Association between PM2.5 components and electrocardiogram abnormalities among elderly individuals: a longitudinal study in Guangzhou, China Authors: Chaoling Zhang, Qin Zhou, Jiamin Chen, Zhengmin Min Qian, Maya Kavuri, Tom Burroughs, Zhonghua Ai, Shengtao Wei, Weiquan Lin, Zhoubin Zhang, Hui Liu, Hualiang Lin Published: 2026 ABSTRACT AbstractBackgroundFine particulate matter (PM2.5) is linked with cardiovascular diseases (CVD), however, its association with electrocardiogram (ECG) abnormality, a prognostic indicator of CVD, remains largely unknown, especially different chemical components of PM2.5. This study aimed to investigate the longitudinal associations of long-term exposure to PM2.5 and its chemical constituents with electrocardiogram abnormalities among elderly individuals in Guangzhou, China.MethodsThe study was based on a survey of elderly people in the communities of Guangzhou from 2018 to 2022. Bilinear interpolation and residential addresses were used to ascertain annual averages of PM2.5 and its components prior to ECG abnormalities. The generalized estimation equations (GEE) models were used to evaluate the association with ECG abnormalities.Results215,073 participants were included in this study. Each participant received at least one electrocardiogram examination, with a total of 691,990 recordings. PM2.5 and its five components (sulfate, nitrate, ammonium, organic matter, and black carbon) were significantly associated with elevated risks of ECG abnormalities. For PM2.5, odds ratio (OR) of ECG abnormalities per interquartile range (IQR) increase was 1.029 (95% confidence interval [CI]: 1.020–1.037), among its components was the highest for ammonium (OR:1.070, 95% CI: 1.062–1.078), followed by nitrate (OR:1.058, 95% CI: 1.050–1.065), sulfate (OR:1.047, 95% CI: 1.040–1.054), black carbon (OR:1.034, 95% CI: 1.027–1.042) and organic matter (OR:1.030, 95% CI: 1.022–1.037).ConclusionOur study provides evidence that PM2.5 and its components might lead to a greater risk of ECG abnormalities, underscoring the clinical value of ECG as an early indicator of potential reversible risks in the cardiovascular system. Strategies that target reducing PM2.5 pollution might improve the health outcomes of elderly people with ECG abnormalities.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12872-026-06015-0.Keywords: PM2.5, Components, Electrocardiogram abnormality, Longitudinal study Introduction Worldwide, cardiovascular diseases (CVD) maintain its position as the primary factor in the increase of disability-adjusted life years (DALYs) and deaths, comprising approximately 437 million DALYs and 19.2 million deaths in 2023. Population aging has been the main driving factor for the increase in the burden of cardiovascular diseases [8]. In addition, CVD imposes a staggering economic burden, estimated at $1.04 trillion in healthcare costs and productivity losses globally [18]. Early detection of subclinical cardiac dysfunction is critical for mitigating this burden, as irreversible myocardial damage often precedes the onset of symptoms. Electrocardiogram (ECG) abnormality, including arrhythmias, ST-segment deviations, and prolonged QT intervals, serve as pivotal biomarkers for identifying early electrophysiological disturbances. Dynamic ECG monitoring has further demonstrated utility in capturing cumulative environmental insults on cardiac electrophysiology, positioning ECG as a cornerstone for preclinical CVD risk stratification [14]. As a common type of environmental particulate matter, fine particulate matter (PM2.5) contains heterogeneous chemical composition, which comprises black carbon (BC), organic matter (OM), sulfate (), nitrate () and ammonium (). PM2.5 has been established as a modifiable environmental determinant of CVD incidence and progression [2, 45]. Exposure to environmental particulate matter was one of the most significant factors contributing to the global disease burden in 2021 [4]. Epidemiological studies have robustly linked long-term PM2.5 exposure to increased risks of ischemic heart disease, myocardial infarction, stroke and congestive heart failure [33, 36, 41]. Mechanistically, PM2.5 induces oxidative stress, systemic inflammation, and endothelial dysfunction, thereby accelerating atherosclerosis, thrombosis, and arrhythmogenesis [21]. The chemical composition of PM2.5 induces differential biological impacts that vary in severity