Spatial-Temporal Variation in Health Impact Attributable to PM_2.5 and Ozone Pollution in the Beijing Metropolitan Region of China

Abstract

This study aimed to estimate and compare the spatial-temporal variation in health impact attributable to PM_2.5, including the major particulate constituents and anthropogenic emission sectors of PM_2.5, and ozone in the Beijing–Tianjin–Hebei (BTH) region using monitoring data from 2013 to 2020. The liquid phase reaction may play an important role in PM_2.5 formation in winter. We estimated that 110,613 [(95% CI): 91,913, 128,615] and 9921 (95% CI: 3325, 13,191) cases of all-cause mortality in 2020 were attributable to exposure to PM_2.5 and ozone in the BTH region, respectively. The control of PM_2.5 pollution is currently a priority over that of ozone. An appropriate co-control policy for PM_2.5 and ozone pollution is necessary for the surrounding areas of Beijing City to protect public health. From 2013 to 2020, the mortality owing to exposure to PM_2.5 dropped significantly. The reduction in carbonaceous components in PM_2.5 can have the most effective health benefits. The top two contributing emission sectors to the mortality from PM_2.5 in Beijing were regional transportation and vehicles which could explain approximately 6.5% and 5.1% of the total mortality, respectively. The mortality owing to PM_2.5 was higher in Beijing than in Tokyo and Bangkok in East Asia.

1. Introduction

Ambient air pollution is a serious environmental risk factor and a leading contributor to mortality. In this study, we focused on two key indicators of ambient air pollution, namely PM_2.5 and ozone, as these are considered to be the most consistent and independent predictors of the health effects of ambient air pollution [1,2]. In 2015, ambient PM_2.5 and ozone were ranked 5th and 34th among risk factors for global deaths, respectively [2]. PM_2.5 has several adverse effects on human health, such as early death from heart or lung disease, non-fatal heart attack, arrhythmia, exacerbation of asthma, lung dysfunction, and increased respiratory symptoms. Ozone is a powerful oxidizing substance that stimulates the respiratory system. Ozone traps air in the alveoli, causing airway muscles to contract. This can lead to wheezing and shortness of breath. PM_2.5 consists of complex compositions, with the health risks varying among constituents, and different proportions of them in PM_2.5 may induce more or less PM_2.5-related effects. It has been reported that the anthropogenic compositions which mainly account for the mass concentration of PM_2.5 including organic carbon (OC) and elemental carbon (EC) and secondary aerosol constituents (SO₄²⁻ and NO₃⁻, etc.) have a positive association with human mortality [3]. The emitted amounts of PM_2.5 and its composition are different among various emission sectors. The magnitudes of the health risk effect largely vary by location due to the difference in the emission sources of PM_2.5 across regions. For policymakers, understanding the contribution of different emission sectors to health effects is needed for effective pollution mitigation efforts.

As reported before, premature mortality owing to PM_2.5 and ozone grew rapidly in Asia in recent years [4,5]. With rapid economic development and the increasing use of fossil fuels, China is facing serious air pollution, accompanied by severe health problems. In China, more than 1.3 billion people are at high risk attributable to PM_2.5, which exceeds the air quality guidelines of the World Health Organization [6]. The Global Burden of Disease (GBD) 2015 estimated that the number of deaths attributable to PM_2.5 exposure (57 µg/m³) in China was 1,139,724 in 2013 [7]. In early 2013, Beijing, Hebei, and other regions in China suffered from persistent haze, which had a serious effect on people’s health and on economic and social development. Public concern plays an important role in regulating social behavior by putting pressure on the government [8]. The same year, the Chinese government released the Action Plan for the Prevention and Control of Air Pollution to resolve this problem. Because of the control policy of PM_2.5 in recent years and also the influence of the COVID-19 pandemic since 2020, the emission from anthropogenic sectors of PM_2.5 is expected to be changed.

Most current studies on health impacts have focused on particulate matter pollution or ozone pollution at the national level [6,9], but limited studies have investigated the spatial distribution of air pollutant concentrations and their corresponding health effects at a regional level which limits the essential information available to local policymakers. In particular, few studies have presented the regional characteristics and driving factors of the spatial distribution of health impacts in China, such as meteorological conditions and population density distribution. Meteorological conditions play essential roles in modulating the atmospheric processes leading to distinct daily, seasonal, and inter-annual variations in air pollution. The changes in local meteorological conditions, such as temperature, humidity, wind, precipitation, radiation intensity, and so on, may serve as indispensable factors associated with the production and removal of pollutants from the atmosphere [10]. Zhang et al. reported that meteorological conditions will favor winter PM_2.5 and summer O₃ extremes under a high emission scenario in the North China Plain [11]. It has been reported that the severe winter haze during the 2013 winter in the Beijing area was driven by stable synoptic meteorological conditions over northeastern China and not by an abrupt increase in anthropogenic emissions [12]. It is important to analyze the effect of meteorological parameters on the time variation and spatial distribution of air pollutants. The particulate matter comes from both primary sources, such as combustion and dust-related sources, and also atmospheric reactions, namely secondary inorganic/organic particles. On the contrary, ozone is typically a secondary pollutant. The trend analysis and comparison of health impact attributable to PM_2.5 and ozone could provide essential information for co-control policy, which has not been studied sufficiently in East Asia. To date, there have been few studies of the links between compositions and source sectors of particulate matter and mortality in developing countries in Asia. In light of the severe pollution and the knowledge gaps mentioned above, in this study, we investigated and compared the annual trend of health impact attributable to the compositions and emission sectors of PM_2.5 in the representative cities in China; this scientific finding has not been achieved at the regional level in the Asian developing countries yet.

