Ambient temperature and cause-specific mortality in Klang Valley, Malaysia: an ecological time-series study of PM₁₀ effect modification, 2008–2023

Introduction

Climate change and rapid urbanisation have heightened concern about environmental determinants of health, particularly ambient temperature and air pollution (1-3). Ambient temperature commonly shows a nonlinear U- or J-shaped association with mortality, reflecting physiological stress at thermal extremes relative to a locally defined minimum mortality temperature (MMT) (1,2,4-7). Particulate matter with an aerodynamic diameter of 10 µm or less (PM₁₀) has also been linked to adverse health outcomes through systemic inflammation, oxidative stress, and impaired cardiopulmonary function, which may increase vulnerability to temperature-related events (8-11). Beyond their independent associations with mortality, evidence suggests that ambient temperature and air pollution may interact, with pollution potentially modifying temperature-related mortality risk (12).

Multi-city studies have reported that heat-related mortality is often higher on days with elevated air pollution concentrations (12-14). Several biological mechanisms may explain this pattern. Higher pollutant concentrations can impair pulmonary function and trigger systemic inflammatory responses, thereby reducing the physiological reserve required to tolerate heat-induced vasodilation and increased cardiac workload (15). Conversely, high ambient temperature may alter the chemical composition or toxicity of particulate matter and increase susceptibility among at-risk populations (16,17). However, the magnitude and direction of this effect modification remain inconsistent across settings, potentially reflecting differences in climate, urban morphology, pollution sources, population adaptation, and baseline vulnerability.

This uncertainty is particularly relevant in Southeast Asia, where persistently high temperatures, rapid urban expansion, and recurrent transboundary haze create substantial potential for joint exposure to heat and air pollution (18,19). However, evidence on temperature-pollution interactions remains dominated by studies from temperate cities, leaving tropical urban centres underrepresented. Regional studies have generally treated air pollution as an independent exposure or confounder rather than as a potential modifier of temperature-related mortality (20-22). Consequently, whether particulate pollution modifies temperature-related mortality in tropical Southeast Asian populations remains insufficiently characterised.

Malaysia provides an opportunity to examine this question over an extended period using PM₁₀, which has greater historical monitoring continuity than PM₂.₅. Although the national air-quality monitoring network has operated since 1995, PM₂.₅ was introduced as a routine monitoring parameter only in 2017 and incorporated into the Malaysian Air Pollutant Index in 2018 (23). Consequently, PM₁₀ has provided the most temporally consistent particulate metric for Malaysian time-series studies extending to earlier years; its use in long-term research reflects monitoring continuity rather than evidence that PM₁₀ is more toxicologically specific or important than PM₂.₅.

Klang Valley is a particularly relevant setting because of its rapid urbanisation, industrialisation and infrastructure expansion. The scale of road traffic is substantial: as of 31 December 2017, 8.3 million motor vehicles were registered across Selangor, Kuala Lumpur and Putrajaya, of which 92% were private vehicles (24). Road traffic, together with industrial and power-generation activities, construction, port operations, biomass burning, resuspended road and soil dust, and episodic transboundary haze, contributes to the region’s complex particulate pollution profile (24-29). The metropolitan core lies between the Titiwangsa Mountains to the north and east and the Strait of Malacca to the west. This geographical setting, together with frequent calm winds and land-sea-breeze circulation, may restrict atmospheric ventilation and favour pollutant accumulation (24,28). Combined with year-round heat and humidity, these characteristics make Klang Valley an informative setting for examining temperature-pollution relationships.

Previous Malaysian time-series studies have established nonlinear associations between ambient temperature and mortality but have not directly evaluated PM₁₀ as a modifier of the temperature-mortality relationship. In Klang Valley, Yatim et al. examined mortality during 2006–2015 and evaluated PM₁₀ and O₃ primarily as potential confounders (21). Sapari et al. subsequently assessed temperature-related mortality in Kedah and Penang during 2011–2019 but excluded PM₂.₅ because more than 70% of its observations were missing (22). Thus, although these studies support the use of PM₁₀ in Malaysian temperature-mortality research and illustrate the limitations of PM₂.₅ monitoring for historical time-series analyses, neither directly tested whether PM₁₀ modified the nonlinear temperature-lag-mortality association. To address this gap, the present study examined the modifying role of PM₁₀ in the temperature-lag-mortality relationship in Klang Valley from 2008 to 2023 across non-accidental, cardiovascular, and respiratory mortality, using interaction models that allowed the temperature-lag-response relationship to vary by PM₁₀ level, a policy-relevant World Health Organization (WHO) PM₁₀ threshold, and alternative PM₁₀ specifications. We present this article in accordance with the STROBE reporting checklist (available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0039/rc).

