Ambient air pollution and daily hospital admissions for mental disorders in Shanghai, China

Science of the Total Environment 613–614 (2018) 324–330

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Ambient air pollution and daily hospital admissions for mental disorders in Shanghai, China Chen Chen a,1, Cong Liu a,1, Renjie Chen a, Weibing Wang a, Weihua Li b, Haidong Kan a,b,⁎, Chaowei Fu a,⁎⁎ a School of Public Health, Key Lab of Public Health Safety of the Ministry of Education and Key Laboratory of Health Technology Assessment of the Ministry of Health, Fudan University, Shanghai 200032, China b Key Lab of Reproduction Regulation of National Population and Family Planning Commission, Shanghai Institute of Planned Parenthood Research, Institute of Reproduction and Development, Fudan University, Shanghai 200032, China.

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Limited evidence on association between air pollution and mental disorders • Increased risk of admissions for mental disorders associated with PM10, SO2, and CO • Associations of air pollutants were generally stronger in warm period.

a r t i c l e

i n f o

Article history: Received 20 April 2017 Received in revised form 24 August 2017 Accepted 11 September 2017 Available online xxxx Editor: D. Barcelo Keywords: Air pollution Mental disorders Hospital admission Time-series study

a b s t r a c t Few studies have investigated the associations between ambient air pollution and mental disorders (MDs), especially in developing countries. We conducted a time-series study to explore the associations between six criteria air pollutants and daily hospital admissions for MDs in Shanghai, China, from 2013 to 2015. The MDs data were derived from the Shanghai Health Insurance System. We used over-dispersed, generalized additive models to estimate the associations after controlling for time trend, weather conditions, day of the week, and holidays. In addition, we evaluated the effect of modification by age, sex, and season. A total of 39,143 cases of hospital admissions for MDs were identified during the study period. A 10-μg/m3 increase in 2-day, moving-average concentration of inhalable particulate matter, sulfur dioxide (SO2), and carbon monoxide was significantly associated with increments of 1.27% [95% confidence interval (CI): 0.28%, 2.26%], 6.88% (95% CI, 2.75%, 11.00%), and 0.16% (95% CI: 0.02%, 0.30%) in daily hospital admissions for MDs, respectively. We observed positive but insignificant associations of fine particulate matter, nitrogen dioxide and ozone. The estimated association of SO2 was relatively robust to the adjustment of simultaneous exposure to other pollutants. We found generally stronger associations of air pollutants with MDs in warm seasons than in cool seasons. There were no significant differences in the associations between different sex and age groups. This study suggested that short-term exposure to air pollution, especially to sulfur dioxide, was associated with increased risk of hospital admissions for MDs in Shanghai, China. © 2017 Elsevier B.V. All rights reserved.

⁎ Correspondence to: H. Kan, P.O. Box 249, 130 Dong-An Road, Shanghai 200032, China. ⁎⁎ Correspondence to: C. Fu, P.O. Box 289, 130 Dong-An Road, Shanghai 200032, China. E-mail addresses: [email protected] (H. Kan), [email protected] (C. Fu). 1 Co-first authors that contributed equally to this work.

http://dx.doi.org/10.1016/j.scitotenv.2017.09.098 0048-9697/© 2017 Elsevier B.V. All rights reserved.

C. Chen et al. / Science of the Total Environment 613–614 (2018) 324–330

325

1. Introduction

2.2. Environmental data

Mental disorder (MD) is a syndrome characterized by clinically significant disturbance in an individual's cognition, emotion regulation, or behavior that reflects a dysfunction in the psychological, biological, or developmental processes underlying mental functioning (American Psychiatric Association, 2013). MDs pose a heavy burden on public health, leading to more than one-fifth of years of life lived with disability worldwide (WHO, 2015). Globally, over 350 million people were estimated to be affected by depression; 48 million people were affected by dementia; and 21 million people were affected by schizophrenia and other subtypes of MDs (WHO, 2016). A full understanding about the risk factors of MDs is of great importance to public health. Previous studies have identified that psychiatric diseases can be induced by genetic (Di Forti et al., 2012; Geschwind and Flint, 2015; Robinson et al., 2016; Whalley, 2016), socio-economic (Fazel et al., 2014; Kawakami et al., 2012), and behavioral risk factors such as smoking (Tobe, 2012). Recently, there is increasing evidence of air pollution hazards on the brain (Block et al., 2012). Until now, only a small fraction of studies has explored air pollution as a novel risk factor for the incidence of MD or its subtypes, such as depressive disorder, completed suicide, panic attacks, etc. (Bakian et al., 2015; Cho et al., 2015; Cho et al., 2014). The associations are biologically plausible, for example, in that environmental factors may cause activation of the immune system, oxidative stress, inflammation, alterations in the concentrations of cerebral neurotransmitters, and eventually lead to mental or behavioral alterations (Deisenhammer, 2003; Kelly, 2003; Xu et al., 2016). As the largest developing country, China is experiencing one of the worst air pollution problems in the world. Several epidemiological studies in China have revealed that MDs are associated with ambient air pollution, but the evidence is limited and somewhat inconsistent (Gao et al., 2017; Tong et al., 2016). For example, a time-series study in Tianjin, China reported significant associations of psychosis with sulfur dioxide (SO2), nitrogen dioxide (NO2), and inhalable particulate matter (PM10 ) (Tong et al., 2016). However, another study with case-crossover design did not observe significant associations between suicide and PM10 or SO2 (Bakian et al., 2015). Besides, there is a lack of studies evaluating the associations between air pollution and specific subtypes of MDs. Therefore, we conducted this time-series analysis to explore the impacts of air pollution on hospital admissions for total MDs and two sub-categories (manic episode and depressive disorder) in Shanghai, the largest city of China.

