1 Introduction

Air pollution is one of the greatest environmental risks to health. More than 90% of the global population 2019 lived in areas where concentrations exceeded the 2005 WHO air quality guideline of 10 µg/m3 (World Health Organization, 2021). The combined effects of ambient and household air pollution are associated with 7 million premature deaths annually (World Health Organization, 2023). From a health perspective, outdoor air pollution is a complex mixture of components that include airborne particulate matter (PM) and gaseous pollutants such as sulfur dioxide (SO2) and nitrogen dioxide (NO2), named primary pollutants (Camacho et al., 2020; Sun et al., 2010). Air pollution is widely recognized as an important causative agent of many non-communicable diseases, including diabetes mellitus, cardiovascular disease, Parkinson’s disease, neurological disorders, cancer, etc. (Meo et al., 2015; Cosselman et al., 2015; Chen et al., 2017; Heusinkveld et al., 2016; Turner et al., 2020). Moreover, air pollution contributes substantially to premature mortality and the disease burden of different populations (Cohen et al., 2017; Lelieveld et al., 2015; Yin, 2023).

Most research on air pollution and health is conducted in North America and Europe, while there needs to be more research on health effects in parts of Africa and Asia (Badida et al., 2023; Zhang et al., 2014). However, the negative impact of air pollution is more massive in low—and middle-income countries (World Health Organization, 2021). As the largest developing country, China has experienced severe ambient air pollution for a long time (Zhang et al., 2019a, b; Wei et al., 2023). Inhalable particles with aerodynamic diameters less than 10 and 2.5 um (PM10 and PM2.5), sulfur dioxide (SO2) and nitrogen dioxide (NO2) are important pollutants of concern to China (Chen et al., 2013; Tian et al., 2019). According to the bulletin on China's ecological environment in 2022, 126 out of 339 cities in China do not fulfill the Chinese air quality standards (Ministry of Ecology and Environment of the People’s Republic of China, 2022). Meanwhile, China is facing a severe challenge of aging. The 2020 Chinese census data showed that the proportion of people aged over 65 had reached 13.5% (Zheng, 2021). They indicate that China will soon transition from an aging into an aged society (Luo et al., 2021).

Aging is a continuous process of progressive decline of the body’s function leading to increased vulnerability, frailty or sensitivity in older people (Poland et al., 2014; Simoni et al., 2015). Previous studies have shown that older people are more vulnerable to air pollution than younger people (Lu et al., 2022; Luo et al., 2017; Qian et al., 2013). It is reported that increased pollution exposures have been associated with increased mortality and hospital admissions/emergency-room visits of the elderly, mainly due to exacerbations of chronic diseases or respiratory tract infections (Simoni et al., 2015). Several studies have shown a statistically significant correlation between air pollution and mortality among the elderly in China (Yang et al., 2012; Cai et al., 2019; Zeng et al., 2017). A study in Shenzhen found that with an increase of 10 µ g/m3 in PM2.5 concentration, the excess risk (ER) of mortality in the elderly is 1.32% (Cai et al., 2019). In Shenyang, a time-stratified case cross-analysis report states that with an increase in PM2.5 per 10 µg/m3 in the air, the excess risk of non-accidental death for elderly people aged 65–74 and ≥ 75 is 0.51% and 0.58%, respectively (Ma et al., 2011). Furthermore, a study in Baotou found that for every 10 µg/m3 increase in PM2.5 and PM10 in the air, the excess risk of non-accidental mortality among people aged 65 and above was 0.104% and 0.045%, respectively (Lu et al., 2022).

