With urbanization and modernization, China has become one of the countries suffering from the most serious air pollution (1, 2). Air pollution has become the fourth leading cause of mortality in China, and it has been related to an increased chance of premature death due to heart disease, stroke, and lung cancer (3–5). The severity of air pollution in China is evident from the frequent occurrence of haze events in recent years, which has raised significant concern (6). Since the beginning of 2013, the incidence and extent of haze have quickly grown with 75% of cities and 8 million people suffering from haze pollution in China, which presents a significant challenge to public health (7).

In response, China has implemented policies such as the Air Pollution Prevention and Control Action Plan (APPCAP) to enhance air quality and lessen the harmful impacts of pollution on human health (8). However, the air quality in Beijing remains critical, with average pollutant concentrations exceeding the World Health Organization's (WHO) targets (9). Recently, a growing number of epidemiological studies have been done in China to gain a comprehensive understanding of the short-term impacts of air pollution on non-accidental mortality (10, 11). Air pollution in China is complexly characterized by high levels of PM_2.5 and O₃ (12). A recent meta-analysis in China confirmed that short-term exposure to O₃ was associated with an elevated risk of cardiovascular mortality (13). Significant links were found between short-term coarse particulate matter exposure and daily non-accidental mortality in a nationwide analysis of 272 Chinese cities (14). However, the findings on gaseous pollutants (SO₂ and NO₂) were not entirely consistent across several studies. For instance, the time-series study conducted in Hefei City showed SO₂ was linked with non-accidental mortality (15), while studies in Northern China and the Pearl River Delta region reported no association between SO₂ and mortality (16, 17). Furthermore, the studies from Hong Kong and Taipei City reported a significant increase in respiratory disease mortality associated with short-term NO₂ exposure (18, 19), while another study conducted in Beijing reported a positive but non-significant association (20). It is noteworthy that very few studies have analyzed directly the association between non-accidental deaths and short-term exposure to air pollutants, where both particulate matter and gaseous pollutants are involved (21, 22). Therefore, it is necessary to conduct more research to better understand the relationship between air pollution and non-accidental mortality in China.

Previous studies have predominantly relied on hospitalization data as the primary data source to investigate the association between air pollution and non-accidental mortality, with limited utilization of emergency department data (23, 24). The study has the advantage of providing the evidence of the association between air pollutants (particulate matter and gaseous pollutants) and emergency non-accidental mortality in Beijing. First of all, emergency data provides faster feedback because emergency services typically respond immediately and record relevant data. This allows us to observe acute health events and mortality cases related to air pollution more promptly. Secondly, since emergency services are accessible to the public and people are more likely to seek emergency assistance during acute conditions, these data can cover a broader range of age groups and populations. Some studies have shown that it is significant to examine emergency department data specifically, as it represents more immediate, acute impacts and may identify vulnerability in specific populations (25, 26). Finally, emergency department data can avoid the interference of cross-regional visits when compared with hospitalization data, and brings the advantage of exposure assessment (27).

As the capital of China, Beijing is situated in northern China, with high population density, high levels of industrialization, long heating hours, and high air pollution levels (28). Therefore, it becomes very important to develop related preventive strategies to protect public health, especially sensitive populations, by estimating the acute health impacts of air pollution. This study performed a time-series analysis to investigate the short-term effects of six air pollutants (PM_2.5, PM₁₀, SO₂, NO₂, O₃, and CO) on non-accidental mortality in Beijing and to explore the populations vulnerable to adverse effects of pollutant exposure.

The total number of non-accidental deaths was 8,676 during 2017–2018, of which 58.37% were males, 77.11% were older than 65 years of age, and 66.83% died from circulatory diseases (Supplementary Table 1). As shown in Table 1, the 24-h average concentrations were 52.58 μg/m³ for PM_2.5, 80.92 μg/m³ for PM₁₀, 5.95 μg/m³ for SO₂, 40.63 μg/m³ for NO₂, 0.87 mg/m³ for CO, and 61.60 μg/m³ for O₃.

