- Email: [email protected]
Differentiating the effects of characteristics of PM pollution on mortality from ischemic and hemorrhagic strokes
Accepted Manuscript Title: Differentiating the effects of characteristics of PM pollution on mortality from ischemic and hemorrhagic strokes Author: Hualiang Lin Jun Tao Yaodong Du Tao Liu Zhengmin Qian Linwei Tian Qian Di Weilin Zeng Jianpeng Xiao Lingchuan Guo Xing Li Yanjun Xu Wenjun Ma PII: DOI: Reference:
S1438-4639(15)00152-2 http://dx.doi.org/doi:10.1016/j.ijheh.2015.11.002 IJHEH 12886
To appear in: Received date: Revised date: Accepted date:
4-7-2015 13-11-2015 14-11-2015
Please cite this article as: Lin, H., Tao, J., Du, Y., Liu, T., Qian, Z., Tian, L., Di, Q., Zeng, W., Xiao, J., Guo, L., Li, X., Xu, Y., Ma, W.,Differentiating the effects of characteristics of PM pollution on mortality from ischemic and hemorrhagic strokes, International Journal of Hygiene and Environmental Health (2015), http://dx.doi.org/10.1016/j.ijheh.2015.11.002 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Differentiating the effects of characteristics of PM pollution on mortality from
2
ischemic and hemorrhagic strokes
3
Hualiang Lin a, Jun Tao b, Yaodong Du c, Tao Liu a, Zhengmin Qian d, Linwei Tian e,
5
Qian Di f, Weilin Zeng a, Jianpeng Xiao a, Lingchuan Guo a, Xing Li a, Yanjun Xu g,
6
Wenjun Ma a,*
us
cr
ip t
4
7 8
a
9
Disease Control and Prevention, Guangzhou, 511430, China
an
Guangdong Provincial Institute of Public Health, Guangdong Provincial Center for
10
b
11
Protection, Guangzhou, China
12
c
Guangdong Provincial Weather Center, Guangzhou, China
13
d
Department of Epidemiology, School of Public Health, Saint Louis University, Saint
14
Louis, MO, USA;
16 17 18 19
M
d
te
Ac ce p
15
South China Institute of Environmental Sciences, Ministry of Environmental
e
School of Public Health, The University of Hong Kong, Hong Kong SAR, China
f
Department of Environmental Health, Harvard School of Public Health, Boston,
USA g
Guangdong Provincial Center for Disease Control and Prevention, Guangzhou,
511430, China
20 21
Corresponding author at: Wenjun Ma, Guangdong Provincial Institute of Public
22
Health, Guangdong Provincial Center for Disease Control and Prevention, Qunxian 1
Page 1 of 34
23
Road, 160, Guangzhou, China. E-mail: [email protected]
24
Abbreviations:
26
PM: particulate matter; OC: organic carbon; EC: elemental carbon; PM10: particulate
27
matter with aerodynamic diameter ≤10µm; PM2.5: particulate matter with
28
aerodynamic diameter ≤2.5µm; QA: quality assurance; QC: quality control; ICD-10:
29
International Classification of Diseases, Tenth Revision; SO2: sulfur dioxide; NO2:
30
nitrogen dioxide; GAM: generalized additive model; PH: public holidays; DOW: day
31
of the week; df: degree of freedom; ER: excess risk; CI: confidence interval; IQR:
32
interquartile range.
M
an
us
cr
ip t
25
Ac ce p
te
d
33
2
Page 2 of 34
ABSTRACT
34
Though increasing evidence supports significant association between particulate
35
matter (PM) air pollution and stroke, it remains unclear what characteristics, such as
36
particle size and chemical constituents, are responsible for this association. We
37
investigated the association between particulate matter with different particle sizes,
38
chemical constituents of fine particulate pollution and mortalities from ischemic
39
stroke and hemorrhagic stroke in Guangzhou, China. A time-series model with
40
quasi-Poisson function was applied to assess the association of PM pollution with
41
different particle sizes and chemical constituents with mortalities from ischemic and
42
hemorrhagic strokes in Guangzhou, Chinathe association of PM with different particle
43
sizes and seven major PM2.5 components with mortality from overall stroke, ischemic
44
stroke and hemorrhagic stroke in Guangzhou, we controlled for potential confounding
45
factors in the model, such as temporal trends, day of the week, public holidays,
46
meteorological factors and influenza epidemic.with adjustment for potential
48 49 50
cr
us
an
M
d
te
Ac ce p
47
ip t
33
confounding factors. We found significant association between stroke mortality and various PM fractions, such as PM10, PM2.5 and PM1, with generally larger magnitudes for smaller particles. For the PM2.5 chemical constituents, we found that organic carbon (OC), elemental carbon (EC), sulfate, nitrate, and ammonium were
51
significantly associated with hemorrhagic stroke mortality. The analysis for specific
52
types of stroke suggested that it was hemorrhagic stroke, rather than ischemic stroke,
53
that was significantly associated with PM pollution. Our study shows that various PM
54
pollution fractions are associated with stroke mortality, and constituents primarily 3
Page 3 of 34
from combustion and secondary aerosols might be the harmful components of PM2.5
56
in Guangzhou, and this study suggests that PM pollution is more relevant to
57
hemorrhagic stroke in the study area, however more studies are warranted due to the
58
underlying limitations of this study.
ip t
55
Keywords: Particulate matter air pollution; stroke; particle size; chemical constituents
us
60
cr
59
61
Ac ce p
te
d
M
an
62
4
Page 4 of 34
62 63
Introduction A great number of epidemiological studies have demonstrated a consistent increased risk from stroke with short term exposure to ambient particulate matter (PM)
65
air pollution, which was generally measured as particulate matter with aerodynamic
66
diameter ≤10µm (PM10) or ≤2.5µm (PM2.5) (Dominici et al., 2006, Williams et al.,
67
2014). Previous studies mainly evaluated the relationship of stroke occurrence with
68
total mass concentration of the particles, fewer have examined the stroke effects of
69
different characteristics of the particles, such as particle size and their chemical
70
constituents (Kettunen et al., 2007), and more importantly limited studies have been
71
conducted to differentiate the effects of these PM characteristics on different stroke
72
types, which presented an obstacle to a better understanding of the biological
73
mechanisms of their association.