Methods

Study design and settings

We conducted an ecological time-series study in Klang Valley, Malaysia, from 1 January 2008 to 31 December 2023. For this study, the Klang Valley study region was operationally defined as the entire state of Selangor (all nine administrative districts), together with the Federal Territories of Kuala Lumpur and Putrajaya (Figure 1). The study period represented the longest interval for which concurrent daily mortality, meteorological, and air pollution data with sufficiently consistent geographic coverage were available. The year 2008 was the earliest common year across the required data sources, whereas 2023 was the latest complete year available when the study period was established and the data were acquired (30-32).

Figure 1 Study area and locations of 12 Malaysian Meteorological Department monitoring stations and eight Department of Environment continuous air-quality monitoring stations across Selangor, Kuala Lumpur, and Putrajaya. The station symbols identify the two monitoring networks used to construct regional daily exposure averages; station coverage was not spatially uniform.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was reviewed and approved by the Medical Research and Ethics Committee (MREC), Ministry of Health Malaysia, with the identification number of ethical approval NMRR ID-24-01742-T9W. Individual consent for this retrospective analysis was waived.

Data sources and eligibility criteria

Daily individual-level mortality records were obtained from the Department of Statistics Malaysia (DOSM). Daily mortality counts comprised medically certified deaths among persons of all ages whose usual residential address was within the study region; deaths were assigned geographically according to residential address rather than place of death. Outcomes included non-accidental mortality [International Classification of Diseases, 10th Revision (ICD-10) codes A00-R99] and the cause-specific subsets of cardiovascular mortality (I00-I99) and respiratory mortality (J00-J99). Records were excluded if they were not medically certified, occurred outside the study period, or did not fall within a prespecified ICD-10 outcome definition.

Daily mean temperature and relative humidity were obtained from 12 Malaysian Meteorological Department (MetMalaysia) monitoring stations: 10 stations distributed across all nine districts of Selangor, one in Kuala Lumpur, and one in Putrajaya. Hourly concentrations of PM₁₀, ozone (O₃), and sulfur dioxide (SO₂) were obtained from eight fixed-site continuous air quality monitoring stations operated by the Department of Environment (DOE). The locations and spatial distribution of both monitoring networks are shown in Figure 1. Station-level observations were aggregated into daily regional averages to represent population-level exposure across the study region. For O₃, the daily maximum 8-hour rolling average was calculated for each station before averaging across the monitoring network, as this metric better represents acute health risk (33,34).

Daily PM₁₀ was classified as ‘below WHO’ (<45 µg/m³) or ‘above WHO’ (≥45 µg/m³) according to the 2021 WHO 24-hour air quality guideline (35). The binary classification was selected to retain policy relevance, and the PM₁₀ distribution showed limited separation between the 25th percentile (30.2 µg/m³) and 75th percentile (49.8 µg/m³). In this context, percentile-based stratification would have produced categories with relatively limited separation across much of the PM₁₀ distribution and less direct policy interpretability. The WHO threshold was therefore selected to distinguish days below the guideline from days at or above it.

Missing environmental observations were imputed using multiple imputation by chained equations (MICE) to create five completed datasets. A quasi-Poisson distributed lag nonlinear model (DLNM) was fitted separately to each dataset, and cumulative log-relative risks and their variances were combined using Rubin’s rules (8,36-38). An 81-day air pollution data gap from 14 April to 3 July 2017, caused by a change in data concessionaire, was retained in the continuous calendar but excluded as outcome-days; consequently, each regression model used 5,763 eligible outcome-days.

Statistical analysis

A quasi-Poisson regression model with a DLNM was used to estimate the association between ambient temperature and mortality. To examine the modifying effect of PM₁₀, we incorporated an interaction between the temperature cross-basis and binary PM₁₀ status. The interaction term allowed the full temperature exposure-lag-response surface to vary by PM₁₀ category, enabling the model to capture potential shifts in both the non-linear exposure-response relationship and the temporal lag structure across different PM₁₀ levels. The model was specified as follows:

log(E(Yt))=αd+cb(Tempt,lag=21)×PM10(binary)+ns(PM10cont,df=3)+ns(Time,df=5/years)+ns(SO2,df=3)+ns(O3,df=3)+ns(RH,df=3)+ns(Rainfall,df=3)+DOW+Holiday+Covid[1]