and mechanism. For example, BC penetrates alveolar barriers, translocates into systemic circulation, and exacerbates oxidative DNA (DeoxyriboNucleic Acid) damage, while secondary inorganic aerosols (e.g., ) disrupt acid–base homeostasis, potentiating endothelial injury [7]. The existing evidence underscores the association between PM2.5 and ECG-derived subclinical endpoints. Epidemiological studies have found that reduced heart rate variability (HRV) is strongly associated with autonomic imbalance and ventricular repolarization instability, respectively, both of which predict adverse cardiovascular outcomes independent of traditional risk factors [32]. A study revealed that PM2.5 exposure correlated with HRV reduction, particularly in metabolically susceptible populations [13]. Experimental models further elucidate that PM2.5 disrupts iron metabolism, inducing ferroptosis and mitochondrial dysfunction in cardiomyocytes, which in turn manifest as repolarization anomalies and arrhythmias [40]. For instance, PM2.5 exposure in murine models elevated cardiac iron accumulation and Nuclear Receptor Coactivator 4 (NCOA4)-mediated ferritinophagy, exacerbating myocardial hypertrophy and ST-segment depression [17]. Previous studies have indicated that there might be a significant correlation between PM2.5 and abnormal electrocardiogram results [47]. However, few longitudinal studies have characterised the differential effects of PM2.5 components on ECG abnormalities in elderly populations. First, most studies focus on aggregate PM2.5 mass, overlooking component-specific effects. Second, elderly populations, who face heightened susceptibility due to age-related cardiovascular remodeling and comorbidities, are underrepresented in existing cohorts [16]. Third, longitudinal data linking prolonged PM2.5 constituent exposure to ECG changes are sparse, limiting causal inference and policy-relevant insights. Consequently, this study aimed to investigate the longitudinal associations of long-term exposure to PM2.5 and its chemical constituents with electrocardiogram abnormalities among elderly individuals in Guangzhou, China. Materials and methods Study subjects and data sources The population was from elderly individuals aged 65 and above in the communities of Guangzhou from 2018 to 2022. Based on the collection, cleaning and integration of the original data, a dynamic health database of Guangzhou elderly population with over 700,000 participants was constructed. Each participant completed a detailed questionnaire at baseline entry, including basic demographic information, lifestyle, physical health status, physical measurements, and supplementary examinations (including ECG examination) [22]. The study conducts longitudinal analysis based on repeated measurement data. The inclusion criteria were: 1) availability of data on physician-verified ECG diagnosis data; 2) a normal ECG result at baseline and at least one follow-up visit. The exclusion criteria were: 1) Participants with inaccessible address or missing environmental exposure (pollutant exposure could not be assessed due to an invalid address) were excluded; 2) Participants have cardiovascular diseases at the baseline; 3) Repetitive or incorrect recording of electrocardiogram or other variable data (Fig. S1). We estimated the required sample size for repeated measurement data in longitudinal study. Liu et al. proposed a sample size calculation method based on the GEE approach for continuous or binary responses. Moreover, a study summarized the parameters required for sample size estimation, including the number of repeated measurements (k), the correlation between repeated measurement results (ρ), statistics significance (α), test power (β), effect size (OR) and the type of working correlation matrix [23, 35]. We give a repeated measurements of 3, a correlation between repeated measurements of 0.6, an effect size of 1.03, a statistics significance set at 0.05 (α = 0.05), and to achieve a statistical power of 80%, a minimum of 97,749 subjects was required, when increasing the power to 90%, the sample size would be 130,858. Accounting for an anticipated attrition rate of 10%, the target sample size was determined to be at least 145,397 participants. Exposure concentration of PM2.5 and its components The daily ambient concentrations of PM2.5 and its chemical constituents were derived from the Tracking Air Pollution in