The association between ambient inhalable particulate matter and the disease burden of respiratory disease: An ecological study based on ten-year time series data in Tianjin, China

The association between ambient inhalable particulate matter and the disease burden of respiratory disease: An ecological study based on ten-year time series data in Tianjin, China

Environmental Research 157 (2017) 71–77 Contents lists available at ScienceDirect Environmental Research journal homepage: www.elsevier.com/locate/e...

390KB Sizes 1 Downloads 19 Views

Environmental Research 157 (2017) 71–77

Contents lists available at ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

The association between ambient inhalable particulate matter and the disease burden of respiratory disease: An ecological study based on ten-year time series data in Tianjin, China

MARK

Qiang Zenga,1, Ziting Wub,1, Guohong Jianga, Xiaoyin Wub, Pei Lia, Yang Nia, Xiuqin Xiongb, ⁎ Xinyan Wangb, Parasatb, Guoxing Lib, , Xiaochuan Panb a b

Tianjin Centers for Disease Control and Prevention, China Department of Occupational and Environmental Health, Peking University, China

A R T I C L E I N F O

A B S T R A C T

Keywords: PM10 Time series study Respiratory disease Disease burden Years of life lost

There is limited evidence available worldwide about the quantitative relationship between particulate matter with an aerodynamic diameter of less than 10 μm (PM10) and years of life lost (YLL) caused by respiratory diseases (RD), especially regarding long-term time series data. We investigated the quantitative exposureresponse association between PM10 and the disease burden of RD. We obtained the daily concentration of ambient pollutants (PM10, nitrogen dioxide and sulphur dioxide), temperature and relative humidity data, as well as the death monitoring data from 2001 to 2010 in Tianjin. Then, a time series database was built after the daily YLL of RD was calculated. We applied a generalized additive model (GAM) to estimate the burden of PM10 on daily YLL of RD and to determine the effect (the increase of daily YLL) of every 10 μg/m3 increase in PM10 on health. We found that every 10 μg/m3 increase in PM10 was associated with the greatest increase in YLL of 0.84 (95% CI: 0.45, 1.23) years at a 2-day (current day and previous day, lag01) moving average PM10 concentration for RD. The association between PM10 and YLL was stronger in females and the elderly (≥65 years of age). The association between PM10 and YLL of RD differed according to district. These findings also provide new epidemiological evidence for respiratory disease prevention.

1. Introduction Air pollution affects a large group of people and is not easy to protect against; thus, it has become the focus of attention in recent years, and the research of air pollution on respiratory health is increasing (Guan et al., 2016). A large number of studies have confirmed that air pollution will adversely affect the respiratory disease health of the exposed population. The National Morbidity Mortality and Air Pollution Study (NMMAPS) in the United States found that an increase in particulate matter with an aerodynamic diameter of less than 10 μm (PM10) concentration (10 μg/m³) was associated with an increase in respiratory mortality with an RR value of 1.013 (95% CI: 1.005–1.020) (USEPA, 2004). Similarly, a study in England and Wales demonstrated that PM10 exposure in 2001 was associated with respiratory and cardiovascular

mortality in 2002–2009 with stronger associations for respiratory disease (OR 1.22 (95% CI: 1.04–1.44)) (Hansell et al., 2016). The results of the Chinese researchers also revealed the adverse effects of PM10 (Fang et al., 2016; Wang et al., 2016; Zhu et al., 2016). For example, China Air Pollution and Health Effects Study (CAPES) found a 10-ug/m3 increase in 2-day moving-average PM10 was associated with a 0.56% (95% CI: 0.31, 0.81) increase of respiratory mortality (Chen et al., 2012). Previous studies concerning the effect of PM10 on health have usually focused on the number of deaths, and less attention was paid to the age of death and composition ratio. Therefore, some information will be lost, and premature death due to the loss of life years (Years of Life Lost, YLL) can explain this discrepancy. Most studies have assessed the relationship between air pollution and total YLL/circulation system disease YLL. Zeng et al. found that the increase in PM10 was

⁎ Correspondence to: Department of Occupational and Environmental Health, School of Public Health, Peking University, No. 38, Xueyuan Road, Haidian District, Beijing 100191, China. E-mail addresses: [email protected] (Q. Zeng), [email protected] (Z. Wu), [email protected] (G. Jiang), [email protected] (X. Wu), [email protected] (P. Li), [email protected] (Y. Ni), [email protected] (X. Xiong), [email protected] (X. Wang), [email protected] (Parasat), [email protected] (G. Li), [email protected] (X. Pan). 1 Contributed equally to this paper.

http://dx.doi.org/10.1016/j.envres.2017.05.004 Received 14 February 2017; Received in revised form 3 May 2017; Accepted 4 May 2017 0013-9351/ © 2017 Elsevier Inc. All rights reserved.

