Association between ambient ne particulate matters and chronic obstructive pulmonary disease (COPD) mortality: an analysis in Southeastern China

Background: The objective of this study was to investigate the association between ambient particulate matters(PMs)and chronic obstructive pulmonary disease (COPD) mortality. Methods: Generalized Additive Mixed Model was employed to investigate the effects of ambient ne and coarse PMs on COPD mortality using 13,066 deaths from 2014 to 2016 among six cities in Zhejiang Province in Southeastern China. Results: The daily average death count due to COPD was 3, varying from 1 to 7among six cities. The daily 24-hour mean concentrations were diverse among cities, from 29.7 to 56.8 µg/m 3 for PM 2.5 , 16.7 to 30.3 µg/m 3 for PM 2.5−10 , and 50.3 to 87.1 µg/m 3 for PM 10 , respectively. The analysis showed that daily exposure to PM 2.5 and PM 10 was associated with increased mortality due to COPD and that weak effects were observed between PM 2.5−10 and COPD mortality. Conclusions: Our results provided evidence that the ne particles in air pollution have stronger functions on adverse health effects other than coarser particles in Southeastern China, which may be considered as a potential clinic target in PM-associated COPD. in lags of 1, 2, 3, 5, 7, 0–1, 0–2, and 0–3 days (ER = 1.03%, 95% CI: 0.361%, 1.70%; ER = 1.09%, 95% CI: 0.427%, 1.75%; ER = 0.868%, 95% CI: 0.214%, 1.53%; ER = 0.671%, 95% CI: 0.015%, 1.33%; ER = 0.751%, 95% CI: 0.096%, 1.41%; ER = 1.10%, 95% CI: 0.299%, 1.91%; ER = 1.55%, 95% CI: 0.668%, 2.44%; and ER = 1.85%, 95% CI: 0.895%, 2.82%, respectively). We also observed a signicant association between COPD mortality and PM 2.5−10 in lags of 2, 0–2, and 0–3 days (ER = 1.09%, 95% CI: 0.0680%, 2.13%; ER = 1.54%, 95% CI: 0.164%, 2.93%; ER = 1.50%, 95% CI: 0.058%, 2.95%, respectively). We also observed a signicant association between COPD mortality and PM 10 in lags of 1, 2, 3, 5, 0–1, 0–2,and 0–3 days (ER = 0.711%, 95% CI: 0.236%, 1.19%; ER = 0.797%, 95% CI: 0.333%, 1.26%; ER = 0.504%, 95% CI: 0.0440%, 0.966%; ER = 0.495%, 95% CI: 0.0370%, 0.955%; ER = 0.764%, 95% CI: 0.189%, 1.34%; ER = 1.09%, 95% CI: 0.468%, 1.71%;and ER = 1.22%, 95% CI: 0.559%,

Zhejiang Province is located in the Yangtze River Delta (YRD) region, which is considered one of the most rapidly developing regions in China.
Due to its urbanized and industrialized processes, Zhejiang Province has severe PM pollution like many Chinese areas [17,18].We selected Hangzhou (HZ), Jinhua (JH), Lishui (LS), Ningbo (NB), Taizhou (TZ), and Zhoushan (ZS), all cities from Zhejiang Province, as survey cities, which account for 54.6% (6/11) of its total cities. The locations of the cities are shown in Fig. 1. The urban areas of each city were selected as survey areas due to the same cover of air-quality monitors and mortality.

Mortality data
Data on mortality were collected from the local mortality register of the Zhejiang Provincial Center for Disease Prevention and Control. Based on the 10th revision of the international classi cation of diseases and related health problems (ICD-10), the deaths due to RDs (RD: ICD-10 codes J00-99) and COPD (COPD: ICD-10 codes J40-44) were selected from January 1, 2014, to December 31, 2016, in HZ and ZS; January 1, 2015, to December 31, 2016, in JH and LS; and January 1, 2016, to December 31, 2016, in NB and TZ. Since deaths due to COPD accounted for 64.8% (13,066/20,168) of mortality due to RD according to the collected data, we selected COPD as the target disease in our study.