China (TAP, http://tapdata.org.cn) database. This comprehensive repository integrates multi-source environmental data, including ground-based monitoring records, satellite remote sensing measurements, high-resolution emission inventories, topographic elevation datasets, and population distribution statistics. The database incorporates an operational Weather Research and Forecasting/Community Multiscale Air Quality Modeling System (WRF/CMAQ) modeling framework, yielding complete daily PM2.5 coverage at 10 km resolution. Validation through bagged cross-validation demonstrated strong predictive performance (R2 = 0.80–0.88; RMSE = 13.9–22.1 μg/m3) across 2013–2020 [5, 42]. The research consortium characterized spatiotemporal patterns of PM2.5 components during 2005–2012 [6]. Model verification confirmed close agreement between TAP-derived chemical components and ground observations. The CF-corrected simulations achieved robust validation metrics (R = 0.52–0.72,NMB ≤|3.23%|,RMSE ≤ 0.07) in 10-fold cross-validation [24]. Temperature(°C) measurements in degrees Celsius were obtained from the ERA5-Land reanalysis product [27], released by the European Centre for Medium-Range Weather Forecasts and containing multiple meteorological variables. Utilizing daily gridded air quality data, we employed bilinear interpolation to refine exposure estimates for study participants. Geographic coordinates derived from residential addresses enabled calculation of inverse-distance weighted averages, incorporating pollution values from the four nearest grid points [1, 46] . Annual mean exposure concentrations were computed by aggregating daily estimates over the calendar year and dividing by the number of corresponding observation days. Pollutants and temperature exposure metrics were derived through identical computational procedures. Electrocardiogram examination assessment The primary study outcome was of the diagnosis based on electrocardiogram examinations conducted by community doctors. Electrocardiogram examinations were conducted in community health service institutions in Guangzhou, and ECG diagnoses were made by community physicians in accordance with clinical guidelines. According to the Minnesota coding [34] and ICD-10, the electrocardiogram was classified as normal or abnormal. Normal electrocardiogram was defined as follows: all waveforms, intervals, amplitudes and shapes are within the normal range; the heart rate is 60–100 beats per minute; the P-R interval is 0.12–0.20 s; and the QRS complex duration is 0.06–0.10 s. Abnormal electrocardiogram was categorized as follows: 1) Major Q wave abnormalities (1–1, 1–2); 2) Minor Q wave abnormalities plus ST-T abnormalities (1–3 + 4–1/4–2/5–1/5–2); 3) Major Isolated ST-T abnormalities (4–1, 4–2, 5–1, 5–2); 4) Complete or intermittent bundle branch block(7–1, 7–2); 5) Left ventricular hypertrophy plus ST-T abnormalities(3–1 + 4–1/4–2/5–1/5–2); 6) Major QT prolongation (QT index ≥ 116%); 7) Atrial Fibrillation or Flutter(8–3); 8) Major atrioventricular conduction abnormalities and 9) other major arrhythmias (8–2, 8–4-2/8–4-1 with heart rate > 140). The clinical doctors who conducted the electrocardiogram measurements received standardized training and checked validity and reliability of ECG diagnosis according to the standardized procedures [39]. Covariates During serial study visits, participant sociodemographic characteristics were prospectively ascertained using standardized instruments, including age (> 75 years and ≤ 75 years), sex (male and female), educational attainment (No formal schooling, Primary, Junior secondary, Senior secondary, and Tertiary education), body mass index (underweight [< 18.5 kg/m2], normal weight [18.5–24.0 kg/m2], overweight [24.1–28.5 kg/m2], obese [> 28.5 kg/m2]), smoking status (Yes or No), drinking status (Yes or No), physical activity level (Physical activity was divided into high, medium and low levels based on questionnaire survey), cardiovascular medication (Yes or No) and comorbidities (Yes or No). As a meteorological factor, temperature was included in the analysis. To control for long-term rhythmic fluctuations in ECG parameters, the survey participation time was converted into a seasonal variable. Based on the actual conditions of the study area, we defined the period from April to September each year as the warm season, while the rest of the time was the cold season. Data analysis Baseline