In view of the important role of prioritized control policy in mitigating the health risk related to PM_2.5 and ozone, we explored the temporal-spatial variation in PM_2.5 and ozone from 2013 to 2020 in the BTH region including Beijing City, Tianjin City, and Hebei Province using monitoring data to understand the status of regional air pollution and the influence of meteorological conditions and estimate the health impact attributable to PM_2.5 and ozone, also compared with that in the other representative regions in East Asia.

2. Materials and Methods

2.1. Study Area

The most polluted area with PM_2.5 in 2015 was located in northern China (Figure 1), where the BTH region is located in the capital metropolitan region of the country. The BTH region covers an area of 216,000 km² and had a population of 110 million in 2020. The BTH region is surrounded by mountains to the west (Taihang Mountains) and north (Yanshan Mountains), the Sea of Bohai to the east, and the Heibei Plain to the south. The BTH region has a typical continental climate with hot and rainy summers, cold and dry winters, and short springs and autumns.

The BTH region was designated as one of the key areas for air pollution prevention and control since the severe haze pollution occurred in early 2013 [13]. Diseases related to air pollution, including heart, cerebrovascular, and respiratory system diseases, were ranked second to fourth in death rate and composition from 10 major diseases in recent years [14].

2.2. Data Collection

PM_2.5, ozone exposure levels, and the number of premature deaths in the BTH metropolitan region were assessed from 2015 to 2020. The daily monitoring data of PM_2.5 and ozone were obtained from the website of the China Ministry of Ecology and Environment (http://datacenter.mee.gov.cn, accessed on 5 April 2021). There are 35, 15, and 53 monitoring sites in Beijing, Tianjin City, and Hebei Province, respectively. Significant agreement was obtained between daily PM_2.5 concentration data from the US Embassy in Beijing and the nearest monitoring site within the study periods (Figure S1). The methodology and QA/QC process for PM_2.5 and ozone measurement in the national monitoring sites are described in the technical manuals of PM_2.5 and ozone monitoring, respectively [15,16,17]. Data on population and local death rate (Table S1) were collected from the Statistical Year Book of China. Local baseline incidences (BIs) in the BTH region were used to reveal the health effects associated with PM_2.5 and ozone pollution. The incidence of mortality was collected from the 2019 Year Book of Health in the People’s Republic of China.

The daily data on temperature, relative humidity, wind direction, wind speed, air pressure, and sunshine hours from meteorological stations in the BTH region from 2015 to 2018 (the meteorological data in this period are available for this study) were collected from the China Meteorological Data Service Center (http://data.cma.cn, accessed on 6 April 2020). Only the dataset of air concentration and meteorological parameters during the simultaneous period (from 2015 to 2018) was used for the meteorological effect discussion based on the seasonal variation and correlation analysis in Section 3.2.

The results of main anthropogenic compositions (OC, EC, SO₄²⁻, and NO₃⁻) (Table S2) and source appointments of PM_2.5 in Beijing (Figure S3) were obtained from the published papers [18,19,20,21,22,23] and the report of Beijing Municipal Ecology and Environment Bureau (BMEEB), respectively. BMEEB conducted the project on source appointments of PM_2.5 in 2012–2013, 2017–2018, and 2020–2021. The results (Figure S3) present the change in emission amounts of source sectors of PM_2.5 in Beijing between 2017 and 2020.

For comparison with the results evaluated in the BTH region, this study also estimated the all-cause mortality in Tokyo, Japan, in 2015 and Bangkok, Thailand, in 2016 attributable to PM_2.5 including particulate compositions and emission sectors. Tokyo is a representative metropolitan city in a developed country in East Asia, and Bangkok is a typical metropolitan city in a developing country in Southeast Asia. The dataset of particulate compositions (OC, EC, SO₄²⁻, and NO₃⁻) and results of appointments in Tokyo and Bangkok were obtained from the report of the Ministry of Environment Japan and the report on “A Study in Urban Air Pollution Improvement in Asia” by the Japan International Cooperation Agency (JICA) Ogata Research Institute [24] and a related paper [25], respectively. The baseline incidence and population data, data source of particulate compositions, and emission sectors are shown in Tables S1 and S2, and Figure S3, respectively.