The Spearman correlation analysis results showed there were strong correlations among the air pollutants (Supplementary Table 2). It is worth noting that PM_2.5 was highly correlated with CO (r = 0.85), and PM₁₀ (r = 0.78). O₃ was negatively correlated with SO₂, NO₂, and CO, respectively (P < 0.01). The overall trend in non-accidental deaths from emergency department visits was U-shaped, and higher in winter (Figure 1). SO₂, NO₂, and O₃ showed seasonal fluctuations over time with a stable overall trend.

The estimated RRs with 95% CIs for non-accidental mortality in single-pollutant models linked to a unit increment of air pollutant concentrations are shown in Table 2. After adjusting for time-trended, daily average temperature, relative humidity, and day of week, there were significant positive relationships between air pollution and non-accidental mortality. Significant associations between exposure to air pollutants (PM_2.5, SO₂, NO₂, and CO) and non-accidental mortality for emergency department visits could be observed at lag 3 days (lag 4 and lag 5 for O₃) in single-day lagged models. In contrast to single-day lags, larger effect estimates were observed in multi-day lags. The strongest effects observed for a unit increase in short-term exposure were 1.008 (95% CI: 1.001, 1.016) for PM_2.5 at lag 04, 1.007 (95% CI: 1.001, 1.013) for PM₁₀ at lag 05, 1.054 (95% CI: 1.009, 1.100) for SO₂ at lag 04, 1.038 (95% CI: 1.013, 1.063) for NO₂ at lag 04, 1.032 (95% CI: 1.011, 1.054) for O₃ at lag 05, 1.034 (95% CI: 1.004, 1.066) for CO at lag 04.

Figure 2 shows the relationship between air pollution and cause-specific mortality in different lag models. We found that these pollutants (PM_2.5 and CO) only have significant positive associations with neoplasms disease mortality at lag 3, the RRs were 1.021 (95% CI: 1.005, 1.038) for PM_2.5 and 1.080 (95% CI: 1.003, 1.163) for CO. Meanwhile, SO₂ (lag 3, lag 03, and lag 04) and O₃ (lag 2, lag 02, lag 04, and lag 05) were only associated with circulatory disease mortality, the RRs were 1.057 (95% CI: 1.002, 1.114) for SO₂ (lag 04) and 1.034 (95% CI: 1.009, 1.061) for O₃ (lag 05). Short-term exposure to NO₂ had the significant impact on circulatory disease mortality in cumulative lag models and the RR was 1.044 (95% CI: 1.014, 1.075) in lag 04. The detailed results were shown in Supplementary Table 3.

The results of the sex-specific analyses in different lag models are shown in Figure 3. Short-term exposure to CO and NO₂ had the significant impact on males, and the RR values were 1.052 (95% CI: 1.020, 1.086) for NO₂ at lag 04 and 1.067 (95% CI: 1.026, 1.110) for CO at lag 04 (the sex modification was statistically significant for CO). Meanwhile, the correlation between O₃ and daily non-accidental deaths in emergency department visits was significantly positive in females, and the RR was 1.053 (95% CI: 1.019, 1.087) at lag 05 (the sex modification was statistically significant). The detailed results of stratified analyses by sex were in Supplementary Table 4.

The outcomes of the stratified analysis by age in different lag models are shown in Figure 4. Short-term exposure to the four air pollutants (PM₁₀, SO₂, NO₂, and O₃) had significant effects on people aged 65 years and older (the age modification was not statistically significant). The strongest effects observed in response to a unit increase in exposure were 1.008 (95% CI: 1.002, 1.014) for PM₁₀ at lag 03, 1.062 (95% CI: 1.007, 1.121) for SO₂ at lag 05, 1.040 (95% CI: 1.012, 1.068) for NO₂ at lag 04, 1.030 (95% CI: 1.006, 1.054) for O₃ at lag 05. Detailed outcomes of the age-specific stratified analysis are shown in Supplementary Table 5.