76 77 78 79
cr
us
an
M
d
te
75
Epidemiological and toxicological studies suggest that smaller particles might be more harmful to human health. Only limited epidemiological studies, however, have
Ac ce p
74
ip t
64
examined the association between stroke and these smaller particles with inconsistent findings. For example, a study examined the relationship between stroke mortality and three size fractions (PM2.5-10, PM1-2.5 and PM1) in Barcelona, Spain, and found that PM1 and PM2.5-10, rather than PM1-2.5, was associated with stroke mortality (Perez
80
et al., 2009). A study from Helsinki, Finland only detected significant stroke effects of
81
PM2.5 and ultrafine particles (PM0.1) in warm season, but not in cold season (Kettunen
82
et al., 2007). And in Copenhagen, Denmark, ultrafine particle, rather than PM10, was
83
found to be associated with ischemic stroke hospitalization without atrial fibrillation 5
Page 5 of 34
84
(Andersen et al., 2010). One recent systematic review suggested that PM2.5 and PM10,
85
rather than PM2.5-10, were associated with stroke mortality (Wang et al., 2014). On the other hand, the current air pollution control guidelines/regulations
87
generally use total mass concentration as the indicator. Although it is important in to
88
protecting human health, more targeted air quality standards need to incorporate PM
89
chemical constituents or emission sources that are more directly related to the health
90
impacts, and this has been viewed as the ultimate goal of air pollution control policies.
91
However, such standards has been hindered by limited information on the toxicity of
92
PM constituents (Dai et al., 2014). PM consists of many chemical components that
93
originate from different sources, such as industrial emission, traffic, biomass burning
94
and coal combustion. Exploring which specific PM component was associated with a
95
given health outcome will also help to explain the underlying mechanism of the PM
96
health effects (Li et al., 2014).
98 99 100 101
cr
us
an
M
d
te
The objective of this study was to investigate the relationship between particulate
Ac ce p
97
ip t
86
matter airPM pollution with different particle sizes (PM10, PM2.5-10, PM2.5, PM1-2.5 and PM1), PM2.5 constituents and mortalities from ischemic stroke and hemorrhagic strokes in Guangzhou, China.
102
Materials and methods
103
Study setting
104
Guangzhou, the capital city of Guangdong Province, is an economic center of
105
south China. Due to the rapid economic development and corresponding rise in 6
Page 6 of 34
energy consumption over the past decades, Guangzhou this city has experienced some
107
of the worstserious air pollution among China’s cities. Guangzhou has a typical
108
subtropical humid-monsoon climate with an average annual temperature of 22 °C and
109
average rainfall of 1500-2000 mm. The residents in urban districts of Guangzhou
110
were selected as the subjects of this study, with a population of about 5.5 million,
111
accounting for 69.7% of the whole population of this city. The urban areas were
112
chosen for two reasons. First, they were close to the air monitoring stations. Second,
113
the mortality data were of high quality because most of the residents living in these
114
districts are permanent residents (Yu et al., 2012).
an
us
cr
ip t
106
116
M
115
Data collection
Daily mortality data from 1 January 2007 to 31 December 2011 were obtained
118
from the death registry system, which included the information of sex, date of death,
119
age at death, residential address, and the underlying cause of death. In Guangzhou, all
121 122 123
te
Ac ce p
120
d
117
deaths were obliged to be reported to the death registry system. Hospital or community doctors were requested to report the cause of death on a death certificate. The government has mandated detailed quality assurance (QA) and quality control (QC) for the death registry. The cause of death was coded using the International
124
Classification of Diseases, Tenth Revision (ICD-10). Mortality from stroke (ICD-10:
125
I60–I66), and sub-categories, including ischemic stroke (ICD-10: I63–I66), and
126
hemorrhagic stroke (ICD-10: I60–I62) were extracted to construct the time series. We
127
also compiled stroke mortality data among the residents near the air monitoring 7
Page 7 of 34
128
station to evaluate the impact of potential exposure misclassification. The air pollution data were collected from two different air monitoring stations.