2.3. Health Impact Assessment

As recommended by the WHO and other studies [26,27,28,29], the linear-log function was used to estimate the all-cause mortality attributable to exposure to PM_2.5 using the following equation:

I = I₀ × exp {β × (C − C₀)}

(1)

where I is the mortality rate and I₀ is the mortality at C₀, C is the annual mean concentration of PM_2.5 and particulate constituents, and C₀ is the threshold concentration. Threshold concentration refers to that below which PM_2.5 had no adverse effect on mortality (in this study, we assumed no threshold for the health impact from PM_2.5 or particulate compositions). The unit-less exposure–response coefficient β is derived from the relative risk (RR) value reported in epidemiological studies, which indicates the proportion of the health effects of change per unit of PM_2.5 [28,29] (Table S3). The annual mean concentration of PM_2.5 was selected as the reference concentration.

The number of all-cause mortality was calculated using the following equation:

Y = E₀ × {(I − I₀)/I} × P = E₀ × {1 − exp [−β × (C − C₀)]} × P

(2)

where Y is the number of deaths caused by PM_2.5, E₀ (‰) is the baseline incidence rate, and P is the population.

By eliminating (or restricting) the emission sectors, many deaths could be avoided [9]. In this study, sector-specific mortality is used to estimate the human health benefit. The sector-specific mortality method first calculates the fractional PM_2.5 concentrations from an emission sector (C_SECTOR), which is equal to the fraction contribution of this sector (Figure S3) multiplied by the PM_2.5 concentration, and then uses this fraction to scale the total mortality estimate.

C_SECTOR = C × Fraction of an emission sector (%)

(3)

Y_SECTOR = E₀ × {1 − exp [−β × (C_SECTOR − C₀)]} × P

(4)

Y_SECTOR is the mortality owing to a specific sector, and C_SECTOR is the concentration of PM_2.5 emitted from a specific emission sector.

To estimate the number of deaths attributable to annual ozone levels exceeding the reference level, we used the following equation:

$$Mortalit{y}_{O_3}\left(i,j,n\right)=Y_0\left(i,j,n\right)\left\{1-exp\left[-{\beta }_{O_3}\Delta O_3\left(i,j,n\right)\right]\right\}$$

(5)

$$\Delta O_3\left(i,j,n\right)={\left[\max 8 h mean-35\right]}_n$$

(6)

where n is the calculation day, Y₀ is the daily incidence of premature mortality at a certain ozone level where there is no clear health effect likely to occur (in this study, it is estimated by multiplying the population by the daily baseline mortality), β is estimated based on an RR value, and ΔO₃ is the change in ozone concentration calculated based on the daily maximum 8 h mean concentrations above 35 ppb [30].

The annual average concentration of PM_2.5 and the 8 h maximum mean concentration of ozone were input into the ArcGIS system to generate health risk values using the Kriging method and then illustrate the geographical distribution of mortality. The ArcGIS was used to generate the spatial distribution of the concentration of PM_2.5 in China, the concentration of PM_2.5 and ozone in the BTH region, and the all-cause mortality attributable to PM_2.5 and ozone in the BTH region. The population distribution in the BTH region was obtained from the Gridded Population of the World (https://sedac.ciesin.columbia.edu/data/set/gpw-v4-admin-unit-center-points-population-estimates-rev11, accessed on 1 June 2020) [31]; the size of the population grid cell was approximately 0.04167 [4,5].

Uncertainty in the assessment of the mortality effects of ambient air pollutants often lies in the exposure assessment and mortality baseline [6]. Because there is no full geographical coverage of the ground monitors, it is impossible to be certain that the estimated exposure coincides with the actual ambient concentrations in a given location. Gridded population data were used in this study, and a lack of discussion on the dynamic changes in population distribution may lead to low assessment accuracy [26]. Even if population exposure is well estimated, individual exposure can vary substantially as a result of differences in concentrations at different places as well as individual activity patterns. Personal monitoring is generally necessary to assess individual-level risk. Additionally, most CR functions do not consider the influence of co-existing air pollutants such as PM_2.5 and ozone. There is a lack of direct epidemiological evidence in China. Studies are urgently needed in heavy pollution regions because the health response per unit change in air pollution at such high levels may differ from that seen in countries with lower pollution levels.