Environmental Research 157 (2017) 71–77

Q. Zeng et al.

Tianjin Environmental Monitoring Center. Daily meteorological data (including temperature and relative humidity) were obtained from the Tianjin Meteorological Bureau. The daily mortality data regarding respiratory death (International Classification of Diseases 10th version (ICD-10: A00-R99)) were obtained from the Death Register and Report Information System, Tianjin Center for Disease Control and Prevention. The information on gender, age, date of death, cause of death and death code were included in this system. Years of Life Lost (YLL) is an indicator for the lost life year because of premature death. We used World Health Organization (WHO) standard life table (Supplemental Table S1), for the sake of comparing with the results in other country and district. YLL for each death was calculated by matching age to the WHO standard life table (Li et al., 2016). Daily YLL were calculated by summing the YLL for all deaths on the same day.

significantly associated with daily death and YLL in Tianjin (Zeng et al., 2016). A study in Guangzhou showed that a 10 μg/m3 increase above the corresponding threshold of 40.4 μg/m3 PM10 was related to a YLL increase of 2.8 years with a lag of 0–1 days for cardiovascular diseases (Yang et al., 2016). Kowalski M et al. reported that a reduction in shortterm exposure to PM10 by 5 μg/m3 resulted in a lower number of yearly non-external deaths (2.6–2.75 per 100,000 inhabitants) (Kowalski et al., 2016). Until now, few scholars have used YLL to evaluate the relationship between PM10 and respiratory system health in China. One study in Ningbo in southern China found that an increase of 10 μg/m3 in PM10 was associated with an increase of 0.74 (95% CI: 0.00–1.48) in respiratory disease YLL (He et al., 2016). However, no similar study has been conducted in northern China. The present situation of the environment in China is not optimistic, especially in three key areas, the Beijing-Tianjin-Hebei district, Pearl River Delta and Yangtze River Delta, where air pollution was significantly more severe than the national average (Ye et al., 2016). The Beijing-Tianjin-Hebei district is located in the heart of the Bohai Sea in northeastern China. Tianjin is the largest coastal city in northern China and is the most important industrial city in the Beijing-Tianjin-Hebei district. In this study, respiratory disease YLL was selected as the health effect indicator, using the atmospheric pollution and respiratory disease death monitoring data to explore the quantitative relationship between PM10 and the respiratory disease burden within the population. This study provides a scientific basis for regionalized environmental management policies and public health strategies for public health interventions in regions with high air pollution levels.

2.2. Statistical analysis The time series study is used for its good performance in demonstrating the exposure-response association between air pollution and health effects. generalized additive model (GAM), a combination of the generalized linear model and additive model, has become a standard method in air pollution epidemiology. This model can fit the air pollutant with some unknown confounders using parametric and nonparametric approaches. For example, we can control the non-linear confounders such as temperature by using a smooth function to estimate the risk of pollutant flexibly. The basic GAM is p

2. Materials and methods

g (μ) = α +

∑ f j (Xj ) j =1

2.1. Data

where μ is the expectation of Y, or μ=E(Y/X1,…,Xp). We collected the air pollutant monitoring data, meteorological data and mortality data from six urban districts in Tianjin (Fig. 1), including the Hedong district, Hebei district, Hexi district, Heping district, Nankai district and Hongqiao district, from January 1st, 2001 to December 31st, 2010. The daily air pollutant monitoring data (including PM10, nitrogen dioxide (NO2) and sulphur dioxide (SO2)) was obtained from the

g(·) is a link function, α is the intercept. fj(·) is a one-variable function of every predictive variable. The main aim of this study was to estimate the association between the concentration of PM10 and YLL of RD. Based on previous literature (He et al., 2016), the identity link function was used, and the basic

Fig. 1. Location of the target districts in Tianjin, China, 2001–2010.