PM pollution and weather variables
Concentrations of PM 2.5 , PM 2.5−10 , and PM 10 were monitored by ten environmental monitoring stations in urban areas of HZ, three urban areas of JH, three urban areas of LS, eight urban areas of NB, three urban areas of TZ, and one urban area of ZS during the study period. Twentyfour-hour means were applied for both air pollutants. The average of xed monitoring stations in each city was used as the daily concentrations of PM 2.5 , PM 2.5−10 , and PM 10 . To allow for the adjustment of weather conditions on mortality, we obtained daily mean temperature, atmospheric pressure, relative humidity, wind speed, and precipitation measurements for each city from the Zhejiang Meteorological Administration.

Statistical analyses
The descriptive statistics were calculated for COPD mortality, PM pollutants, and meteorological factors. The ANOVA test was used to compare the differences among cities with those variables, and the Spearman's correlation was used to evaluate the correlations among those variables.
We examined the acute effects of PM pollutants (PM 2.5 , PM 2.5−10 ,and PM 10 ) on COPD mortality using the Generalized Additive Mixed Model, which was used to account for the random effect of a city to the additive predictor [19,20]. A quasi-Poisson distribution was applied to account for overdispersion of daily deaths from COPD. In our model, several confounders were considered. Firstly, we adjusted possible variations in a week using dummy variables for day of the week (DOW). Then time trends, seasonal patterns, and weather conditions were controlled using penalized smoothing splines to exclude potential nonlinear effects on health [15,21]. The degrees of freedom (df) for the splines were determined via generalized cross validation (GCV) [22,23]. To examine the association with lag effect, we used the concentrations of PM pollutants from the current day (lag0) to the previous six days (lag6), as well as moving the average of current and previous 1,2, and 3 day: lag01, lag02, lag03. We presented the results as Excess Risk (ER) with a 95% con dent interval (CI) for COPD mortality per 10 µg/m 3 increase of PM pollutants. The model used appeared as follows: Log [E(Y t )] = α + βPM t−i + s (time, df) + s (X t , df) + re(city) + DOW where E(Y t ) is the expectation of the number of daily COPD mortality at day t; αis the intercept; PM t−i is the concentration of PM pollutants in lag(i) day, i = 0 to 6 and 01 to 03; β is the regression coe cient; s() refers to the function based on penalized smoothing splines; df is degrees of freedom; time represents temporal trend; X t refers to the meteorological factors, such as daily mean temperature, atmospheric pressure, relative humidity, wind speed and precipitation; re(city)represents the random effect of city using categorical variables. DOW represents the dummy variables for the day of the week.
We further examined the nonlinear associations between PM pollutants and COPD mortality [24]. The lag time was selected according to the minimum values of AIC in the single lag model. To exhibit the nonlinear response more clearly, we generated expose-response curves to visualize the effects of PM pollutants on COPD mortality using the natural splines.

