Short-term effects of fine particulate matter on non-accidental and circulatory diseases mortality: A time series study among the elder in Changchun

Background and objectives Fine particulate matter (PM2.5, particulate matter with an aerodynamic diameter less than or equal to 2.5 μm) has multiple adverse effects on human health, especially on the respiratory and circulatory system. The purpose of this study was to evaluate the short-term effect of PM2.5 on the mortality risk of non-accidental and circulatory diseases, and to explore the potential effect modification by sex, education and death location. Methods We collected daily mortality counts of Changchun (China) residents, daily meteorology and air pollution data, from January 1, 2014, to January 1, 2017. We focused on the elderly (≥65 years old) population who died from non-accidental causes and circulatory diseases, and stratified them by sex, education, and death location. A generalized additive Poisson regression model (GAM) was used to analyse the impact of air pollutants on mortality. We fit single pollutant models to examine PM2.5 effects with different lag structures of single-day (distributed lag:lag0-lag3) and multi-day (moving average lag: lag01-lag03). To test the sensitivity of the model, a multi-pollutant model was established when the PM2.5 effect was strongest. Results In the single pollutant models, an increment of PM2.5 by 10 μg/m3 at lag0-3 was associated with a 0.385% (95% CI: 0.069% to 0.702%) increase in daily non-accidental mortality and a 0.442% (95% CI: 0.038% to 0.848%) increase in daily circulatory disease mortality. NO2 (lag1) and O3 (lag0, lag1, lag2, lag01,lag02, lag03) were associated with daily non-accidental death and NO2 (lag1, lag3, lag03) and O3 (lag0, lag1, lag01,lag02, lag03) were associated with daily circulatory disease mortality. In the co-pollutant models, the risk estimates for PM2.5 changed slightly. The excess mortality risk of non-accidental and circulatory diseases was higher for women, people with low education, and died outside hospital. Conclusions We found that short-term exposure to PM2.5 increased the mortality risk of non-accidental and circulatory diseases among the elderly in Changchun. Women, people with low education and died outside hospital are more susceptible to PM2.5. NO2 and O3 were also associated with an increase in mortality from non-accidental and circulatory diseases and the O3 is a high effect.

a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 PM 2.5 changed slightly. The excess mortality risk of non-accidental and circulatory diseases was higher for women, people with low education, and died outside hospital.

Conclusions
We found that short-term exposure to PM 2.5 increased the mortality risk of non-accidental and circulatory diseases among the elderly in Changchun. Women, people with low education and died outside hospital are more susceptible to PM 2.5 . NO 2 and O 3 were also associated with an increase in mortality from non-accidental and circulatory diseases and the O 3 is a high effect.

1.Introduction
Ambient air pollution (AAP), which has a substantial impact on human health, has become a global public health risk [1]. Data from the World Health Organization (WHO) show that 98% of cities in the world's low-and middle-income countries experience high levels of air pollution [2]. Fine particulate matter (PM 2.5 , particulate matter with an aerodynamic diameter less than or equal to 2.5 μm) is a principal air pollutant [3]. PM 2.5 is a mixture of various compounds that can enter major organ systems through the lungs, and it has multiple adverse effects on human health, especially on the respiratory and circulatory systems [4][5][6][7][8]. Numerous epidemiologic studies have reported the short-term effects of PM 2.5 [9][10][11][12]. However, short-term effects varied by exposure levels, air pollutant, population characteristics, and geographic location [13].
The WHO air quality guidelines (AQGs) set an annual mean of 10 μg/m 3 and a 24-hour mean of 25 μg/m 3 for PM 2.5 , and the risks of short-term and long-term mortality can be significantly reduced at below-mean concentrations [14]. Globally, the population-weighted average concentration of PM 2.5 continues to increase, and 87% of people worldwide were exposed to PM 2.5 levels greater than 10 μg/m 3 in 2013 [15].
In China, PM 2.5 pollution is more severe. Only 4% of Chinese people live in areas that meet the WHO AQGs [15]. In 2013, 760,000 people in China died from exposure to PM 2.5 [16]. As the capital city of Jilin Province (China), Changchun is the largest automobile industrial city in China and the birthplace of China's automobile, film, optical, bio-pharmaceutical, and track bus industries. In 2017, the total area was 20,565 square kilometres, with a total registered population of 7.489 million and an urban population of 4.383 million. The energy source of Changchun mainly consists of coal, making it a soot polluted city. In recent years, with the increase in car ownership, there is a trend towards mixed pollution. Due to the long heating period, the air pollution in Changchun is more serious in the winter. No study on the relationship between short-term exposure to PM 2.5 and mortality risk has been conducted in Changchun. Since the elderly are more vulnerable to air pollution [17,18], we therefore conducted a time-series study of the elderly in Changchun to evaluate the short-term effect of PM 2.5 on non-accidental and circulatory disease mortality, and explored the potential effect modification by sex, education and death location.