72

Environmental Research 157 (2017) 71–77

Q. Zeng et al.

function by GAM was as follows:

E (Yi ) = α +

n

Table 1 Descriptive statistic of the daily YLL of RD in Tianjin, 2001–2010.

m

∑i −1 βi Xi + ∑ j −1 f j Zj )

(1)

where Yi is the observed daily YLL at day i, and for every i, Yi obeys the normal distribution with a population mean of E(Yi). E(Yi) is the expected value of the observed daily YLL at day i. X is the explanatory variable that may exert linear effect on the dependent variables, and it is the air pollutant indicator here. β is the coefficient of the explanatory variable, f is the nonparametric smooth function (penalized spline), Z is the variable that may exert a non-linear effect on the dependent variables, such as time, temperature and relative humidity. Specifically, 7 degrees of freedom (df) per year was used for time, and 3 df was used for temperature and relative humidity. Considering the lagging effects of temperature, 14-day moving average temperature was used. Detailed information is described in our previous study. The effect of air pollutants on daily YLL is measured by the increase in YLL, which is the β value from the model (1). Additionally, the increase of YLL is years of life lost for the whole population. A large number of studies in air pollution epidemiology shown that daily mortality is closely associated with the concentration of air pollutants that day and one day before (Schwartz, 2000). Thus, the average value of the two concentrations is usually adopted in most studies as the exposure concentration (Touloumi et al., 2004). We used a moving average concentration of the current day and the day before that day in our main analyses. We also examined the association using the concentration of air pollutants the current day (lag0) and the day before that day (lag1). We also investigated the validity of the linearity assumption on the relationship between PM10 and mortality. We replaced the linear term of the PM10 concentrations with a smooth function with 3 df using penalized spline. The likelihood ratio test with 2 df (which compares the original model with the smoothed model) was used. In addition, we stratified analyses by sex and age (≥65 and < 65 years). We also determined the association at a district level (Hedong district, Hebei district, Hexi district, Heping district, Nankai district and Hongqiao district). We used the auto correlation function (the lag time is 30 days) to determine whether the model is efficient to control the auto correlation of the time series to test the adequacy of the model. We also conducted a multi-pollutant model to test the stability of the results. We also change the df of relative humidity, temperature and time trends to test the robustness of results. In addition, we estimated the association between daily mortality and PM10. All data processing work, statistical analyses and model construction were conducted with the “mgcv” package using the statistical software R (version3.1.2).

Groups

Subgroups

Total Death

Daily Death

Total YLL

Daily YLL

Gender

Male Female elder (≥65y) Non-elder (< 65y) Hedong Hebei Hexi Heping Nankai Hongqiao

13,510 12,398 23,648 2260

3.7 ± 2.3 3.4 ± 2.3 6.5 ± 3.4 0.6 ± 0.8

237,099 204,032 351,644 89,487

64.9 ± 43.7 55.9 ± 41.4 96.3 ± 54.1 24.5 ± 34.3

4466 4604 6144 2945 4597 3152 25,908

1.2 ± 1.2 1.3 ± 1.3 1.7 ± 1.6 0.8 ± 1.0 1.3 ± 1.3 0.9 ± 1.0 7.1 ± 3.6

79,509 80,069 102,104 45,667 78,072 55,711 441,131

21.8 ± 23.6 21.9 ± 25.5 28.0 ± 30.3 12.5 ± 17.0 21.4 ± 23.6 15.3 ± 20.0 120.8 ± 67.4

Age

District

Total RD: respiratory disease. YLL: Years of life lost.

