Ambient PM_2.5 and acute incidence of myocardial infarction in China: a case-crossover study and health impact assessment

Abstract

Background and purpose: 

Evidence investigating the association between ambient fine particulate matters (PM_2.5) and acute incidence of myocardial infarction in developing countries is limited. This study aims to investigate linear and nonlinear patterns for the association between PM_2.5 and acute incidence of myocardial infarction based on multicounty registry dataset and evaluate the reduction of premature myocardial infarction incidence under different pollution control objectives in China.

Methods: 

Thirty-six thousand six hundred and seventy-nine registered myocardial infarction incidence cases from 15 Chinese counties from January 1, 2013, to December 31, 2017, were obtained. We adopted a time-stratified case-crossover design with conditional logistic regression models.

Results: 

With a 10 μg/m³ increase in PM_2.5 exposure, there was an increase of 0.98% (95% CI: 0.40%–1.57%) in acute incidence risk of myocardial infarction. The corresponding values for males and individuals aged over 74 years were 1.58% (95% CI: 0.82%–2.35%) and 1.19% (95% CI: 0.35%–2.05%) respectively, indicating higher risks than other groups. The nonlinear concentration–response curve indicated a steeper slope under daily exposure below 50 μg/m³ and the marginal avoided premature morbidity became larger under the current air quality standard.

Conclusions: 

The robust findings from this study may suggest the necessity for a continuous reduction of PM_2.5 exposure concentration from the perspectives of public health.

INTRODUCTION

Myocardial infarction (MI) is one of the most common acute cardiac events with a high annual incidence of 10.6 million worldwide^[1], leading to severe damage of the heart and resulting in a high rate of mortality and disability^[2]. Epidemiological studies have provided evidence that exposure to ambient fine particle matter (PM_2.5) may have a compelling impact on MI compared with exposure to other air pollutants^[2,3]. Therefore, understanding the overall relationships between PM_2.5 and acute incidence of MI is imperative for generating implications needed for identifying public health priorities and implementing health intervention actions, particularly in the context of widely spreading air pollution and increasing disease burden of cardiac disease^[1]. Since the beginning of the 21st century, emerging multicenter studies in developed countries have explored the association between short-term exposure to PM_2.5 and the incidence risk of MI mostly by either using hospital admission or emergency room visit data^[2,4–6]. The latest research in the United States reported a 0.11% increase in PM_2.5-induced MI hospital admission risk based on the national Medicare inpatient claims^[7]. However, this study focused on the situation in a developed country with a relatively low and stable air pollution level. Scientific evidence regarding the association between short-term exposure to PM_2.5 variation and acute incidence of MI in heavily polluted developing countries is still inadequate. In China, long-lasting air pollution and a high prevalence of cardiovascular disease are two co-existing public health issues^[8,9], emphasizing the significance of exploring the impact of PM_2.5 on MI incidence. A large-scale study covering the wide variance of the air pollution level in China could contribute to drawing the complete association between PM_2.5 and MI risk, especially capturing the feature of the curve relationship. However, strong evidence from large-scale studies has been very limited^[10–13]. The latest published 184-city research study in China that reported PM_2.5-related hospital admission risk of various causes of cardiovascular disease unfortunately lacked an analysis on the association between PM_2.5 and MI^[14]. Additional research is urgently needed to understand how MI incidence is influenced by widespread PM_2.5 pollution from the perspective of public health protection.

Previous studies generally utilized databases comprising hospital admission or emergency visit data as the only outcome describing the acute incidence of MI. A common limitation of these studies was that their use of the outcome measure may underestimate MI incidence risk and even the impact of air pollution because hospital admission or emergency cases usually could not represent the whole population in the studied area^[13]. Deviation on incidence date existed by using hospital admission data since there was always a delay from the onset of symptoms to the date of hospital admission in China. In addition, findings are less clear about the overall shape of the concentration–response patterns between short-term PM_2.5 exposure and the acute incidence of MI. Although broad epidemiological evidence has focused on a linear association^[15], a small number of studies from different locations have reported a nonlinear concentration–response curve between air pollution and cardiovascular risk^[16,17], indicating the necessity of exploring the association pattern between PM_2.5 and MI incidence risk from the perspective of supporting effective policy and regulatory actions.