characteristics across the complete study cohort were summarized using descriptive analytical methods, with missing covariate data handled through multiple imputation techniques. This study analyzed the correlation between PM2.5, its components and temperature. Adjustment covariates included demographic factors (age, sex, education), meteorological variables (temperature), time variable (season), behavioral indicators (smoking, physical activity, alcohol consumption), cardiovascular medication, and comorbidities (diabetes, hypertension and hyperlipidaemia). To account for autocorrelation in longitudinal measurements, this study employed the generalized estimating equation (GEE) to assess the association between PM2.5 and its components, and ECG abnormalities [10, 49]. Variance inflation factor (VIF) was used to assess multicollinearity within the model that included multiple PM2.5 components. This study designed three progressively adjusted models: Model 1 only corrected for age and gender; Model 2 added additional variables, including educational level, season and temperature to Model 1; Model 3 was the final analysis model, which further included smoking, drinking, physical activity level, BMI, cardiovascular medication, and comorbidities variables. Nonlinear exposure–response relationships for PM2.5 constituents were characterized using restricted cubic spline. Stratification analysis by sex, age, and administrative district was conducted to examine effect heterogeneity [25]. Sex was classified as male and female, and age was divided into two groups: > 75 years and ≤ 75 years. The administrative districts covered all districts of Guangzhou, China. Sensitivity analyses encompassed five approaches: 1) Sequential modeling with three tiers of covariate adjustment to assess robustness; 2) Time-window sensitivity testing using 1-, 2-, and 3-year moving averages preceding baseline; 3) Using arrhythmias as secondary outcome to verify associations with PM2.5 components; 4) Multi-pollutant model included not only PM2.5 and its components, but also incorporates nitrogen dioxide or ozone for analysis; 5) The highest 1% of pollutant exposure values were trimmed to estimate the exposure–response relationship. Using R statistical software (v4.3.1), we employed geepack for generalized estimating equation models and ggplot2 for figure generation. A statistics significance of p < 0.05 (two-sided) was applied. Results Descriptive analysis at baseline Between 2018 and 2022, the study included 215,073 participants in Guangzhou and had a total of 691,990 ECG recordings (Figure S1). Within the elderly cohort, most participants were aged 75 years or younger (82.9%), females predominated (61.1%), and most participants had completed secondary education or higher (Table 1). Table 1Participants’ characteristicCharacteristicsCategoriesStatisticsNo. of participants215,073Age, n (%) > 75 years36,768 (17.1) ≤ 75 years178,305 (82.9)Sex, n (%)Male83,598 (38.9)Female131,475 (61.1)Educational level, n (%)High school36,106 (16.8)Illiterate21,409 (9.9)Middle school43,596 (20.3)Primary school70,717 (32.9)University or higher42,834 (19.9)Missing411 (0.2)BMI, n (%)Normal136,198 (63.3)Obese8383 (3.9)Overweight60,891 (28.3)Underweight9366 (4.4)Missing235 (0.1)Physical activity level, n (%)High86,280 (40.1)Moderate70,993 (33.0)Low53,871 (25.0)Missing3929 (1.9)Smoking, n (%)No181,994 (84.6)Yes32,972 (15.3)Missing107 (0.1)Drinking, n (%)No195,455 (90.9)Yes19,310 (9.0)Missing308 (0.1)Season, n (%)Warm season136,354 (63.4)Cold season78,719 (36.6)Cardiovascular medication, n (%)No214,966 (99.9)Yes107 (0.1)Diabetes, n (%)No164,794 (76.6)Yes50,279 (23.4)Hypertension, n (%)No76,334 (35.5)Yes138,739 (64.5)Hyperlipidaemia, n (%)No115,300 (53.6%)Yes99,773 (46.4%) Participants’ characteristic Table 2 shows characteristics of pollutants and temperature in this study. Interrelationships among PM2.5, its chemical constituents, and ambient temperature are illustrated in Fig. 1. Pairwise correlation analyses revealed strong collinearity between total PM2.5 concentration and constituents. Furthermore, inverse associations were observed between temperature and both PM2.5 and its constituent species. The VIFs for all pollutants were extremely high (ranging from 39.17 to 199.65), indicating severe multicollinearity (Table S2). Therefore, we entered each pollutant separately into the model in the subsequent analyses. Table 