Mortality data.
Daily mortality data for residents in Changchun from January 1, 2014, to January 1, 2017, was obtained from the Jilin Provincial Center for Disease Control and Prevention. The anonymous records included age, sex, date of death, and the underlying cause of death. The underlying causes of death are classified by the International Classification of Diseases (ICD-10), in which non-accidental mortality are A00-R99 and circulatory system diseases are I00~I99. To determine the daily death toll, we selected all the elderly (�65 years old) who 'died of non-accidental causes and those who died of circulatory diseases and stratified them by sex, education and death location.
2.1.2. Air pollution and meteorology data. Daily air pollution data were obtained from ten national environmental monitoring stations in Changchun, including PM 2.5 , particulate matter with an aerodynamic diameter of equal or less than 10 μm (PM 10 ), sulphur dioxide (SO 2 ), nitrogen dioxide (NO 2 ) and ozone (O 3 ) (average concentration over 8 hours). Each indicator has at least 27 daily averages per month (at least 25 daily averages in February), and at least 324 daily averages per year. Daily meteorology data were obtained from the Changchun meteorological bureau, including average relative humidity and mean temperature. The annual data loss rate is less than 5%, and each index has at least 347 daily mean values per year. Audits of the validity, accuracy, normalization of the data, and of data quality can be used for analysis.

Statistical analysis
Spearman's rank correlation test was used to evaluate the correlation between air pollutants and meteorological conditions. The generalized additive Poisson regression model (GAM) was used to analyse the impact of air pollutants on the mortality of people. We applied smoothing spline functions to control the effects of confounding factors such as daily mean temperature, relative humidity, secular trend and seasonality on population mortality. The day of the week (DOW) was included in the multiple regression model as a dummy variable. The basic model was as follows: -ν is the degree of freedom. The degree of freedom was selected according to minimum Akaike information criterion (AIC). We applied the following degrees of freedom (df): 7/ year for the time trend and 3 for mean temperature and average relative humidity; -dow is the day of the week; -α is the intercept We fit single pollutant models to examine PM 2.5 effects with different lag structures of single-day (distributed lag: lag0-lag3) and multi-day (moving average lag: lag01-lag03). In singleday models, lag0 and lag1 for example correspond to the pollution concentration on the day and the day before. In multi-day lag models, lag03 for example corresponds to a 4-day moving average pollutant concentration of the current and previous 3 days. To test the sensitivity of the model, a multi-pollutant model was established when the PM 2.5 effect was strongest. Considering that there may be collinearity among air pollutants, other pollutants (PM 10 , SO 2 , NO 2 , O 3 ) are entered in the multi-pollutant model step-by-step. The excess risk (ER) was used to evaluate the increased death risk of non-accidental and circulatory diseases with each increase of 10μg/m 3 of air pollutants (ER = (RR-1)×100%; RR, relative risk). We also stratified the association of daily mortality with PM 2.5 by sex, education, and death location (died in hospital versus outside of hospital).
SPSS(Version 17.0; IBM Corp, Armonk, New York, USA) was used for descriptive analysis and Spearman correlation analysis, and R software (version 3.5.0; http://www.r-project.org) was used for time series analysis, with test degree α = 0.05.