In sensitivity analysis, our findings were robust by adjusting the dfs to control for long-time trends (10–14 df per year), temperature (2–7 df), and relative humidity (2–7 df ; Supplementary Tables 6–8). After controlling for other pollutants, the short-term effects of air pollutants on non-accidental mortality in co-pollutant models are shown in Table 3 and Supplementary Table 9. The lag periods used in models were determined based on the maximum lag effect between non-accidental mortality and air pollution in single-pollutant models (lag 05 for PM₁₀ and O₃, lag 04 for other air pollutants). After adjusting for other pollutants in two-pollutant models, the positive associations between air pollutants and daily non-accidental deaths from emergency department visits remained significant, which demonstrated the relative robustness of the estimated effects. After adjusting for temperature of longer lag days (up to 28 days), the effect estimates changed slightly (Supplementary Table 10).

Figure 5 shows the E-R association between air pollutant levels (lag 05 for PM₁₀ and O₃, lag 04 for other air pollutants) and non-accidental mortality in emergency department visits. An approximate linear effect of NO₂ was found for non-accidental mortality. Non-accidental morbidity monotonously increased with the concentrations of NO₂. Meanwhile, the E-R curve for O₃ was U-shaped and it showed potential risk within the concentration of < 30 or > 75 μg/m³.

So far as we know, there are few studies examining the short-term effects of gaseous pollutants and particulate matter on non-accidental mortality based on data from emergency department visits in Beijing. Our data were obtained through the biggest emergency center in Beijing, which could more reasonably and accurately reflect the acute health effects from ambient pollution (26). Short-term exposure to air pollutants, especially gaseous pollutants, was demonstrated to have a positive relationship with non-accidental mortality. Meanwhile, short-term exposure to NO₂, SO₂, and O₃ increased the risk of circulatory mortality. Additionally, stronger associations were presented in females for O₃ while in males for CO. People aged 65 and over are strongly associated with ambient air pollution.

Gaseous pollutants had greater effects on non-accidental mortality than particulate matter in this study. In previous studies, such results were also found (31, 36). This may be related to a variety of factors including residents' awareness of self-protection, exposure levels, biological mechanisms, and population susceptibility. Gaseous pollutants are more likely to cause acute diseases, such as acute myocardial infarction, acute respiratory infections, and ischemic heart disease (37–39). Some studies have demonstrated that long-term exposure (e.g., several years) to particulate matter increases the risk of mortality to an even larger extent than exposure over a few days (40, 41). It has been shown that long-term exposure to PM_2.5 is related to accelerated coronary artery calcification or increased the risk to plaque rupture (42). One study concluded that people had less awareness of protection against ozone when compared to particulate matter (43). The E-R associations of gaseous pollutant concentrations with non-accidental mortality were positive. It was interesting to note that the E-R curve for O₃ was U-shaped, which showed a potential risk in the concentration range of < 30 or >75 μg/m³. Similar results were found in a study conducted in Delhi, India (44). This could be related to seasonal fluctuations in O₃, with higher concentrations in summer. Synergy effects between temperature and O₃ for non-accidental mortality are also biologically plausible (45, 46). The State Council of China issued the APCAP in 2013, but the annual average concentrations of NO₂ and O₃ in China have not changed significantly (47). This study indicates that emission control efforts for gaseous pollutants and people's awareness of protection should be strengthened in the future.

The relationship between short-term exposure to ambient gaseous pollution and non-accidental deaths has been extensively explored in developed countries. Similar to our research results, several previous meta-analyses based on North America and Europe investigated the relationship between ambient ozone and non-accidental mortality which demonstrated a statistically significant risk effect for non-accidental mortality (48, 49). There was a high positive association between most air pollutants (PM_2.5, SO₂, NO₂, and CO) and daily mortality from diabetes in a study conducted in Montreal, Quebec, Canada (50). Similar outcomes related to non-accidental mortality have been discussed in China. A study conducted in Hefei, an inland city in China, showed that per 10 μg/m³ increment in SO₂ (lag 03) and NO₂ (lag 01) were significantly associated with 4.93% (95% CI: 1.94, 8.00), 2.11% (95% CI: 1.18, 3.05) increase of daily non-accidental deaths (15). Similar to our findings, a study exploring the acute effects of SO₂ and NO₂ on mortality found non-accidental mortality in Beijing raised 0.60% (95% CI: 0.26, 0.95) for every 10 μg/m³ increase in daily NO₂ concentration (20). Meanwhile, a meta-analysis of the acute effects of ambient ozone on mortality in Chinese cities showed that the percent change for non-accidental mortality was 0.42% (95% CI: 0.32, 0.52) with an increase of 10 μg/m³ in the maximum 8-h average concentration of O₃ (51). Variations in effect estimates are caused by many factors, including variation across the populations, various analysis strategy, and issues related to data quality and measurement error (50).