3. Results and Discussion

3.1. Spatial and Temporal Variation in PM_2.5 and Ozone Concentrations

The temporal trends of PM_2.5 and ozone in the BTH region were analyzed using the Mann–Kendall method. From 2013 to 2020, the PM_2.5 concentration shows a significant declining trend in Beijing City (S = −8.0 μg/m³/yr, p < 0.005), Tianjin City (S = −6.6 μg/m³/yr, p < 0.005), and Hebei Province (S = −8.6 μg/m³/yr, p < 0.01) and almost achieved the National Standard Grade II in 2020. From 2013 to 2020, the concentration of PM_2.5 dropped by 57% from 89 to 38 μg/m³. It has been reported that the concentration of PM_2.5 in the whole of China decreased by 32% from 2013 to 2017 [26]. From 2013 to 2015, the ozone concentration increased in Beijing and then showed a slightly decreasing trend (S = −4.0 μg/m³/yr, p < 0.01). From 2013 to 2020, the ozone concentration decreased by 7% in Beijing City; on the contrary, the ozone concentration increased by 26% and 4% in Tianjin City and Hebei Province, respectively. The ozone concentration was higher than the standard grade II from 2017 onward in the BTH region (Figure 2).

The spatial distributions of PM_2.5 and ozone in the BTH region in 2018 are shown in Figure 3. More serious PM_2.5 pollution was observed in the southern areas of the BTH region. The distribution of PM_2.5 concentration was consistent with that of a recent study [26]. Using the national monitoring data, this study presented the spatial distribution of PM_2.5 concentration in China in 2015 (Figure 1), which shows the most severe PM_2.5 pollution was observed in the Center-East region. In this region, the PM_2.5 pollution was higher in Henan, Shanxi, Shandong, and Hebei Provinces than in Beijing–Tianjin City. Due to the industrial and energy structures, the emissions of different air pollutants vary very much among cities. Figure S2 shows the amount of PM_2.5 emissions in the Center-East region in China [32]. Beijing and Tianjin are the largest metropolitan cities, where heavy industries are no more the major sectors of the economy. The PM_2.5 emissions were lower in both cities than in other cities. Tangshan City of Hebei Province, the largest heavy industrial city in the region, has the country’s largest iron and steel production capacity and the largest PM_2.5 emissions (208,463 t/yr), which were significantly higher than the emissions of other cities in the region [32]. Cangzhou City of Hebei Province also has high PM_2.5 emissions of 85,407 t/yr [32]. The much larger emissions of major air pollutants in Hebei Province can result in higher PM_2.5 concentrations observed in Hebei Province than in Beijing and Tianjin City.

A high concentration of ozone was also found in the southern area of Hebei Province and Tianjin City. In contrast, the ozone concentration in downtown Beijing was lower than that in the surrounding suburban areas, such as the western area of Hebei Province and Tianjin City. In the atmosphere, ozone can oxidize NO into NO₂ (O₃ + NO→O₂ + NO₂), and the low concentration of ozone in the urban area of Beijing City could result from significant NO emissions from vehicle exhaust.

3.2. Relationship between Air Pollutants and Meteorological Parameters

The PM_2.5 concentration was higher in winter and lower in summer in the BTH region (Figure 4). Overall, the PM_2.5 concentrations showed inverse variation with temporal changes in wind speed, ambient temperature, sunshine hours, and precipitation. However, ozone seasonal variation showed higher concentrations during summer when the temperature, humidity, rainfall, and wind speed were higher than in winter, probably because of the more intensive photochemical reactions in summer than in other months.

As shown above, the highest concentrations of PM_2.5 and ozone were observed in winter and summer, respectively; therefore, the effects of meteorological parameters on PM_2.5 and ozone pollution in December (

Table 1. Spearman correlation among daily means of PM_2.5, ozone, and meteorological parameters in December from 2015 to 2018.

	PM2.5	Wind Speed	Relative Humidity	Air Temperature	Sunshine Hours	Air Pressure	Ozone
PM2.5	1	−0.578 **	0.823 **	0.255 **	−0.560 **	−0.402 **	−0.538 **
Wind speed		1	−0.620 **	0.051	0.462 **	0.172	0.713 **
Relative humidity			1	0.17	−0.672 **	−0.366 **	−0.672 **
Air temperature				1	0.013	−0.641 **	0.119
Sunshine hours					1	0.218 *	0.655 **
Air pressure						1	0.195 *
Ozone							1

** p < 0.01; * p < 0.05.

) and August (

Table 2. Spearman correlation among daily means of PM_2.5, ozone, and meteorological parameters in August from 2015 to 2018.