PM10 (lag01) and burden of RD. Table 4 shows the quantitative relationship between PM10 and burden of RD. For the total effect, significant positive effects were observed in all lag models, whereas lag01 showed the greatest effects on YLL of RD, which was 0.84 (95% CI: 0.45, 1.23) years per 10 μg/m3 increase. To further investigate the health effect of PM10, we explored the relationship between PM10 and YLL from RD in different subgroups. The effects of PM10 by different lag structures showed significant effects on daily YLL in females. For males, only PM10 on lag0 had a significant effect on YLL of RD, which increased by 0.23 (95% CI: 0.00, 0.45) years per 10 μg/m3 increase. For females, the greatest effect of PM10 on YLL appeared in lag01 (0.59 (95% CI: 0.34, 0.83) years) per 10 μg/m3 increase. The effects of PM10 on daily YLL of RD in both age groups were statistically significant. For the elderly population, a maximum effect was found in lag01 where the YLL increase was 0.59 (95% CI: 0.28, 0.89) years per 10 μg/m3 increase. For the non-elderly population, the maximum effect was found in lag01 where the effect was 0.26 (95% CI: 0.03, 0.48) years per 10 μg/m3 increase. We found that the effects of PM10 on daily YLL caused by RD were higher in the elderly population (≥65 years) than that in the younger population (< 65 years). The effects in six urban districts showed different trends. There was no statistical significance in the Hexi and Hongqiao districts. The greatest effect in Heping district was observed in lag1, with an increase in YLL of 0.09 (95% CI: 0.01, 0.17) years. The most significant effects in the other three districts were all found in lag01 where the effect values (YLL increment) in the Nankai, Hedong and Hebei districts were 0.24 (95% CI: 0.09, 0.38), 0.22 (95% CI: 0.08, 0.37) and 0.17 (95% CI: 0.02, 0.32) years per 10 μg/m3 increase, respectively. The residual autocorrelation function (ACF) plots of the models of the impact of PM10 concentration (lag01) on daily YLL were verified (Supplemental Figs. S1, S2). The results indicated that the established models of the effect of atmospheric PM10 on daily YLL comply with the modelling requirement. In addition, after we changed the degrees of freedom for relative humidity, temperature and time-trend, the association between PM10 and RD YLL was robust when compared with the former results (Supplemental Tables S2, S3). We also conducted the multi-pollutant model analysis (Supplemental Table S4). The results in the multi-pollutant model were similar with that in the single-pollutant model. The effects of PM10 on RD YLL and mortality are both significant (Supplemental Table S4).

3. Results The mortalities of RD were 25,908 in the central city of Tianjin during the study period (2001–2010), which caused 120.8 years of life lost (YLL) in daily level (Table 1). The PM10 concentration of the central city of Tianjin was defined as the mean PM10 concentration measured by the data from atmospheric monitoring sites in 6 urban areas (Hedong, Hebei, Hexi, Heping, Nankai, and Hongqiao) in Tianjin. The daily mean PM10 concentration was 117.6 μg/m³ (2001–2010). The daily temperature and humidity means in the corresponding period were 13.3 ℃ and 63.5%, respectively (Table 2). The monthly YLL, meteorological index and air pollution all showed obvious seasonal trends (Fig. 2). The Spearman correlation analyses between ambient pollutants and meteorological data are shown in Table 3. The correlation between gaseous air pollutants and mean temperature was strong, and the relationship between PM10 and mean temperature was relatively weak. We used GAM to describe the relationship between PM10 and YLL from RD. Fig. 3 shows the linear exposure-response curve between

4. Discussion In this study, Tianjin was selected as a typical city with a high level of air pollution. Based on the data of air pollution and population death data in Tianjin from 2001 to 2010, this study has determined new epidemiological evidence to prevent the adverse effects of air pollution. 73

Environmental Research 157 (2017) 71–77

Q. Zeng et al.

Table 2 Descriptive analyses of daily PM10 concentration and meteorological conditions for Tianjin, 2001–2010. Variables

District

Mean ± SD

Min

25th

Median

75th

Max

PM10 (μg/m3)

Total Hedong Hebei Hexi Heping Nankai Hongqiao Total Hedong Hebei Hexi Heping Nankai Hongqiao Total Hedong Hebei Hexi Heping Nankai Hongqiao Total Total

117.6 ± 70.9 114.7 ± 74.8 120.4 ± 77.3 111.7 ± 73.8 125.3 ± 79.0 113.9 ± 70.8 119.7 ± 72.3 51.7 ± 20.7 40.2 ± 21.2 47.3 ± 27.7 50.7 ± 24.8 57.8 ± 27.7 54.0 ± 25.0 59.9 ± 27.2 71.6 ± 60.5.3 66.3 ± 60.3 64.3 ± 61.7 70.3 ± 65.5 84.2 ± 83.4 71.1 ± 66.9 73.3 ± 67.6 63.5 ± 20.0 13.3 ± 11.1

9.7 1.0 8.0 5.0 5.0 9.0 7.0 8.7 3.0 1.0 2.0 2.0 3.0 3.0 4.3 1.0 2.0 1.0 2.0 2.0 2.0 7.0 −14.0

71.5 67.0 70.0 63.0 74.0 70.0 71.0 37.3 26.0 28.0 35.0 39.0 37.0 42.0 30.7 26.0 23.0 30.0 28.0 27.0 27.0 49.3 3.0

