Association between atmospheric particulate matter and emergency room visits for cerebrovascular disease in Beijing, China

Abstract

Purpose

The association between atmospheric particulate matter and emergency room visits for cerebrovascular disease were evaluated in Beijing.

Methods

A generalized additive model was used to evaluate the associations between particulate matter and cerebrovascular disease, based on the daily data of meteorological elements, PM concentrations, and emergency room (ER) visits for cerebrovascular disease in Beijing from 2009 to 2012. Long-term trends and the effects of holidays, the day of the week, and confounding factors were controlled to determine the lag effect at 0–6 days. Single- and double-pollutant models were employed for different age and sex groups.

Results

The effect of PM_2.5 concentration on the number of daily ER visits for cerebrovascular disease was much stronger than that of PM₁₀ concentration. PM_2.5 and PM₁₀ had maximum RR values of 1.096 and 1.054 at lag 6 for patients aged 61–75 years. For each inter-quartile range (IQR) increase in PM₁₀ concentration, the maximum RR values for the total, males, females, aged 15–60 years, aged 61–75 years, and aged > 75 years were 1.024, 1.044, 1.043, 1.038, 1.054, and 1.032, respectively. For each IQR increase in PM_2.5 concentration, the maximum RR values for the total, males, females, aged 15–60 years, aged 61–75 years, and aged > 75 years were 1.038, 1.064, 1.076, 1.054, 1.096, and 1.049, respectively. The RR values of the double-pollutant models were lower than those of the single-pollutant models.

Conclusion

This study showed that the effects of PM pollution on cerebrovascular disease were different among different gender and age groups, and aged 61–75 years were mostly sensitive to particulate matters. The effects of PM_2.5 on cerebrovascular disease were stronger than those of PM₁₀. Our results can provide scientific evidence for the local government to take effective measures to improve air quality and the health of residents.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40201-021-00776-w.

Introduction

Numerous epidemiological studies have proven that atmospheric particulate matter (PM) is closely related to cardiovascular disease [16–18]. Cardiovascular and cerebrovascular diseases are among the leading causes of death and disability worldwide [16, 17, 47]. According to the latest report from the American Heart Association, nearly 17.3 million people die globally each year from cardiovascular disease, accounting for 30% of global mortality [7, 39].

Research has demonstrated that PM_2.5 exposure increases the risks of respiratory, cardiovascular, and cerebrovascular diseases and can even cause death [44]. Many studies [35, 52] related to PM exposure and cerebrovascular disease have been conducted in Taiwan and elsewhere, and obvious differences have been found in various regions. A statistically significant association was observed between daily mean PM_2.5 concentration and all-cause mortality in Madrid, Spain [26]. The relative risk (RR) of mortality due to circulatory disease was 1.088 (1.041–1.135) among people over the age of 75 years with a 25 μg/m³ increase in PM_2.5. However, no correlation was observed between PM₁₀ and mortality. A study conducted in Santiago, Chile also reported that the risk of hospital emergency room (ER) admissions for cerebrovascular causes increased by 1.29% [95% confidence interval (CI): 0.552%–2.03%] for every 10 μg/m³ increase in PM_2.5 concentration [37]. However, another study from Madrid found no statistically significant associations between PM_2.5 and other ischemic heart diseases or cerebrovascular disease [38]. A study conducted in Rome revealed a significant correlation between PM₁₀ and cerebrovascular disease and reported that Saharan dust also increased the effect of PM₁₀ on cerebrovascular disease [1].

Research in China has focused more on cardiovascular disease than on cerebrovascular disease. Many studies have demonstrated that exposure to increased concentrations of PM₁₀ and PM_2.5 increases the risk of death due to cardiovascular disease [36, 48]. A study conducted in Beijing reported that with every 10 μg/m³ increase in PM_2.5, the associated risks of death from cerebrovascular disease and from problems with the entire circulatory system increased by 0.75% and 1.38%, respectively [12]. However, studies have limited their focus to specific subgroups of cerebrovascular disease.