However, no previous meta-analyses focus on the effect of different air pollutants on elderly mortality. Only reviews and meta-analyses have demonstrated the acute effects of PM2.5 (Shang et al., 2013; Xia et al., 2019), PM10 (Jin et al., 2016; Shang et al., 2013), SO2 (Shang et al., 2013), and NO2 (Shang et al., 2013) on mortality rates for all age groups. Shang et al. (2013) showed that with 10 µg/m3 increases in gaseous pollutants PM2.5, PM10, SO2 and NO2, the total mortality risk in China were 0.38%, 0.32%, 0.81% and 1.30%. Xia et al. (2019) found that with an increase of PM2.5 in the air by 10 µg/m3, the excess risk of non-accidental death in the whole population in China will increase by 0.73%. Jin et al. (2016) concluded that with the increase of PM10 in the atmosphere by 10 µg/m3, the excess mortality of the whole population in China was 0.29%.

To our knowledge, this study is the first to systematically review existing literature regarding the impact of short-term exposure ambient to environmental air pollution on mortality among elderly adults in China. It synthesized scientific evidence examining the relationship between PM2.5, PM10, SO2, NO2 and excess mortality risk of the elderly in China. This study's findings help to provide theoretical evidence for the health risk assessment of air pollution of the elderly in China and other low- and middle-income countries. It also identified the limitations and gaps in this field that warranted future research.

2 Methods

Systematic review and meta-analysis procedures were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA, 2020).

2.1 Study Selection Criteria

Studies that met all of the following criteria were included in the review: (1) study designs: observational studies, such as time series or case-crossover studies; (2) exposures: specific air pollutants (PM2.5, PM10, SO2, NO2); (3) population: People aged only 65 and above; (4) outcome: mortality excess risk; (5) pollution model: single pollution model; (6) statistic methods: generalized linear model (GLM) or generalized additive model (GAM); (7) article types: peer-reviewed publications.

Studies that met any of the following criteria were excluded from the review: (1) no accurate data or the data cannot be converted to an excess risk (ER) value with a 95% CI; (2) no data for elderly individuals; (3) articles not written in Chinese or English; (4) death caused by disease, such as respiratory system disease; (5) study samples enrolled from a specific setting, for example, hospitals and nursing homes, etc.

2.2 Search of the Literature

Keywords and abstracts were searched in English databases of PubMed, Scopus, Web of Science (WOS) and the Chinese database of China National Knowledge Infrastructure (CNKI). The time window of search is from the inception of an electronic bibliographic database to 10th June 2023. The search algorithm includes all possible combinations of the following keywords: (1) PM2.5, PM10, SO2, NO2, ambient particulate matter, air pollutant, air quality and air pollution, (2) mortality, death or die, (3) elderly, old adults or older people, and (4) China or Chinese. Titles and abstracts of the articles identified through the keyword search were screened against the study selection criteria. Potentially relevant articles were retrieved for evaluation of the full text.

A reference list search (i.e., backward reference search) and cited reference search (i.e., forward reference search) were conducted based on the full-text articles meeting the study selection criteria that were identified from the keyword search. Articles identified from the backward and forward reference search were further screened and evaluated using the same study selection criteria. Reference searches were repeated on all newly identified articles until no additional relevant article was found.

2.3 Data Extraction

Standard data extraction tables were used to collect the following data for each included study: (1) first author and year of publication; (2) research time; (3) research area; (4) concentration of air pollutants; (5) statistical methods; (6) experimental design; (7) ER value and 95% confidence interval.

2.4 Study Quality Assessment

Quality assessment was performed to analyze the quality and risk of bias of the literature included in our review since no recognized quality assessment tools for time series and case cross study exists. We drew on the New Castle Ottawa Scale (Peterson et al., 2011) and the Cochrane risk of bias tool (Higgins et al., 2011), as applied in previous studies (Mustafic et al., 2012; Zhang et al., 2021). The quality of each study was evaluated on the basis of three aspects (the quality of air pollution data [0–1 point], the quality of mortality data [0–1 point], and the extent of adjustment for potential confounders [0–3 points]). For the quality of air pollution data, 1 point was awarded if ≤ 25% of daily pollutant data was missing, and 0 points were awarded if > 25% of daily pollutant data was missing. For the quality of mortality data, 1 point was awarded to studies with causes of mortality coded according to the International Classification of Disease, Revision 9 (ICD-9) or Revision 10 (ICD-10), whereas studies that did not meet this criterion were scored 0. For the adjustments for potential confounders, 3 points were awarded to studies that simultaneously considered meteorological factors, day of week and time trend, holiday effects or influenza epidemics. 2 points were awarded to studies that considered meteorological factors, day of the week and time trends effects, but did not consider holiday effects or influenza epidemics. 1 point were awarded to studies that only considered time trends effects. 0 points were given to studies that did not consider any of the confounders as mentioned above.