Daily air quality data were obtained from the database of the Shanghai Environmental Monitoring Center. We evaluated six criteria air pollutants: NO 2 , SO 2 , carbon monoxide (CO), ozone (O3), PM10 and fine particulate matter (PM2.5). Air pollutant concentrations were measured at 10 fixed-site monitoring stations in the urban area of Shanghai. According to the Chinese government's rules, these monitors are mandated not to be in the direct vicinity of apparent emission sources (traffic, industry, and boilers, etc.), so that they may represent the general air pollution levels in urban environments. Pollutants were measured using different methods: the method of Tapered Element Oscillating Microbalance for PM10 and PM2.5, the ultraviolet fluorescence method for SO2 and O3, the chemiluminescence method for NO2, and the infrared absorption method for CO. All these measurements were operated under the China National Quality Control (GB3095-2012). We averaged the daily concentrations of air pollutants from the 10 monitoring stations as the proxy for the general exposure for all populations. We calculated maximum 8-h averages for O3 and daily 24-h averages for the remaining pollutants. Daily meteorological data, including daily mean temperature and relative humidity, were obtained from a fixed station (Xujiahui) operated by the Shanghai Meteorological Bureau.

2. Methods and materials 2.1. Data collection Daily hospital admissions for MDs were collected from the database of the Shanghai Health Insurance System, which provides compulsory universal health insurance to most of the residents in Shanghai (The coverage rate was N 96% in 2014) (Peng et al., 2017). Computerized records of daily hospital admissions are available for each contracted hospital. We obtained the daily number of hospital admissions for MDs during the study period (January 1, 2013, to December 31, 2015) from this system. MDs were defined according to the 10th version of the International Classification of Diseases, with codes F01–F99. We further considered two categories of MDs: manic episode (ICD: F30) and depressive disorder (ICD: F32–33). The Institutional Review Board at the School of Public Health, Fudan University, approved the study protocol (No. 2014-07-0523) with a waiver of informed consent, because all data were analyzed at aggregate level and no participants were contacted.

2.3. Statistical analysis Hospital admissions for MDs were linked with air pollutant concentrations by date. We applied a time-series approach to analyze the data, which has the advantage of automatically controlling for time-invariant confounders at population level. Specifically, as daily hospital admissions for MDs approximately followed a quasi-Poisson distribution, we used the overdispersed generalized additive model (GAM) to analyze the association between daily hospitalizations of MDs and each air pollutant. Consistent with many previous time-series studies in this area, we incorporated several covariates in the main model: (1) a natural smooth spline function of calendar time with 7 degrees of freedom (df) per year to control for long-term and seasonal trends longer than two months (Chen et al., 2010; Peng et al., 2006; Zanobetti and Schwartz, 2009); (2) natural smooth functions with 6 df for the present-day mean temperature and 3 df for the present-day relative humidity to exclude potential nonlinear confounding effects of weather conditions (Chen et al., 2014; Peng et al., 2006); (3) a factor variable for day of the week (DOW) to account for the variation of hospital admissions within a week and (4) a binary dummy variable for public holidays to adjust for the holiday effects (Kan et al., 2008). We used a 2-day moving average of current- and previous-day (lag 01) concentrations of air pollutants in our main model because this lag often produced the largest effect estimate in previous studies (Chen et al., 2012; Kan et al., 2008). To explore the lag patterns in the impacts of air pollution, we further introduced both single-day lags from 0 to 6 and a 7-day moving average of the current and previous 6 days (lag 06), using the same models. The main model is described as follows: Log E(Yt) = Zt + s (day, df) + s (temperature, df) + s (relative humidity, df) + holiday + DOW + α, where E (Yt) denotes the estimated daily hospital admissions for MDs; Zt indicates the pollutant concentrations on day t; s denotes a natural spline function; df is the degrees of freedom; and α is the intercept. We plotted the exposure–response (E–R) relationship curves between 6 air pollutants and hospital admissions for MDs by adding a natural spline function with 3 df for each pollutant term in the above main model (Cao et al., 2012). In addition, we conducted stratification analyses to explore the potential effect of modification by age (≤44, 45–64, and ≥65), sex, and season (cool: October to March; warm: April to September). We further evaluated the statistical significance for the differences in estimates

