Control of chronic obstructive pulmonary disease in urban populations: findings from a cross-sectional prevalence survey in Shenzhen, China

The prevalence of chronic obstructive pulmonary disease (COPD) among urban populations is generally lower than rural residents, but the disease burden is still high. We conducted a cross-sectional prevalence survey of COPD among residents aged ≥40 years in an emerging city Shenzhen, China from September 2018 to June 2019. Through multi-stage stratified random sampling, a total of 4157 eligible participants were invited to complete a questionnaire and to take the spirometry test; 3591 with available data were enrolled in the final analysis. Individuals were diagnosed with COPD if the post-bronchodilator FEV1/FVC ratio was less than 0.7. The estimated standardized prevalence of COPD among residents over 40 years old in Shenzhen was 5.92% (95% confidential intervals [CI] 4.05–8.34). Risk factors for COPD included elder age (adjusted odds ratio 1.206, 95% CI 1.120–1.299 per 10-year increase), smoking over 20 pack-years (1.968, 1.367–2.832), history of chronic bronchitis (1.733, 1.036–2.900) or asthma (4.920, 2.425–9.982), and exposure to higher annual minimum concentrations of ambient SO2 (1.156, 1.053–1.270 per 1-μg/m3 increase). Among 280 spirometry-diagnosed patients, most (221, 78.93%) patients were classified as mild COPD (GOLD stage I). This survey found that the prevalence of COPD in Shenzhen is low and most patients had mild symptoms, thus recommended screening using spirometry in primary health care to detect early-stage COPD. Increased risk from the exposure to air pollutants also indicated the urgent need for environmental improvement in city settings.


Introduction
Chronic obstructive pulmonary disease (COPD) is one of the leading causes of death in the world, and its prevalence will further increase in the future because of the increasing smoking rates in developing countries and the aging population in high-income countries, as forecasted by the World Health Organization (WHO) (Halpin et al. 2021). A countrywide survey in China during 2014-2015 (Fang et al. 2018) showed that the prevalence of COPD among residents aged > 40 years old is 13.6%, which was significantly higher than that in 2007 (8.2%) (Zhong et al. 2007). The majority of patients detected by active screening were in the early stage of the disease with no symptoms or mild symptoms (Fang et al. 2018). Early intervention for these patients can effectively control the progress of the disease and reduce mortality (Zhou et al. 2017). While the prevalence of COPD in cities is generally lower than that in rural areas of China, better medical systems and equipment resources provide a solid foundation for the screening of early-stage patients and their long-term intervention. Understanding the characteristics of COPD patients and related risk factors in city settings is essential and a prerequisite for disease control.
The national survey  indicated that underweight, smoking, and high ambient concentrations of particulate matter with a diameter less than 2.5 (PM 2.5 ) increased the risk of COPD. Chinese cities have made great progress in controlling smoking and improving people's nutritional level, but air pollution has become a major threat (Guan et al. 2016). Limited data have shown how air pollutants impact COPD prevalence in China. To investigate the characteristics and risk factors for COPD in Shenzhen, China, we conducted a crosssectional epidemiological survey and collected the annual data on air quality for the analysis.
Shenzhen is an economically developed coastal city in southern China and attracted a large number of young internal migrants to this city for more work opportunists and higher income. It has a better environment than most inland areas. The annual average value of PM 2.5 is 19 μg/m 3 , and the comprehensive air quality index (AQI) ranks in the top ten cities in the country (Shenzhen Municipal Ecological Environment Bureau 2020). The prevalence of COPD in Shenzhen was expected to be low for its young population and good air quality. A pilot survey (Chen et al. 2017) in one district showed a COPD prevalence of 9.6% among residents over 40 years old. People who have related symptoms or exposure to risk factors would be suggested to take the spirometry test every year; however, seldom migrants actively take the examination due to mild symptoms or restricted access to medical services. More comprehensive strategies in Shenzhen are needed to better diagnose and manage COPD patients. Based on the epidemiological survey in all the ten districts of Shenzhen, we aim to estimate the overall prevalence of COPD and analyze patients' clinical characteristics, assess the risk of relevant demographic factors and environmental exposure, and formulate control strategies for the city.