Studies that obtained the entire score in all three aspects were considered of high quality, studies that did not score in any of these aspects were considered of low quality, and studies in between were considered to be of intermediate quality.

2.5 Statistical Analysis

Meta-analysis was performed to estimate the effect of pollutants (PM2.5, PM10, SO2, NO2) and mortality excess risk among elderly Chinese individuals (age ≧65). The level of heterogeneity represented by the I2 index was interpreted as modest (I2 ≤ 25%), moderate (25% < I2 ≤ 50%), or considerable (I2 > 75%) (An et al., 2018; Higgins et al., 2003). Sensitivity analyses were assessed by repeating meta-analysis, removing one by one; that is, each study was excluded to test if individual studies had an effect on the pooled estimates (Deng et al., 2021). Publication bias was assessed by the funnel plots with Begg’s and Egger’s tests (Egger et al., 1997; Sterne & Egger, 2001; Racine et al., 2021).

All statistical analyses were conducted using the Stata 14.2 SE version (StataCorp, College Station, TX). Specific STATA commands included “metan” and “meta bias.” All analyses used two-sided tests; p-values less than 0.05 were considered statistically significant.

3 Results

3.1 Study Selection

Figure 1 shows the study selection flow chart. We identified 1722 articles through keyword and reference searches, including 513 articles written in Chinese from CNKI and 1209 articles written in English from PubMed, Scopus and WOS. After removing duplicates, 1449 unique articles entered title and abstract screening, by which 1308 articles were excluded. The full texts of the remaining 141 articles were reviewed against the study selection criteria. Of these, 115 articles were excluded. Reasons for exclusion included: 54 articles did not have original data, 32 studies did not have data on the elderly, and 29 studies only did not have mortality data of non-accident. In total, 26 articles were included in the review, including 18 studies written in English and 8 in Chinese.

Fig. 1
figure 1

Flow chart of literature screening

Full size image

3.2 Basic Characteristics of the Selected Studies

Table 1 summarizes the essential characteristics of the 26 studies included in the review. Twenty-one articles included other age groups and elderly individuals; only 5 (Chen et al., 2019; Huang et al., 2020; Qu et al., 2018; Zeng et al., 2019) articles exclusively examined older adults. All studies were published in the past twenty years, and the study areas included multiple cities in the north and south in China. Among these studies, there are 16, 14, 8 and 8 studies, including PM2.5, PM10, SO2 and NO2, respectively. Four studies adopted a time-stratified case-crossover study design, and others adopted a time-series design. The statistical models applied included the generalized additive model (GAM) and generalized linear model (GLM).

Table 1 Basic information of the literature on the effect of short-term exposure to air pollutants on death in the elderly
Full size table

3.3 Meta-Analysis of the Impact of Four Pollutants on Elderly Mortality

Meta-analysis was conducted to estimate the correlation of four pollutants on mortality among the elderly in China. The analysis used excess risk included in each study. When the heterogeneity of literature in different pollution groups is greater than 75%, the data analysis adopts a random effect model. Figure 2 show the forest map generated by the meta-analysis of different pollutants. Table 2 show that the I2 values of PM2.5, PM10, SO2, and NO2 are 95.5%, 96.6%, 91.5%, and 82.6%. The combined ER values of PM2.5, PM10, SO2, and NO2 were 0.96 (95%CI: 0.63–1.28; p < 0.01), 0.53 (95%CI: 0.29–0.77; p < 0.01), 0.92 (95%CI: 0.44–1.41; p < 0.01) and 1.43 (95%CI: 0.92–1.94; p < 0.01), respectively. These findings indicate that for every 10 µg/m3 increase in the concentration of PM2.5, PM10, SO2, and NO2 in the atmosphere, the excess risk of mortality among the elderly was 0.96%, 0.53%, 0.92% and 1.43%, respectively.