130
The two stations were surrounded by residential areas where there were neither major
131
industrial sources nor local fugitive dust sources (Fig. 1). An automatic air monitoring
132
system was installed on the rooftop of Panyu Meteorological Centre (Station 1) to
133
measure daily air pollution from 1 January 2009 through 31 December 2011,
134
including PM10, PM2.5 and PM1, and also gaseous air pollutants including sulfur
135
dioxide (SO2) and nitrogen dioxide (NO2). Measurements of PM10, PM2.5 and PM1
136
were performed using a GRIMM Aerosol Spectrometer (Model 1.108, Grimm Aerosol
137
Technik GmbH, Ainring, Germany). Details about the air pollution collection and
138
quality control have been introduced elsewhere. In brief, the GRIMM model 1.108
139
monitor is designed to measure particle size distribution and the mass concentration
140
based on a light scattering measurement of individual particles in the collected air
141
samples. Thus, the particle density, determined by chemical composition, could affect
143 144 145
cr
us
an
M
d
te
Ac ce p
142
ip t
129
the concentration of PM mass. However, the PM mass concentrations are converted to a mass distribution using a density factor corresponding to the GRIMM established “urban environment” factor. Thus, the measured PM mass should be accurate (Grimm and Eatough 2009). We estimated PM2.5-10 concentrations by subtracting daily PM2.5
146
from PM10, and PM1-2.5 by subtracting PM1 from PM2.5. When no concentration was
147
measured on some observation days, they were treated as missing data; during the
148
measurement period of 2009-2011, the proportion of missing data for the particle
149
concentration was very low (ranging from 1% to 2%). 8
Page 8 of 34
Daily PM2.5 chemical composition data were obtained from another automatic
151
monitoring system, which was located on the rooftop of the South China Institute of
152
Environmental Sciences (Station 2). PM2.5 samples were collected on 47 mm quartz
153
microfiber filters (Whatman International Ltd, Maidstone, England, QMA) using a
154
sampler (BGI Incorporated, Waltham, MA, US, Model PQ200) operating at a flow
155
rate of 16.7 L min-1. For this study, we measured seven major PM2.5 chemical
156
constituents for four months (January, April, July and November) of each year from
157
2007 through 2010. These measures included organic carbon (OC), elemental carbon
158
(EC), and five water-soluble ions (nitrate (NO3–), sulfate (SO42–), ammonium (NH4+),
159
sodium ions (Na+), and chloride ion (Cl–)). Details of the measurement and QA/QC
160
procedure can be found elsewhere (Tao et al., 2012).
M
an
us
cr
ip t
150
We also collected daily meteorological data, including daily mean temperature
162
(°C) and relative humidity (%) from Guangzhou Weather Station. Approval to conduct
163
this study was granted by the Medical Ethics Committee of Guangdong Provincial
165 166 167
te
Ac ce p
164
d
161
Centre for Disease Control and Prevention.
Statistical analysis
Due to different time periods to measure the concentrations of different particle
168
fractions and PM2.5 chemical constituents, we constructed two datasets for the data
169
analyses. The first one involved daily mass concentrations of PM10, PM2.5-10, PM2.5,
170
PM1-2.5, and PM1 for 1 January 2009 through 31 December 2011 and the second
171
included daily concentrations of PM2.5 chemical constituents for January, April, July 9
Page 9 of 34
172
and November of 2007-2010. We examined the short-term association between daily PM pollution and stroke
174
mortality using generalized additive models (GAM) with a quasi-Poisson link
175
function to account for over-dispersion in stroke mortality. In the model, we
176
controlled for public holidays (PH) and day of the week (DOW) using categorical
177
variables. To reduce the issues associated with multiple testing and model
178
specifications, we followed some previous studies to select model specification a
179
priori and the degree of freedom (df) for seasonal patterns and long-term trends and
180
other meteorological variables (Peng et al., 2006). We applied 8 df per year for time
181
trends to filter out the information at time scales of longer than two months, 6 df for
182
mean temperature of the current day (Temp0) and moving average of previous three
183
days (Temp1–3) to account for the potential nonlinearity in the relationship between
184
temperature and mortality, and 3 df for current day’s humidity (Humidity0).
185
Smoothing spline functions were used for all nonlinear time varying variables.
187 188 189 190
cr
us
an
M
d
te
Ac ce p
186
ip t
173
We established the core model for stroke mortality without air pollutants in the
model. We used residual plots to check whether there were discernable patterns for the core model. The core model can be specified as: log[E(Yt)] =α+s(t, df=8/year) + s(Temp0, df =6) + s(Temp1–3, df=6) + s(Humidity0, df=3) + β1*DOW+β2*PH
191
where E(Yt) is the expected stroke mortality count on day t, α is the model intercept,
192
s() indicates a smoother based on penalized splines, df is the degree of freedom, t
193
represents time to adjust for long-term trend and seasonality, Temp0 is the mean 10
Page 10 of 34
temperature on the current day, Temp1–3 means the moving average for the previous 3
195
days’ temperature, Humidity0 presents the humidity on the current day, PH represents
196
a binary variable for the public holiday, DOW is an indicator for day of week, β is the
197
regression coefficient.
ip t
194
After the core model was established, we included the 5 particle size fractions in
199
the models individually to analyze the association between these pollutants and stroke
200
mortality. Because the particulate pollutants were highly inter-correlated (see Table 2),
201
we did not include these pollutants in the same model.
an
us
cr
198
For the association between PM2.5 chemical constituents and stroke mortality, we
203
included month and year, instead of time sequence, in the model to control for
204
seasonal and long-term trends, as we only have data of four months each year. For the
205
chemical constituents which were significantly associated with stroke mortality, we
206
found that they were highly correlated with PM2.5 mass concentration (Table s1), so
207
we did not include PM2.5 mass in the multivariate model in order to avoid collinearity.
209 210 211
d
te
Ac ce p
208
M
202
To examine the temporal relationship of PM pollution with stroke mortality, we
fitted the models with different lag structures from the current day (lag0) up to three lag days (lag3), as previous studies in China showed little evidence of association with a lag beyond 3 days (Yu et al., 2012). We also examined the stroke mortality impacts
212
of multi-day lags (moving averages for the current day and the previous 1, 2 and 3
213
days: lag01, lag02, and lag03, respectively). We reported the estimates at the lag with
214
the most certain results for each PM size and PM2.5 components. To testify the
215
linearity assumption of the relationship between the logarithm of daily stroke 11
Page 11 of 34
216
mortality and PM, we graphically examined the dose-response curve derived from a
217
smoothing function (Kan et al., 2007).