326

C. Chen et al. / Science of the Total Environment 613–614 (2018) 324–330

Table 1 Summary statistics of environment and health data during the study period (January 1, 2013 to December 31, 2015). Mean SD Mental disorders (N) 36 28 Manic episode 1 1 Depressive disorder 2 2 Others 33 26 Age(N) ≤ 44 8 6 45–64 15 14 ≥ 65 13 10 Gender(N) Male 18 16 Female 18 13 Season(N) a 36 27 Cool 36 28 Warmb Air pollutant concentrations (24-h average, μg/m3) 56 37 PM2.5 76 46 PM10 19 13 SO2 46 21 NO2 O3 100 44 CO 820 309 Weather conditions (24-h average) Temperature(°C) 17 9 Humidity (%) 72 13

Min P25

Median

P75

Max

1 0 0 1

14 0 0 13

36 1 1 33

46 2 2 43

291 7 11 281

0 0 0

3 5 7

8 14 12

12 19 16

37 164 103

0 0

7 8

18 18

23 24

175 116

1 1

14 14

37 36

46 47

291 244

7 6 6 5 11 364

30 45 11 31 69 606

45 63 15 42 96 735

70 97 22 57 124 949

255 305 103 143 266 2281

−1 31

10 63

18 73

24 81

35 98

Abbreviations: SD, standard deviation; P (25), 25th percentile; P (75), 75th percentile; PM2.5, particulate matter with an aerodynamic diameter less than or equal to 2.5 μm; PM10, particulate matter with an aerodynamic diameter less than or equal to 10 μm; O3, ozone; SO2, sulfur dioxide; NO2, nitrogen dioxide; CO, carbon monoxide. a Cool season: from October to March. b Warm season: from April to September.

^ 1−Q2) ^ ± across strata by calculating 95% confidence intervals (CI) as (Q qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ^ 1 and Q2 ^ are the estimates for two categories, ^2 þ SE ^2, where Q 1.96 SE 1 2 ^2 and SE ^2 are their standard errors (Chen et al., 2012). and SE 1 2 We conducted two sensitivity analyses. First, we fitted two-pollutant models to test the robustness for the effect estimate of a pollutant after controlling for the concomitant exposure to other pollutants, using the same parameter settings as in the main model. Second, we changed the df from 5 to 10 per year to check whether the results were robust to varying smoothness of time. The statistical tests were two-sided, and associations of P b 0.05 were considered statistically significant. All analyses were conducted using R software (version 3.3.2) with the mgcv package. Effect estimates were described as percent changes and 95% confidence intervals (CIs) in daily hospitalizations for MDs per 10-μg/m3 increase in PM2.5, PM10, O3, SO2, NO2, and CO.