Fig. 2
figure 2figure 2

Forest plot of the effect of air pollutants on the excess risk of elderly mortality in China

Full size image
Table 2 The combined estimates for ER (%) and 95% Confidence Intervals (95% CI) for a 10 µg/m3 increase in PM2.5, PM10, SO2, and NO2 of elderly mortality in China
Full size table

3.4 Subgroup Analysis

The included articles were grouped according to the concentration of air pollution for subgroup analysis. The GB3095-2012 standard annual average first level concentration limits NO2 (40 µg/m3) of the ministry of Ecology and Environment of the People's Republic of China were used as segmentation points for NO2 (Ministry of Ecology & Environment of the People's Republic of China, 2012). Moreover, since the average concentrations of PM2.5, PM10 and SO2 included in the literature are almost all higher than the annual average first-level concentration limits (PM2.5: 15 µg/m3, PM10: 40 µg/m3, and SO2: 20 µg/m3), the World Health Organization's 24-h second level concentration limit for PM2.5 (50 µg/m3) and PM10 (100 µg/m3) and the Air Quality Guidelines (AQG) level value for SO2 (40 µg/m3) were selected as the segmentation point (World Health Organization, 2021).

Table 3 shows the results of the subgroup analysis of the correlation of different concentrations of pollutants on elderly mortality. Under different concentrations of PM2.5, PM10, and NO2, an increase of 10 µg/m3 in PM2.5, PM10, and NO2 was significantly correlated with an increase in elderly mortality. Meanwhile, compared to the higher concentration group, in the lower concentration group, with each increase of 10 µg/m3 in PM2.5, PM10, and NO2, the excess risk mortality rate of the elderly is higher. Moreover, subgroup analysis of SO2 shows that in a low-concentration environment, the increase of SO2 concentration by 10 µg/m3 is not significantly related to elderly mortality, while in high concentration environment, the increase of SO2 concentration is significantly related to elderly mortality.

Table 3 Subgroup analysis
Full size table

3.5 Publication Bias Test

Begg and Egger's methods were used in this study, and the results are shown in Table 4. The publication bias of air pollutants were not significant (P > 0.05). In addition, the funnel plot was drawn with logER as the abscissa and its standard error as the ordinate. It is found that the funnel plot is roughly symmetrical, indicating that there is no publication bias (Fig. 3). Therefore, there was no evidence of publication bias in the included literature.

Table 4 Results of publication bias test
Full size table
Fig. 3
figure 3figure 3

Funnel chart of pollutants associated with air pollutants on the mortality of elderly individuals in China

Full size image

3.6 Sensitivity Analysis

A sensitivity analysis was performed by repeating the meta-analysis to check the robustness of the combined results. Each of the initial studies was excluded, and the impact on the overall results was investigated according to the difference between the new and the initial effect values. Figure 4 shows that in the analysis of PM2.5, PM10, SO2, and NO2, the estimates obtained were essentially consistent with the initial results when any study was excluded, and changes only occurred in a narrow interval.

Fig. 4
figure 4figure 4

Sensitivity analysis of air pollutants on the mortality of elderly individuals in China

Full size image

3.7 Study Quality Assessment

Table 5 reports the quality assessment of the included 26 studies. Five studies were classified as high quality, 21 were classified as medium quality, and none were classified as low quality. Overall, the quality of these studies was good, with high and intermediate quality accounting for 86.2%.