219
ip t
218
Sensitivity analysis
The sensitivity of the key findings was assessed in terms of the degrees of
221
freedom in the smooth function of time trends (6-9 per year) and meteorological
222
variables, including mean temperature (df = 5-9) and relative humidity (df = 3-7). To
223
check the potential exposure misclassification resulting from the pollution data from
224
one individual air monitoring station, we compiled stroke mortality data among the
225
residents near the air monitoring stations, and did a similar analysis by restricting
226
daily mortality data to those residents (with a mean distance of about 10-15 km from
227
the monitoring stations, Fig. 1).
te
d
M
an
us
cr
220
All analyses were conducted using the “mgcv” package in R (version 3.1.0; R
229
Development Core Team, Vienna, Austria). We reported the results as excess risk (ER,
230 231 232 233
Ac ce p
228
with 95% CI) in daily stroke mortality for an interquartile range (IQR) increase of PM pollution. Statistical significance was defined as p< 0.05.
Results
234
From 1 January 2007 to 31 December 2011, we recorded a total of 9,066 stroke
235
deaths, the corresponding annual stroke mortality rate was 33.21 per 100,000
236
population. oOf whichthem, 5113 were ischemic stroke and 3953 were hemorrhagic
237
stroke. On average, there were 5.0 stroke deaths per day (Table 1). 12
Page 12 of 34
During the period of 2009-2011, daily concentrations of PM10, PM2.5-10, PM2.5,
239
PM1-2.5 and PM1 in Guangzhou were 56.0, 14.6, 41.4, 4.4 and 37.0 µg/m3, respectively
240
(Table 1). PM2.5 accounted for a substantive fraction of PM10 concentration: the ratio
241
of PM2.5 to PM10 ranged from 40.5% to 98.1%, with an average of 76.0%; and PM1
242
also accounted for a substantial fraction of PM2.5 (89.3% on average, ranging from
243
35.5% to 99.3%). Daily mean concentrations of NO2 and SO2 were 44.1 and 37.1
244
µg/m3, respectively (Table 1). Various PMs were strongly correlated with each other
245
(Table 2). PM10 was strongly associated with PM2.5 (correlation coefficient, r = 0.97)
246
and with PM1 (r = 0.96). PM2.5 and PM1 were highly correlated (r = 0.99). Correlation
247
coefficients for PM with gaseous pollutants and weather factors ranged from low to
248
moderate.
M
an
us
cr
ip t
238
Over the 4 years (2007-2010), the daily averaged concentrations were 10.9 µg/m3
250
for OC, 5.6 µg/m3 for EC, 7.7 µg/m3 for nitrate, and 17.5 µg/m3 for sulfate (Table 1).
251
The other contributors to PM2.5 were ammonium (5.2 µg/m3), chloride (1.8 µg/m3)
253 254 255 256
te
Ac ce p
252
d
249
and sodium (1.8 µg/m3).
Generally, moderate to high correlations (r = 0.21–0.94) were observed for PM2.5
with the PM2.5 chemical constituents (supplemental Table s1). A significant correlation was observed among the PM2.5 components with the exception of sodium and EC.
257
Fig. 2 presents the risk estimates for different particle fractions. We found PM10,
258
PM2.5 and PM1 were associated with stroke mortality. The excess risk (ER) for per
259
IQR increase in lag 2 day’s PM10, PM2.5, and PM1 were 7.18% (95% CI: 1.62%, 13
Page 13 of 34
13.05%), 6.62% (95% CI: 1.28%, 12.25%), and 8.74% (95% CI: 1.33%, 16.68%),
261
respectively. We observed significant risk estimates for various PM fractions on
262
hemorrhagic stroke mortality; however, this analysis did not find any association
263
between PM pollutants and ischemic stroke mortality.
ip t
260
Fig. 3 illustrates the percentage change in overall stroke mortality for per IQR
265
increase in each of the PM2.5 chemical constituents, obtained from single-pollutant
266
and multi-pollutant models. There was a positive and statistically significant
267
association between overall stroke mortality and sulfate, nitrate, ammonium, OC, and
268
EC in both single-pollutant and multi-pollutant models. We found most of the
269
components had the highest impact at lag 3 day in single-day effects. In the
270
single-pollutant model, an IQR increases in the 3-day lagged concentration of sulfate
271
(14.6 ug/m3), nitrate (8.9 ug/m3), ammonium (5.1 ug/m3), OC (6.9 ug/m3), and EC
272
(4.1 ug/m3) corresponded to a 1.39% (95% CI: 0.34%, 2.45%), 1.34% (95% CI:
273
0.12%, 2.58%), 2.59% (95% CI: 0.96%, 4.25%), 1.67% (95% CI: 0.72%, 2.63%), and
275 276 277
us
an
M
d
te
Ac ce p
274
cr
264
2.71% (95% CI: 1.05%, 4.39%) increase in stroke mortality, respectively. In multi-pollutant models with SO2 or NO2 being adjusting for, we found similar effect estimates for various constituents. Consistent results were observed for hemorrhagic stroke, while no association was detected for ischemic stroke.
