Long‐term effects of air pollution on hospital admissions and mortality for chronic obstructive pulmonary disease in Beijing, China

Abstract Objective We aimed to clarify the association between air pollution and hospital admissions for chronic obstructive pulmonary disease (COPD) and mortality in Beijing, China. Methods In this retrospective study, we recruited 510 COPD patients from 1 January 2006 to 31 December 2009. The patient data were obtained from the electronic medical records of Peking University Third Hospital in Beijing. Air pollution and meteorological data were obtained from the Institute of Atmospheric Physics of the Chinese Academy of Sciences. Monthly COPD hospital admissions, mortality and air pollution data were analysed using Poisson regression in generalised additive models adjusted for mean temperature, pressure and relative humidity. Results There were positive correlations between sulfur dioxide (SO2), particulate matter with an aerodynamic diameter ≤ 10 μm (PM10) and COPD hospital admissions in the single‐pollutant model. An increase of 10 μg/m3 in SO2 and PM10 were associated with an increase of 4.053% (95% CI: 1.470–5.179%) and 1.401% (95%CI: 0.6656–1.850%) in COPD hospital admissions. In the multiple‐pollutant model [SO2 and nitrogen dioxide (NO2) combinations], there was only a positive correlation between SO2 and COPD hospital admissions. An increase of 10 μg/m3 in SO2 were associated with an increase of 1.916% (95% CI: 1.118–4.286%) in COPD hospital admissions. There was no correlation between three pollutant combinations and COPD hospital admissions. We did not find correlations between air pollution and COPD mortality in either single‐ or multiple‐pollutant models. Conclusions SO2 and PM10 may be important factors for the increase in COPD hospital admissions in Beijing, China.