In this model, E(Yt) denotes the expected mortality count on day t; αd is the outcome-specific intercept; cb is the cross-basis function for daily mean temperature, modelling both the nonlinear exposure-response and distributed lag effects over 0–21 days; PM₁₀ denotes the binary indicator based on the WHO 24-hour threshold; and ns(...) denotes natural cubic splines for covariates. Potential confounding by atmospheric and meteorological factors was controlled using natural cubic splines with 3 degrees of freedom (df) for daily mean SO₂, O₃, rainfall, relative humidity, and PM₁₀. Long-term and seasonal trends were controlled using a natural cubic spline of time with 5 df per year. The model also included indicators for day of week (DOW), public holidays (Holiday), and the coronavirus disease 2019 (COVID-19) pandemic period (18 March 2020 to 31 May 2023). The cumulative exposure-response functions were centred at the observed median daily mean temperature (27.4 ℃), providing a common data-supported reference across outcomes and PM₁₀ strata. In DLNM analyses, centring defines the reference value for relative-risk (RR) estimation and affects the interpretation, but not the fitted exposure-lag-response association (39).

Prespecified summary RRs were estimated using distinct lag windows. Following established temperature-mortality contrasts, the higher-temperature (heat) summary estimate compared the 99th with the 90th percentile of daily mean temperature and was accumulated over lags 0–3, while the lower-temperature (cold) summary estimate compared the 1st with the 10th percentile and was accumulated over lags 0–28 (21,22,40). These percentile contrasts were used for the summary estimates reported in Table 1 and the related sensitivity analyses. The overall and PM₁₀-stratified cumulative exposure-response curves depict associations across the observed temperature distribution and were centred at 27.4 ℃. Interaction estimates were reported for the Below WHO and Above WHO PM₁₀ categories. All hypothesis tests were two-sided, and P<0.05 was considered statistically significant.

Sensitivity analysis

Sensitivity analyses assessed the robustness of the findings by varying the degrees of freedom for the time trend, temperature spline, and lag spline from 2 to 5 for non-accidental, cardiovascular, and respiratory mortality models, respectively (Tables S1-S3). Additional models used daily minimum and maximum temperatures instead of mean temperature (Table S4). The cumulative lag-window sensitivity analysis evaluated heat over lags 0–1, 0–3, 0–5, and 0–7 and cold over lags 0–14, 0–21, and 0–28 under both the full and reduced covariate specifications (Table S5). Lag-response curves were displayed through lag 28, with lags 0–3 marked as the primary heat window (Figure S1), and temperature-lag surfaces were displayed over lags 0–21 (Figure S2). Additional PM₁₀ interaction sensitivity analyses used quartiles, extreme pollution days, and continuous linear and nonlinear modification (Table S6); stability across pre-pandemic and COVID-19 periods was also evaluated (Figure S3). Covariate selection was guided by quasi-Akaike information criterion (qAIC) comparisons (Table S7), while established meteorological and pollution confounders were retained in the full primary model on epidemiological grounds. All analyses were conducted in R version 4.4.2 using the dlnm and mice packages.

Results

Descriptive statistics

A total of 293,432 non-accidental deaths were recorded in the Klang Valley during the 5,844 calendar days from 2008 to 2023. Cardiovascular and respiratory deaths accounted for 31.3% and 19.1% of total mortality, respectively. The region had a consistently warm thermal profile, with a mean daily temperature of 27.41 °C and a standard deviation (SD) of 0.85 °C, and a narrow range between the 1st (25.3 °C) and 99th (29.3 °C) percentiles. Mean relative humidity was 79.3%. The mean daily PM₁₀ concentration was 43.0 µg/m³, with a maximum of 382.5 µg/m³ recorded during a severe haze episode. The PM₁₀ concentration exceeded the 2021 WHO 24-hour guideline of 45 µg/m³ on 1,952 days, representing 33.4% of the study period. On these days, the mean PM₁₀ concentration was 63.5 µg/m³. The mean daily maximum 8-hour O₃ concentration was 67.0 µg/m³. Table 2 presents the summary statistics for mortality, meteorological variables, and air pollutants.