278
Supplemental Fig. s1 and s2 illustrate the effects of various PM2.5 chemical
279
constituents on mortalities from ischemic and hemorrhagic strokes. Consistent results
280
were observed for hemorrhagic stroke, while no association was detected for ischemic
281
stroke. 14
Page 14 of 34
Fig. 4 shows the exposure-response relationships of PM10, PM2.5 and PM1 with
283
stroke mortality. An approximately linear relationship was observed, with no clear
284
evidence of obvious threshold concentrations below which PM concentration had no
285
significant impact on stroke mortality.
ip t
282
Sensitivity analyses found that the estimates obtained from this study were
287
relatively robust to the degrees of freedom of the smoothing adjustment for time trend
288
and weather variables. And when we only used the mortality from residents
289
geographically close to the air monitoring station, we found comparable estimates
290
(supplemental Fig. s3 and s4).
an
us
cr
286
292
M
291
Discussion
There has been limited information on the association between characteristics of
294
PM pollution and stroke mortality (Kettunen et al., 2007). To our knowledge, this
295
might be the first study to simultaneously quantify the effects of particle size and
297 298 299
te
Ac ce p
296
d
293
chemical constituents of PM pollution on stroke mortality. We found a significant association between stroke mortality and PM10, PM2.5, and PM1, and we found that PM2.5 chemical constituents, mainly combustion-related and secondary particle components, exhibited associations with stroke mortality; and our analysis found
300
there was significant association of PM pollution with hemorrhagic stroke mortality,
301
but not ischemic stroke mortality.
302
The finding of this study was generally consistent with previous studies (Perez et
303
al., 2009, Andersen et al., 2010, Liu et al., 2013). For example, a case-crossover study 15
Page 15 of 34
in Barcelona, Spain, also found that PM1 was associated with cardiovascular mortality,
305
with ER for a 10 µg/m3 increase in PM1 on lag day 1 being 2.8% (95% CI: 0.0%-5.8%)
306
(Perez et al., 2009). Two recent studies from Beijing and Shenyang, China also
307
demonstrated significant cardiovascular impacts of PM0.1 and PM0.5, respectively (Liu
308
et al., 2013, Meng et al., 2013). Significant mortality effects of smaller particles were
309
also observed in other areas (Andersen et al., 2010, Belleudi et al., 2010) and in one
310
recent systematic review (Wang et al., 2014). On the other hand, some studies
311
reported that coarse and fine particles, but not ultrafine particles, might account for
312
the association between PM pollution and human health (Andersen et al., 2008), and
313
some studies did not find any significant effect of PM pollution on occurrence of
314
stroke (Villeneuve et al., 2012).
M
an
us
cr
ip t
304
This study found that ambient levels of EC and OC were among the toxic
316
components with respect to stroke mortality. This is in agreement with a recent
317
meta-analysis of short-term associations between ambient EC and cardiovascular
319 320 321
te
Ac ce p
318
d
315
mortality (Smith et al., 2009). Several other studies have also reported significant associations between cardiovascular health and OC/EC. A study from California found that an increase in daily ambient OC and EC were associated with higher cardiovascular mortality risk (Ostro et al., 2007), and further analysis suggested
322
people with lower education levels were the most vulnerable population (Ostro et al.,
323
2008). Significant cardiovascular effects have also been reported in Atlanta, US and
324
Xi’an City, China (Metzger et al., 2004, Cao et al., 2012). Several possible biological
325
pathways have been suggested to explain this association. For example, one study 16
Page 16 of 34
from Germany found higher levels of OC or EC could lead to changes in myocardial
327
repolarization, which potentially increase the risk of sudden death (Henneberger et al.,
328
2005). The ambient EC exposure has also been linked with ST-segment depression
329
among the elderly (Gold et al., 2005).
ip t
326
Our analysis found a significant association between stroke mortality and
331
secondary aerosol components of PM2.5, including sulfate, nitrate, and ammonium.
332
Sulfates have long been blamed as the main toxic component in fine particulates.
333
Consistent with our study, Mar et al. found increased cardiovascular mortality
334
associated with sulfate in Phoenix, US (Mar et al., 2000). On the other hand, a few
335
other studies did not find significant association between sulfate and human health
336
(Fairley 1999, Cao et al., 2012). Further examination of this issue is warranted.
337
Nitrate was positively associated with stroke mortality in our study. Consistent
338
findings were also reported in California, US (Fairley 2003), the Netherlands (Hoek et
339
al., 2000), and Xi’an, China (Cao et al., 2012). However, Klemm and co-authors did
341 342 343
us
an
M
d
te
Ac ce p
340
cr
330
not identify a significant association between nitrate and mortality in Atlanta, US (Klemm et al., 2004). The significant relationship between ammonium and stroke mortality observed in this study was in accordance with the findings from several recent studies. Peng’s study including 119 US communities observed a significant
344
association in single-pollutant models, and the association was not statistical
345
significant after adjusting for co-pollutants (Peng et al., 2009). In Xi’an, higher
346
mortality was also associated with ammonium, but the association was not statistically
347
significant (Huang et al., 2012). 17
Page 17 of 34
Several mechanisms have been proposed for the relationship between particulate
349
pollution and stroke mortality. The effects on ischemic stroke might be through
350
increased plasma viscosity (Yorifuji et al., 2011). While a few pathways have been
351
proposed for their effects on hemorrhagic stroke. First, when inhaled in the pulmonary
352
tract, the particles could cause pulmonary inflammation, oxidative stress, heart rate
353
variability, and alterations in cardiovascular function; local inflammation may further
354
contribute to a systemic inflammatory state, which in turn was able to activate
355
hemorrhagic pathways, deteriorate vascular function, and possibly promote rupture of
356
atherosclerotic plaques (Brook et al., 2004); second, direct ischemic damage to blood
357
vessels induced by PM pollution might result in brain hemorrhage; third, PM
358
pollution has been linked with atherosclerosis and hypertension, which may cause
359
rupture of brain vessels and hemorrhagic stroke. Furthermore, the stronger association
360
for hemorrhagic stroke than ischemic stroke might be due to existence of other
361
environmental and genetic risk factors, such as hypertension, dietary pattern, or
363 364 365
cr
us
an
M
d
te
Ac ce p
362
ip t
348
physical activity, which might be associated with hemorrhagic stroke in Guangzhou. We found significant effects of PM pollution on hemorrhagic stroke, but not
ischemic stroke, and this finding was consistent in the results of particle sizes and chemical constituents. Previous findings about the association of air pollution with
366
hemorrhagic and ischemic strokes remained inconsistent among different countries.