Here, we performed a multicenter analysis on the associations between short-term PM_2.5 exposure and the acute incidence of MI in China using the MI incidence registry database based on a case-crossover design. Both linear and nonlinear patterns of the associations were investigated. The avoided premature MI incidence attributed to the reduction of PM_2.5 exposure in the studied regions was estimated to provide supportive policy implications.

MATERIALS AND METHODS

Environmental exposure data

Daily ambient PM_2.5 and other air pollutants concentration data from 2013 to 2017 from 15 Chinese counties was obtained from the National Air Pollution Monitoring System (http://www.cnemc.cn/sssj). The county-specific daily exposure was averaged from the hourly concentration data of all the fixed monitoring stations within each county. To control for the potential confounding effects of meteorological factors, we also obtained the same-day daily mean temperature and relative humidity data from the China Meteorological Data Sharing Service System (http://data.cma.cn). MI cases from the same county were matched with county-level environmental exposure data.

Myocardial infarction incidence data

The first-ever MI incidence cases from 2013 to 2017 were extracted from the Chinese Environmental Public Health Tracking System (CEPHT)^[18], which is a network integrating datasets of environmental factors and various relative health outcomes. MI incidence data collected in CEPHT was originally generated by local medical organizations reporting cases for all kinds of ischemic heart disease through their disease incidence registry system, which was built and managed by the local centers for disease control and prevention (local CDCs). Information on the reported cases was collected when the cases first arrived at the local medical organizations with the onset of symptoms, which could correctly record the incidence date since the average time from the onset of symptoms and hospital arrival was approximately 4 h^[19]. Reported MI cases were diagnosed by cardiologists when the case met the standard clinical guidelines with acute symptoms, high-sensitivity cardiac troponins test, and cardiac imaging in all studied regions. We then extracted data of each diagnosed MI case by using codes I21 to I23 according to the International Classification of Diseases, 10th revision (ICD-10). Demographics including sex and age were extracted from the dataset.

Statistical analysis

This study employed a time-stratified case-crossover design by restricting the control periods to the same day of the week, month, and year as the case day to control for time-associated confounders^[20]. Conditional logistic regression analysis was adopted and the main models were constructed as follows: (1) to adjust for the nonlinear effects of weather conditions, we employed a smoothing function using natural splines with three degrees of freedom (df) for mean temperature and relative humidity; (2) to investigate the delayed effect of exposure, we separately fitted the models considering different lag times up to 3 days prior to the date of the incident (single lag 1-lag 3, cumulative lag 01-lag 03). We then chose the lag exposure window according to the optimal estimate among all the exposure–response relationships of different lags, based on which we also performed subgroup analyses by sex and age by separately fitting models for each subgroup to explore modification effects and identify susceptible populations. Furthermore, to model the shape of the exposure–response relationships between PM_2.5 and MI incidence, we also set up a one-basis function by applying a natural cubic spline of PM_2.5 with three dfs^[21].

To examine the robustness of the main model, we performed several sensitivity analyses including: (1) fitting two-pollutant models by adding ozone (O₃) and nitrogen dioxide (NO₂) in the main model because of their significant correlation between daily PM_2.5 (for O₃: coefficient = −0.08, P < 0.001; for NO₂: coefficient = 0.57, P < 0.001); (2) adjusting dfs for controlling meteorological factors; and (3) adopting the symmetric referent period^[20,22]. A percentage increase in the acute incidence risk with a per 10 μg/m³ increase in PM_2.5 concentration and 95% confidence intervals (95% CI) was observed.

Based on the concentration–response curve, we evaluated the annual avoidance of premature MI morbidity associated with per 10 μg/m³ reduction of daily concentration of PM_2.5 from 160 to 10 μg/m³. The annual avoidance of the morbidity burden for these 15 counties in 2016 was evaluated as follows:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{\Delta }\mathrm{I}\mathrm{n}\mathrm{c}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{c}{\mathrm{e}}_{\mathrm{i}}=\mathrm{R}\times \mathrm{P}\mathrm{O}\mathrm{P}\times \mathrm{E}\mathrm{R}{\mathrm{C}}_{\mathrm{i}},}\end{array}}$$

where $\mathrm{\Delta }\mathrm{I}\mathrm{n}\mathrm{c}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{c}{\mathrm{e}}_{\mathrm{i}}$ was the avoided premature morbidity under different reductions of PM_2.5 concentration, R was the baseline incidence rate, approximately 79 incidents per 100,000 people, calculated from 15 counties using their total population and acute incidence of MI, POP was the total population in 15 counties in 2015, $\mathrm{E}\mathrm{R}{\mathrm{C}}_{\mathrm{i}}$ was the attributable percentage changes in incidence risk associated with a 10 μg/m³ increase of PM_2.5 derived from the curve.