Temporal trends and covariate correlations

The time-series plot (Figure 2) showed consistent seasonal fluctuations in mortality, with a pronounced increase in non-accidental mortality during the COVID-19 pandemic period. PM₁₀ concentrations showed a relatively low baseline with intermittent high-magnitude peaks, particularly during transboundary haze episodes in 2013, 2015, and 2019. The correlation matrix (Figure 3) showed a strong negative correlation between mean temperature and relative humidity (r=−0.76). O₃ was moderately correlated with temperature and PM₁₀, suggesting that these exposures may share some meteorological drivers while still representing distinct environmental conditions.

Figure 2 Time series of daily cause-specific mortality, mean temperature, and PM₁₀ concentrations in Klang Valley, 2008–2023. PM₁₀, particulate matter with an aerodynamic diameter of 10 µm or less.

Figure 3 Spearman correlation coefficients between air pollutants and meteorological parameters in Klang Valley, 2008 to 2023. PM₁₀, particulate matter with an aerodynamic diameter of 10 µm or less.

Exposure-response and lag-response relationships

The primary cumulative exposure-response curves, estimated over lags 0–21 and centred at the median temperature of 27.4 °C, showed distinct nonlinear patterns across the three mortality outcomes (Figure 4). Cardiovascular mortality exhibited a reverse J-shaped association, with risk increasing steeply below the reference temperature and remaining comparatively flat above it. In contrast, non-accidental and respiratory mortality exhibited U-shaped associations, with the increase at higher temperatures being most pronounced for respiratory mortality.

Figure 4 Cumulative relative risks for cardiovascular, non-accidental, and respiratory mortality across the daily mean temperature distribution, centred at the median temperature of 27.4 ℃. The dashed vertical line indicates the reference temperature; shaded bands show 95% confidence intervals. RR, relative risk.

Table 1 presents the cumulative relative risk (RR) contrasts for the prespecified higher- and lower-temperature windows. All six overall estimates were greater than 1.00 and statistically significant (all P≤0.006). The largest overall estimate was observed for respiratory mortality at lower temperatures [RR 1.253; 95% confidence interval (CI): 1.111–1.412; P<0.001]. The lag-response functions indicated that higher-temperature associations were concentrated at or shortly after exposure, whereas lower-temperature associations developed after the initial lags and extended over a longer period (Figure S1).

The estimates were robust across the alternative cumulative lag windows and the full and reduced covariate specifications (Table S5). All higher- and lower-temperature estimates remained greater than 1.00 and statistically significant, although their magnitudes increased as the cumulative windows were extended. The increase across longer higher-temperature windows was most evident for respiratory mortality, whereas cardiovascular higher-temperature estimates were smaller and changed comparatively little. Lower-temperature estimates increased progressively from lags 0–14 to lags 0–28 for all three outcomes.

Effect modification by PM₁₀ concentration

The cumulative curves for the below WHO and above WHO PM₁₀ categories broadly overlapped across non-accidental, cardiovascular, and respiratory mortality (Figure 5). The lag-specific summary estimates in Table 1 were also similar between the two categories. The absence of consistent curve separation indicated limited evidence of PM₁₀ effect modification.

Figure 5 Cumulative relative risks for temperature and mortality stratified by daily PM₁₀ category, centred at the median temperature of 27.4 ℃. Lines show the Below WHO category, PM₁₀ <45 µg/m³, and the above WHO category, PM₁₀ ≥45 µg/m³; the dashed vertical line indicates the reference temperature, and shaded bands show 95% confidence intervals. PM₁₀, particulate matter with an aerodynamic diameter of 10 µm or less; RR, relative risk; WHO, World Health Organization.

All stratum-specific P values were <0.001 except for the cardiovascular higher-temperature summary estimate in the below WHO (P=0.004) and above WHO (P=0.34) categories and the respiratory higher-temperature estimate in the above WHO category (P=0.001). These P values test each stratum-specific RR against 1.00 rather than the difference between PM₁₀ categories. Confidence intervals overlapped between categories for every outcome and summary contrast, further supporting limited evidence of effect modification.

Alternative PM₁₀ interaction specifications did not provide consistent evidence of effect modification. Although the quartile-based interaction model yielded a statistically significant likelihood ratio test (P=0.04), the curves did not show a monotonic exposure-response gradient, and the finding was not reproduced by the extreme-day or spline-based analyses. The continuous-modifier result was close to, but did not meet, the conventional significance threshold (P=0.051). Taken together, these analyses supported limited and non-robust modification by PM₁₀.