| INTRODUCTION
Chronic obstructive pulmonary disease (COPD) is one of the most common chronic respiratory disease and a major health problem in the developed and developing countries. According to the China Pulmonary Health (CPH) study, the prevalence of COPD in Chinese people aged ≥20 years was 8.6%, and aged ≥40 years old is 13.7%. 1 COPD affects nearly 100 million adults and is the third leading cause of death in China. 2 Tobacco smoking is by far the most common factor in the etiology of COPD. However, additional causal associations have also been found among both smokers and nonsmokers, including dust, gas, vapor, and biological and chemical exposures. 3,4 Of all the other risk factors, particulate air pollution is now recognised as a potential environmental risk factor that causes many health hazards for respiratory diseases, especially COPD.
Numerous epidemiological studies have demonstrated that short-and long-term exposure to air pollution is associated with COPD hospital admissions, acute exacerbation and mortality. [5][6][7] Dong et al. 8 demonstrated that a 10-μg/m 3 increase in PM 10 , SO 2 and NO 2 was associated with a 0.25% (95%CI: 0.01-0.49%), 1.67% (95%CI: 0.54-3.93%) and 1.37% (95%CI: 0.25-2.51%) increase in COPD outpatient visits, respectively. Zhang et al. 9 found that every 10-μg/m 3 increase in pollutants, the majority of the summary estimates for COPD hospital admissions were in the order of (ozone 3) O 3 > (particulate matter with aerodynamic diameter ≤ 2.5 um) PM 2.5 > NO 2 > PM 10 > SO 2 . However, inconsistent and different results have been found for air pollution and COPD hospital admissions and mortality in different countries and even within the same regions. In 2012, our team conducted a meta-analysis of PM 10 and COPD hospitalisations and mortality. We found that a 10-μg/m 3 increase in PM 10 was associated with a 2.7% (95%CI: 1.9-3.6%) increase in COPD hospitalisations and a 1.1% (95% CI: 0.8-1.4%) increase in COPD mortality. 10 In 2020, we conducted another meta-analysis of PM 2.5 and COPD hospitalisations and mortality. We found that a 10-μg/m 3 increase in PM 2.5 was associated with a 2.5% (95%CI:1.8-3.2%) increase in COPD hospitalisations and a 1.5% (95% CI: 0.9-2.2%) increase in COPD mortality. 11 However, most of the aforementioned investigations have been performed in America and Europe, where air pollutant levels are much lower than those in developing countries. Few studies conducted in China have explored the relationship between air pollution and COPD hospital admissions and mortality. Rapid development in Beijing has resulted in a significant increase in air pollution. Therefore, it is important to determine the effects of air pollution on COPD. Beijing, the capital of China, has a population of over 11 million and is located at 39 56 0 N 0 and 116 20 0 E 0 . The Haidian District is located in the northwest of Beijing, which belongs to the city center, and there are many tall buildings that can block air flow. Haidian District is socioeconomically highly developed and has a heavy and complex flow of traffic, as well as a large floating population. Air pollution levels are very high in the Haidian District. Most residents of Haidian District depend on the Peking University Third Hospital as a health resource. Data on COPD hospital admissions and mortality at this hospital can thus reflect the general health condition of the Beijing population. The first author was a graduate student at Peking University Third Hospital in Haidian District from 2009 to 2012, and COPD data could therefore be obtained during the study period.
In this study, we assessed the effects of air pollution on hospital admissions and mortality among patients with COPD at Peking University Third Hospital, from 1 January 2006 to 31 December 2009 in Beijing, China. We assessed the patterns of exposure-response relationships for three pollutants (monthly SO 2 , NO 2 and PM 10 ) and COPD hospital admissions and mortality in single and multiple-pollutant models. The inclusion criteria were (1) patients with a diagnosis of COPD according to the American Thoracic Society (ATS) 12 and Global Initiative for Obstructive Lung Disease (GOLD) 13 or according to the International Classification of Diseases, 10th Revision 14 ; (2) living in the district for 4 years and had been away for less than 6 months during the study period; (3) patients with complete basic information, such as residential address and phone numbers; and (4) all the participants that lived more than 500 m away from the main road without exposure to fuels or occupational exposure. The residence and surrounding environment of all participants were confirmed via phone.
The exclusion criteria were as follows: (1) unclear diagnosis of COPD; (2) presence of complications, such as malignant tumour, asthma, tuberculosis (TB), severe cardiovascular and cerebrovascular diseases, hepatic and renal insufficiency; and (3) missing or insufficient basic information.
As this was a retrospective study, we waived the requirement for ethical approval and informed patient consent from the Ethics Committee of Peking University Third Hospital.

| Air pollution and meteorological data collection
Monthly records of PM 10 , SO 2 and NO 2 in the Haidian District in Beijing from 1 January 2006 to 31 December 2009 were obtained from the Institute of Atmospheric Physics, Chinese Academy of Science. Meteorological data, including the daily mean temperature, pressure and relative humidity were obtained from the Institute of Atmospheric Physics, Chinese Academy of Sciences.

| Statistical analysis
To estimate the effects of air pollution on COPD hospital admissions and mortality, we used Poisson regression in generalised additive models (GAM) with R version 3.6.0 software for Windows (www.r-project.org). The models were adjusted for the daily mean temperature, pressure and relative humidity. The choice of degrees of freedom (df) for the smoothing spline of any weather variable was between 3 df. Before the analysis, we first fitted the original data using log 10 -trasformed to improve normality and stabilise variance. We fitted nonparametric smoothing terms (using the smoothing spine function) for daily mean temperature, pressure and relative humidity. After controlling for the weather variables, we introduced each pollutant separately into the single-pollutant model. We also fitted models with different combinations of pollutants (two and three pollutants per model) to assess the stability of the effect, which was analysed in the multiple-pollutant model.
The monthly number of patients hospitalised for COPD relative to the general population was a small probability event with an approximate Poisson distribution. The statistics approximate at Poisson distribution. We used the model described below to explore the relationship between air pollutants and COPD hospital admissions and mortality.
where t refers to the month of observation, and Yt refers to monthly cases of hospital admissions and mortality because of COPD, respectively. Log[E (Yt)] denotes the estimated monthly cases in month t, X refers to air pollutants (PM 10 , SO 2 and NO 2 ), β refers to the effect value, time refers to the unit of year and month, Z t refers to meteorological data, and s denotes the cubic smoothing spline. Because the data were small, a p-value of 0.05 was deemed significant. This retrospective study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. 15 3 | RESULT