The seventh population census of China in 2020 showed that the proportion of people over 60 years has increased by 5.44%, indicating that the aging of the population has intensified. Cerebrovascular disease has a high prevalence, high disability and mortality rate, which does great harm to the elderly. In the current study, we used time-series analysis to evaluate the short-term effects of PM (PM₁₀ and PM_2.5) on the number of ER visits for cerebrovascular disease in Beijing from 2009 to 2012. Age- and sex-specific analyses were performed for different lag effects (1–6 days) of PM. Single-pollutant models and double-pollutant models were also developed to examine the adverse health effects of PM pollutants.

Data and methods

Study area

Beijing (39°54′N, 116°23′E) is the political and cultural center of China. Air pollution, particularly PM_2.5 exposure, causes a variety of health problems [19]. With the rapid development of society and the economy, the type of air pollution in Beijing has changed from a soot type to a composite type of soot and motor vehicle emissions. The aging of the urban population has made circulatory disease one of the main health concerns among residents.

Three large general top-level hospitals located in the metropolitan area of Beijing were selected for this study, and the daily mean PM₁₀ and PM_2.5 concentrations were obtained from the Beijing Environmental Monitoring Center, with included data recorded at eight fixed monitoring stations (Fig. 1). The eight monitoring stations are located at 39°56′N, 116°18′E; 40°1′N, 116°24′E; 39°56′N, 116°52′E; 39°48′N, 116°28′E; 39°52′N, 116°17′E; 40°N, 116°13′E; 39°6′N, 116°9′E and 39°44′N, 116°8′E, respectively. And the locations of the station comply with the Ambient Air Detection Code, and there is no pollution source around 50 m. Our meteorological elements data were sourced from the Beijing Meteorological Bureau.

Fig. 1.

Locations of the eight fixed air pollutant monitoring stations, the meteorological station, and the three hospitals in Beijing, China

Data collection

Daily meteorological data from January 1, 2009, to December 31, 2012, were collected from the Beijing Meteorological Bureau, including the daily minimum (maximum) temperatures (°C), daily average temperature (°C), relative humidity (%), daily average precipitation, average wind speed (m/s), daily maximum wind speed (m/s), maximum wind direction, and sunshine hours (h).

Daily air pollutant data were obtained from Beijing monitoring Center and consisted of daily PM_2.5 and PM₁₀ concentrations. We used daily average concentrations of eight fixed monitoring stations. The tapered element oscillating microbalance (TEOM) method was used to measure the 24-h mean concentrations of PM_2.5 and PM₁₀ (Franklin, MA, USA) [33]. The method of measuring according to the national standard, ensures the accuracy, continuity and integrity of the data. There were several zero number and missing value in particulate matter concentration. So we revised the zero number and supplemented the missing value by linear interpolation.

The number of ER visits for cerebrovascular disease was collected from three comprehensive hospitals in Beijing for a period spanning 2009 to 2012. We obtained information on 14,509 cases of cerebrovascular disease. The records were classified according to the disease classification criteria outlined in the 10th revision of the International Statistical Classification of Diseases and Related Health Problems, and information regarding patient visits for cerebrovascular disease to the ERs of three hospitals was summarized. Cases with missing information were removed. The patients admitted to the ER were divided into male and female groups as well as three age groups (15–60, 61–75, and > 75 years).

Statistical methods

A generalized additive model (GAM) was developed to evaluate the influences of PM pollution on ER visits for cerebrovascular disease. The usual form of the GAM model is as follows:

$$\mathrm{Log}\left[\mathrm{E},\left({\mathrm{Y}}_k\right)\right]=\alpha +DOW+{\beta X}_k+s\left(t,i,m,e,,,d,f\right)+s\left(Z_k,,,d,f\right)$$ 1

In Eq. (1), Y_K indicates the number of admissions on day k, E(Y_K) is the expected number of hospital visits on day k, α is the intercept, X_K denotes the concentration of pollutants on day k, s is a nonparametric spline-smoothing function, K is the number of people admitted on day k, DOW is the day of the week (a dummy variable), time is the calendar time, Z_K is element k of the meteorological elements, df is the degree of freedom, and β is the regression coefficient.