Discussions
In the present study, investigation using 13,066 deaths from 2014 to 2016 among six cities in Zhejiang Province revealed that daily exposure to PM 2.5 and PM 10 was associated with increased mortality due to COPD, and that fewer effects were observed between PM 2.5−10 and COPD mortality, suggesting that the ne particles in air pollution have stronger functions on adverse health effects other than coarser particles, which may be considered as a potential clinic target in PM-associated COPD.
Our previous study provided evidence that both the mortality rates and outpatient visits for RDs were signi cantly associated with air pollution among which PM 2.5 plays a crucial role in two typical cities (HZ and ZS) in Zhejiang Province [16]. In order to further investigate the effects of particulates with different diameters, more data with a longer period from six cities with different air conditions in Zhejiang Province were collected. Our results showed that the average death rate due to COPD varied sharply among the six cities, which may be due to differences in economic level, medical level, or population size of those cities, although differences in PM pollutants among six cities may contribute to the change in health effects to some extent. Therefore, we compared the death counts in the same cover of air-quality monitors strati ed by the city and applied the GAMM to account for the random effects from a different city. The positive correlation between COPD mortality and the concentrations of PM 2.5, PM 2.5−10, and PM 10 preliminarily indicated the increase in concentrations of PM pollutants may result in the increase in mortality counts due to COPD. The signi cant effects of meteorological factors on COPD mortality were also found. Thus, we controlled the effects of weather conditions in our model with covariates.
Both PM 2.5 and PM 10 signi cantly affected the COPD mortality in this study, which was similar to some previous epidemiologic studies. The signi cant association with COPD mortality was observed in two cohort studies from 34 cities in the United States (ER:22% for PM 10 ) [26] and Norway(Male ER: 29% for PM 10 and 27% for PM 2.5 ; Female ER: 6% for PM 10 and 9% for PM 2.5 ) [27]. A case-crossover study from Barcelona found increased PM 10 associated with the higher mortality of COPD(ER:11%) [28]. Another time series study from 10 US cities [29] and Hong Kong [30] also found the increased COPD mortality risk was associated with PM 10 (ER: 1.7% for the Unite States and 1.0% for Hong Kong). On the other side, several studies only observed signi cant associations in speci c groups. A time series study in Netherlands reported the effect of PM 10 on COPD mortality was found in some age groups [31]. Moreover, a cross-sectional study in Japan also only observed the signi cant association between PM 2.5 , PM 10 pollutants, and COPD mortality in females [22]. Compared with the ER values in previous studies, we found the values in cohort and case-crossover studies were largely higher than those in a time series design, which was in part due to the bigger power of the test in cohort and case-crossover design. Meanwhile, the ER values in this time series study were consistent with the results from other time series studies [29,30]. Best lags for PM 2.5 and PM 10 were lag 2 for single-day and lag 03 for multiple-day, respectively. Those were similar to the Hong Kong study [30], and differ from the US study [29], which suggested delay effects are varied in different regions partly owing to spatial variation of pollutants.
The particulate air pollution problem is one of the severest problems that a country normally would be facing during an industrialization process and social development. As one of the largest developing countries in the world, China is going through this particular process, and Zhejiang Province is China in miniature in a sense, as it is located in Yangtze River Delta (YDR) region, one of the fastest developing areas in the country. Therefore, analysis using data from six cities with varying environment conditions in Zhejiang Province may provide solid evidence to estimate the relationship between particulate air pollution and health problems in China to a certain extent.
As we all know, PM 2.5 was the most widely used risk indictor when the disease burden of ambient air pollution was estimated, but adverse exposure was related to increased mortality of respiratory and cardiovascular diseases, but the relationship became weaker and yielded less precise effect estimates after adjustment for PM 2.5 and publication bias [13]. These results are consistent with our nding that COPD mortality was associated to PM 2.5 and PM 10 , but weak and less association between PM 2.5−10 and COPD mortality was observed after adjustment for PM 2.5 , indicating that ne particulates rather than coarser ones may have stronger effects on COPD patients. Similar results were also observed in another study based on data from 272 cities throughout China. It found signi cant associations between short-term PM 2.5−10 exposure and daily non-accidental and cardiopulmonary mortality, but a city-speci c situation was observed [12].
The limitations of the present analysis should also be noted. One limitation of our study was that although Zhejiang Province is a good representative area in China, only six cities in Zhejiang Province were recruited. Therefore, more cities in Zhejiang Province and even more provinces should be included in further studies. Another limitation was that the study period in each city was not identical due to the data system of each city. Therefore, data from longer periods are better to be used as the basis of analysis. Additionally, data of morbidity rate and outpatient visits should also be analyzed in future studies.

Conclusions
In conclusion, signi cant and strong associations between COPD mortality and PM 2.5 as well as PM 10 were observed using data from six cities in Zhejiang Province in the present study. PM 2.5−10 was also signi cantly related to COPD mortality but with much weaker effects. Our results provided evidence that ne particulate matter has worse adverse health effects on respiratory disease from an epidemiologic point of view.

Declarations
Availability of data and materials  Figure 1 Map of Zhejiang Province, China, highlighting the cities of Hangzhou, Jinhua, Lishui, Ningbo, Taizhou, and Zhoushan as survey areas. Note:

Figures
The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Excess risk (95% con dent intervals) of COPD mortality associated with a 10 μg/m3 increase in PM pollutants in Zhejiang Province, 2014-2016.

Figure 3
Exposure-response curves for daily concentrations of PM pollutants associated with COPD mortality in Zhejiang Province, 2014-2016