101.5 98.0 103.0 97.0 108.0 100.0 105.5 48.0 37.0 42.7 47.0 54.0 50.0 57.0 49.0 46.0 45.0 50.0 53.0 49.0 50.0 66.0 14.0

144.3 142.7 149.0 142.0 153.2 140.0 149.0 61.8 51.0 62.0 63.0 72.0 67.0 74.8 95.7 85.2 84.0 85.0 115.0 91.0 97.0 79.0 23.0

1154.8 1140.0 801.0 1434.0 1340.0 1260.0 954.0 186.2 170 241 184 229 230 234 468.3 494 469 546 578 548 551 100.0 33.0

NO2 (μg/m3)

SO2 (μg/m3)

Relative humidity (%) Temperature (℃)

SD: standard deviation PM10: particulate matter with aerodynamic diameter less than 10 μm.

We found that every 10 μg/m3 increase in PM10 was associated with an increases in YLL of 0.84 (95% CI: 0.45, 1.23) years with a 2-day moving average PM10 concentration (lag01). Females and elderly individuals were the most susceptible. The YLL was used as an indicator to assess the impact of air

pollution on the burden of disease in the population by taking not only the number of deaths into account but also the age at death. Because this approach considers life expectancy at different ages, it can provide a comprehensive scientific basis for policy setting and resource allocation (Bell et al., 2004; Li et al., 2016). Some previous studies abroad

Fig. 2. Boxplots of monthly years of life lost (YLL), meteorological index and air pollutants in Tianjin, China, 2000–2010. Note: Temperature (°C); Relative humidity (%); PM10 (μg/m3); SO2 (μg/m3); NO2 (μg/m3); YLL (years).

74

Environmental Research 157 (2017) 71–77

Q. Zeng et al.

Table 3 Spearman correlations between daily ambient pollutants and meteorological conditions in Tianjin, 2001–2010. Variables

PM10 (μg/m3)

NO2 (μg/m3)

SO2 (μg/m3)

Temperature (°C)

Relative humidity (%)

PM10 (μg/m3) NO2 (μg/m3) SO2 (μg/m3) Temperature (°C) Relative humidity (%)

1.00 0.54** 0.39** −0.08** 0.06*

1.00 0.71** −0.44** 0.07

1.00 −0.68** −0.00

1.00 0.22**

1.00

150

* P < 0.05. ** P < 0.01.

50 −100

−50

0

YLL

100

contamination (Bell et al., 2013; Guo et al., 2013; Kan et al., 2010; Qian et al., 2007; Wong et al., 2010; Zheng et al., 2013). For example, Guo et al. found that airborne particulate matter in Beijing had a significantly greater impact on daily YLL in women than in men (9.3 vs. 6.5) (Guo et al., 2013). Similarly, Wang X et al. in Beijing and Phung D et al. in Vietnam found higher risks due to the PM10 exposure for females relative to males (Phung et al., 2016; Wang et al., 2016). A study in Lanzhou showed more significant associations between PM10 and emergency room visits on dust days for elderly females (Ma et al., 2016). Similar results have been found in time series studies of mortality from air pollution in humans (Kan et al., 2010; Qian et al., 2007; Wong et al., 2010; Zheng et al., 2013). This may due to differences in the physiological structures. The lungs and airway diameters for females are narrower than in men, which may increase the sensitivity of the female airway (Bennett et al., 1996; Brown et al., 2002; Valavanidis et al., 2008). The particles themselves can not only be deposited into the human respiratory tract and lungs but can also carry toxic and harmful substances to the respiratory tract and lungs, which can impact human health (Brown et al., 2002; Valavanidis et al., 2008). Vrijens et al. (2016) found that there may be some differences between men and women in the expression of genes associated with PM10, which provided new insight and rationale to study the reasons behind these differences. All of these results suggest that females should pay more attention to protect themselves from atmospheric particulate pollution. This study found that the effect value of PM10 on daily YLL in the elderly population in Tianjin was higher than that in the non-elderly population. Previous studies have shown that older people are more sensitive to air pollution (Bell et al., 2013; Guo et al., 2013; Kan et al., 2010; Lu et al., 2015; Sandstrom et al., 2003), which leads to a greater threat for older people even exposed to lower concentrations of air pollutants (Torjesen, 2015). For example, Zeng et al. (2016) found the

0

200

400 PM10

600

800

Fig. 3. The exposure-response curve between PM10 (lag01, μg/m3) on daily YLL caused by RD in Tianjin, 2001–2010.