3. Results Table 1 summarizes descriptive statistics of this study. A total of 39,143 cases of hospital admissions for MDs occurred over the study period, with a daily average of 36 cases. Males and the elderly above 65 years of age accounted for 50.5% and 36.5%, respectively. Hospital admissions for MDs were slightly higher in warm seasons (51.1%) than in cool seasons (48.9%). Daily mean concentrations were 56 μg/m 3 for PM2.5, 76 μg/m3 for PM10, 19 μg/m3 for SO 2, 46 μg/m3 for NO2, 100 μg/m3 for O3, and 0.8 mg/m3 for CO. There were strong correlations among various air pollutants except for O 3 (Table S1). The mean of daily temperature and relative humidity were 17 °C and 72%, respectively. Table 2 shows the estimates in single-pollutant models using different lag days. Overall, we found statistically significant associations of PM 10 , SO 2, and CO with MD hospitalizations. The associations remained statistically significant only in the first two or three days and the lag 01 day generally produced the largest estimates. The associations of PM2.5, NO2 and O3 with MD hospitalizations were generally positive but statistically insignificant in all lags we examined. For example, a 10-μg/m3 increase in concentrations (lag 01 day) of PM2.5 , PM10 , SO2 , NO2, O3 and CO was associated with increments of 1.09% (95% CI: –0.10%, 2.28%), 1.27% (95% CI: 0.28%, 2.26%), 6.88% (95% CI: 2.75%, 11.00%), 1.88% (95% CI: –0.40%, 4.16%), 0.34% (95% CI: –1.08%, 1.75%), and 0.16% (95% CI: 0.02%, 0.30%) in daily hospital admissions for MDs, respectively. For specific subtypes of MDs, we observed positive associations of PM 10, SO2, and CO with depressive disorders, as well as positive associations of SO2 and CO with manic episodes; however, these associations were not statistically significant (Table S2). Fig. 1 shows the E–R curves for the associations between each pollutant at lag 01 day and daily hospital admissions for MDs. The E–R relationships for SO 2 and CO were almost linear, showing no thresholds for their associations with MD hospitalizations. For the curve of PM10, we observed a relatively flat slope at low concentrations, and then a drastic increase at concentrations N 100 μg/m3 . The E–R curve for PM2.5 showed a linear and moderately positive association. The curve for NO2 showed a steep slope at concentrations N60 μg/m3. The curve for O3 had a steep slope at concentrations between 50 μg/m 3 and 100 μg/m 3 and became flat at higher concentrations. Table 3 summarizes the results for possible effect of modification by age, sex, and season. The associations of PM10, SO2, and CO became insignificant among the young (b45 years old), but were similar between those of 45–64 years and those of 65 years and older. The associations between air pollution and MD hospitalizations were similar between males and females. In season-specific analysis, the associations were more prominent in warm seasons than in cool seasons, especially for

Table 2 Percent increase (means and 95% confidence intervals) in daily hospitalizations for mental disorders associated with a 10-μg/m3 increase in PM2.5, PM10, SO2, NO2, O3, and CO, using different lag days in single-pollutant models. Lag

PM2.5

PM10

SO2

NO2

O3

CO

0 1 2 3 4 5 6 01 06

0.72 (−0.26, 1.71) 0.74 (−0.25, 1.73) 0.61 (−0.40, 1.63) −0.82 (−1.87, 0.23) −0.70 (−1.70, 0.30) 0.03 (−0.96, 1.02) 0.63 (−0.37, 1.63) 1.09 (−0.10, 2.28) 1.12 (−1.12, 3.36)

0.85 (0.03, 1.67) 0.86 (0.04, 1.68) 0.58 (−0.24, 1.40) −0.59 (−1.43, 0.25) −0.62 (−1.44, 0.19) −0.11 (−0.93, 0.70) 0.02 (−0.81, 0.84) 1.27 (0.28, 2.26) 0.64 (−1.03, 2.32)

4.65 (1.17, 8.14) 4.76 (1.39, 8.12) 3.34 (−0.03, 6.71) −0.67 (−4.14, 2.79) −1.45 (−4.80, 1.90) −0.70 (−4.01, 2.62) 0.17 (−3.22, 3.57) 6.88 (2.75, 11.00) 4.41 (−1.78, 10.60)

1.11 (−0.84, 3.05) 1.72 (−0.28, 3.72) 0.82 (−1.18, 2.81) −1.05 (−3.04, 0.93) −1.61 (−3.54, 0.32) −0.92 (−2.83, 0.98) 0.55 (−1.36, 2.45) 1.88 (−0.40, 4.16) 0.21 (−3.44, 3.85)

−0.19(−1.35, 0.98) 0.60 (−0.47, 1.67) 0.63 (−0.35, 1.62) −0.31 (−1.27, 0.65) 0.14 (−0.78, 1.06) −0.35 (−1.29, 0.59) −0.15 (−1.08, 0.79) 0.34 (−1.08, 1.75) 0.82 (−1.30, 2.95)

0.09 (−0.03, 0.21) 0.13 (0.01, 0.25) 0.17 (0.05, 0.29) −0.02 (−0.15, 0.10) −0.09 (−0.22, 0.03) −0.02 (−0.14, 0.10) 0.04 (−0.09, 0.16) 0.16 (0.02, 0.30) 0.20 (−0.06, 0.45)

Abbreviations as in Table 1. The statistically significant estimates are highlighted in bold. Lag 01, the moving average concentrations on the present day and previous day. Lag 06, the moving average concentrations on the present day and previous 6 days. Note: The model covariates include long-term and seasonal trends, temperature, humidity, day of week, and holiday.