	PM2.5	Wind Speed	Relative Humidity	Air Temperature	Sunshine Hours	Air Pressure	Ozone
PM2.5	1	−0.245 **	0.424 **	0.235 **	−0.236 **	−0.196 *	0.474 **
Wind speed		1	−0.134	0.015	0.079	−0.003	−0.161
Relative humidity			1	−0.243 **	−0.676 **	−0.124	−0.189 *
Air temperature				1	0.314 **	−0.461 **	0.549 **
Sunshine hours					1	−0.017	0.449 **
Air pressure						1	−0.155
Ozone							1

** p < 0.01; * p < 0.05.

) are discussed in more detail as follows. The distributions of the wind direction and speed in December and August 2018 are shown in Figure 5. In December, the highest correlation was observed between the PM_2.5 concentrations and relative humidity, with a correlation coefficient of 0.83 (p < 0.01), whereas it was negatively associated with wind speed and sunshine hours, with a correlation coefficient of −0.45 (p < 0.01) and −0.58 (p < 0.01), respectively (Table 2). This suggests that higher humidity may play an important role in PM_2.5 formation (haze pollution) during winter in Beijing, and the wind has a significant dilution effect on PM_2.5. The severe air pollution in the winter season was due to the high emission of PM_2.5 from coal combustion for heating and unfavorable meteorological conditions. It is easy to cause accumulation of local primary pollutants and chemical formation of secondary pollutants in winter under unfavorable meteorological conditions such as small near-surface wind speed, higher relative humidity, near-surface inversion, and lower boundary layer height. It has been reported that the occurrence of the severe winter haze during 2012–2013 resulted from stable synoptic meteorological conditions over a large area of northeastern China [12]. In contrast, the ozone concentrations in December were positively associated with wind speed and sunshine hours, with correlation coefficients of 0.70 (p < 0.01) and 0.62 (p < 0.01), respectively, but negatively associated with relative humidity (correlation coefficient, −0.65 (p < 0.01)). The inverse correlation of PM_2.5 and ozone with relative humidity, wind speed, and sunshine hours suggests different effects of atmospheric reaction and transportation on PM_2.5 and ozone pollution. Atmospheric gaseous reactions, which account for the secondary formation of ozone, are positively affected by longer sunshine hours (higher air temperatures) and lower humidity. Contrarily, higher humidity plays a more important role in the liquid phase reaction for the formation of PM_2.5 (haze pollution) in winter. Song et al. [33] reported that the heterogeneous production of hydroxymethanesulfonate (HMS) by SO₂ and formaldehyde is favored under winter haze conditions in northern China owing to high aerosol water content. The wind had a clear dilution effect on PM_2.5 but resulted in an increase in the ozone concentration in winter. In December, the predominant wind direction at high speeds in the BTH region was from north to south (Figure 5). This indicates that the wind in winter diluted the PM_2.5 concentration in Beijing but transported the PM_2.5 pollution to the southern area of Hebei Province, where the PM_2.5 concentration was highest in the BTH region.

In August, the highest correlation was observed between ozone and ambient air temperature and sunshine hours with correlation coefficients of 0.52 (p < 0.01) and 0.47 (p < 0.01), respectively. This indicates an enhanced contribution of photochemical reactions to ozone formation in the summer. Overall, PM_2.5 concentrations were weakly correlated with meteorological parameters in August. In this month, the wind speed was low, and the main wind direction was from east to west in the southern area of the BTH region (Figure 5). Wind did not cause ozone pollution from Beijing and Tianjin to the southern area of Hebei, and the high concentration of ozone in the southern area of BTH mainly resulted from local contributions.

3.3. Health Impact Attributable to PM_2.5 and Ozone in the BTH Region

3.3.1. Heath Impact Attributable to the Concentration of PM_2.5 and Ozone

The estimated number of all-cause deaths associated with exposure to PM_2.5 and ozone and the corresponding confidence interval from 2013 to 2020 in the BTH region are shown in Figure 6. From 2013 to 2020, a significant declining trend of mortality due to PM_2.5 was observed in Beijing City (Sen’s Slop = −2397, p < 0.01), Tianjin City (Sen’s Slop = −1895, p < 0.01), and Hebei Province (Sen’s Slop = −11,653, p < 0.01). In the BTH region, a significant increasing trend of mortality attributed to ozone was found in Tianjin City (Sen’s Slop = 67 p < 0.05). The trend of mortality attributable to ozone in Beijing City and Hebei Province was not clear.