have used YLL as an indicator to evaluate the impact of air pollution (Hänninen et al., 2014; Lelieveld et al., 2015; Yoon et al., 2015), but these data are still very scarce in China. Based on the data in Beijing, Guo et al. found that an interquartile range (IQR) increase in PM10 was related to a non-accidental YLL increase of 15.8 years during 2004–2008 (Guo et al., 2013). Scholars in Nanjing found that IQR increases in the two-day PM10 average were significantly related to non-accidental YLL increases of 20.5 (95% CI: 6.3–34.8) years during 2009–2013 (Lu et al., 2015). Similar results were also found in a study using data from 2009 to 2013 in the city of Ningbo (He et al., 2016). This study confirmed the effects of PM10 on respiratory disease YLL using long-term time series data. To further explore the gender differences in the disease burden caused by the main pollutants in Tianjin, the studied population was stratified by gender. The results showed that the effect of PM10 on the daily YLL was greater for females. Previous studies have also shown that females are more sensitive than males to airborne particulate

Table 4 Daily YLL increment of RD caused by PM10 (per 10 μg/m³) in total population and different subgroups in Tianjin, 2001–2010. Groups

Subgroups

YLL increment (person years) lag0

Gender Age District

Male Female elder (≥65y) Non-elder (< 65y) Hedong Hebei Hexi Heping Nankai Hongqiao

Total

0.23 0.41 0.49 0.14 0.19 0.05 0.07 0.04 0.16 0.08 0.63

(0.00, 0.45)* (0.20, 0.61)* (0.24, 0.74)* (−0.04, 0.33) (0.07, 0.31)* (−0.07, 0.18) (−0.08, 0.22) (−0.04, 0.13) (0.04, 0.28)* (−0.02, 0.19) (0.31, 0.95)*

RD: respiratory disease. YLL: years of life lost. lag0: the concentration of PM10 the current day. lag1: the concentration of PM10 the day before the current day. lag01: the moving average of lag0 and lag1. * P < 0.05.

75

lag1

lag01

0.12 (−0.10, 0.34) 0.39 (0.18, 0.59)* 0.31 (0.05, 0.56)* 0.20 (0.02, 0.39)* 0.12 (−0.00, 0.24) 0.17(0.05, 0.30)* −0.00 (−0.15, 0.15) 0.09 (0.01, 0.17)* 0.16 (0.04, 0.28)* −0.03 (−0.14, 0.07) 0.51 (0.19, 0.83)*

0.26 0.59 0.59 0.26 0.22 0.17 0.05 0.10 0.24 0.04 0.84

(−0.01, 0.52) (0.34, 0.83)* (0.28, 0.89)* (0.03, 0.48)* (0.08, 0.37)* (0.02, 0.32)* (−0.14, 0.23) (−0.00, 0.20) (0.09, 0.38)* (−0.09, 0.16) (0.45, 1.23)*

Environmental Research 157 (2017) 71–77

Q. Zeng et al.

and Prevention for providing air pollution data, meteorological data and mortality data. The study was supported by the Natural Science Foundation of China [Grant no.: 81372950]; Key Scientific and Technological Project on Health of Tianjin [Grant no.: 16KG170]. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript

associations of PM10 on daily death counts and YLL were stronger in elderly people (≥65 years) than in younger people (< 65 years). He et al. (2016) in Ningbo also found that the increase in PM10 concentration in the elderly population (> 65 years) was significantly higher than that in the non-elderly population (≤65 years). However, some researchers found reverse results. For example, the results of Lu et al. (2015) showed that the effect of air pollutants on YLL in the population under 75 years of age in Nanjing was significantly higher than that in people aged 75 years and above. The discrepancy of the study results may be due to the considerable life expectancy at death and standard life tale should be used in further analysis. Therefore, it is necessary to carry out large-scale epidemiological studies to explore the impact of air pollution on YLL, such as the long-time series and multi-city research. In the district level analysis, we found that the Naikai district showed the most significant statistical association among the six urban districts, which implies that the greatest adverse effect of PM10 exists in this part of the city. We assume that the variations in pollutant characteristics may exist even within the central area of a city. It is widely acknowledged that the toxicity of air pollution is not only related to the concentration but also depends on the components of particulate matter and geographic factors (Wu et al., 2014; Zhang et al., 2012). The results suggested that further studies are needed to quantify the PM10-YLL of RD with high resolution within urban areas. Few studies have focused on the effects of PM10 on the respiratory disease burden. For the first time, a 10-year long-term data series in Tianjin was used to establish the exposure-response relationship between air pollutants and YLL. The confidence interval of the effect value from this long-term series study is narrow, and therefore, the results are more stable and accurate. Nevertheless, the exposure data of air pollutants in this study are derived from the fixed monitoring sites, and there is only one fixed monitoring point in each urban district of Tianjin, which could not accurately present the individual exposure concentration and may underestimate the effect (Ni et al., 2016; Zeger et al., 2000). In addition, fallacies may exist (Sedgwick, 2015), ecological studies generally examine groups of different people, but the individual micro-behavioural patterns may be significantly different from the macro-behavioural patterns of the whole group (Sedgwick, 2014).