Discussion

This ecological time-series study identified distinct temperature-mortality patterns across non-accidental, cardiovascular, and respiratory mortality in Klang Valley. The cumulative association was reverse J-shaped for cardiovascular mortality and U-shaped for non-accidental and respiratory mortality, with higher-temperature associations occurring predominantly at earlier lags and lower-temperature associations extending over a longer period. Cardiovascular mortality showed the clearest increase at lower temperatures, whereas respiratory mortality showed a pronounced early higher-temperature association while also exhibiting an elevated cumulative association at lower temperatures. The lower-temperature findings should nevertheless be interpreted with some caution because the non-accidental and respiratory estimates attenuated when greater flexibility was allowed in the temperature spline, whereas the cardiovascular lower-temperature association was more robust across model specifications. PM₁₀-stratified curves and summary estimates were broadly similar, and alternative interaction analyses did not show a consistent or monotonic pattern of modification. These findings extend previous Malaysian evidence by evaluating PM₁₀ as a modifier of the temperature-lag-mortality relationship rather than solely as a potential confounder, while also examining outcome-specific lag patterns over a longer study period.

The temperature-mortality patterns were broadly consistent with existing literature but refine the evidence for Malaysia. In Klang Valley during 2006–2015, Yatim et al. reported nonlinear associations, relatively immediate heat effects, delayed cold effects, and changes of less than 4% in temperature estimates after adjustment for PM₁₀ and O₃ (21). The present study supports the same general temporal pattern while providing clearer evidence of cumulative cold associations with cardiovascular and respiratory mortality over lags 0–28. Direct numerical comparison is inappropriate, however, because the studies used different temperature reference points, cumulative lag windows, and model specifications. Recent evidence from Kedah and Penang similarly showed immediate heat-related associations for non-accidental and respiratory mortality, with weaker cardiovascular heat associations (22). A time-series study in Manila likewise reported increased respiratory mortality above its MMT, supporting heat sensitivity in another tropical Southeast Asian setting, although its temperature contrast and study period differed from those used in the present study (41). The finding that respiratory mortality had the largest heat estimate is also consistent with a multicountry analysis in which heat-related risks were highest for respiratory mortality, whereas cold-related risks were highest for cardiovascular mortality (42). Furthermore, a recent study across six tropical climate regions in Thailand found stronger and more prolonged cold-wave than heat-wave mortality effects, particularly for cardiovascular mortality, although its event-based temperature definitions differed from the continuous cumulative association evaluated here (43). These findings reinforce that cooler temperatures can remain epidemiologically important in tropical populations. Immediate heat associations may reflect rapid thermoregulatory strain, dehydration, and respiratory stress, whereas delayed cold associations may involve vasoconstriction, sympathetic activation, increased blood viscosity, and thrombotic processes (4,44,45); these mechanisms were not directly evaluated in the present study.

The PM₁₀ findings require cautious interpretation in relation to previous studies. Yatim et al. evaluated PM₁₀ and O₃ as potential confounders, so the small changes after pollutant adjustment do not constitute a formal test of effect modification (21). A systematic review and meta-analysis reported stronger heat-related all-cause and non-accidental mortality at higher PM₁₀ and O₃ levels, but modification was not consistently demonstrated for cause-specific mortality, heatwaves, or cold effects, and heterogeneity across studies was substantial (3). Subsequent multicity analyses have provided suggestive evidence that particulate pollution can intensify heat-related mortality, including studies across 620 cities globally and 482 cities examining cardiorespiratory mortality (13,46). A recent study in 10 Indian cities also reported synergistic heat-PM₂.₅ associations, although it examined a finer particle fraction and used different exposure and lag specifications (47). In contrast, the present study showed broadly overlapping temperature-mortality curves below and above the WHO PM₁₀ guideline and no consistent evidence across alternative PM₁₀ specifications. Although the quartile-based interaction test reached statistical significance and the continuous linear interaction was close to the conventional significance threshold, neither finding was supported by a consistent monotonic pattern across the sensitivity analyses. The appropriate interpretation is therefore that evidence of PM₁₀ modification was limited and non-robust, rather than that temperature and PM₁₀ act independently. Detecting interaction may also have been limited by information loss from dichotomising PM₁₀, the exposure range observed in Klang Valley, and exposure measurement error arising from regional averages derived from fixed monitoring stations.