2Characteristics of pollutants and temperatureMinp25MedianMeanp75MaxIQRPM2.5(μg/m3)17.9730.9234.8734.2736.7975.075.86BC(μg/m3)1.101.691.861.862.013.510.32OM(μg/m3)5.188.249.018.999.6419.001.40(μg/m3)3.695.886.236.336.8011.960.92(μg/m3)2.654.144.404.494.8517.020.70(μg/m3)1.932.973.173.273.5310.560.56Temperature(°C)2.7322.3422.5622.5422.8024.100.46 Characteristics of pollutants and temperature Correlation between PM2.5, its components, and temperature. Abbreviations: PM2.5, particulate matter with an aerodynamic diameter ≤ 2.5 μm; BC, black carbon; , nitrate; , ammonium; OM, organic matter; , sulfate Exposure–response function estimation By plotting the increase of per interquartile range (IQR) of PM2.5 constituents against the corresponding odds ratio (OR) value, it can be observed that the exposure–response relationships for each pollutant all demonstrated nonlinear patterns (Fig. 2). The histogram displays the concentration distribution of pollutant exposures. Significant positive correlations between PM2.5 components and ECG abnormalities were observed above specific exposure concentration levels. For instance, as the black carbon concentration rose to 1.99 μg/m3, the trend became monotonically increasing. The exposure–response relationship between PM2.5, its components, and ECG. Abbreviations: A to F illustrate the exposure–response relationship between PM2.5, its components, and ECG. The black dotted line in A marks the inflection point concentration. In B to F when the exposure concentration is higher than the exposure concentration indicated by the black dotted line, there is a significant positive correlation between the PM2.5 component and ECG Association between PM2.5 constituents and ECG abnormalities Figure 3 presents ORs and 95% confidence intervals (95% CIs) for ECG abnormalities associated with per IQR increases in PM2.5 constituents. All components demonstrated positive associations with elevated ECG abnormalities risk. Ammonium showed the strongest association (OR:1.070; 95% CI: 1.062–1.078; IQR: 0.56 μg/m3), followed sequentially by nitrate (OR:1.058; 95% CI: 1.050–1.065; IQR: 0.71 μg/m3), sulfate (OR:1.047; 95% CI: 1.040–1.054; IQR: 0.92 μg/m3), black carbon (OR:1.034; 95% CI: 1.027–1.042; IQR: 0.32 μg/m3), and organic matter (OR:1.030; 95% CI: 1.022–1.037; IQR: 1.40 μg/m3). Comparatively, PM2.5 mass concentration exhibited a more modest association (OR:1.029; 95% CI: 1.020–1.037; IQR: 5.86 μg/m3). Relationship between PM2.5, its components, and ECG. Abbreviations: Model 1 has been standardized for age and gender; Model 2 adds educational level, seasonal and temperature factors to Model 1; Model 3 further incorporates smoking, drinking, physical activity level, BMI, cardiovascular medication, and comorbidities variables (diabetes, hypertension and hyperlipidaemia) to Model 2 Sensitivity analyses and stratification analysis The comparative patterns between the five constituents and ECG abnormalities remained consistent across 1-, 2-, and 3-year exposure windows. For PM2.5 and sulfate, the strongest associations emerged at the 2-year exposure duration, whereas black carbon, organic matter, ammonium, and nitrate demonstrated progressively intensifying effects from 1- to 3-year exposure periods (Fig. 4).With arrhythmia as the outcome, the effect magnitude (OR) for PM2.5 and its chemical constituents consistently exceeded that observed for ECG abnormality (Table S3). Adjustment for nitrogen dioxide markedly elevated the association estimates for PM2.5 and its components (except for nitrate and ammonium) compared to unadjusted values, whereas adjustment for ozone did not significantly change the estimates (Table S4-S5). After trimming extreme high exposure observations above the 99th percentile, the exposure–response relationship was similar to the non-linear pattern observed before trimming (Figure S2). Association between per IQR increase of PM2.5, its components, and ECG under different exposure windows Except for nitrate and ammonium, other components exhibited stronger associations among males compared to females. Among participants > 75 years, all constituents demonstrated equivalent or higher risks of electrocardiograph abnormality relative to their younger counterparts (≤ 75 years) (Table S6). PM2.5 and its components were positively and significantly associated with ECG abnormalities in Tianhe, Yuexiu, Huadu, and Conghua