During the process of model establishment, the hysteresis effect of pollutants on cerebrovascular disease must be considered. We developed lag models for 1–6 days. After model establishment, the Akaike information criterion (AIC) was used to perform a simulated goodness-of-fit. Minimum principle of AIC was used to determine the df. By using the single-pollutant models, we determined the optimal lag time of the variables. PM₁₀ and PM_2.5 were added into the double-pollutant models according to the optimal lag time to evaluate the main risks affecting the number of PM pollutants on cerebrovascular disease.

According to the regression coefficient estimated using the GAM model, the relative risk (RR) of the natural logarithm of daily ER visits for cerebrovascular disease and the 95% confidential intervals (CIs) were calculated as a unit increase of pollutant concentration (interquartile range, IQR) to quantitatively evaluate the health effects of PM pollutants. All analyses were conducted in R 3.6.2 software (R Foundation for Statistical Computing, Vienna, Austria) via “mgcv” package.

$$RR=\exp (\beta \times IQR)$$ 2

$$\mathrm{CI}\left(95,\%\right)=\exp [(\beta \pm 1.96\times SE)\times IQR]$$ 3

Results

Table 1 presents the descriptive statistics for meteorological elements, PM concentrations, and ER visits. During the study period, the daily average number of ER visits for cerebrovascular disease was 10, and the highest number of ER visits per day was 29. The daily average PM₁₀ and PM_2.5 concentrations were 130.06 μg/m³ and 70.71 μg/m³, respectively. The average daily temperature and relative humidity were 13.08 °C and 50.54%, respectively, and the minimum (maximum) temperature was − 12.50 °C (34.50 °C), indicating that Beijing has a typical moderate continental monsoon climate.

Table 1.

Descriptive statistics of ER visits, PM concentrations, and meteorological elements

	Annual	Cold	Warm	Median	SD	Min	Max	P25	P50	P75
Number of daily ER visits for cerebrovascular disease	10	9	10	9	5	0	29	6	9	13
PM10 (μg/m3)	130.06	132.87	128.07	112.0	87. 7	0	563.3	67.0	112.0	172.77
PM2.5 (μg/m3)	70.71	74.16	68.27	58.0	56.8	0	381.6	28.3	58.0	98.03
Temperature (°C)	13.08	0.93	21.66	15.10	11.60	-12.50	34.50	1.70	15.10	24.00
Relative Humidity (%)	50.54	42.89	55.94	52.0	20.07	9.0	97.0	34.0	52.0	67.0
Pressure (hPa)	1012.32	1020.93	1006.23	1011.8	10.19	989.7	1037.3	1003.9	1011.8	1020.4
Velocity (m/s)	2.23	2.27	2.20	2.1	0.92	0.5	6.4	1.6	2.1	2.7

Figure 2 presents a time series of cerebrovascular disease and PM concentrations. From 2009 to 2012, the number of ER visits for cerebrovascular disease exhibited an increasing trend by year.

Fig. 2.

Time series of PM concentrations and number of ER visits for cerebrovascular disease

Figure 3 shows that both PM₁₀ and PM_2.5 concentrations were the highest in summer, and that the lowest in winter and spring, respectively.

Fig. 3.

Seasonal distribution of PM concentration (spring: March–May; summer: June–August; autumn: September–November; winter: December–February)

The correlation coefficient between cerebrovascular disease and PM₁₀ concentration was 0.046, which failed to reach statistical significance. The correlation coefficient between cerebrovascular disease and PM_2.5 concentration was 0.061 (p < 0.05). The same correlation coefficient was obtained for the correlation between cerebrovascular disease and temperature. Cerebrovascular disease was also positively correlated with relative humidity and negatively correlated with air pressure and wind speed (p > 0.05) (Table 2).

Table 2.

Results of Spearman correlation analysis of the association of ER visits for cerebrovascular disease with PM concentrations and meteorological factors

	Number of ER visits for cerebrovascular disease	PM10	PM2.5	Temperature	Relative Humidity	Pressure	Velocity
Number of ER visits for cerebrovascular disease	1.000
PM10	0.046	1.000
PM2.5	0.061*	0.879**	1.000
Temperature	0.061*	0.161**	0.140**	1.000
Relative Humidity	0.045	0.389**	0.520**	0.359**	1.000
Pressure	-0.020	-0.258**	-0.230**	-0.869**	-0.343**	1.000
Velocity	-0.043	-0.316**	-0.418**	0.012	-0.450**	-0.037	1.000

^* indicates p < 0.05, ^** indicates p < 0.01

The effect of PM concentrations on the number of ER visits exhibited a slight decreasing and then increasing trend. When the PM₁₀ concentration was higher than 200 μg/m³ or the PM_2.5 concentration was greater than 150 μg/m³, the logRR noticeably increased, indicating that the number of ER visits for cerebrovascular disease increased as the PM concentration increased (Fig. 4). We also attempted to add the influence of gaseous pollutants (SO₂ and NO₂) to the model, but the impact was extremely small (see Supporting Information for details).