Study site
Shenzhen is a coastal city in southern China, and its economy has continuously and prosperously developed over the past 40 years of reform and opening up. A large number of internal migrants have poured into this special economic zone, constituting nearly two-thirds of the 15 million residents in 2020 (Shenzhen Statistics Bureau, NBS Survey Office in Shenzhen 2019). This migration resulted in a relatively young residential population in Shenzhen, with an average age of 32.5 years (Shenzhen Statistics Bureau, NBS Survey Office in Shenzhen 2019). The city has a tropical oceanic climate with warm and humid air throughout the year. The average temperature in 2019 is 24°C, and only one day in the whole year had a temperature below 10°C (Shenzhen Meteorological Bureau 2020). Shenzhen also has good environmental conditions with 43.4% land of the green area and was titled "the National Forest City" in 2018.

Sampling methods
In this COPD prevalence survey from September 2018 to June 2019, we enrolled eligible participants ≥ 40 years old with residence at least 6 months in Shenzhen using the multi-stage stratified cluster random sampling method. The numbers of communities of each district were determined based on the number of resident populations. A total of 40 communities from ten districts were selected, and about 120 households were selected from each sampled community. The Kish Grid method was used to randomly select one resident over 40 years old from each household for investigation. Excluded were patients with mental illness or cognitive impairment, cancer, high paraplegia, pregnant or breastfeeding women, and other subjects who cannot cooperate to complete the epidemiological investigation.
This survey assumes that the prevalence rate is 9.6% based on the previous survey in Shenzhen (Chen et al. 2017). Under the condition that the statistical error probability (α) is 0.05 and the allowable error is 80%, the sample size of each group is estimated to be 1852. Stratified by gender and considering the non-response rate of 15%, the total sample size of the survey is estimated to be 4260.

Spirometry test and parameters
Qualified subjects would undergo the pulmonary function test by spirometry and take a structured questionnaire survey after signing the informed consent form. The trained and certified staff used the spirometer (BTL Science and Technology (Shenzhen) Co., Ltd.) to perform pulmonary function examination and bronchodilation test on all the included subjects. The measurement indicators mainly include the forced expiratory volume in one second (FEV1), forced vital capacity (FVC), and 1-s rate (FEV1/FVC). Each respondent should complete three valid pre-bronchiectasis pulmonary function tests and then perform a bronchodilation test 15 min after inhaling 400 mg of the bronchodilator salbutamol aerosol. Subjects allergic to salbutamol or resting heart rhythm higher than 100 bpm were excluded.
According to the 2017 Global Chronic Obstructive Pulmonary Disease Action (GOLD) guidelines , we defined the diagnostic criteria for COPD as the post-bronchodilator FEV1/FVC ratio < 0.7. The severity of limitation of lung function among patients of spirometryconfirmed COPD was classified according to the percentage of FEV1 in the predicted value (pv): mild (GOLD stage I: FEV1/pv ≥ 80%), moderate (GOLD stage II: 50% ≤ FEV1/ pv < 80%), severe (GOLD stage III: 30% ≤ FEV1/pv < 50%), and extremely severe (GOLD stage IV: FEV1/pv < 30%). Patients who were diagnosed with COPD would have further chest X-ray examination and COPD Assessment Test (CAT) (Jones et al. 2009).

Epidemiological investigation and exposure assessment
Face-to-face surveys were conducted using a structured questionnaire to obtain patients' basic demographic information, knowledge of COPD, respiratory disease history, family disease history, respiratory symptoms, exposure to smoking, and other risk factors. The height and weight of the participants were measured at the survey, and the body mass index (BMI) was calculated as the weight in kilograms divided by height in meters squared. According to the GOLD document , the main types of risk factors for COPD are as follows: "indoor biofuel exposure" is defined as households using biofuels for cooking or heating; "coal exposure" as the households using coal fuel for cooking or heating; and "workplace exposure to dust or chemicals" as the exposure in the workplaces for at least 12 months to dust in the air or harmful chemical gases.
Annual levels of air pollutants were detected by the environmental monitoring stations in each district. The indicators included the air quality index (AQI) and the ambient concentrations of six air pollutants, including PM 2.5 , PM 10 , SO 2 , NO 2 , CO, and O 3 . The original monitoring data were obtained from the China National Environmental Monitoring Center, which contained the 24-h measurement values from 11 monitoring points throughout 2018. We took the mean of the measured values from 8:00 a.m. to 8:00 p.m. as the daily environmental exposure for the daytime (Guo et al. 2018), and the period also included rush hours in Shenzhen when people most often travel for work (Chen et al. 2020). We calculated the means of daily exposure throughout the year for all the air quality indicators to represent their average annual exposure. The minimum and maximum daily values in the year of each indicator were also assessed for their impact on COPD. For the sensitivity analysis, the mean of the 24-h measurements was taken as the daily exposure and was used to calculate the average, minimum, and maximum annual exposure as mentioned above.