367
Most such studies from western countries have showed significant association with
368
ischemic but not hemorrhagic stroke, while studies from Asian countries have
369
suggested an association between air pollution and hemorrhagic stroke (Yorifuji et al., 18
Page 18 of 34
2014). For example, a Japanese study also reported a stronger effect of PM pollution
371
on hemorrhagic stroke than ischemic stroke (Yorifuji et al., 2011), while studies from
372
US and European countries suggested that PM pollution was associated with ischemic
373
stroke, rather than hemorrhagic stroke (Le Tertre et al., 2002, Wellenius et al., 2005).
374
Our study, from a more in-depth analysis based on different particle size fractions and
375
chemical compositions, supported the previous observations.
cr
us
376
ip t
370
A few limitations should be noted. First, we calculated PM2.5-10 and PM1-2.5 by subtracting PM2.5 from PM10, and PM1 from PM2.5, which was subjected to two
378
sources of measurement errors. Second, the study design was ecological in nature,
379
which did not allow us to explore individual-based association and control for
380
potential confounder at the individual level, such as smoking, hypertension,
381
occupational air pollution exposure, etc. Third, we used the data from one single
382
monitoring station, which may introduce exposure misclassification though the
383
sensitivity analysis by restricting the analysis to residents close to the monitoring
385 386 387
M
d
te
Ac ce p
384
an
377
station indicated the robustness of risk estimates. Fourth, the stroke mortality rate in the study area was relatively lower than that in other industrialized cities in China (Chen et al., 2013), indicating the possibility of under-reporting of the mortality data. Furthermore, we did not have the information of other pollutants, such as ultrafine
388
particles and ozone, which might also relate to stroke mortality. The onset date of
389
stroke symptom likely preceded date of death by a few days or longer, therefore the
390
analysis using stroke mortality may under-estimate the association (Wellenius et al.,
391
2005). 19
Page 19 of 34
392
394
Conclusion In summary, the present study suggests particle size and chemical composition
ip t
393
might be important determinant of the association between PM pollution and stroke
396
mortality, and this study suggests that PM pollution might be more relevant to
397
hemorrhagic stroke in the study area, however more studies are warranted due to the
398
underlying limitations of this study.
us
cr
395
an
399
Acknowledgments We sincerely thank those who participated in data collection and
401
management of this study. This study was supported by Guangdong Medical Research
402
Foundation (B2013077).
406 407 408 409
d None.
Ac ce p
405
Conflict of interest
te
403 404
M
400
References
Andersen, Z.J., Olsen, T.S., Andersen, K.K., Loft, S., Ketzel, M., Raaschou-Nielsen, O., 2010. Association between short-term exposure to ultrafine particles and hospital admissions for stroke in Copenhagen, Denmark. European Heart Journal 31(16):
410
2034-2040.
411
Andersen, Z.J., Wahlin, P., Raaschou-Nielsen, O., Ketzel, M., Scheike, T., Loft, S.,
412
2008. Size distribution and total number concentration of ultrafine and accumulation
413
mode particles and hospital admissions in children and the elderly in Copenhagen, 20
Page 20 of 34
Denmark. Occupational and Environmental Medicine 65(7): 458-466.
415
Belleudi, V., Faustini, A., Stafoggia, M., Cattani, G., Marconi, A., Perucci, C.A.,
416
Forastiere, F., 2010. Impact of Fine and Ultrafine Particles on Emergency Hospital
417
Admissions for Cardiac and Respiratory Diseases. Epidemiology 21(3): 414-423.
418
Brook, R.D., Franklin, B., Cascio, W., Hong, Y., Howard, G., Lipsett, M., Luepker, R.,
419
Mittleman, M., Samet, J., Smith, S.C., 2004. Air pollution and cardiovascular disease:
420
A Statement for Healthcare Professionals From the Expert Panel on Population and
421
Prevention Science of the American Heart Association. Circulation 109(21):
422
2655-2671.
423
Cao, J., Xu, H., Xu, Q., Chen, B., Kan, H., 2012. Fine Particulate Matter Constituents
424
and Cardiopulmonary Mortality in a Heavily Polluted Chinese City. Environmental
425
Health Perspectives 120(3): 373-378.
426
Chen, R., Zhang, Y., Yang, C., Zhao, Z., Xu, X., Kan, H., 2013. Acute Effect of
427
Ambient Air Pollution on Stroke Mortality in the China Air Pollution and Health
429 430 431
cr
us
an
M
d
te
Ac ce p
428
ip t
414
Effects Study. Stroke.
Dai, L., Zanobetti, A., Koutrakis, P., Schwartz, J.D., 2014. Associations of Fine Particulate Matter Species with Mortality in the United States: A Multicity Time-Series Analysis. Environ Health Perspect 122(8): 837-842.