Fig. 4.

Exposure–response relationship between PM (PM₁₀ and PM_2.5) concentration and number of ER admissions for cerebrovascular disease from 2009 to 2012 in Beijing

As displayed in Fig. 5, overall, the effect of PM₁₀ was only statistically significant at lag 3 (3 days), and the RR was 1.024, indicating that for every IQR increase in PM₁₀ concentration, the number of ER visits increased by 2.4%. PM_2.5 had a statistically significant effect from lag 1 to lag 3, with the RR reaching its maximum of 1.038 at lag 2 (i.e., the number of ER visits increased by 3.8% with an IQR increase in the PM_2.5 concentration). PM₁₀ and PM_2.5 had significant effects on male patients at lag 1, lag 2, and lag 4, and the maximum RR value of PM₁₀ for male patients was 1.044 (95% CI: 1.015, 1.073), observed at lag 1. The maximum RR value of PM_2.5 was 1.064 (95% CI: 1.022, 1.108), observed at lag 2. For female patients, the RR values were statistically significant at lag 3 and lag 6. The maximum RR value was observed at lag 6. For every IQR increase in PM₁₀ and PM_2.5 concentrations, the number of ER visits increased by 4.3% and 7.6%, respectively.

Fig. 5.

RR (95% CI) of ER visits of different groups in the single-pollutant models with an IQR increase in PM₁₀ and PM_2.5 concentrations

Table 3 presents the RR of ER visits according to the double-pollutant models with an IQR increase in PM₁₀ and PM_2.5 concentrations. The effects of PM₁₀ and PM_2.5 on patients aged 15–60 years were the greatest at lag 2 (p < 0.01). The corresponding RR values for PM₁₀ and PM_2.5 were 1.038 (95% CI: 1.003, 1.074) and 1.054 (95% CI: 1.001, 1.111), respectively. The greatest RR for the subgroup of patients aged 61–75 years was observed at lag 1, and each IQR increase in PM₁₀ and PM_2.5 concentrations corresponded to RR values of 1.054 (95% CI: 1.018, 1.092) and 1.096 (95% CI: 1.038, 1.157) respectively. For patients older than 75 years, the highest RR for PM₁₀ was 1.030 (95% CI: 0.995, 1.070), observed at lag 3, and that for PM_2.5 was 1.049 (95% CI: 0.993, 1.108), observed at lag 6. Generally, PM_2.5 had a greater impact than PM₁₀ did on the number of ER visits for cerebrovascular disease.

Table 3.

RR of ER visits in double-pollutant models with an IQR increase in PM₁₀ and PM_2.5 concentrations

	Pollutants (lag days)	Single—pollutant model	Double—pollutant model
Total	PM10 (3)	1.024 (1.004 ~ 1.045) *	1.017 (0.993 ~ 1.041)
	PM2.5 (2)	1.038 (1.005 ~ 1.072) *	1.025 (0.988 ~ 1.064)
Male	PM10 (1)	1.044 (1.015 ~ 1.073) *	1.031 (0.999 ~ 1.063)
	PM2.5 (2)	1.064 (1.022 ~ 1.108) *	1.045 (0.998 ~ 1.093)
female	PM10 (3)	1.041 (1.009 ~ 1.073) *	1.037 (1.005 ~ 1.069) *
	PM2.5 (6)	1.076 (1.028 ~ 1.127) *	1.070 (1.022 ~ 1.121) *
15-60yrs	PM10 (2)	1.038 (1.003 ~ 1.074) *	1.045 (0.967 ~ 1.1280)
	PM2.5 (2)	1.054 (1.001 ~ 1.111) *	0.991 (0.881 ~ 1.115)
61-75yrs	PM10 (1)	1.054 (1.018 ~ 1.092) *	0.985 (0.909 ~ 1.0660)
	PM2.5 (1)	1.096 (1.038 ~ 1.157) *	1.129 (0.998 ~ 1.2770)
> 75yrs	PM10 (3)	1.032 (0.995 ~ 1.070)	1.029 (0.992 ~ 1.068)
	PM2.5 (6)	1.049 (0.993 ~ 1.108)	1.045 (0.989 ~ 1.104)

^* indicates p < 0.05.