C. Chen et al. / Science of the Total Environment 613–614 (2018) 324–330 Fig. 1. The concentration–response relationship curves between air pollutants (lag 01 day) and daily hospitalizations for mental disorders. Abbreviations as in Table 1. The x-axis is 2-day, moving average concentrations of air pollutants; the y-axis is the estimated percent changes in hospital admissions for mental disorders; the solid line represents the mean estimate and the dashed lines represent the 95% confidence intervals.

327

328

C. Chen et al. / Science of the Total Environment 613–614 (2018) 324–330

SO2 and CO. The results of effect modification using different lag days are presented in Figs. S1–S3. In the first sensitivity analysis (Table 4), the associations of SO2 with MD hospitalizations remained relatively robust after adjustment for all other pollutants. The associations of PM10 and CO were statistically significant only when controlling for O3. The results for two-pollutant models using different lag days are presented in Table S3. In the second sensitivity analysis, the estimates varied little using alternative smoothness of time (Table S4).

4. Discussion To the best of our knowledge, this is one of the few studies in developing countries to explore the associations between air pollution and hospitalizations for MDs. Overall, our study suggested that short-term exposure to air pollutants, including PM 10 , SO 2 , and CO, was significantly associated with increased risks of hospital admissions for MDs. The associations between SO2 and MD hospitalizations were robust to the adjustment of co-pollutants in two-pollutant models. The associations were stronger in warm seasons than in cool seasons. The associations between air pollutants and MD hospitalizations varied by sex and age subgroups, but the differences were not statistically significant. Our findings added to the limited evidence regarding the impacts of air pollutants on the incidence of MDs in developing countries. We observed significant associations between hospital admissions for MDs and air pollutants (i.e., PM 10 , SO 2 , and CO). These findings were generally consistent with previous studies, although not all studies reported the associations of the same pollutants on the same subcategories of MDs. For example, a time-series study in Tianjin reported that hospital admissions for MDs were significantly associated with PM10 and SO2 (Tong et al., 2016). A timeseries study in Beijing also reported significant association between PM 10 and hospital admissions for MDs (Gao et al., 2017). However, we found positive but insignificant associations of PM 2.5 and NO 2 with MD hospitalizations, whereas other studies reported significant associations of specific subtypes of MDs with NO 2 (Bakian et al., 2015; Cho et al., 2014; Lin et al., 2016; Szyszkowicz et al., 2009) and PM 2.5 (Bakian et al., 2015; Kim et al., 2010). In addition, we found generally positive but insignificant associations between air pollutants and hospital admissions for two subtypes of MDs (manic episode and depression). The differences between our results and previous findings may be due to different study designs, categories of MDs, study populations, and the characteristics of air pollution mixtures. We cannot exclude the possibility that some other subtypes of MDs, which were not examined in the current study, may be significantly affected by air pollution.

We found that the associations of SO2 and CO with MD hospitalizations were stronger in warm seasons than in cool seasons, which is consistent with a Canadian study (Szyszkowicz et al., 2009). The stronger associations in warm seasons may be attributable to higher personal exposure to ambient air pollutants in relation to more outdoor activities and natural ventilation (Zeger et al., 2000). However, other studies have demonstrated different seasonal patterns. For example, Tong et al. found a stronger association of psychosis morbidity with SO2 and NO2 in cool seasons than in warm seasons in Tianjin (Tong et al., 2016). Likewise, Lin et al. suggested stronger associations between suicide rates and air pollution in cool seasons in Guangzhou (Lin et al., 2016). These differences may be due to the complex components of air pollution, climatic conditions, and exposure patterns in different regions. Therefore, the exact seasonal patterns are still uncertain and merit further investigations. It is of great significance to understand which air pollutant is more important than the others in triggering MD hospitalizations. In two-pollutant models, the associations of SO2 with MD hospitalizations remained robust after the adjustment for co-pollutants. The seemingly independent association between SO2 and MD hospitalizations may be explained by the oxidative damage and increased lipid peroxidation levels in the brain according to an experimental study in mice (Meng and Zhang, 2003). However, given the correlations among various pollutants, it is difficult to disentangle the association of a specific air pollutant. Exploration of the shape of E–R relationships is crucial for public health assessment, especially in developing countries with severe air pollution problems. In the present study, we did not observe threshold concentrations below which CO, SO2, NO2 and O3 were not associated with MD hospitalizations. Although E–R relationships may vary by a number of factors such as locations, air pollution mixture, climatic characteristics, and population sensitivity, there are still public health implications of the current findings that air pollution levels should be continuously lowered to protect the MD patients and reduce the risks of MD exacerbations. Although the exact mechanisms are unclear, the associations between air pollution and increased MD hospitalizations are somewhat biologically plausible. It has been indicated that air pollutants may lead to biological changes in the brain, such as change in brain activity, inflammatory reactions, and pathological changes in brain tissues (Xu et al., 2016). Studies have also suggested that the development of MDs is dependent on the interaction between genetic and environmental factors, which may be mediated by the activation of the immune system, oxidative stress and inflammation. For example, an animal-based study found that a 4-week exposure to particulate matter upregulated neuroinflammatory cytokines, including tumor necrosis factor tumor necrosis factor-α and interleukin − 1β, and further provoked depressive-like responses (Hogan et al., 2015). Oxidative stress has been recognized as one of the main pathways by