| Characteristics of study subjects
During the 4-year study period, 510 patients with COPD were eligible for inclusion. Of these, 95 COPD patients were dead during the study period.
The study included 368 male patients (72.16%) and 142 female patients (27.84%). The mean body mass index (BMI) was 22.68 ± 4.16 kg/m 2 (range: 13.85-41.02 kg/m 2 ). The proportion of the group with overweight was 25.89%. The proportion of the group with lower education (less than senior high school) was 69.22%. The proportion of the group with a high education (senior high school and above) was 30.78%. The mean number of smoking packs was 14 packs/year, and the mean age was 76.52 ± 7.98 year (range:44-90 year). The major complication among the COPD patients were hypertension and diabetes. The mean monthly number of COPD hospitalisations was 8.65 ± 3.26 (range:2-17). Among all cases, 95 patients died by 31 May 2010. The mean monthly COPD mortality was 1.9 ± 1.63 (range:0-9). The mean FEV 1 /FVC ratio was 65.24 ± 2.19%. The mean FEV 1 % pred was 61.12 ± 18.16%. The severity of COPD patients were graded according to the 2023 GOLD guidelines. 16 The patients with moderate and severe COPD were predominant throughout the population. The mean FEV 1 was 1.67 ± 0.32 L. The mean FVC was 2.85 ± 0.74 L. Further clinical characteristics of patients are reported in Tables 1 and 2.

| Exposure to pollutants
The monthly mean concentration of PM 10 was 141.53 ± 41.20 μg/m 3 . The monthly mean concentration of SO 2 and NO 2 were 44.19 ± 39.09 μg/m 3 and 54.67 ± 12.25 μg/m 3 , respectively. Yearly mean concentration of PM 10 and NO 2 were all over the China Grade II standard for ambient air quality (PM 10 70 μg/m 3 and NO 2 40 μg/m 3 ) (GB3095-2012). The SO 2 was within the China Grade II standard for ambient air quality (60 μg/m 3 ). The mean temperature was 13.63 ± 11.41 (À12-40 C). The mean pressure and relative humidity were 10123.26 ± 91.57 (9854-10 453 0.1 hPa) and 52 ± 15.81% (4-181%), respectively ( Table 3). Table 4 shows the results of Spearman correlations between meteorological variables and air pollutants over the 4-year study period. It showed there was significant correlation among the pollutants and meteorological variables. SO 2 was correlated with NO 2 , PM 10 and meteorological variables ( p < 0.05). NO 2 was correlated with SO 2 , PM 10 and meteorological variables (p < 0.05). Relative humidity was also correlated with temperature and pressure(p < 0.05). Figure 1 showed the time trends of air pollution and COPD hospital admissions and mortality. They all showed relatively stable seasonal trends annually, and they were higher in cold season and lower in warm season. Ever year from November to February of the next year, COPD hospital admissions, COPD mortality and the three air pollutants concentrations were all at the peak.
In the multiple-pollutant models, COPD hospital admissions was only positively associated with SO 2 and NO 2 combinations (SO 2 : t = 3.926, p = 0.045*; NO 2 : t = 1.935, p = 0.345). An IQR change of SO 2 (48.58 μg/m 3 ) was associated with an increase of 9.31% (5.43-20.82%) in hospital admissions because of COPD (Table 5 and Figure 3). There was no correlation between three pollutant combinations and COPD hospital admissions (Table 5).

| Effect on COPD mortality
We used the single-pollutant model to assess the effects among SO 2, NO 2 and PM 10 with COPD mortality. There was no correlation between air pollution and COPD mortality in the single-pollutant model (Table 5 and Figure 4). In the multiple-pollutant models, there was no correlation between two pollutant combinations, three pollutant combinations and COPD mortality ( Table 5).
The single and multiple models were all adjusted for the mean temperature, pressure and relative humidity.