Data processing and statistical analysis
The original questionnaire data is entered into EpiData, exported to Excel format, and saved after data cleaning. Weighted by the gender-age ratio in the 2017 Shenzhen Population Survey (Shenzhen Statistics Bureau, NBS Survey Office in Shenzhen 2017), the standardized prevalence and 95% confidence intervals (CI) were estimated overall and for subgroups. The chi-square test was used to compare the differences in prevalence between groups, and the Cochran-Armitage trend test was used for trend analysis. Univariable logistic regression was used to assess the impact on COPD of demographic characteristics, personal and family disease history, and exposure of risk factors. Variables with a P value < 0.2 were included in the multivariable regression, and stepwise regression was used to include the variables. In the final model, variables with a P value < 0.05 were regarded to have significant correlation with COPD. Statistical analysis was performed using R (https://www.r-project.org/; version 4.0.3).

Basic characteristics
A total of 4,157 people were investigated in the survey, and 3591 (86.48%) were included for analysis who had available data of both epidemiological data and pulmonary function tests. The participants included 1603 (44.6%) males, the average age was 55.25 ± 9.46 years, the average BMI was 24.44 ± 3.20 kg/m 2 , and the average annual exposure to PM 2.5 was 24.24 ± 2.08 μg/m 3 . Low proportions of participants reported ever being diagnosed with other lung diseases, including 150 (3.27%) with chronic bronchitis, 50 (0.80%) with asthma, 66 (1.29%) with tuberculosis, and 88 (2.70%) who were hospitalized for lung diseases during childhood. Regarding exposure to smoking, a total of 556 (20.19%) people were current smokers, of which the vast majority (531, 95.50%) were men and only 25 (4.50%) were women; 102 (2.13%) participants reported that their mother smoked during pregnancy. Regarding the exposure to other risk factors, 669 (18.75%) people were exposed to dust or chemicals in workplaces, 9 (0.23%) people to biofuels at home, and 12 (0.41%) people to coal at home. The basic demographic characteristics and family history of the study population were listed in Table 1.
The prevalence of COPD also showed a difference in different areas of Shenzhen. The crude prevalence in the central area was the highest, reaching 9.34% (7.89-10.95), and lowest in the eastern region (5.44%, 3.14-8.69), of which the prevalence in Pingshan District was the lowest of only Note: Continuous variables were expressed as mean ± standard variation (m ± sd), and category data as counts (proportion) 3.51% (Fig. 2a). Residents in the central area were elder than in northern or eastern areas, which probably lead to the higher crude prevalence of COPD. However, standardized estimates still indicated the discrepancy among COPD prevalence in different areas. The standardized prevalence in central, northern, and eastern areas were, respectively, 7.59% (5.02-10.93), 5.60% (3.80-7.93), and 3.86% (0.85-11.61) (Fig. 2b).

Risk factors for COPD
To evaluate the impact on COPD of the air conditions including AQI and the ambient concentrations of six major air pollutants, we performed univariable logistic regression to include the annual average as well as the minimum and maximum daytime exposure of each indicator (Table 3). The results showed that the maximum value of PM 10 , the average and minimum value of SO 2 , and the average value of CO were significantly correlated with COPD. After standardization by gender and age, the effect of the maximum value of PM 10 and the minimum value of SO 2 remained statistically significant. The risk of COPD increased by 3.8% (0.5-7.1) as the maximum concentration of PM 10 increased every 10 μg/m 3 . For every 1 μg/m 3 increase in the minimum concentration of SO 2 , the risk of COPD increased by 11.8% (2.4-22.1). In the sensitivity analysis using 24-h estimates as the daily exposure, the impact of these indicators on COPD was mostly similar, with a statistical significance in the minimum SO 2 exposure (P=0.030) and a slight significance in the minimum (P=0.082) and maximum of NO 2 exposure (P=0.088). The results of univariable logistic regression on other risk factors showed that smoking, education level, medical insurance type, lung disease history, and environmental exposure were significantly correlated with COPD (Table 4). As shown in the final multivariable logistic regression model, the risk of COPD increased by 1.206 (1.120-1.299) times as every 10 years of age increased and by 1.156 (1.053-1.270) times per every 1 μg/m 3 increase of the annual minimum air SO 2 concentration; people who smoked greater than 20 packyears (OR 1.968, 95% CI 1.367-2.832), whose mother smoked during pregnancy (OR 1.881, 95% CI 1.039-3.405), or who had been diagnosed with chronic bronchitis (OR 1.733, 95% CI 1.036-2.900) or asthma (OR 4.920, 95% CI 2.425-9.982) significantly had a higher risk of COPD (Table 4).