Table 3 Age-, gender-, and season-specific percent increase (means and 95% CI) in daily hospitalizations for mental disorders associated with a 10-μg/m3 increase in concentrations (lag 01) of PM2.5, PM10, SO2, NO2, O3, and CO in single-pollutant models. Age (years)

PM2.5 PM10 SO2 NO2 O3 CO

Gender

Season

≤44

45–64

≥65

Male

Female

Cool

Warm

0.39 (−0.82, 1.60) 0.64 (−0.38, 1.65) 3.67 (−0.54, 7.88) 0.31 (−2.02, 2.64) 0.24 (−1.18, 1.66) 0.07 (−0.08, 0.22)

1.38 (−0.24, 3.00) 1.48 (0.13, 2.84) 8.84 (3.23, 14.44) 2.80 (−0.31, 5.90) 0.26 (−1.69, 2.20) 0.20 (0.00, 0.39)

1.22 (−0.06, 2.50) 1.43 (0.36, 2.49) 6.74 (2.26, 11.21) 1.85 (−0.59, 4.29) 0.51 (−1.01, 2.03) 0.17 (0.02, 0.33)

1.27 (−0.18, 2.72) 1.36 (0.15, 2.58) 7.64 (2.61, 12.67) 2.17 (−0.62, 4.97) 0.56 (−1.19, 2.30) 0.16 (−0.02, 0.33)

0.91 (−0.22, 2.05) 1.17 (0.22, 2.12) 6.08 (2.12, 10.04) 1.59 (−0.58, 3.76) 0.12 (−1.22, 1.46) 0.16 (0.03, 0.30)

0.71 (−0.60, 2.02) 1.11 (0.02, 2.21) 4.75 (0.31, 9.19)⁎

1.78 (−0.71, 4.27) 1.34 (−0.76, 3.45) 17.94 (5.94, 29.93)⁎ 0.63 (−3.99, 5.25) 0.27 (−1.41, 1.95) 0.35 (0.05, 0.65)

Abbreviations as in Table 1. CI: confidence interval. The statistically significant estimates are highlighted in bold. Note: The model covariates include long-term and seasonal trends, temperature, humidity, day of week, and holiday. ⁎ Statistically significant for between-group difference (P b 0.05).

1.67 (−0.94, 4.29) 0.19 (−2.71, 3.10) 0.09 (−0.07, 0.25)

C. Chen et al. / Science of the Total Environment 613–614 (2018) 324–330 Table 4 Percent increase (means and 95% confidence intervals) in daily hospitalizations for mental disorders associated with a 10-μg/m3 increase in concentrations (lag 01) of PM2.5, PM10, SO2, NO2, O3, and CO in single and two-pollutant models.

PM10

SO2

CO

Model

Mean (95% CI)

– Adjusted for SO2 Adjusted for NO2 Adjusted for O3 Adjusted for CO – Adjusted for PM2.5 Adjusted for PM10 Adjusted for NO2 Adjusted for O3 Adjusted for CO – Adjusted for PM2.5 Adjusted for PM10 Adjusted for SO2 Adjusted for NO2 Adjusted for O3

1.27 (0.28, 2.26) −0.03 (−1.66, 1.60) 1.43 (−0.08, 2.93) 1.25 (0.21, 2.28) 1.18 (−0.72, 3.08) 6.88 (2.75, 11.00) 10.47 (3.60, 17.33) 7.04 (0.21, 13.87) 10.93 (4.21, 17.65) 6.28 (2.15, 10.42) 8.83 (1.84, 15.82) 0.16 (0.02, 0.30) 0.22 (−0.12, 0.56) 0.01 (−0.26, 0.29) −0.08 (−0.32, 0.16) 0.16 (−0.07, 0.40) 0.15 (0.00, 0.30)