Several strengths and limitations should be considered when interpreting these findings. Strengths include the 16-year daily time-series, the use of DLNMs to represent nonlinear and delayed associations, prespecified higher- and lower-temperature contrasts with different cumulative lag windows, multiple imputation of missing environmental observations with pooled inference, and sensitivity analyses across alternative model specifications. However, regional averages from fixed monitoring stations could not capture within-region spatial variation, indoor or occupational exposures, individual time-activity patterns, or air-conditioning use, and such exposure measurement error may be particularly important when assessing interaction. Age-, sex-, and socioeconomic-stratified analyses were not conducted, limiting assessment of susceptible population groups and individual-level effect heterogeneity. PM₂.₅ was not consistently available throughout the study period, and the absence of NO₂ from the primary adjustment set leaves the possibility of residual co-pollutant confounding. In addition, the binary WHO PM₁₀ threshold may have obscured modification occurring only at higher concentrations or along a nonlinear concentration gradient, while the sensitivity of some lower-temperature estimates to spline flexibility indicates uncertainty at the sparsely observed lower end of the temperature distribution. These findings nevertheless suggest that temperature-health preparedness in tropical urban settings should consider risks at both lower and higher temperatures and should not depend solely on PM₁₀ alerts. Future studies incorporating spatially resolved PM₂.₅ and particle-composition data, population vulnerability indicators, and flexible joint exposure-response models could better characterise temperature-air pollution interactions in tropical populations.

Conclusions

Ambient temperature was nonlinearly associated with non-accidental, cardiovascular, and respiratory mortality in Klang Valley, with outcome-specific exposure-response shapes and lag patterns. Cardiovascular mortality showed a stronger and more prolonged association with lower temperatures, whereas respiratory mortality showed a more immediate higher-temperature association while remaining elevated at lower temperatures. Similar PM₁₀-stratified associations and the lack of consistent findings across alternative interaction models provided limited and non-robust evidence that PM₁₀ modified these temperature-mortality associations. Temperature-health preparedness should therefore address risks associated with both lower and higher temperatures, while further studies using more spatially resolved particulate exposure and population vulnerability data are needed to clarify joint temperature-air pollution effects.

Acknowledgments

The authors would like to acknowledge the Director General of Health Malaysia and the Ministry of Health Malaysia for granting permission to conduct this study. The authors would like to thank the Department of Environment, Malaysia for providing air pollution data, the Malaysian Meteorological Department for providing meteorological data, and the Department of Statistics Malaysia (DOSM) for providing mortality data for this study.

Footnote

Provenance and Peer Review: This article was commissioned by the Guest Editor (Mellissa Withers) for the series “Towards Planetary Health Equity: A Global Call for Shared Solutions” published in Journal of Public Health and Emergency. The article has undergone external peer review.

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0039/rc

Data Sharing Statement: Available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0039/dss