432
Dominici, F., Peng, R.D., Bell, M.L., Pham, L., McDermott, A., Zeger, S.L., Samet,
433
J.M., 2006. Fine particulate air pollution and hospital admission for cardiovascular
434
and respiratory diseases. JAMA: The Journal of the American Medical Association
435
295(10): 1127-1134. 21
Page 21 of 34
Fairley, D., 1999. Daily mortality and air pollution in Santa Clara County, California:
437
1989-1996. Environmental Health Perspectives 107(8): 637-641.
438
Fairley, D., 2003. Mortality and air pollution for Santa Clara County, California,
439
1989–1996. Revised analyses of time-series studies of air pollution and health.
440
Special report. Boston: Health Effects Institute 97: 106.
441
Gold, D.R., Litonjua, A.A., Zanobetti, A., Coull, B.A., Schwartz, J., MacCallum, G.,
442
Verrier, R.L., Nearing, B.D., Canner, M.J., Suh, H., 2005. Air pollution and
443
ST-segment depression in elderly subjects. Environmental Health Perspectives 113(7):
444
883- 887.
445
Grimm, H., Eatough, D.J., 2009. Aerosol measurement: the use of optical light
446
scattering for the determination of particulate size distribution, and particulate mass,
447
including the semi-volatile fraction. Journal of the Air & Waste Management
448
Association 59(1): 101-107.
449
Henneberger, A., Zareba, W., Ibald-Mulli, A., Rückerl, R., Cyrys, J., Couderc, J.-P.,
451 452 453
cr
us
an
M
d
te
Ac ce p
450
ip t
436
Mykins, B., Woelke, G., Wichmann, H.-E., Peters, A., 2005. Repolarization changes induced by air pollution in ischemic heart disease patients. Environmental Health Perspectives 113(4): 440- 446.
Hoek, G., Brunekreef, B., Verhoeff, A., Wijnen, J.v., Fischer, P., 2000. Daily mortality
454
and air pollution in the Netherlands. Journal of the Air & Waste Management
455
Association 50(8): 1380-1389.
456
Huang, W., Cao, J., Tao, Y., Dai, L., Lu, S.-E., Hou, B., Wang, Z., Zhu, T., 2012.
457
Seasonal Variation of Chemical Species Associated With Short-Term Mortality 22
Page 22 of 34
Effects of PM2.5 in Xi’an, a Central City in China. American Journal of
459
Epidemiology 175(6): 556-566.
460
Kan, H., London, S.J., Chen, G., Zhang, Y., Song, G., Zhao, N., Jiang, L., Chen, B.,
461
2007. Differentiating the effects of fine and coarse particles on daily mortality in
462
Shanghai, China. Environ Int. 33(3): 376-384.
463
Kettunen, J., Lanki, T., Tiittanen, P., Aalto, P.P., Koskentalo, T., Kulmala, M., Salomaa,
464
V., Pekkanen, J., 2007. Associations of fine and ultrafine particulate air pollution with
465
stroke mortality in an area of low air pollution levels. Stroke 38(3): 918-922.
466
Klemm, R., Lipfert, F., Wyzga, R., Gust, C., 2004. Daily mortality and air pollution in
467
Atlanta: two years of data from ARIES. Inhalation toxicology 16(s1): 131-141.
468
Le Tertre, A., Medina, S., Samoli, E., Forsberg, B., Michelozzi, P., Boumghar, A.,
469
Vonk, J., Bellini, A., Atkinson, R., Ayres, J., 2002. Short-term effects of particulate air
470
pollution on cardiovascular diseases in eight European cities. Journal of epidemiology
471
and community health 56(10): 773-779.
473 474 475
cr
us
an
M
d
te
Ac ce p
472
ip t
458
Li, P., Xin, J., Wang, Y., Li, G., Pan, X., Wang, S., Cheng, M., Wen, T., Wang, G., Liu, Z., 2014. Association between particulate matter and its chemical constituents of urban air pollution and daily mortality or morbidity in Beijing City. Environmental Science and Pollution Research 22(1): 358-368.
476
Liu, L., Breitner, S., Schneider, A., Cyrys, J., Brüske, I., Franck, U., Schlink, U.,
477
Marian Leitte, A., Herbarth, O., Wiedensohler, A., 2013. Size-fractioned particulate
478
air pollution and cardiovascular emergency room visits in Beijing, China.
479
Environmental Research 121: 52-63. 23
Page 23 of 34
Mar, T.F., Norris, G.A., Koenig, J.Q., Larson, T.V., 2000. Associations between air
481
pollution and mortality in Phoenix, 1995-1997. Environmental Health Perspectives
482
108(4): 347-353.
483
Meng, X., Ma, Y., Chen, R., Zhou, Z., Chen, B., Kan, H., 2013. Size-Fractionated
484
Particle Number Concentrations and Daily Mortality in a Chinese City.
485
Environmental Health Perspectives 121(10): 1174-1178.
486
Metzger, K.B., Tolbert, P.E., Klein, M., Peel, J.L., Flanders, W.D., Todd, K.,
487
Mulholland, J.A., Ryan, P.B., Frumkin, H., 2004. Ambient air pollution and
488
cardiovascular emergency department visits. Epidemiology 15(1): 46-56.
489
Ostro, B., Feng, W.-Y., Broadwin, R., Green, S., Lipsett, M., 2007. The effects of
490
components of fine particulate air pollution on mortality in California: results from
491
CALFINE. Environmental Health Perspectives 115(1): 13-19.