Abbreviations as in Table 1. The statistically significant estimates are highlighted in bold. Note: The model covariates include long-term and seasonal trends, temperature, humidity, day of week, and holiday.

which air pollutants cause damage to cardiovascular and respiratory systems (Kelly, 2003). Likewise, it may also be hypothesized that air pollution may impair the nervous system through oxidative stress pathways. Several limitations of the present study should be noted. First, as done in most previous time-series studies, we used the measurements from fixed-site air quality monitors instead of personal measurements, resulting in inevitable exposure misclassification in the present analysis. However, this resultant non-differential error was reported to produce an underestimate on the associations of ambient air pollution (Zeger et al., 2000). Second, we could not obtain data on more specific MD subtypes, leading to the failure in a comprehensive analysis on air pollution and MD hospitalizations. Third, limited by the data availability, we could not analyze the short-term associations between air pollution and emergency-department visits for specific MDs. Fourth, we could not exclude repeated or scheduled hospitalization for MDs, which are virtually not affected by air pollution. However, it is not a severe problem in that scheduled visits are not common in China and, further, we did not find significant associations by using longer lag days (i.e., up to 1 week, Table 2). Fifth, our analysis focused on only one highly polluted Chinese city; thus, the generalizability of our results is limited. Finally, because of the small number of manic episode and depression, we failed to conduct stratification analyses by different age groups and gender. 5. Conclusions This study suggested that short-term exposure to air pollution, especially to SO2, might significantly increase the risks of hospital admissions for MDs. Our study contributed to the increasing evidence about the adverse effects of air pollution on mental health, which merits further investigations and considerations in environmental protection and public health interventions for MD patients. Acknowledgements The study was supported by the Public Welfare Research Program of National Health and Family Planning Commission of China (201502003), National Natural Science Foundation of China (91643205), and Shanghai 3-Year Public Health Action Plan (GWTD2015S04 and 15GWZK0202). The authors declared no conflicts of interests.

329

Appendix A. Supplementary data Results of Pearson correlation coefficients among air pollutants (Table S1), effect estimates of air pollutants on subtypes of mental disorders (Table S2), results of the two sensitivity analyses (Table S3, Table S4) and results of effect modification using different lag days by age, sex, and season (Figs. S1–S3) are available. Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j. scitotenv.2017.09.098.