The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and was approved by the Ethical Review Committee of National Institute of Environmental Health, Chinese Center for Disease Control and Prevention (NO.: 202102; date: February 25, 2021). Due to the retrospective nature of the study, informed consent was waived by the Ethical Review Committee of National Institute of Environmental Health, Chinese Center for Disease Control and Prevention. The R software 4.1.2 (Lucent Technologies, Murray Hill, New Jersey, USA) was used for statistical analysis. ArcGIS 10.2 (Esri, Redlands, California, USA) was used to produce the map of study sites.

RESULTS

Overall, 36,679 MI events were identified from 123,572 registered ischemic heart disease cases from 15 counties (Supplementary Figure 1, https://links.lww.com/CARDIOPLUS/A10). Table 1 indicates an average concentration of PM_2.5 exposure across all the counties at 58 μg/m³, and Supplementary Table 1, https://links.lww.com/CARDIOPLUS/A10, indicates a high exposure variation widely ranged from 2 to 567 μg/m³. Supplementary Figure 2, https://links.lww.com/CARDIOPLUS/A10, presents PM_2.5 exposure level of 15 counties covering eight provinces from northern to southern China.

Table 1. - Descriptive statistics of the myocardial infarction incidence and exposure from 2013 to 2017

Item	Value
Total incidence of myocardial infarction, n	36,679
Cases, n	36,679
Controls, n	138,830
Sex, n (%)
Male	23,072 (62.9)
Female	13,607 (37.1)
Age, y
Median	74
Mean (SD)	72 (14)
Age, n (%)
0–74 y	19,152 (52.2)
> 74 y	17,510 (47.7)
Environmental exposure, mean (SD)
Daily average PM2.5, μg/m3	57.9 (40.1)
O3, μg/m3	78.2 (46.4)
NO2, μg/m3	40.4 (24.0)
Daily average temperature, °C	16.1 (9.9)
Daily relative humidity, %	71.8 (16.5)

PM_2.5: Fine particulate matter; SD: Standard deviation.

We first summarized the results of the lag structure of the association between PM_2.5 and the acute incidence risk of MI by simultaneously estimating the effect from lag days 0 to 3. Figure 1 indicates the percentage change in the acute incidence risk of MI associated with the PM_2.5 concentration on different lag days. Significant acute effects were observed on the exposure day and short lag days. Considering a 10 μg/m³ increase in PM_2.5 concentration, the largest percentage increase was 0.98% (95% CI: 0.40%–1.57%) (P = 0.001) on day lag 02, but the effects eventually declined. Sensitivity analysis in Supplementary Table 2, https://links.lww.com/CARDIOPLUS/A10, demonstrated the robustness of the main model. We then evaluated the effect modification by sex and age, as presented in Figure 2. With a 10 μg/m³ increase in day lag 02’s PM_2.5 concentration, the incidence risk in the male group significantly increased by 1.58% (95% CI: 0.82%–2.35%) (P < 0.001), while that in the female group increase by only 0.06% (95% CI: −0.87–1.00%) (P = 0.065), without statistical significance. Additionally, Figure 2 presents a suggestive difference in the percentage change between the over 74 years old age group (1.19% [95% CI: 0.35%–2.05%], P = 0.005) versus the 0 to 74 years old age group (0.77% [95% CI: −0.04%–1.60%], P = 0.907).

Figure 1.: Percent change in the MI incidence associated with each 10 μg/m ³ increase in PM _2.5 concentrations on different lag days. A, Single day lag. B, Cumulative lag. MI: myocardial infarction; PM_2.5: fine particulate matter.

Figure 2.: Percent change in MI incidence in different subgroups associated with a 10 μg/m ³ increase in PM _2.5 exposure during the same and the previous 2 days (lag 02). MI: myocardial infarction; PM_2.5: fine particulate matter.