| DISCUSSION
In this study, we present one of the few reports regarding the effects of air pollution on hospital admissions and mortality because of COPD in Beijing, China. We found that over the 4-year study period, COPD hospital admissions and mortality showed relatively stable seasonal trends annually. They were higher in cold season and lower in warm season. In recent decades, air pollution has cause considerable public concern in China, especially in Beijing. Yearly concentrations of NO 2 and PM 10 exceeded the Chinese ambient air quality standards and were far higher than those reported in the USA and Europe. Yearly concentrations of SO 2 were higher than those reported in the USA and Europe, too. 17 Luo et al. 18 found that PM 10 pollution occurred mainly in Spring and Winter. Cichowicz et al. 19 reported that PM 10 , SO 2 , NO 2 and CO were higher in winter. And the higher levels of pollution in winter month are presented by PM 10 concentration that exceed the limit values. This study indicated that PM 10 and NO 2 were higher than SO 2 . The reason may be the source of PM 10 and NO 2 is from motor vehicle  emissions, coal burning and frequent dust storms, while combustion of sulfur-contain fuels and production of waste gas in chemical industry dominated SO 2 emission. Beijing has a heavy industrial structure and massive motor traffic, resulting in large quantities of industrial emissions for PM 10 and NO 2 . Air pollution is worse in winter because of coal burning for heating in Beijing.
Pollutants are not easy to spread because winter is dominated by downdraft, and temperature inversion is easy to occur. Fortunately, people paid increasing attention to the effects of air pollution, and we observe changes of air pollution in Beijing through the efforts of the government. Therefore, we aimed to update the association between air pollution and COPD hospital admissions and mortality and compared the effect after 10 years in Beijing, China. Unfortunately, the data of 10 years late have not been fully collected. This study showed that an increase of 10 μg/m 3 in SO 2 and PM 10 was associated with an increase of 4.053% (95% CI: 1.470-5.179%) and 1.401% (95%CI: 0.6656-1.850%) in COPD hospital admissions, respectively. In the multiple-pollutant model (SO 2 and NO 2 combinations), there was only a positive correlation between SO 2 and COPD hospital admissions. An increase of 10 μg/m 3 in SO 2 was associated with an increase of 1.916% (95% CI 1.118%-4.286%) in COPD hospital admissions. This findings were similar to the previous studies. Dabrowiecki P et al. 20 reported that during 21 days after exposure, an increase of 10 μg/m 3 in SO 2 and PM 10 was associated with an increase of 14.5% (95% CI: 3.8-26.2%) and 2.8% (95%CI: 0.8-4.9%) in COPD hospital admissions, respectively. Ghanbari et al. 21 reported that during 1 year exposure, an increase of 10 μg/m 3 in SO 2 was associated with an increase of 0.5% (95% CI: 0%-1%) in COPD hospital admissions. Studies indicated that short-term exposure to air pollution may also indicate risk of COPD hospital admissions. Li et al. 17 reported that each 10 μg/m 3 increase in SO 2 and PM 10 concentrations corresponded to an increase   23 This study did not found correlation between NO 2 and COPD hospital admissions. However, many studies conducted NO 2 and COPD hospital admissions. Li et al. 17 reported that each 10 μg/m 3 increase in NO 2 concentrations corresponded to an increase in COPD hospitalisation of 3.03% (95% CI: 1.82-4.26%) at lag 0-6 days. In Iran, Hanieh et al. found that a 10 μg/m 3 increase in NO 2 concentrations corresponded to an increase of 4.9% (95% CI: 1.0-9.0%) in COPD hospital admissions. 24 We also found no association between the air pollutants investigated and COPD mortality. Yan et al. reported a significant positive association of SO 2, NO 2 and PM 10 with deaths because of COPD. In their single pollutant model, each 10 μg/m 3 increase in SO 2, NO 2 and PM 10 levels increased COPD mortality by 4.299%(95%CI:0.978-7.729%), 1.816%(95%CI: 0.515-3.313%) and 0.583%(95%CI: 0.055-1.113%) at lag0-3, respectively. 25 Chen et al. had reported that an increase in SO 2 (8 μg/m 3 ) and NO 2 (8 μg/ m 3 ) was associated with an increase of 4.3% (95%CI: 2.1-6.4%) and 3.6% (95%CI: 1.7-5.6%) in COPD mortality in patients aged 60 years and older. 26 Meng et al. 27 reported that an increase of 10 μg/m 3 in SO 2, NO 2 and PM 10  not find association between PM 10 and COPD mortality in the United States. 29 The conflicting findings in comparison with our study might be because of the different chemical composition of air pollutants in different regions, which may result in different effects of COPD hospitalisations and mortality.