Table 5 Quality assessment of the included studies
Full size table

4 Discussion

This study is the first attempt to systematically review the correlation of air pollution on mortality among Chinese elderly. The results showed that with the increase of PM2.5, PM10, SO2 and NO2 concentrations by 10 µg/m3, the excess risk mortality of the elderly was 0.96%, 0.53%, 0.92% and 1.43%, respectively. There is no meta-analysis specifically targeting the elderly group in previous studies, but some surveys conducted in other countries have obtained similar results. A study in California shows that the excess risk mortality of the elderly was 0.6%, with an increase in PM2.5 concentration per 10 µg/m3 (Ostro et al., 2006). A study conducted in Seoul analyzed the health burden of related mortality caused by PM2.5 and concluded that PM2.5 contributed as much as 0.62% to the total mortality of the elderly (Jung et al., 2019). In a study in Singapore, the impact of air particles on the health of the population aged 65 years old and above was assessed. It was found that the risk of non-accidental death increased by 0.771% and 0.955%, respectively, in 0–5 days for every 10 µg/m3 increase in the concentration of PM10 and PM2.5 (Yap et al., 2019). Furthermore, A study observed that the percent increase in daily non-accidental mortality on each additional consecutive day with PM10 was 0.77% in Japan (Kim et al., 2019). Although the regions and air concentrations of the above studies varied, they all obtained relatively consistent results.

We found a more considerable negative correlation between air pollution and mortality among the elderly compared with the previous meta-analysis of ambient air pollutants and all-age mortality in China. Previous meta-analyses have found that as PM2.5, PM10, SO2, and NO2 in the air increase by 10 µ g/m3, the excess risk of non-accidental mortality for all age groups in China will increase by 0.73% (Xia et al., 2019), 0.29% (Jin et al., 2016), 0.81%, and 1.30% (Shang et al., 2013). These data are lower than the results of this study on older adults.

The subgroup analysis found that the influence of air pollution concentration on the excess risk mortality of the elderly is not a simple linear relationship. This is consistent with the conclusions of many previous studies (Burnett et al., 2014; Pope et al., 2011; Xia et al., 2019). In a meta-regression study, Burnett et al. (2014) found a more significant correlation at lower concentrations, and the PM2.5-mortality association was non-linear and more complex than described by a single unknown parameter. This may be the influence of other relevant factors in addition to air pollution concentration. In an integrated evaluation study of the impact of ambient air pollution on the risk of cardiovascular death, Pope et al. (2011) found that both exposure duration and intensity of air pollution impacted the risk of death. When the population is exposed to low to moderate pollution, a very steep and almost linear exposure–response relationship is observed, while at very high exposure, the exposure–response function is flat or stable, and there are likely essential risk trade-offs between duration and intensity of exposure. Moreover, a meta-analysis in China shows that the effect of daily average concentration has great regional differences (Xia et al., 2019). These studies provide a new perspective on the relationship between air pollution and mortality. Although the current linearity and nonlinearity have yet to be thoroughly examined or supported, more reasonable methods should be used to analyze the excess risk mortality of air pollution.

In addition, this study still has some limitations. First, other air pollutants such as CO and O3 may also affect the death of the elderly. However, due to a lack of research literature and inconsistent data, the analysis of the impact of increased CO and O3 on elderly mortality was not included. Second, this study only includes the effect estimates of the single pollutant model without considering the potential joint effects and collinearity among multiple pollutants. Moreover, in addition to air pollutants' concentration, the impact of temperature, duration of exposure, season and region on elderly mortality has also been put forward in previous studies. However, due to the limitations of the research scope, these contents were not analyzed.

5 Conclusion

This study synthesized scientific evidence examining the correlation of air pollutants on the excess risk of mortality among the elderly in China. The meta-analysis found that with the increase of PM2.5, PM10, SO2, and NO2 concentration by 10 µg/m3, the excess risk of death of the elderly was 0.96%, 0.53%, 0.92% and 1.43%, respectively. At different concentrations, the impact of air pollutants on the excess risk of death of the elderly is different. Future studies can also examine two or more air pollution models and the impact of air pollution on mortality in the elderly under different temperatures, duration of exposure, seasons and regions.