Descriptive statistics
From January 2014 to January 2017, a total of 46,780 non-accidental deaths were reported among the elderly (�65 years old) in Changchun, of which 26,498 died of circulatory diseases. The mean daily deaths from non-accidental and circulatory diseases were 32.0 and 18.1, respectively. The daily average temperature was 7.0˚C, and the daily average relative humidity was 60.1%. The daily mean concentration of PM 2.5 was 55.9 μg/m 3 , with a Standard deviation (SD) of 49.5 μg/m 3 . For the other air pollutants, the daily mean concentrations of PM 10 PM 2.5 was positively correlated with PM 10 (r = 0.83), SO 2 (r = 0.52) and NO 2 (r = 0.72). O 3 , mean temperature and relative humidity were negatively correlated with PM 2.5 ( Table 2). Table 3 shows the excess mortality risk of non-accidental and circulatory diseases for the elderly due to PM 2.5 for each 10 μg/m 3 increase on different lag days in single pollutant models. PM 2.5 was associated with daily non-accidental mortality on lag1 and lag2 days, and the largest risk estimates were found with lag03; PM 2.5 was associated with daily circulatory disease mortality on lag1, the largest risk estimates were also found with lag03. Fig 1 shows the excess mortality risk to the elderly population caused by per 10 μg/m 3 increase in air pollutions. NO 2 (lag1) and O 3 (lag0, lag1, lag2, lag01,lag02, lag03) were associated with daily non-accidental death; Fig 2 shows the excess mortality risk to the elderly who die of circulatory diseases caused by per 10 μg/m 3 increase in air pollution. NO 2 (lag1, lag3, lag03) and O 3 (lag0, lag1, lag01,lag02, lag03) were associated with daily circulatory disease mortality.

Co-pollutant model
In the single pollutant models, an increment of PM 2.5 by 10 μg/m 3 at lag0-3 was associated with a 0.386% (95% CI: 0.07% to 0.70%) increase in daily non-accidental mortality and a 0.44% (95% CI: 0.04% to 0.85%) increase in daily circulatory disease mortality; NO 2 was associated with increased daily circulatory disease mortality, and O 3 was associated with increased daily non-accidental mortality and daily circulatory disease mortality. In the co-pollutant models, the risk estimates for PM 2.5 changed slightly (Table 4).

Subgroup analysis
The excess mortality risk of non-accidental and circulatory diseases was higher for women, people with low education and died outside hospital than it was for men, poeple with high education and died in hospital (Table 5).

Discussion
Our study analysed the short-term effects of PM 2.5 on the elderly in Changchun. This study is the first study to examine the relationship between PM 2.5 and population mortality in Jilin Province. We found that PM 2.5 was associated with an increase in mortality from non-   accidental and circulatory diseases. There appears to be a greater impact of PM 2.5 on mortality from non-accidental and circulatory diseases on women, poeple with low education and died Table 3. Association between 10 μg/m 3 increase in PM 2.5 and increase in deaths by lags.  Short-term effects of fine particulate matter outside hospital. NO 2 and O 3 were also associated with an increase in mortality from non-accidental and circulatory diseases. The effect of O 3 is higher than other pollutants. The estimated effects in our study of PM 2.5 on non-accidental mortality risk were lower than effects observed in a worldwide meta-analysis of a 1.04% (0.52%-1.56%) per 10 μg/m 3 increase in PM 2.5 [19]. This discrepancy could be due to differences in population composition, geographical location and PM 2.5 sources. However, compared with studies of Chinese populations, the estimated effects in our study are higher. For example, a recent Chinese multicity study found that the risk of non-accidental death increased by 0.22% for every 10 μg/m 3 increase in PM 2.5 (lag01) [20]. In another study, mortality risks for non-accidental deaths increased by 0.25% at lag days 0-1 for every 10 μg/m 3 increase in PM 2.5 [21]. The larger effect estimates observed in our study may be due to the age of the subject. We studied the elderly and the elderly make up a high-risk group [22,23]. The short-term association of PM 2.5 with mortality from circulatory diseases was consistent with previous epidemiological studies [17]. PM 2.5 may reduce cardiac parasympathetic input leading to decreased heart rate variability Short-term effects of fine particulate matter (HRV) [24,25]. Additionally, the direct translocation of PM 2.5 into the circulatory system can lead to an acute cardiovascular response [26][27][28].