In this study, the majority of the summary estimates for COPD hospital admissions were in the order of SO 2 > PM 10 > NO 2. The yearly mean concentration of SO 2 was low compared with the other two pollutants, but SO 2 had a stronger effect on COPD hospital admissions. The reason for this phenomenon remains unclear. Regional differences may contribute to these results. In our study, the small sample size and single study area may be the reason for our results. SO 2 was mainly came from burning of fossil fuels. It is very soluble and can be absorbed more when people move. The patients with COPD often have difficulty in breathing; they are more sensitive and suffer more damage. It can exert immediate effects on the respiratory tract at early contact. Particulate matter exposures may cause production of reactive oxygen and inflammatory factors in alveolar macrophages in humans. NO 2 exposures can exacerbate existing respiratory disease by impairing the functions of epithelial cells and alveolar macrophages, contributing to airway inflammation. 30,31 The specific mechanism of air pollution and COPD is unclear. It is suggested that the exacerbation of inflammation, oxidative stress and immunosuppression may play important roles in COPD. [32][33][34] Further studies are needed to demonstrate the mechanism of air pollution and COPD.
Our study had several limitations. First, this study was conducted in one region of China and in one hospital, and included a small number of patients with COPD. Second, other important unknown and unmeasured factors may have affected the results. For example, the data on body mass index (BMI), smoking, socioeconomic status and education level collected in this study may play an important role as effect modifiers. Finally, personal exposure to outdoor and indoor air pollution was inconsistent among the selected COPD patients. The last but not the least, personal exposure to outdoor air pollution and indoor air pollution may not accurately represented exposure levels in general population.
Fortunately, people in China have begun consideration to the health effects of air pollution and COPD, especially PM 2.5 (particulates with an aerodynamic diameter of ≤2.5um). Many studies have suggested that the adverse effects on COPD because of to PM 2.5 are greater than those of PM 10. 35-37 In 2020, our team had conducted a meta-analysis on the association between PM 2.5 and COPD hospitalisations and mortality. 10 Additionally, our findings regarding the effect of PM 2.5 on lung function in COPD are currently being summarised. Elucidating the mechanism of the effects of air pollution in COPD will be F I G U R E 3 Air pollutants and COPD hospital admissions (multiple-pollutant model: SO 2 and NO 2 ). SO 2 , sulfur dioxide; NO 2 , nitrogen dioxide; COPD, chronic obstructive pulmonary disease.
the main direction of our future research. Air pollution control is also a topic priority in the future.

| CONCLUSION
This study supported the association between air pollution and COPD hospital admissions. SO 2 and PM 10 may be important factors for the increase in COPD hospital admissions in Beijing, China. Air pollution is high, and it emphasises the need to reduce air pollution in Beijing, China. Further detailed studies are needed to confirm the present findings and to clarify the mechanisms.

AUTHOR CONTRIBUTION
Rui-xia, Zhu conducted literature collection and selection. Rui-xia Zhu wrote the manuscript. Jin Chen provided valuable advice on the entire article.
included as a co-author. The first author obtained permission from Professor Chen and Peking University to exclude mentors from this paper. The authors thank Xin-Jie Hui for providing statistical advice during the preparation of this manuscript. We also sincerely thank all patients who participated in this study.