The short-term effect on MI incidence presented a nonlinear pattern with two-stage features divided by a cutoff point at approximately 50 μg/m³: a steeper slope indicating sharply rising risk within 0 to 50 μg/m³ and a nearly stable risk under relatively high exposure over 50 μg/m³ were observed (Figure 3). Significant increased risk was observed under the primary concentration limits of PM_2.5 in the Chinese Air Quality Standard (CAQS) at 75 μg/m³. Figure 4 presents the avoided premature morbidity from reducing each 10 μg/m³ of PM_2.5 daily concentration from 160 to 10 μg/m³. In detail, as derived from the exposure–response curve between PM_2.5 and MI incidence risk, a larger avoidance of health burden could be achieved with continuous reduction, especially at lower concentration. Marginal avoidance of morbidity became larger approximately under the current CAQS limits of 75 μg/m³. The largest avoided morbidity occurred when PM_2.5 concentration was reduced under the WHO Air Quality Guideline (AQG 2005 version) at 25 μg/m³.

Figure 3.: Concentration–response curve between PM _2.5 concentration and MI incidence risk. PM_2.5: fine particulate matter; MI: myocardial infarction.

Figure 4.: Annual reduction of premature morbidity associated with a 10 μg/m ³ decrease in daily PM _2.5 concentration under different proposed control strategies in 2016. Labels under each pillar represent PM_2.5 concentration reduction at different levels, for example, “160-150” indicates controlling PM_2.5 from 160 to 150 μg/m³, “20-10” indicates reducing PM_2.5 concentration from 20 to 10 μg/m³. PM_2.5: fine particulate matter.

DISCUSSION

Linear association between PM_2.5 and myocardial infarction incidence risk

Significant acute effects were observed on very short lag days. The highest percentage increase was 0.98% (95% CI: 0.40%–1.57%) (P = 0.001) considering a 10 μg/m³ increase in PM_2.5 concentration exposure in 2 days before the events (lag 02), followed by reduced risk with increased lag time. Considering other studies in Chinese regions, due to the lack of multicenter reports, comparison was only available among single-location studies. Our estimates derived from multiple-center datasets presented a much stronger positive association than the result reported from a single-city research study considering the association between hospital admissions and daily PM_2.5 (0.46% per 10 μg/m³ increase of PM_2.5)^[11]. Possible reasons include the increased statistical power brought by larger study samples and the validity of using the incidence registry dataset due to its full coverage of MI incidents. Compared to related global studies, our estimation of the change in risk is within the range of the pooled changes in acute incidence risk of MI associated with daily PM_2.5 exposure reported as 1.0% (95% CI: 0.3%–1.7%)^[2] and 2.4% (95% CI: 0.7%–4.1%)^[3]. Solid evidence in this study indicated the necessity to pay attention on the immediate increased risk induced by PM_2.5, which may occur via influencing acute abnormal cardiac autonomic function and induced systemic inflammation^{[2,11,23–26]}, and lead to significant changes in imbalance in the appropriate interplay among the key components that maintain normal cardiac function.

Susceptible groups

Elderly individuals over 74 years old and males were more susceptible to PM_2.5, for whom the estimated increments of MI incidence risk were 1.19% (95% CI: 0.35%–2.05%) (P = 0.005) and 1.58% (95% CI: 0.82%–2.35%) (P < 0.001), respectively. This result was consistent with previous findings regarding effect modifications by sex^[11,12] and age^[10]. We observed a higher risk in males than in females, there are three possible reasons: (1) more males (11.8%) than females (8.0%) are at high cardiovascular risk in the whole Chinese population; (2) more males drink and smoke (56.5% and 34.1%, respectively) than females (2.3% and 3.2%, respectively), which are two important risk factors for cardiovascular disease; and (3) although Chinese females have higher prevalence rates of hypertension, diabetes, and overweight, they also have higher percentages of taking medicines than males, which may be helpful to reduce the risk^[27]. Furthermore, as reported in the latest review of epidemiological cardiovascular features in China, the aging population will lead to a rapid increase in the annual morbidity of MI^[9]. Therefore, males and elderly individuals are vulnerable to acute incidence risk of MI due to their high susceptibility to both internal and external factors. Under the current level of air pollution, the protection of these susceptible populations needs to be considered. Our findings provide support for identifying new public health intervention objectives.