We estimated that 110,613 [(95% CI): 91,913, 128,615] and 9921 (95% CI: 3325, 13,191) cases of all-cause mortality in 2020 were attributable to exposure to PM_2.5 and ozone, respectively, in the whole BTH region. Although the health effect owing to PM_2.5 exposure decreased significantly from 2013 to 2020, it was still more serious than that of ozone. As seen in other health impact assessments in China [34], the present study also shows that the effect of PM_2.5 on human health was more than 10 times higher than the effect of ozone, and thus the control of PM_2.5 pollution is still a priority over ozone in the BTH region. As discussed before, it is worth noting that the PM_2.5 concentration in winter mainly accounted for the annual PM_2.5 pollution, and thus the control of PM_2.5 in Beijing during winter has a significant effect on mitigating the corresponding health risk. Our results show that the number of all-cause deaths from 2013 to 2020 owing to exposure to PM_2.5 dropped by 51%, 47%, and 62% in Beijing, Tianjin, and Hebei Province, respectively. The all-cause mortality attributable to ozone exposure decreased by 7% and 10% from 2013 to 2020 in Beijing City and Hebei Province, respectively, but increased by 43% in Tianjin City (Figure 6). The increase in mortality owing to ozone in Tianjin city is probably due to the increase in ozone concentrations during the study period. The ratio of mortality caused by PM_2.5 to the mortality attributable to ozone shows a significant declining trend in the BTH region (Figure 6). In 2013, this ratio observed in Tianjin City and Hebei Province was approximately 1.7 times higher than the ratio observed in Beijing. On the contrary, from 2017, the relative stranded deviation of this ratio in the BTH region became smaller to approximately 9%. This suggests that the relative contribution to health risks owing to ozone pollution becomes more significant and especially more attention is required in Tianjin City and Hebei Province. An appropriate co-control policy for PM_2.5 and ozone pollution in the surrounding area of Beijing is necessary to protect public health. From the perspective of health impact, the policy of emission control has different effects on the mitigation of health risks owing to PM_2.5 and ozone pollution, because the sources and atmospheric behavior of each are quite different.

The relative contribution of mortality owing to PM_2.5 and ozone exposure was estimated based on the number of deaths in the BTH region as follows. On average, between 2013 and 2020, the fraction of all-cause mortality owing to PM_2.5 exposure contributed to the number of deaths in Beijing City, Tianjin City, and Hebei Province was 21.7%, 22.2%, and 26.8%, respectively. The proportions of total mortality attributable to ozone pollution in Beijing City, Tianjin City, and Hebei Province were 1.8%, 1.5%, and 1.6%, respectively. From 2013 to 2020, the proportion of all-cause mortality owing to PM_2.5 exposure decreased from 28.8% to 13.5% in Beijing City, from 30.6% to 16.7% in Tianjin City, and from 44.3% to 19.8% in Hebei Province. From 2013 to 2020, the proportion of all-cause mortality attributable to ozone exposure changed from 1.8% to 1.5%, 1.2% to 1.8%, and 1.5% to 1.6% in Beijing City, Tianjin City, and Hebei Province, respectively.

To compare with the health impact owing to PM_2.5 in Beijing City, this study also estimated the mortality due to PM_2.5 was 6612 (95% CI: 5423–7788) in Tokyo, Japan, in 2015 and 3649 (95% CI: 3000–4287) in Bangkok, Thailand, in 2016. The contribution of the all-cause mortality owing to PM_2.5 to the total death number was 5.9% in Tokyo and 8.3% in Bangkok, and those ratios were approximately four to three times lower than the ratios observed in Beijing. The control of PM_2.5 has important meaning for human health benefit in China.

The spatial distribution shows that the all-cause mortality owing to exposure to both PM_2.5 and ozone was higher in the urban areas of Beijing City and Tianjin City and the southern area of Hebei Province (Figure 7). The all-cause mortality from PM_2.5 and ozone in Tianjin City and Hebei Province showed similar spatial distributions compared with the pattern of PM_2.5 and ozone concentrations. The high all-cause mortality in Tianjin City and Hebei Province was dependent on both the high population density and higher concentrations of PM_2.5 and ozone. Although lower concentrations of both PM_2.5 and ozone were observed in the downtown area of Beijing City (Figure 3), higher all-cause mortality owing to both PM_2.5 and ozone was found in the downtown area of Beijing (Figure 7), where the population density was also high. The spatial distribution of all-cause mortality caused by exposure to both PM_2.5 and ozone in Beijing City was more influenced by the distribution of population density. Because the predominant wind direction during winter was from north to south, the control of PM_2.5 pollution in Beijing City during winter has an important effect on the mitigation of pollution and corresponding health risks in Hebei Province.

3.3.2. Health Effect Attributable to Main Compositions in PM_2.5

The mortality attributable to PM_2.5 compositions is shown in Figure 8. Although the mortality due to PM_2.5 in Beijing decreased significantly from 2013 to 2020, the mortality due to compositions in PM_2.5 showed a different change. The mortality due to OC in PM_2.5 decreased from 2014, and the trend of mortality attributable to OC was comparable to that of PM_2.5. On the contrary, the mortality due to EC in PM_2.5 did not show a clear trend. The mortality due to SO₄²⁻ and NO₃⁻ in PM_2.5 increased from 2016.