districts, indicating that older adults in these districts might be high-risk populations (Table S7). Discussion This study employed physician-diagnosed electrocardiographic anomalies as a composite clinical endpoint to assess cardiovascular effects of targeted PM2.5 species. It was found that PM2.5 constituent exposure was a significant predictor of ECG pathology risk. More specifically, water-soluble ionic fractions such as , and were all strongly associated with poor ECG outcomes. Addressing critical knowledge gaps, the longitudinal analysis dissected early cardiac electrophysiological alterations of persistent PM2.5 constituent exposure, focusing on electrocardiographic precursors to cardiovascular morbidity. Our findings underscored that future air pollution control policies should extend beyond regulating total PM2.5 mass to specifically targeting the sources of key chemical constituents. In particular, strengthened controls on precursors of sulfate and nitrate (such as sulfur dioxide and nitrogen oxides) from urban, industrial, and transportation sources are warranted. Aligned with prior epidemiological work, our findings revealed a nonlinear exposure–response relationship between PM2.5 components and ECG abnormalities, with evidence of the concentration threshold below which no significant positive correlation was observed [10]. Accumulating evidence has demonstrated that short- and long-term PM2.5 exposure is significantly associated with increased cardiovascular morbidity and mortality, yet the specific shape of the exposure–response curve and the presence of potential thresholds remain less characterized for ECG-based electrophysiological endpoints [37, 43]. Chronic exposure to OM, BC, , and , consistently predicted prolonged electrocardiographic intervals in multicenter investigations [3]. Acute BC exposure effects manifested as ST-depression [9] and altered ventricular repolarization [11]. However, prior literature has shown inconsistencies regarding the impacts of major inorganic PM2.5 constituents on mortality and cardiovascular risk [12, 44]. Overall, most studies have assessed pollutant-related ECG abnormalities in participants with pre-existing heart conditions [20] or in elderly populations [9, 26]. Notably, our study contributed new evidence regarding the association between key water-soluble ionic species (sulfate, nitrate and ammonium) and elevated odds of electrocardiographic anomalies, a relationship that has received limited attention in the existing literature. Several biological pathways may explain the observed associations. It has been suggested that BC may increase cardiovascular risk and hypertension through alterations in DNA methylation and oxidative stress [30]. Moreover, a study suggested that exposure to secondary inorganic aerosols (i.e., , , ) can activate the hypothal­amic–pituitary–adrenal (HPA) axis [31], which may adversely affect the cardiovascular system. These sulfate, ammonium and nitrate ions are typically formed from sulfur and nitrogen oxides, which are emitted as gases from fossil fuel combustion and vehicular traffic emissions. Major inorganic constituents constitute a significant mass fraction of PM2.5, disproportionately contributing to its toxicological burden [12, 19, 38]. We utilized high-resolution ambient pollution data and bilinear interpolation to achieve precise exposure estimates at the individual level, and achieved a significant enhancement in statistical power through a larger sample size. However, several limitations warrant consideration. First, it did not account for indoor pollutant exposure, which is a significant determinant of personal exposure profiles. Individuals primarily reside in enclosed environments where indoor pollution sources, notably cooking emissions, tobacco smoke, and household cleaning agents, may substantially reshape exposure profiles and compound disease burdens [15, 26, 48]. This limitation is particularly relevant for semi-volatile components like ammonium and nitrate, which exhibit complex indoor-outdoor partitioning dynamics [28, 29]. The absence of indoor exposure quantification may lead to exposure misclassification, especially for populations with limited ventilation or significant indoor pollution sources. Second, there was a lack of formal inter-observer validation for ECG readings. The absence of a standardized process to assess consistency across different readers may