References American Psychiatric Association, 2013. Diagnostic and Statistical Manual of Mental Disorders. 5th ed. American Psychiatric Publishing, Arlington. Bakian, A.V., Huber, R.S., Coon, H., Gray, D., Wilson, P., McMahon, W.M., et al., 2015. Acute air pollution exposure and risk of suicide completion. Am. J. Epidemiol. 181, 295–303. Block, M.L., Elder, A., Auten, R.L., Bilbo, S.D., Chen, H., Chen, J.C., et al., 2012. The outdoor air pollution and brain health workshop. Neurotoxicology 33, 972–984. Cao, J., Xu, H., Xu, Q., Chen, B., Kan, H., 2012. Fine particulate matter constituents and cardiopulmonary mortality in a heavily polluted Chinese city. Environ. Health Perspect. 120, 373–378. Chen, R., Chu, C., Tan, J., Cao, J., Song, W., Xu, X., et al., 2010. Ambient air pollution and hospital admission in Shanghai, China. J. Hazard. Mater. 181, 234–240. Chen, R., Kan, H., Chen, B., Huang, W., Bai, Z., Song, G., 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. Chen, R., Li, T., Cai, J., Yan, M., Zhao, Z., Kan, H., 2014. Extreme temperatures and out-ofhospital coronary deaths in six large Chinese cities. J. Epidemiol. Community Health 68, 1119–1124. Cho, J., Choi, Y.J., Suh, M., Sohn, J., Kim, H., Cho, S.K., et al., 2014. Air pollution as a risk factor for depressive episode in patients with cardiovascular disease, diabetes mellitus, or asthma. J. Affect. Disord. 157, 45–51. Cho, J., Choi, Y.J., Sohn, J., Suh, M., Cho, S.K., Ha, K.H., et al., 2015. Ambient ozone concentration and emergency department visits for panic attacks. J. Psychiatr. Res. 62, 130–135. Deisenhammer, E.A., 2003. Weather and suicide: the present state of knowledge on the association of meteorological factors with suicidal behaviour. Acta Psychiatr. Scand. 108, 402–409. Di Forti, M., Iyegbe, C., Sallis, H., Kolliakou, A., Falcone, M.A., Paparelli, A., et al., 2012. Confirmation that the AKT1 (rs2494732) genotype influences the risk of psychosis in cannabis users. Biol. Psychiatry 72, 811–816. Fazel, S., Geddes, J.R., Kushel, M., 2014. The health of homeless people in high-income countries: descriptive epidemiology, health consequences, and clinical and policy recommendations. Lancet 384, 1529–1540. Gao, Q., Xu, Q., Guo, X., Fan, H., Zhu, H., 2017. Particulate matter air pollution associated with hospital admissions for mental disorders: a time-series study in Beijing, China. Eur. Psychiatry 44, 68–75. Geschwind, D.H., Flint, J., 2015. Genetics and genomics of psychiatric disease. Science 349, 1489–1494. Hogan, M.K., Kovalycsik, T., Sun, Q., Rajagopalan, S., Nelson, R.J., 2015. Combined effects of exposure to dim light at night and fine particulate matter on C3H/HeNHsd mice. Behav. Brain Res. 294, 81–88. Kan, H., London, S.J., Chen, G., Zhang, Y., Song, G., Zhao, N., et al., 2008. Season, sex, age, and education as modifiers of the effects of outdoor air pollution on daily mortality in Shanghai, China: the public health and air pollution in Asia (PAPA) study. Environ. Health Perspect. 116, 1183–1188. Kawakami, N., Abdulghani, E.A., Alonso, J., Bromet, E.J., Bruffaerts, R., Caldas-de-Almeida, J.M., et al., 2012. Early-life mental disorders and adult household income in the world mental health surveys. Biol. Psychiatry 72, 228–237. Kelly, F.J., 2003. Oxidative stress: its role in air pollution and adverse health effects. Occup. Environ. Med. 60, 612–616. Kim, C., Jung, S.H., Kang, D.R., Kim, H.C., Moon, K.T., Hur, N.W., et al., 2010. Ambient particulate matter as a risk factor for suicide. Am. J. Psychiatry 167, 1100–1107. Lin, G.Z., Li, L., Song, Y.F., Zhou, Y.X., Shen, S.Q., Ou, C.Q., 2016. The impact of ambient air pollution on suicide mortality: a case-crossover study in Guangzhou, China. Environ. Health 15, 90. Meng, Z., Zhang, B., 2003. Oxidative damage of sulfur dioxide inhalation on brains and livers of mice. Environ. Toxicol. Pharmacol. 13, 1–8. Peng, R.D., Dominici, F., Louis, T.A., 2006. Model choice in time series studies of air pollution and mortality. J. R. Statist. Soc. A Statist. Soc. 169, 179–198. Peng, Z., Wang, Q., Kan, H., Chen, R., Wang, W., 2017. Effects of ambient temperature on daily hospital admissions for mental disorders in Shanghai, China: a time-series analysis. Sci. Total Environ 590–591, 281–286. Robinson, E.B., St Pourcain, B., Anttila, V., Kosmicki, J.A., Bulik-Sullivan, B., Grove, J., et al., 2016. Genetic risk for autism spectrum disorders and neuropsychiatric variation in the general population. Nat. Genet. 48, 552–555. Szyszkowicz, M., Rowe, B.H., Colman, I., 2009. Air pollution and daily emergency department visits for depression. Int. J. Occup. Med. Environ. Health 22, 355–362. Tobe, E., 2012. Progressive neuropsychiatric and brain abnormalities after smoke inhalation. BMJ Case Rep. 2012.

330

C. Chen et al. / Science of the Total Environment 613–614 (2018) 324–330

Tong, L., Li, K., Zhou, Q., 2016. Season, sex and age as modifiers in the association of psychosis morbidity with air pollutants: a rising problem in a Chinese metropolis. Sci. Total Environ. 541, 928–933. Whalley, K., 2016. Psychiatric disorders: linking genetic risk to pruning. Nat. Rev. Neurosci. 17, 199. WHO, 2015. Health in 2015: From MDGs to SDGs. World Health Organization, Geneva, p. 156. WHO, 2016. WHO Fact Sheet on Mental Disorders. http://www.who.int/mediacentre/ factsheets/fs396/en/, Accessed date: 1 June 2016.

Xu, X., Ha, S.U., Basnet, R., 2016. A review of epidemiological research on adverse neurological effects of exposure to ambient air pollution. Front. Public Health 4, 157. Zanobetti, A., Schwartz, J., 2009. The effect of fine and coarse particulate air pollution on mortality: a national analysis. Environ. Health Perspect. 117, 898–903. Zeger, S.L., Thomas, D., Dominici, F., Samet, J.M., Schwartz, J., Dockery, D., et al., 2000. Exposure measurement error in time-series studies of air pollution: concepts and consequences. Environ. Health Perspect. 108, 419–426.