Peer Review File: Available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0039/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0039/coif). The series “Towards Planetary Health Equity: A Global Call for Shared Solutions” was commissioned by the editorial office without any funding or sponsorship. The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was reviewed and approved by the Medical Research and Ethics Committee (MREC), Ministry of Health Malaysia, with the identification number of ethical approval NMRR ID-24-01742-T9W. Individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Cheng C, Liu Y, Han C, et al. Effects of extreme temperature events on deaths and its interaction with air pollution. Sci Total Environ 2024;915:170212. [Crossref] [PubMed]
2. Hansel NN, McCormack MC, Kim V. The Effects of Air Pollution and Temperature on COPD. COPD 2016;13:372-9. [Crossref] [PubMed]
3. Hu X, Han W, Wang Y, et al. Does air pollution modify temperature-related mortality? A systematic review and meta-analysis. Environ Res 2022;210:112898. [Crossref] [PubMed]
4. Alahmad B, Khraishah H, Royé D, et al. Associations Between Extreme Temperatures and Cardiovascular Cause-Specific Mortality: Results From 27 Countries. Circulation 2023;147:35-46. [Crossref] [PubMed]
5. Arisco NJ, Sewe MO, Bärnighausen T, et al. The effect of extreme temperature and precipitation on cause-specific deaths in rural Burkina Faso: a longitudinal study. Lancet Planet Health 2023;7:e478-89. [Crossref] [PubMed]
6. 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. [Crossref] [PubMed]
7. Jacobson LDSV, Oliveira BFA, Schneider R, et al. Mortality Risk from Respiratory Diseases Due to Non-Optimal Temperature among Brazilian Elderlies. Int J Environ Res Public Health 2021;18:5550. [Crossref] [PubMed]
8. Ibrahim MF, Hod R, Ahmad Tajudin MAB, et al. Children's exposure to air pollution in a natural gas industrial area and their risk of hospital admission for respiratory diseases. Environ Res 2022;210:112966. [Crossref] [PubMed]
9. Kermani M, Arfaeinia H, Masroor K, et al. Health impacts and burden of disease attributed to long-term exposure to atmospheric PM10/PM2.5 in Karaj, Iran: effect of meteorological factors. International Journal of Environmental Analytical Chemistry 2022;102:6134-50.
10. Rahim NNA, Yahya A, Mahiyuddin WRW. Association Between Particulate Matter (PM10) Exposure and Cardiorespiratory Hospital Admissions: A Time Series Analysis. American Journal of Environmental Sciences 2021;17:23-30.
11. Tapia V, Steenland K, Vu B, et al. PM2.5 exposure on daily cardio-respiratory mortality in Lima, Peru, from 2010 to 2016. Environ Health 2020;19:63. [Crossref] [PubMed]
12. Analitis A, De' Donato F, Scortichini M, et al. Synergistic Effects of Ambient Temperature and Air Pollution on Health in Europe: Results from the PHASE Project. Int J Environ Res Public Health 2018;15:1856. [Crossref] [PubMed]
13. Stafoggia M, Michelozzi P, Schneider A, et al. Joint effect of heat and air pollution on mortality in 620 cities of 36 countries. Environ Int 2023;181:108258. [Crossref] [PubMed]
14. Zafeiratou S, Samoli E, Analitis A, et al. Modification of heat-related effects on mortality by air pollution concentration, at small-area level, in the Attica prefecture, Greece. Environ Health 2024;23:10. [Crossref] [PubMed]
15. Burkart K, Meier F, Schneider A, et al. Modification of Heat-Related Mortality in an Elderly Urban Population by Vegetation (Urban Green) and Proximity to Water (Urban Blue): Evidence from Lisbon, Portugal. Environ Health Perspect 2016;124:927-34. [Crossref] [PubMed]
16. Areal AT, Zhao Q, Wigmann C, et al. The effect of air pollution when modified by temperature on respiratory health outcomes: A systematic review and meta-analysis. Sci Total Environ 2022;811:152336. [Crossref] [PubMed]
17. Li J, Woodward A, Hou XY, et al. Modification of the effects of air pollutants on mortality by temperature: A systematic review and meta-analysis. Sci Total Environ 2017;575:1556-70. [Crossref] [PubMed]
18. Rackow B, König HH, Wall M, et al. The interaction between air pollution, weather conditions, and health risks: a systematic review. Sci Total Environ 2025;996:180080. [Crossref] [PubMed]
19. Wu K, Ho HC, Su H, et al. A systematic review and meta-analysis of intraday effects of ambient air pollution and temperature on cardiorespiratory morbidities: First few hours of exposure matters to life. EBioMedicine 2022;86:104327. [Crossref] [PubMed]
20. Denpetkul T, Phosri A. Daily ambient temperature and mortality in Thailand: Estimated effects, attributable risks, and effect modifications by greenness. Sci Total Environ 2021;791:148373. [Crossref] [PubMed]
21. Yatim ANM, Latif MT, Sofwan NM, et al. The association between temperature and cause-specific mortality in the Klang Valley, Malaysia. Environ Sci Pollut Res Int 2021;28:60209-20. [Crossref] [PubMed]