The optimal lag days in the single-pollutant models were used in the double-pollutant models. The RR values of the double-pollutant models were lower than those of the single-pollutant models. Of the double-pollutant models, the PM₁₀ model for patients aged 15–60 years and the PM_2.5 model for patients aged 61–75 years had higher RR values than the single-pollutant models did, with RR values of 1.045 (0.967–1.128) and 1.129 (0.998–1.277), respectively. The double-pollutant model for female patients produced the most significant results among all of the subgroups. For every IQR increase in PM₁₀ and PM_2.5 concentrations, the number of ER visits among female patients increased by 3.7% and 7.0%, respectively.

Discussion

In this study, the exposure–response relationship indicated that the RR increased significantly for PM₁₀ concentrations higher than 200 μg/m³. When the PM_2.5 concentration was greater than 150 μg/m³, the upward trend was even more pronounced. We investigated the effects of PM₁₀ and PM_2.5 exposure on the pathogenesis of cerebrovascular disease in different subgroups and discovered significant lag effects and specific differences. These findings are in contrast to those of some other studies that have reported no significant association between PM exposure and cerebrovascular disease. For example, a study conducted in Sweden demonstrated that PM exposure had no significant effect on cerebrovascular disease (RR = 0.97, 95% CI: 0.88, 1.07) [45]. Zanobetti et al. [53] even reported that in the United States, elevated concentrations of atmospheric pollutants caused a significant reduction in disease incidence. However, research conducted in some parts of North America has demonstrated that cardiovascular mortality is associated with environmental pollution in the long term, with an RR of 1.06 (95% CI: 1.00, 1.13) [15]. Studies in China [22] and South Korea [42] showed that exposure to fine particulate matter was associated with excess mortality from cerebrovascular diseases. A study in the United States discovered more significant associations between fine PM exposure and cerebrovascular disease than those reported in the present study (hazard ratio, 1.15,95% CI: 1.03, 1.28 [34]. Similar results were reported by Khaniabadi et al. [28] in western Iran and Fiorito et al. [16, 17] in Italy. Gu et al. [18] reported that in 248 Chinese cities, a 10 μg/m³ increase in PM_2.5 concentration was significantly associated with a 0.19% (95% CI: 0.13%, 0.25%) increase in the number of hospital admissions for cerebrovascular disease. However, a large case–crossover study discovered no correlation between changing PM levels and the number of hospital admissions for cerebrovascular disease in Edmonton, Canada [46]. PM may not contribute to an overall increase in cerebrovascular disease but may instead only trigger events in vulnerable populations [3]. Bates et al. [5] identified oxidative stress as a potential mechanism of action for PM toxicity. Although reactive oxygen species are generally continually produced in vivo, antioxidant enzymes regulate their production to remain below a certain level. However, when available in abundance, they have been shown to modify or degenerate biological macromolecules and induce cell dysfunction or death [27]. Many harmful substances such as inorganic compounds, organic polycyclic aromatic hydrocarbons, pathogenic microorganisms and heavy metals in particulate matter could be deposited in bronchioles and lungs [54]. Even smaller components could pass through the lung stroma and affect blood circulation, including counts of platelets, fibrinogen levels and red or white blood cells, which are risk factors for cerebrovascular disease [14, 49].