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.envres.2017.05.004. References Bell, M.L., Dominici, S.J., Schwartz J, F., 2004. Time-series studies of particulate matter. Annu. Rev. Public Health 247–280. Bell, M.L., et al., 2013. Evidence on vulnerability and susceptibility to health risks associated with short-term exposure to particulate matter: a systematic review and meta-analysis. Am. J. Epidemiol. 178, 865–876. Bennett, W.D., et al., 1996. Variability of fine particle deposition in healthy adults: effect of age and gender. Am. J. Respir. Crit. Care Med. 153, 1641–1647. Brown, J.S., et al., 2002. Ultrafine particle deposition and clearance in the healthy and obstructed lung. Am. J. Respir. Crit. Care Med. 166, 1240–1247. Chen, R., et al., 2012. Association of particulate air pollution with daily mortality: the China air pollution and health effects study. Am. J. Epidemiol. 175, 1173–1181. Fang, X., et al., 2016. Bayesian model averaging method for evaluating associations between air pollution and respiratory mortality: a time-series study. BMJ Open. 6, e011487. Guan, W.J., et al., 2016. Impact of air pollution on the burden of chronic respiratory diseases in China: time for urgent action. Lancet 388, 1939–1951. Guo, Y., et al., 2013. The burden of air pollution on years of life lost in Beijing, China, 2004–08: retrospective regression analysis of daily deaths. BMJ 347, f7139. Hänninen, O., et al., 2014. Environmental burden of disease in Europe: assessing nine risk factors in six countries. Environ. Health Perspect. 122, 439–446. Hansell, A., et al., 2016. Historic air pollution exposure and long-term mortality risks in England and Wales: prospective longitudinal cohort study. Thorax 71, 330–338. He, T., et al., 2016. Ambient air pollution and years of life lost in Ningbo, China. Sci. Rep. 6, 22485. Kan, H., et al., 2010. Part 1. A time-series study of ambient air pollution and daily mortality in Shanghai, China. Res. Rep. Health Eff. Inst. 17–78. Kowalski, M., et al., 2016. Health benefits related to the reduction of PM concentration in ambient air, Silesian Voivodeship, Poland. Int. J. Occup. Med. Environ. Health 29, 209–217. Lelieveld, J., et al., 2015. The contribution of outdoor air pollution sources to premature mortality on a global scale. Nature 525, 367–371. Li, G., et al., 2016. Disease burden of ischaemic heart disease from short-term outdoor air pollution exposure in Tianjin, 2002–2006. Eur. J. Prev. Cardiol. 23, 1774–1782. Lu, F., et al., 2015. Short-term effects of air pollution on daily mortality and years of life lost in Nanjing, China. Sci. Total Environ. 536, 123–129. Ma, Y., et al., 2016. Association between ambient air pollution and emergency room visits for respiratory diseases in spring dust storm season in Lanzhou, China. Int. J. Environ. Res. Public Health 13, E613. Ni, Y., et al., 2016. The exposure metric choices have significant impact on the association between short-term exposure to outdoor particulate matter and changes in lung function: findings from a panel study in chronic obstructive pulmonary disease patients. Sci. Total Environ. 542, 264–270. Phung, D., et al., 2016. Air pollution and risk of respiratory and cardiovascular hospitalizations in the most populous city in Vietnam. Sci. Total Environ. 557–558, 322–330. Qian, Z., et al., 2007. Short-term effects of gaseous pollutants on cause-specific mortality in Wuhan, China. J. Air Waste Manag Assoc. 57, 785–793. Sandstrom, T., et al., 2003. The need for a focus on air pollution research in the elderly. Eur. Respir. J. (Suppl. 40), 92s–95s. Schwartz, J., 2000. The distributed lag between air pollution and daily deaths. Epidemiology 11, 320–326. Sedgwick, P., 2014. Ecological studies: advantages and disadvantages. Bmj 348, g2979. Sedgwick, P., 2015. Understanding the ecological fallacy. Bmj 351, h4773. Torjesen, I., 2015. Even low levels of air pollution are linked to more deaths in over 65s, study finds. BMJ 350, h3043. Touloumi, G., et al., 2004. Analysis of health outcome time series data in epidemiological studies. Environmetrics 15, 101–117. USEPA, 2004. Air Quality Criteria for Particulate Matter. Washington DC. Valavanidis, A., et al., 2008. Airborne particulate matter and human health: toxicological assessment and importance of size and composition of particles for oxidative damage and carcinogenic mechanisms. J. Environ. Sci. Health C Environ. Carcinog. Ecotoxicol. Rev. 26, 339–362. Vrijens, K., et al., 2016. Sex-specific associations between particulate matter exposure and gene expression in independent discovery and validation cohorts of middle-aged men and women. Environ. Health Perspect. Wang, X., et al., 2016. Spatiotemporal analysis for the effect of ambient particulate matter on cause-specific respiratory mortality in Beijing, China. Environ. Sci. Pollut. Res. Int.