For cause-specific mortality, there was a significant association between short-term exposure to NO₂, SO₂ and O₃ on circulatory disease mortality. The relationships of increased NO₂, SO₂, and O₃ with increased cardiovascular mortality were also observed in exist evidence (51–54). The biological mechanisms contributing to the link of NO₂, SO₂, and O₃ to circulatory disease mortality pathologies may involve complex processes. Increase in oxidant stress appears to be a broadly applicable mechanism, regardless of the type of pollutants that contributes to the adverse effects of air pollution. NO₂, SO₂, and O₃ induce oxidative stress, thereby triggering inflammatory responses and gene activation, potentially leading to endothelial dysfunction, atherothrombotic alterations, metabolic dysregulation, and the development of cardiometabolic diseases (55–58). Researches have shown that O₃ can lead to an imbalance in heart rate variability and an increased sensitivity to myocardial calcium load, which can trigger vascular and cardiac injury, ultimately manifesting as cardiovascular disease (59, 60). SO₂ can reduce the permeability of red blood cells and lead to organ shortages, while short-term exposure to it can lead to a decreased in cardiac vagal control measures (61). NO₂ is considered a proxy for traffic-related pollutants and is often found in conjunction with traffic noise, which is also associated with adverse cardiovascular health outcomes (62).

In stratified analysis, the effect estimates for CO were consistently higher in males, while the effect estimates for O₃ were consistently higher in females. The sex modification was statistically significant for CO and O₃. Several studies showed that males were more sensitive to air pollutants (32, 43, 63). The smoking rate for males is significantly higher than for females in China. Even after quitting smoking, impaired lung function may not recover to normal quickly, and may have synergistic effects with air pollution (64). Meanwhile, some studies have suggested that the increased susceptibility of females to the adverse effects of O₃, which could be related to differential regulation of the lung immune response (65). The reasons for gender vulnerability to air pollution were not clear and need further investigation. In the stratified analyses by age, the association between short-term exposure to air pollutants and non-accidental deaths seemed to be evident in older people. Several foreign studies have also reported increased susceptibility in older people (66, 67). Studies confirm that certain physiological regulatory functions diminish with age, which may lead to an increased health risk associated with those older than 65 years (68). Some studies have indicated that current air quality guidelines are designed to protect the general population but not enough to protect older people (69). Therefore, we call for more studies to investigate the susceptibility of older people to air pollutant concentrations in different countries or different regions with the aim of determining whether this finding can be generalized across different climatic and air pollution characteristics.

Several limitations exist within this study that should be considered. First, we use data on air pollution from the average of fixed-site air quality–monitoring stations to represent exposure at the individual level, which can lead to exposure misclassification. Secondly, this study fails to control for social and demographic factors (e.g., education level and socioeconomic status) which could influence the demographic composition and mortality (70, 71). Thirdly, this is an ecological research limiting causal inference. This needs to be addressed by further toxicological or epidemiological studies. Finally, it was a single-center study with limited study time. Therefore, we encourage further large-scale studies, especially in developing countries.

Our findings demonstrated that short-term exposure to ambient pollutants, especially gaseous pollutants, increased the risk of non-accidental mortality in emergency department visits to Beijing. Additionally, stronger associations were presented in females for O₃ while in males for CO. People aged 65 and over are strongly associated with ambient air pollution. The study provides health authorities with critical information on the acute adverse effects of air pollution, which will contribute to the development of protective measures in China.

The data that support the findings of this study are available from the corresponding authors, upon reasonable request.

SW and YZ: conceptualization, formal analysis, and writing—original draft preparation. XL: software and writing—review and editing. JZ: methodology. NZ: validation and visualization. YG: validation. JC, YLi, and YLy: data curation. ZC: writing—original draft preparation. CL: project administration. JG: software and funding acquisition. WZ: methodology and resources. JM: investigation, resources, and supervision. All authors contributed to the article and approved the submitted version.