The results of sex-specific analysis in this study revealed that male patients were severely affected by PM exposure after 1–4 days, whereas women were affected after 3 and 6 days (p < 0.01). The RR values for female patients were significantly higher than those for male patients. In agreement with our study, a study conducted in Norway showed that for every increase of 10 ug/m³ in PM₁₀ concentration, the mortality rates of male and female patients with cardiovascular disease respectively increased by 1.09 (95% CI: 1.04, 1.15) and 1.11 (95% CI: 1.04, 1.19) [40]. Similar results were reported in the United States [52]. Studies have also indicated that female patients are more vulnerable than male patients are to PM_2.5, which is in agreement with the results of the present study [8, 35]. However, a retrospective cohort analysis in four cities in China revealed that the RR values of PM₁₀ for male patients were much higher than those for female patients [50]. Although few studies have explored differences between the sexes in terms of response to PM exposure, Kim and Hu [30] suggested that female and male patients differ with respect to their respiratory anatomy and physiology as well as the extent to which they are affected by lung particle deposition; they also reported that coarse particles in the proximal airway impose a greater burden on local tissue in women than in men during everyday activities, regardless of breathing patterns. The decrease of estrogen in females over 45 years old was not conducive to the metabolism of cholesterol and lipoprotein [4]. This might result in a higher incidence of respiratory irritation or illness in women than in men in environments with elevated levels of coarse particles [30].

Age-specific analysis indicated that the RR values of PM were the highest for individuals aged 61–75 years, followed by those aged 16–60 years and finally those older than 75 years. Kim’s single-pollutant (PM_2.5) model produced results indicating that in three cities in South Korea, the number of ER visits for cardiovascular disease was lower among individuals older than 65 years than among all age groups [32]. For cerebrovascular disease, the effect estimates concerning individuals in 248 Chinese cities were reported to be significantly higher in people aged 65–74 years than in other age groups [18]. This result is consistent with the findings of the current study. A multicity study conducted by Pascal et al. [43] in France suggested that patients aged 15–74 years had RR values that were much higher than those of patients older than 74 years. Studies in China [56] and Japan [51] also indicated that the elderly tended to have high effect estimates with gaseous pollutants. On this topic, one study reported no differences between age groups [50]. However, most relevant research has shown that older adults are more vulnerable to the harmful effects of pollutants than younger populations are because they may already have multiple comorbidities, such as atherosclerosis, which is often asymptomatic; moreover, PM exposure may exacerbate underlying ischemic heart disease, thereby triggering potentially fatal coronary events [6, 20]. The lower RR values for the oldest age group might be due to the decrease of respiratory rate and long-term indoor time in the elderly group [24].

A significant lag effect of PM on cerebrovascular disease was observed in this study. In the entire population, the effect of PM₁₀ on cerebrovascular disease was only significant at lag 3, with an RR value of 1.024. PM_2.5 exhibited a significant effect from lag 1 to lag 3, reaching its maximum RR value of 1.038 at lag 2. Similar to our study, a European study reported a 0.76% increase in cardiovascular death with every increase of 10 μg/m³ in PM₁₀ concentration at lag 0–1 [2, 31]. A study conducted in the Ina area of Japan showed that the optimal lag time for suspended PM for cerebral infarction was lag 2, which is consistent with the results of the current study [21]. PM_2.5 exhibited a more acute effect than PM₁₀, with a statistically significant effect from lag 1 to lag 3 days. In line with the results of our study, those of other studies have indicated that the effects of PM_2.5 begin at lag 0 in Taiwan [35]. This might be due to the enrichment of fine particles in the body, which increases the cumulative concentration. However, a study in Korea reported a longer lag effect of PM_2.5 on cardiovascular disease that began at lag 4 [32]. Our results indicate that PM_2.5 has a greater RR value for cerebrovascular disease than PM₁₀ does, which is consistent with the research results of Khosravi et al. [29] for a tourist city in the Middle East. The cumulative results of a study conducted in Vienna also indicated that the RR value of PM_2.5 was much greater than that of PM₁₀ [41], whereas a study conducted in Taiwan reported that the RR value of PM₁₀ was greater than that of PM_2.5 [10]. These findings may be attributable to the ability of PM_2.5 to penetrate the alveoli for direct entry into the body and the body’s subsequent inability to easily excrete it [55]. Compared with PM₁₀, PM_2.5 has smaller particle size and larger specific surface area, which makes it easier to absorb harmful substances [54]. Fine particles in the atmosphere can cause inflammatory reactions, fibrinolysis or coagulation dysfunction, endothelial dysfunction, and myocardial ischemia [55].