Clinical features of COPD patients
Among the 280 patients diagnosed with COPD, most (221, 78.93%) patients were classified as mild COPD (GOLD stage I), 46 (16.43%) patients were moderate COPD (GOLD stage II), and only 13 cases (4.64%) were severe COPD (GOLD stage III/IV). Among 235 patients who underwent COPD Assessment Test (CAT), 15 patients (6.4%) reported no related symptoms, and the other 220 patients (93.6%) reported either respiratory symptoms (cough, sputum, sputum, chest tightness, wheezing, restricted activity) or nonrespiratory symptoms (restricted going-out, poor sleep, and fatigue). The above eight symptoms were evaluated on a scale of 0-5, and the results showed that the median CAT score of these patients was 5 points (interquartile range 2-9 points). Among the investigated patients, 24.7% (58/235) had a total CAT score of ≥ 10 points and were regarded to have severe symptoms, as defined by GOLD (Singh et al. 2019). Lowering the thresholds to detect symptomatic patients, we found that 55.7% (131/235) patients had a total score of over 5 points, and 82.6% (194/235) had a total score of over 2 points, where patients were clinically considered to be symptomatic (Fig. 3a). Among patients with COPD of GOLD stage I, II, and III or above, the proportions of patients with severe symptoms (CAT score ≥ 10) were 18.6%, 38.5%, and 69.2%, respectively. About half of mild (52.5%) and moderate (59.0%) COPD patients gained a CAT score of ≥ 5 points, while most (92.3%) severe patients could be detected under this threshold. The proportions of patients who were clinically considered to be symptomatic (CAT score ≥ 2) were 79.8%, 89.7%, and 100.0%, respectively (Fig. 3b). Nearly half of the patients (44.3%, 104/235) reported at least four symptoms. The proportions of patients reporting ≥ four symptoms among mild, moderate, and severe COPD were 42.6%, 43.6%, and 69.2%, respectively ( Fig. 3c/d). The most frequently reported  (Fig. 3e). Among the above four most common symptoms, patients with any of them accounted for 88.5% (208/235).