In estimating organic matter (OM), an OC to OM conversion factor of 1.6 was adopted for the aerosols [35]. On average, from 2016 to 2019, the fractional contribution of main components to PM_2.5 mass gave the order of OCM (28%, OCM = OC × 1.6), NO₃⁻ (30%), SO₄²⁻ (21%), and EC (5%). On the contrary, the ratio of mortality due to PM_2.5 compositions to the mortality owing to PM_2.5 showed the ranking of OC (10.4%), EC (8.7%), NO₃⁻ (6.0%), and SO₄²⁻ (4.8%). The mortality due to EC (1780, 95% CI: 680–2868) was 76% of that of OC (2294, 95% CI: 803–3766) and 1.4–1.8 times higher than that due to NO₃⁻ (1230, 95% CI: 148–2583) and SO₄²⁻ (1012, 95% CI: 517–1502). Although the fractional contribution of EC to mass was the smallest, the contribution of EC to the mortality was second highest. The human health effect attributable to EC in PM_2.5 is much more sensitive than OC, SO₄²⁻, and NO₃⁻ in PM_2.5; therefore, the regulations on the reduction of EC emission have the most significant effect on mitigating the health risk.

For comparison, this study estimated that the number of mortality cases due to OC, EC, SO₄²⁻, and NO₃⁻ was 556 (95% CI: 193–916), 796 (95% CI: 303–1287), 281 (95% CI: 143–418), and 152 (95% CI: 18–323), respectively, in Tokyo in 2015. The mortality due to EC was 1.4 times higher than that due to OC and three to five times higher than that of SO₄²⁻ and NO₃⁻ in Tokyo. This study estimated that the mortality due to EC (870, 95% CI: 332–1400) in Bangkok was nearly two times higher than that of OC (465, 95% CI: 162–765), followed by SO₄²⁻ (67, 95% CI: 34–100) and NO₃⁻ (11, 95% CI: 1–22). These results indicate that the particulate EC has a more important health effect than OC, SO₄²⁻, and NO₃⁻ in Tokyo and Bangkok, and on the contrary, OC has the highest contribution to mortality in Beijing. The ratio of mortality due to OC to the total mortality showed the ranking of Beijing (2.3%), Bangkok (1.1%), and Tokyo (0.5%). The ratio of EC gave the order of Beijing (2.0%), Bangkok (2.0%), and Tokyo (0.7%). The ratio of SO₄²⁻ followed the ranking of Beijing (0.8%), Tokyo (0.3%), and Bangkok (0.2%). The ratio of NO₃⁻ followed the rank of Beijing (0.9%), Tokyo (0.1%), and Bangkok (0.02%). The relative contribution of mortality due to the main particulate constituents to the total death number in Beijing was higher than in the other metropolitan regions in East Asia (Figure S4).

In this study, the all-cause mortality attributable to the sum of major compositions in PM_2.5 was smaller than the mortality attributable to PM_2.5 mass. One possible reason is that there might be other compositions whose health impact was not included such as PAHs which are typical carcinogenic substances. Another possible reason might be the underestimate of relative risk values of particulate components which may also result in the smaller values of mortality attributable to compositions in PM_2.5. This should be investigated in future studies.

3.3.3. Health Effect Owing to Emission Sectors of PM_2.5 in Beijing

There are many complex sources of particulate air pollution including human activities (transportation, biomass open burning, fossil fuel combustion, industrial activity and residential combustion, etc.) and natural sources (forest fire, sea salt, etc.). Figure 9 shows the comparison of the mortality owing to PM_2.5 (left axis, column) and the ratio of the mortality owing to each source sector of PM_2.5 to the total death number (right y-axis, symbol) in 2017 and 2020. The mortality attributable to PM_2.5 from each sector clearly decreased, especially from the combustion source. All of the ratios from each sector decreased significantly from 2017 to 2020.

The most important source sectors that contributed to the mortality owing to PM_2.5 were regional transportation, vehicles, and dust-related, so the control of regional PM_2.5 has important meaning for human health in Beijing. Regarding the average of results in 2017 and 2020, around 6.5% of total mortality in Beijing was from regional transportation (7029, 95% CI: 5770–8272). If the vehicles and dust-related sectors are removed, the number of averted average all-cause deaths is estimated as 5599 (95% CI: 4592–6597) from vehicles and 1769 (95% CI: 1446–2091) from the dust-related sector, respectively. The numbers of the averted vehicle sector and coal combustion could account for 5.1% and 0.4% of the total death number in Beijing. Considering the local emission, control of vehicles has the most important effect on the mortality attributable to PM_2.5 in Beijing. Because of the strict control of coal combustion for heating in the north of China, the contribution of coal combustion becomes less important to the mortality due to PM_2.5. This study also reported the number of deaths attributable to PM_2.5 from vehicles was 664 (95% CI: 542–786) in Tokyo in 2015 and 1061 (95% CI: 867–1253) in Bangkok in 2016. Compared with Tokyo and Bangkok, the ratio of mortality due to PM_2.5 from vehicle exhaust to the total death number was highest in Beijing (5.1%), followed by Bangkok (2.4%) and Tokyo (0.6%). The PM_2.5 from vehicle exhaust resulted in higher mortality in Beijing than in the other two metropolitan regions in East Asia.