Various factors influence the health effects of PM. Table 4 presents a comparison of sex-specific, age-specific, and lag effects in different regions [6, 10, 20, 29, 40, 43]. Variations among observations made in different cities are to be expected, but the research results for each location still provide insight into the effects in the local area.

Table 4.

Comparisons of sex-specific, age-specific, and lag effects in different regions

Region	Male	Female	Age		Lag
Norway	1.09	1.11
France			Aged 15–74 1.039	Aged ≥ 75 1.011
China	1.31	1.10	Aged < 60 1.21	Aged ≥ 60 1.20	3
US	1.0043	1.0084	Aged 65–74 1.0055	Aged 75–84 1.0054	0
Australia	1.0099	1.0321	Aged 35–64 1.0026	Aged ≥ 65 1.0241	2
Iran	1.02	1.06	Aged 16–65 1.00	Aged > 65 1.02	4

We discovered that the RR values of the double-pollutant models were generally lower than those of the single-pollutant models, except for the PM₁₀ model for individuals aged 15–60 years and the PM_2.5 model for individuals aged 61–75 years. However, only the double-pollutant model for female patients produced statistically significant results, with RR values for PM₁₀ and PM_2.5 of 1.037 (1.005–1.069) and 1.070 (1.022–1.121), respectively. A study of two pollutants in the Netherlands reported that when ultrafine particles were paired with coarse PM, the hazard ratio for cerebrovascular disease decreased toward the null value for both pollutants, which is consistent with our results for cerebrovascular disease [13]. When other pollutants (NO₂, SO₂, CO, and O₃) were introduced into the model, PM_2.5 no longer exhibited any effect on mortality among residents with cardiovascular and cerebrovascular diseases [13]. A possible explanation for this result is that the presence of other confounding factors affects the interaction of air pollutants, resulting in a weakening and no longer significant relationship of PM concentration with death due to cardiovascular and cerebrovascular diseases [11].

In the current study, we identified a significant association between PM (PM_2.5 and PM₁₀) concentration and the number of ER visits for cerebrovascular disease. However, this study had certain limitations that should be addressed. First, we did not use O₃ data, meaning that the model did not consider the influence of ozone, although many reports have suggested that ozone has an effect on cerebrovascular disease. A study conducted in Canada reported that an increase of 10 ppb in ozone exposure was associated with an increase of 1.013 (95% CI 0.996, 1.03) to 1.058 (95% CI 1.034, 1.082) in the hazard ratio for cerebrovascular disease [9], but no statistically significant association was detected in a study from Prague, Czech Republic [23]. When O₃ is included in a GAM, the RR of cerebrovascular disease is decreased or remains the same [25, 48]. Therefore, the lack of O₃ data may have led to errors in the results.

Second, investigating the relationship between PM concentrations and the number of daily ER visits is difficult without information regarding the PM chemical composition. Third, this research only explored single-day lag effects. The multiday cumulative lag effect must be further investigated to reveal the cumulative effect of pollutants on health. Moreover, PM_2.5 and PM₁₀ are strongly correlated, and this knowledge may have led to selection bias in the double-pollutant models. Further research is necessary to develop multiple pollutant models and explore the cumulative lag effects of air pollution on cerebrovascular disease.

Conclusions

There was a significant relationship between particulate matters and cerebrovascular disease. PM exposure increased the risk of cerebrovascular disease, especially among individuals aged 61–75 years and female patients. The effects of PM_2.5 on cerebrovascular disease were found to be stronger than those of PM₁₀, and obvious lag effects were observed. Our results can provide scientific evidence for disease prevention and control. The local government must implement measures to reduce the impact of PM on human health.

Author contribution

Yuxia Ma designed and carried out the research; Jianding Zhou and Yifan Zhang set up models; Bowen Cheng and Fengliu Feng analyzed data; Hang Wang and Jiahui Shen collected and assembled data; Yan Chen collected data; Yuxia Ma and Jianding Zhou wrote the manuscript. Yuxia Ma and Bowen Cheng revised the manuscript.

Declarations

Conflict of interest

The authors declare they have no competing financial interests.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.