RD, Weder AB, Rajagopalan S. “Environmental hypertensionology” the effects of environmental factors on blood pressure in clinical practice and research. J Clin Hypertens (Greenwich) 2011;13:836−42.
- 28. Cui J, Muller MD, Blaha C, et al. Seasonal variation in muscle sympathetic nerve activity. Physiol Rep 2015;3:e12492.
- 29. Woeckel M, Schneider A, Cyrys J, et al. Ambient air temperature and temperature variability affecting blood pressure: a repeated-measures study in Augsburg, Germany. Environ Res Health 2023;1:035001.
- 30. Widlansky ME, Vita JA, Keyes MJ, et al. Relation of season and temperature to endothelium-dependent flow-mediated vasodilation in subjects without clinical evidence of cardiovascular disease (from the Framingham Heart Study). Am J Cardiol 2007;100:518−23.
- 31. Gasparrini A, Guo Y, Hashizume M, et al. Mortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet 2015;386:369−75.
- 32. Zheng S, Wang MZ, Cheng ZY, et al. Effects of outdoor temperature on blood pressure in a prospective cohort of Northwest China. Biomed Environ Sci 2021;34:89−100.
- 33. Fossion R, Rivera AL, Estañol B. A physicist’s view of homeostasis: how time series of continuous monitoring reflect the function of physiological variables in regulatory mechanisms. Physiol Meas 2018;39:084007.
- 34. Yang L, Li L, Lewington S, et al. Outdoor temperature, blood pressure, and cardiovascular disease mortality among 23 000 individuals with diagnosed cardiovascular diseases from China. Eur Heart J 2015;36:1178−85.
- 35. Gronlund CJ, Sheppard L, Adar SD, et al. Vulnerability to the cardiovascular effects of ambient heat in six us cities: results from the multi-ethnic study of atherosclerosis (MESA). Epidemiology 2018;29:756−64.
- 36. Bhattarai S, Tandstad B, Shrestha A, et al. Socioeconomic status and its relation to hypertension in rural Nepal. Int J Hypertens 2021;2021:5542438.
- 37. Allen L, Williams J, Townsend N, et al. Socioeconomic status and non-communicable disease behavioural risk factors in low-income and lower-middle-income countries: a systematic review. Lancet Glob Health 2017;5:e277−89.
- 38. Dhungana RR, Pandey AR, Bista B, et al. Prevalence and associated factors of hypertension: a community-based cross-sectional study in municipalities of Kathmandu, Nepal. Int J Hypertens 2016;2016:1656938.
- 39. Mueller NT, Noya-Alarcon O, Contreras M, et al. Association of age with blood pressure across the lifespan in isolated Yanomami and Yekwana Villages. JAMA Cardiol 2018;3:1247−9.
- 40. Khajavi A, Zayeri F, Ramezankhani A, et al. Age and aging effects on blood pressure: 15 years follow-up of Tehran lipid and glucose study. J Clin Hypertens (Greenwich) 2021;23:1205−11.
- 41. The China PEACE Collaborative Group. Association of age and blood pressure among 3.3 million adults: insights from China PEACE million persons project. J Hypertens 2021;39:1143−54.