Lag days ER of daily non-accidental death count (%) ER of daily circulatory diseases death count (%)
For the metric of mortality from non-accidental and circulatory diseases, women, people with low education and died outside hospital were found to have increased susceptibility to PM 2.5 . Previous studies on the gender-specific health effects of air pollution are lacking, and the patterns are not conclusive. For example, Franklin et al. studied over 1.3 million deaths in 27 US communities and found that women were more susceptible than men to the mortality effects of PM 2.5 [29]. Hong et al. also found that elderly woman were more susceptible than elderly men to the mortality effects of PM 10 [30]. Similarly, a statement from the American Table 4. Association between a 10 μg/m 3 increase in PM 2.5 (lag 03 day) and an increase in deaths using the co-pollutant model. Heart Association (AHA) concluded that particulate matter exposure may contribute to higher cardiovascular mortality in women [31]. There are some reasons that may explain the increased vulnerability of women. Inhaled particles could deposit regionally in women. [32]; women have fewer red blood cells than men, so women are more vulnerable to the toxic effects of air pollution [33]; in addition, the high reactivity of the airway to oxidants, hormonal status, smoking rates, and even a relatively low socioeconomic status are also possible causes [34]. However, the exact reasons for the gender-specific effects of air pollution on health are unclear and deserve further investigation.
Our study found that the impact on non-accidental and circulatory disease mortality from PM 2.5 is higher for those at a lower education level than for those with a higher education level, which is consistent with previous observations on individual education level [35]. In epidemiological studies, education has been used as a surrogate indicator of socioeconomic status (SES) [36]. It is well known that SES can affect health indicators such as mortality [37]. People with a lower SES are more likely to be exposed to air pollutants and are more likely to suffer from diseases linked to air pollution that confer a greater risk of dying [38][39][40]. People of a lower SES may also receive inferior medical treatment and less health care, making them more susceptible to the effects of air pollution [18]. In addition, people with a lower SES may have more limited access to fish, fresh fruits and vegetables, resulting in a reduced intake of protective fatty acids and vitamins [41].
Our results showed that the impact of PM 2.5 on mortality from non-accidental and circulatory diseases appears to be greater among people that die outside of a hospital. People outside hospital are less likely to acquire knowledge about PM 2.5 prevention and have more exposure risks. In addition, it is more difficult for people outside hospital to receive timely treatment to alleviate a condition in the event of an outbreak.
Changchun is located in northeast China and has significant seasonal differences. Although the annual average concentration of gaseous pollutants such as SO 2 , NO 2 and O 3 is not high, both SO 2 and NO 2 maintain relatively high levels during heating periods in the winter, while O 3 is relatively high in the summer, which exposes residents to persistent hazards from different air pollutants in different seasons. Our results showed that NO 2 and O 3 can increase the risk of non-accidental death in the elderly, and there was also a lag effect. Previous studies have also reached the same conclusion [42]. With co-pollutant adjustment in multi-pollutant models, the effect of PM 2.5 on mortality changed slightly. This result suggests that our model is relatively stable.
The findings support public policy-related activities focused on the reduction of PM 2.5 levels and the improvement of target daily PM 2.5 standards. The estimation of the short-term mortality risk due to short-term exposure to PM 2.5 can also help in planning appropriate medical interventions. Some limitations exist in our study: (1) Our study was based in a single city, Changchun, and generalizations need to be made in conjunction with other evidence. (2) Although it is considered reasonable to use the ambient pollutant concentration of PM 2.5 to represent individual exposure [43], there may be errors in the PM 2.5 exposure of different subgroups [44]. Further studies of the relationship between PM 2.5 and mortality are needed.

Conclusions
We found that short-term exposure to PM 2.5 increased the mortality risk of non-accidental and circulatory diseases among the elderly in Changchun. Women, people with low education and died outside hospital are more susceptible to PM 2.5 . NO 2 and O 3 were also associated with an increase in mortality from non-accidental and circulatory diseases and the O 3 is a high effect.
Supporting information S1