Nonlinear association between PM_2.5 and myocardial infarction incidence risk

The concentration–response relationship was nonlinear across the full range of exposures, and there was no threshold for the positive association between PM_2.5 concentration and acute incidence risk of MI. Within a low level of exposure ranging from 0 to 50 μg/m³, the association was approximately linear, and the slope was steeper, in contrast to the flat trend under a relatively high level of exposure. This pattern is similar to the results of previous studies focusing on air pollution and all-cause mortality at the global^[16] and national levels^[17]. This finding suggests the great necessity of paying attention to the possible health risk even when the PM_2.5 concentration is below the Chinese Air Quality Standard (CAQS) at 75 μg/m³. From the perspective of public health, we recommend reviewing the rationale for the current standard of 24-h average PM_2.5 concentration (75 μg/m³). This standard may not be safe enough to protect people, especially the susceptible group with a potential risk of cardiac disease.

Avoided premature incidence attributed to PM_2.5 mitigation

Ultimately, health benefits could be obtained by all reduction strategies. In detail, our study found larger marginal health benefits could be effectively obtained with each 10 μg/m³ reduction of PM_2.5 concentration within lower exposure levels under the WHO Interim Target 2 (IT-2) limits of PM_2.5 (50 μg/m³) in China. The most significant increase in avoided premature morbidity could be achieved under a strategy of reducing the PM_2.5 daily average concentration below the WHO AQG (25 μg/m³). Although this is indeed a difficult objective to meet in the near future, positive mitigation policies should continue to focus on gradually achieving co-benefits of a high-quality environment and public health. This deep implication from our findings could support objectives for PM_2.5 mitigation from the perspectives of public health.

One strength of our study is that county-level incidence registry data were highly representative of the population-based MI incidence, adding significance to the comprehensive measurement of MI acute incidence from the whole study population. The exclusive use of hospital admission data cannot avoid the inherent limitations of representativeness of the real incidence^[23,24,28]. As reported in Liu et al.’s study^[24], the difficulty in distinguishing local and nonlocal patients, misclassification of diseases admission delays due to fully occupied beds in top-ranked hospitals in China, and the use of hospital admission data instead of the time of symptom onset before hospital admission could cause bias in the analysis. This study has two limitations. First, we used aggregated, county-level daily concentrations of PM_2.5 averaged from fixed monitoring stations within each specific county to approximately represent individual exposure^[29]. Berkson and classical measurement errors were inevitable, and the effect of PM_2.5 may be underestimated^[30]. Second, the limited data availability prevented the analysis of possible modification effects by other important factors, such as history of chronic diseases^[4,10,31,32], unhealthy lifestyle habits (smoking)^[4,10,33], and cardiometabolic risk factors (body mass index, blood pressure, total cholesterol, etc), as well an analysis of MI survivors in susceptible subgroups^[26]. However, the case-crossover study design adopted in this study can support each MI case serve as his/her own control to remove unmeasurable individual characteristics that were stable within a short-term window (eg, disease history, smoking habit, other unhealthy lifestyle, common variability in cardiometabolic factors, indoor exposure). Therefore, it can help to eliminate part of the uncertainties brought by the individual confounders.

This study provided robust evidence on linear and nonlinear associations between short-term PM_2.5 exposure and MI incidence in developing countries. Although PM_2.5 exposure poses a relatively low risk of non-communicable disease compared to other lifestyle factors^[34], it is a top public health concern that cannot be ignored because there are large populations widely exposed to various levels of air pollution^[35,36], as it is difficult to effectively avoid exposure when the ambient air is polluted, especially in developing countries. Effort on PM_2.5 control is still needed to achieve effective cardiovascular disease prevention in China and other regions with similar air pollution and cardiac epidemiology conditions.

FUNDING

This work was funded by grants from an Open fund by Jiangsu Key Laboratory of Atmospheric Environment Monitoring and Pollution Control (Grant: KHK2108) and the National Natural Science Foundation of China (Grant: 92143202). Jie Ban was supported by the Li Foundation Climate Change Fellowship Program at the Yale Center on Climate Change and Health.

AUTHOR CONTRIBUTIONS

JB and TL participated in research design; JB, RM, QW, CC, QS, YW, and JH contributed to data collection and verification; JB, RM, and AL participated in data analysis; JB, AL, and TL participated in manuscript writing. All authors reviewed and approved the manuscript.

CONFLICTS OF INTEREST STATEMENT

The authors declare that they have no financial conflict of interest with regard to the content of this manuscript.

DATA SHARING STATEMENT

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.