Discussion
This cross-sectional survey estimated that the standardized prevalence of COPD among urban residents over 40 years old in Shenzhen is 5.92% (95% CI 4.05-8.34), which was greatly lower than the average level of 13.6% across China (Fang et al. 2018). Risk factors including elder age, smoking, history of lung diseases, and exposure to higher ambient concentrations of SO 2 were found significantly associated with the diagnosis of COPD. Most of the diagnosed patients had mild symptoms and were classified as GOLD stage I. Given that early intervention will be the most beneficial for patients with mild COPD (Zhou et al. 2017), we suggest using spirometry in primary care facilities of resourceful cities in order to promptly detect early-stage patients and effectively control the disease progression. Fig. 3. Related symptoms and COPD Assessment Test (CAT) score of COPD patients. The total proportion of symptomatic patients whose CAT score and total score are higher than each threshold (a) and the stratification by GLOD level (b); cumulative proportions of patients reporting different numbers of CAT items (c) and the stratified by GLOD levels (d); (e) the number of people reporting various CAT items.
In our study, the prevalence of COPD among the urban populations over 40 years old in Shenzhen was far lower than the prevalence rate of 13.6% obtained from the 2014 to 2015 national epidemiological survey (Fang et al. 2018) and also lower than the global estimate in 2010 of 11.7% (Adeloye et al. 2015). The prevalence of COPD in other cities in China showed similar levels, such as 6.5% in Shanghai (Zhong et al. 2007), 5.5% in urban areas of Shanxi Province (Zhong et al. 2007), or 6.0% in Heilongjiang Province (Han et al. 2015). More young populations in cities could be one of the reasons for the lower overall prevalence of COPD. Besides, we also observed a significant decrease in COPD prevalence among different age groups, especially among elderly populations-12.7% and 14.9% in Shenzhen versus 21.2% and 35.5% nationally among people aged 60-69 and aged ≥70 years, respectively . We assume that better air quality and improved indoor environment in cities have long-term benefits and prevent people from chronic lung diseases including COPD (Faizan & Thakur 2019). As statistics in our study, 18.75% of investigated people reported exposure to dust or chemicals at workplaces, and only 0.23% and 0.41% were, respectively, exposed to biomass or coal during cooking or heating in households. These figures were all extraordinarily lower than the national levels (46.6%, 40.7%, and 34.3%, respectively) (Fang et al. 2018).
Smoking (Biener et al. 2019), the well-known risk factor for pulmonary diseases, also showed a lower level in Shenzhen than nationally (28.3% versus 40.2%).
Of note, shown were differences in the prevalence of COPD in various regions of Shenzhen. The standardized prevalence in the central region (7.6%) was almost twice that in the eastern coastal areas (3.9%). The lower population density, fewer cars, and better air quality in the eastern region probably led to this difference. Although PM 2.5 , the main indicator for air quality, did not show a significant impact on COPD in our data, we found that exposure to higher ambient concentrations of SO 2 increased the risk of COPD, and its annual minimum value presented the most sensible impact. This result was in line with a recent observation from an Australian cohort, where exposure to higher concentrations of SO 2 was associated with higher incidence of COPD in women of different age groups (Hendryx et al. 2019). SO 2 is one of the components of air pollutants resulting from industrial processes or fossil fuel combustion, which might resource from the crowded traffic in city centers or dense factories in Shenzhen. The spatial analysis in Mongolia showed that the ambient concentrations of SO 2 and other traffic-related pollutants were higher in city centers or near main roads (Huang et al. 2013). Other air pollutants might also increase the risk of COPD, as presented by a study in the UK that higher concentrations of PM 2.5 , PM 10 , and NO 2 were found significantly associated with a higher prevalence of COPD (Doiron et al. 2019). Vehicle emissions could be the main source of air pollution and potential reasons for respiratory diseases in city settings. To improve the environment in cities, policies were implemented including restrictions of vehicle population and optimization of transportation structure (Sun et al. 2020). Comprehensive strategies for the development of healthy cities were also urgently announced (Guan et al. 2016).
Most of the patients (78.93%) diagnosed in this survey were mild COPD and classified as GOLD stage I. The proportion was even higher than that obtained from the national survey in 2014-2015 (56.4%) . COPD patients detected by screening in other countries are mainly moderate-level patients. For example, 30.6% and 51.4% of COPD patients diagnosed in a survey in Poland were classified as mild (GOLD stage I) and moderate level (GOLD stage II), respectively (Bednarek et al. 2008). Early diagnosis and treatment of patients with mild COPD can more effectively control their lung function status and significantly reduce their progression and deterioration (Guirguis-Blake et al. 2016). As recommended by GOLD (Vestbo et al. 2013), spirometry is considered a necessary tool for screening mild COPD patients, which is a basic tool for COPD diagnosis as well as the grading of lung function. In our survey, the proportion of people who have undergone spirometry before our investigation was very low, and even among COPD patients, the proportion was still less than 10%. This indicated that the current screening in Shenzhen is insufficient to timely detect COPD patients in an early stage. For early diagnosis and effective control, spirometry, which helps to identify early-stage COPD patients and to intervene promptly, is necessary for the routine examinations of people over 40 years old, as suggested by the Health Regulations in Shenzhen Special Economic Zone (Standing Committee of Shenzhen Municipal People's Congress 2020).
The CAT score can be used to effectively screen high-risk groups with respiratory symptoms or restricted daily activities (Jones et al. 2009), but the overall CAT score of the patients diagnosed in this survey was generally low. Half of the patients obtained less than 5 points, suggesting that the threshold of CAT score should be lowered in screening COPD patients. For example, if the CAT total score is two or more, 82.6% of mild COPD patients would be recommended to further examinations for confirmation. Other studies (Houben-Wilke et al. 2018;Jones et al. 2011) agreed that the CAT score is useful in screening COPD patients, and Kon S. and his colleagues (Kon et al. 2014) also proposed that a reduction of more than 2 points could assess the effect of the intervention in COPD patients. However, some patients who had no self-reported symptoms or who had a total CAT score of fewer than two points can only rely on spirometry for screening.
This study had some limitations. Firstly, we used the national air surveillance data which included only 11 monitors in Shenzhen. Data from the municipal-level monitors can provide more accurate measurements of personal exposure. Then, this study only involved residents of a specific city in China where numerous young people migrate from rural areas to this well-developed metropolis. Further research is needed to characterize the prevalence of obstructive pulmonary diseases in other cities. And a cost-effectiveness analysis will also support the involvement of different screening strategies.

Conclusions
This cross-sectional survey found a relatively low prevalence of COPD among the urban population over 40 years of age in Shenzhen. Spirometry and a low threshold 2-point CAT score are recommended for COPD screening. Further efforts to improve air quality, including optimizing transportation and industrial processes, are needed in cities.