4. Conclusions and Policy Implications

In summary, we analyzed the temporal trends of PM_2.5 and ozone in the BTH region in northeast China using monitoring data from 2013 to 2020. The concentration of PM_2.5, decreased by approximately 57% in the BTH region from 2013 to 2020, and the concentration of PM_2.5 in Beijing almost achieved the national Standard Grade II in 2020. However, the ozone concentration did not show a clear trend in this region and was higher than the standard grade II since 2017.

More serious PM_2.5 and ozone pollution was observed in the southern area of Hebei Province. In winter, the predominant wind direction from north to south in the BTH region with high wind speed diluted the PM_2.5 concentration in Beijing City but transported PM_2.5 pollution to the southern area of the BTH region. In winter, the control of PM_2.5, emission sources in the urban area of Beijing City may have a significant effect on the reduction in PM_2.5 pollution in the southern suburb region of BTH, consequently mitigating the corresponding health risks. The inverse correlation of PM_2.5 and ozone with meteorological parameters suggest that atmospheric gaseous reactions, which mainly account for the secondary formation of ozone in summer and the liquid phase reaction owing to higher humidity, play an important role in the formation of PM_2.5 or named haze pollution in winter.

The all-cause mortality owing to exposure to PM_2.5 and ozone showed a similar pattern of spatial distribution in the BTH region. Higher all-cause mortality owing to both PM_2.5 and ozone was found in the downtown area of Beijing City, where the population density was also high. The spatial distribution of all-cause mortality caused by exposure to both PM_2.5, and ozone in Beijing City was more significantly affected by the distribution of population density. In contrast, the spatial distribution of all-cause mortality in Tianjin City and Hebei Province was affected by the distribution of both population and atmospheric concentrations of PM_2.5 and ozone.

Overall, all-cause mortality owing to PM_2.5 was significantly higher than that owing to ozone. On average, from 2013 to 2020, all-cause mortality owing to PM_2.5 and ozone exposure accounted for 23.6% and 1.6% of the total number of deaths, respectively, in the BTH region. In view of the health impact, the control of PM_2.5 pollution still has priority over ozone in the present situation. The proportion of all-cause mortality owing to PM_2.5 exposure to the number of deaths was highest in Hebei Province (26.8%) and the proportion of ozone was highest in Beijing City (1.8%). Our results show that the number of all-cause deaths from 2013 to 2020 owing to PM_2.5 dropped by 51%, 47%, and 62% in Beijing City, Tianjin City, and Hebei Province, respectively. On the contrary, from 2013 to 2020 the all-cause mortality attributable to ozone only slightly decreased by 7% in Beijing City and 10% in Hebei Province but increased by 43% in Tianjin City. Although the mortality owing to PM_2.5 exposure decreased significantly during the study period, it was still more serious than that of ozone. The proportion of all-cause mortality owing to ozone pollution in the BTH region has not decreased significantly in recent years. The average ratio of mortality caused by PM_2.5 to the mortality attributable to ozone from 2013 to 2020 was 11.8, 15.6, and 17.8 in Beijing City, Tianjin City, and Hebei Province, respectively, and the ratios show a significant declining trend in the BTH region. This suggests that the health risks owing to ozone pollution, especially in Tianjin City and Hebei Province, need to gain more attention. An appropriate co-control policy for PM_2.5 and ozone pollution is necessary for the surrounding areas of Beijing City to protect public health.

The control of the carbonaceous components in PM_2.5 has important meaning for human health benefit. Although the contribution of EC to the particulate mass was much smaller than that of ions, the mortality due to EC was much higher than ions. The human health effect attributable to EC in PM_2.5 is more sensitive than OC, SO₄²⁻, and NO₃⁻ in PM_2.5, and, therefore, the control policy of EC emission has the most significant effect on mitigating the health risk. The major contributing emission sector to the mortality of PM_2.5 in Beijing was regional transportation which could explain approximately 6.5% of the total mortality in Beijing. The vehicle sector was the most important local source which accounted for 5.1% of total mortality in Beijing. The contribution of coal combustion becomes less important to the mortality due to PM_2.5 in Beijing. The relative contribution from mortality due to PM_2.5 and the main particulate constituents to the total death number in Beijing was higher than in Tokyo and Bangkok, and the PM_2.5 from vehicle exhaust resulted in higher mortality in Beijing than in the above two metropolitan regions in East Asia.