22. Sapari H, Ismail R, Mahiyudin WRW, et al. Association of Daily Temperature on Non-Accidental and Specific-Cause Mortality in Northern Malaysia: A Time-Series Study. Climate 2026;14:139.
23. Tan Z, Latif MT, Ashfold MJ. Assessment of Malaysia-wide PM2.5 Forecasts from a Global Model. Aerosol Air Qual Res 2023;23:220444.
24. Fandi NFM, Jalaludin J, Latif MT, et al. BTEX Exposure Assessment and Inhalation Health Risks to Traffic Policemen in the Klang Valley Region, Malaysia. Aerosol Air Qual Res 2020;20:1922-37.
25. Cheong KH, Ngiam NJ, Morgan GG, et al. Acute Health Impacts of the Southeast Asian Transboundary Haze Problem-A Review. Int J Environ Res Public Health 2019;16:3286. [Crossref] [PubMed]
26. Ismain SHA, Salleh SA, Sham NM, et al. Spatial Distribution of Particulate Matter (PM2.5) in Klang Valley using Inverse Distance Weighting Interpolation Model. IOP Conf Ser: Earth Environ Sci 2023;1217:012033.
27. Latif MT, Dominick D, Hawari NSSL, et al. The concentration of major air pollutants during the movement control order due to the COVID-19 pandemic in the Klang Valley, Malaysia. Sustain Cities Soc 2021;66:102660. [Crossref] [PubMed]
28. Suhaimi NF, Jalaludin J, Abu Bakar S. The Influence of Traffic-Related Air Pollution (TRAP) in Primary Schools and Residential Proximity to Traffic Sources on Histone H3 Level in Selected Malaysian Children. Int J Environ Res Public Health 2021;18:7995. [Crossref] [PubMed]
29. Mohd Shafie SH, Mahmud M, Mohamad S, et al. Influence of urban air pollution on the population in the Klang Valley, Malaysia: a spatial approach. Ecological Processes 2022;11:3.
30. DOSM. Current Population Estimates, 2024. Department of Statistics Malaysia. 2024. Accessed June 3, 2026. Available online: https://www.dosm.gov.my/portal-main/release-content/current-population-estimates-2024.
31. MetMalaysia. Laman Web Rasmi Jabatan Meteorologi Malaysia. Heat Wave Status. 2024. Accessed January 17, 2024. Available online: https://www.met.gov.my/.
32. Environmental Data Center - Department of Environment. October 4, 2021. Accessed August 12, 2026. Available online: https://www.doe.gov.my/en/environmental-data-center/.
33. Lange SS. Comparing apples to oranges: Interpreting ozone concentrations from observational studies in the context of the United States ozone regulatory standard. Sci Total Environ 2018;644:1547-56. [Crossref] [PubMed]
34. Li X, Chen Q, Zheng X, et al. Effects of ambient ozone concentrations with different averaging times on asthma exacerbations: A meta-analysis. Sci Total Environ 2019;691:549-61. [Crossref] [PubMed]
35. World Health Organization W. WHO global air quality guidelines: particulate matter (‎PM2.5 and PM10)‎, ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. 2021. Accessed January 17, 2023. Available online: https://www.who.int/publications-detail-redirect/9789240034228.
36. Hua V, Nguyen T, Dao MS, et al. The impact of data imputation on air quality prediction problem. PLoS One 2024;19:e0306303. [Crossref] [PubMed]
37. Joel LO, Doorsamy W, Paul BS. A comparative study of imputation techniques for missing values in healthcare diagnostic datasets. Int J Data Sci Anal 2025;20:6357-73.
38. Kebalepile MM, Dzikiti LN, Voyi K. Using Diverse Data Sources to Impute Missing Air Quality Data Collected in a Resource-Limited Setting. Atmosphere 2024;15:3.
39. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med 2010;29:2224-34. [Crossref] [PubMed]
40. Guo Y, Gasparrini A, Armstrong BG, et al. Heat Wave and Mortality: A Multicountry, Multicommunity Study. Environ Health Perspect 2017;125:087006. [Crossref] [PubMed]
41. Seposo XT, Dang TN, Honda Y. Evaluating the Effects of Temperature on Mortality in Manila City (Philippines) from 2006-2010 Using a Distributed Lag Nonlinear Model. Int J Environ Res Public Health 2015;12:6842-57. [Crossref] [PubMed]
42. Scovronick N, Sera F, Vu B, et al. Temperature-mortality associations by age and cause: a multi-country multi-city study. Environ Epidemiol 2024;8:e336. [Crossref] [PubMed]
43. Khunthong P, Vittayapraparat P, Puangthongthub S. Regional impact of extreme temperatures on cardiovascular and respiratory mortality in Thailand: cold wave burden across diverse tropical climate zones. J Expo Sci Environ Epidemiol 2026;36:983-97. [Crossref] [PubMed]
44. Hundessa S, Huang W, Zhao Q, et al. Global and Regional Cardiovascular Mortality Attributable to Nonoptimal Temperatures Over Time. J Am Coll Cardiol 2024;83:2276-87. [Crossref] [PubMed]
45. Li Y, Wu J, Xu Y, et al. Cold exposure and the cardiovascular system: from physiological adaptation to pathological risk. Front Physiol 2025;16:1740919. [Crossref] [PubMed]
46. Rai M, Stafoggia M, de'Donato F, et al. Heat-related cardiorespiratory mortality: Effect modification by air pollution across 482 cities from 24 countries. Environ Int 2023;174:107825. [Crossref] [PubMed]
47. de Bont J, Rajiva A, Mandal S, et al. Synergistic associations of ambient air pollution and heat on daily mortality in India. Environ Int 2025;199:109426. [Crossref] [PubMed]