5. Conclusions The association of PM10 with YLL of RD is well demonstrated by GAM. This study found there were some differences in the associations between different genders, age groups and districts. We used YLL data in relation to outcomes, which is more accurate to reflect the quantitative relationship between PM10 and RD. Currently, air pollution is a serious environmental problem in East Asia, as China and India and has already seriously threatened the health of public. These findings provide new epidemiological evidence for respiratory disease prevention suitable for regions with high air pollution levels. Conflict of interest The authors declare no conflicts of interest. Funding The study was supported by the Natural Science Foundation of China [Grant no.: 81372950]; Key Scientific and Technological Project on Health of Tianjin [Grant no.: 16KG170]. Acknowledgements The authors thank Tianjin Environmental Monitoring Center, Tianjin Meteorological Bureau and Tianjin Center for Disease Control 76

Environmental Research 157 (2017) 71–77

Q. Zeng et al.

Zeger, S.L., et al., 2000. Exposure measurement error in time-series studies of air pollution: concepts and consequences. Environ. Health Perspect. 108, 419–426. Zeng, Q., et al., 2016. The short term burden of ambient particulate matters on nonaccidental mortality and years of life lost: a ten-year multi-district study in Tianjin, China. Environ. Pollut. 220 (Pt A), 713–719. Zhang, Y., et al., 2012. The spatial characteristics of ambient particulate matter and daily mortality in the urban area of Beijing, China. Sci. Total Environ. 435–436, 14–20. Zheng, S., et al., 2013. Short-term effects of gaseous pollutants and particulate matter on daily hospital admissions for cardio-cerebrovascular disease in Lanzhou: evidence from a heavily polluted city in China. Int. J. Environ. Res. Public Health 10, 462–477. Zhu, Y.H., et al., 2016. Temperature modifies the acute effect of particulate air pollution on mortality in Jiang'an district of Wuhan. Zhonghua Yu Fang Yi Xue Za Zhi. 50, 519–24.

23, 10946–10956. Wong, C.M., et al., 2010. Part 5. Public health and air pollution in Asia (PAPA): a combined analysis of four studies of air pollution and mortality. Res. Rep. Health Eff. Inst. 377–418. Wu, S., et al., 2014. Association of cardiopulmonary health effects with source-appointed ambient fine particulate in Beijing, China: a combined analysis from the Healthy Volunteer Natural Relocation (HVNR) study. Environ. Sci. Technol. 48, 3438–3448. Yang, J., et al., 2016. Estimating years of life lost from cardiovascular mortality related to air pollution in Guangzhou, China. Sci. Total Environ. 15, 1566–1572. Ye, Q., et al., 2016. Haze is a risk factor contributing to the rapid spread of respiratory syncytial virus in children. Environ. Sci. Pollut. Res. Int. 23, 20178–20185. Yoon, S.J., et al., 2015. Measuring the environmental burden of disease in South Korea: a population-based study. Int. J. Environ. Res. Public Health 12, 7938–7948.

77