introduce variability in ECG interpretation, potentially affecting the reliability of the outcome measures. Additionally, potential exposure misclassification and measurement error in ECG assessment may also exist. Third, potential survival bias may exist, as the sickest patients may have been lost to follow-up due to severe outcomes such as death or hospitalization. The observed associations between PM2.5 exposure and ECG might be underestimated, as the most severe cases potentially caused by the exposure are missing from the final analysis. Fourth, although this study adjusted for some major confounding factors, residual confounding may still exist from unmeasured factors such as genetic susceptibility, psychological stress, and socioeconomic status, which may introduce some bias in effect estimation. In conclusion, our study shows that long-term exposure to PM2.5 chemical species is significantly associated with ECG abnormalities. Indoor monitoring or developing microenvironmental exposure models need to be integrated to better capture total personal exposure burdens. Additionally, the pathophysiological processes that connect ECG abnormalities to cardiac electrophysiological disturbances, along with their subsequent effects on extra-cardiac systems including the cardiovascular system itself, require further elucidation. Current evidence remains insufficient to identify population subgroups most susceptible to cardiac manifestations from PM2.5 component exposure, highlighting further research in the general population is needed to investigate how existing cardiac conditions may interact with PM2.5 exposure to influence the risk of ECG abnormalities. Conclusion Our study demonstrates that long-term exposure to PM2.5 chemical species, particularly organic matter, black carbon, sulfate, nitrate and ammonium, might increase risk of ECG abnormalities. Future research should further explore the pathophysiological process by which PM2.5 components cause cardiac electrophysiological disorders. The formulation of air pollution control policies should specifically target the sources of key hazardous chemical components. Supplementary Information Supplementary Material 1. Acknowledgements We extend our sincere appreciation to all co-authors for their collective efforts and diverse yet integral contributions. Authors’ contributions Chaoling Zhang : Methodology, Formal Analysis, Writing-Original Draft. Qin Zhou : Investigation, Writing-Review & Editing. Jiamin Chen : Resources, Writing-Review & Editing. Zhengmin (Min) Qian : Writing-Review & Editing. Maya Kavuri : Writing-Review & Editing. Tom Burroughs : Writing-Review & Editing. Zhonghua A i: Data Curation, Writing-Review & Editing. Shengtao Wei : Data Curation. Weiquan Lin : Resources, Supervision. Zhoubin Zhang : Project administration, Supervision. Hui Liu : Supervision, Funding Acquisition. Hualiang Lin : Conceptualization, Writing-Review & Editing, Supervision. Funding This work was supported by the National Science and Technology Major Project of China (2024ZD0524500, 2024ZD0524502); National Natural Science Foundation of China (72104061); the Key Project of Medicine Discipline of Guangzhou (No.2025–2027-12). Data availability The datasets generated and analyzed during the current study are not publicly available due to some sensitive information at present but are available from the corresponding author on reasonable request. Declarations Ethical approval and Informed consent Ethics Committee of the Guangzhou Center for Disease Control and Prevention granted ethical approval for this study (GZCDC-ECHR-2023P0081) and trial registration was completed at China Clinical Trial Registry (http://www.chictr.org.cn/, ChiCTR2400089945). The requirement for informed consent was waived and approved by Ethics Committee of the Center for Disease Control and Prevention of Guangzhou (GZCDC-ECHR-2023P0081) because this observational study posed no risk to participants. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Chaoling Zhang and Qin Zhou contributed equally to this work. Contributor Information Hui Liu, Email: gzcdc_liuhui@163.com. Hualiang Lin, Email: linhualiang@mail.sysu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets generated and analyzed during the current study are not publicly available due to some sensitive information at present but are available from the corresponding author on reasonable request.