492
Ostro, B., Feng, W.-Y., Broadwin, R., Malig, B., Green, R., Lipsett, M., 2008. The
493
impact of components of fine particulate matter on cardiovascular mortality in
495 496 497
cr
us
an
M
d
te
Ac ce p
494
ip t
480
susceptible subpopulations. Occupational and environmental medicine 65(11): 750-756.
Peng, R.D., Bell, M.L., Geyh, A.S., McDermott, A., Zeger, S.L., Samet, J.M., Dominici, F., 2009. Emergency Admissions for Cardiovascular and Respiratory
498
Diseases and the Chemical Composition of Fine Particle Air Pollution. Environ
499
Health Perspect. 117(6): 957-963.
500
Peng, R.D., Dominici, F., Louis, T.A., 2006. Model choice in time series studies of air
501
pollution and mortality. J. Roy. Statist. Soc. Ser. A 169(2): 179-203. 24
Page 24 of 34
Perez, L., Medina-Ramon, M., Kunzli, N., Alastuey, A., Pey, J., Perez, N., Garcia, R.,
503
Tobias, A., Querol, X., Sunyer, J., 2009. Size fractionate particulate matter, vehicle
504
traffic, and case-specific daily mortality in Barcelona, Spain. Environmental Science
505
& Technology 43(13): 4707-4714.
506
Smith, K.R., Jerrett, M., Anderson, H.R., Burnett, R.T., Stone, V., Derwent, R.,
507
Atkinson, R.W., Cohen, A., Shonkoff, S.B., Krewski, D., Pope, C.A., Thun, M.J.,
508
Thurston, G., 2009. Public health benefits of strategies to reduce greenhouse-gas
509
emissions: health implications of short-lived greenhouse pollutants. The Lancet
510
374(9707): 2091-2103.
511
Tao, J., Shen, Z., Zhu, C., Yue, J., Cao, J., Liu, S., Zhu, L., Zhang, R., 2012. Seasonal
512
variations and chemical characteristics of sub-micrometer particles (PM1) in
513
Guangzhou, China. Atmospheric Research 118: 222-231.
514
Villeneuve, P.J., Johnson, J.Y., Pasichnyk, D., Lowes, J., Kirkland, S., Rowe, B.H.,
515
2012. Short-term effects of ambient air pollution on stroke: who is most vulnerable?
517 518 519
cr
us
an
M
d
te
Ac ce p
516
ip t
502
Science of the Total Environment 430: 193-201. Wang, Y., Eliot, M.N., Wellenius, G.A., 2014. Short‐term Changes in Ambient Particulate Matter and Risk of Stroke: A Systematic Review and Meta‐analysis. Journal of the American Heart Association 3(4): e000983.
520
Wellenius, G.A., Schwartz, J., Mittleman, M.A., 2005. Air pollution and hospital
521
admissions for ischemic and hemorrhagic stroke among medicare beneficiaries.
522
Stroke 36(12): 2549-2553.
523
Williams, M.L., Atkinson, R.W., Anderson, H.R., Kelly, F.J., 2014. Associations 25
Page 25 of 34
between daily mortality in London and combined oxidant capacity, ozone and
525
nitrogen dioxide. Air Quality, Atmosphere & Health 7(4): 407-414.
526
Yorifuji, T., Kawachi, I., Sakamoto, T., Doi, H., 2011. Associations of outdoor air
527
pollution with hemorrhagic stroke mortality. Journal of occupational and
528
environmental medicine 53(2): 124-126.
529
Yorifuji, T., Suzuki, E., Kashima, S., 2014. Cardiovascular Emergency Hospital Visits
530
and Hourly Changes in Air Pollution. Stroke 45(5): 1264-1268.
531
Yu, I.T.S., Zhang, Y.H., San Tam, W.W., Yan, Q.H., Xu, Y.J., Xun, X.J., Wu, W., Ma,
532
W.J., Tian, L.W., Tse, L.A., Lao, X.Q., 2012. Effect of ambient air pollution on daily
533
mortality rates in Guangzhou, China. Atmos Environ 46: 528-535.
cr
us
an
M
Ac ce p
te
d
534 535
ip t
524
26
Page 26 of 34
535
Figure legends:
536
Fig. 1. Geographical distribution of air pollution monitoring stations in Guangzhou,
538
China.
ip t
537
cr
539
Fig. 2. Percent increase (ERR with 95% CI) in stroke mortality for per IQR increase
541
in particulate pollutants with different lag days (single lags for the current day (lag0)
542
to 3 days before the current day (lag3) and multiday lags for the current day and prior
543
1 day before (lag01), 2 days (lag02), or 3 days (lag03)).
an
us
540
M
544
Fig. 3. Percent increase (ERR with 95% CI) in overall stroke mortality for an
546
interquartile increase in PM2.5 chemical constituents with different lag days (single
547
lags for the current day (lag0) to 3 days before the current day (lag3) and multiday
548
lags for the current day and prior 1 day before (lag01), 2 days (lag02), or 3 days
550 551 552
te
Ac ce p
549
d
545
(lag03)).
Fig. 4. Smoothing plots of daily PM concentration against stroke mortality. X-axis is the PM concentration (1 µg/m3). Confounding factors included DOW, public holidays,
553
time trend, influenza epidemic, the current day’s mean temperature, previous three
554
days’ moving average of mean temperature, relative humidity, SO2 and NO2.
555
27
Page 27 of 34
555
Tables
556
Table 1
558
Summary statistics of daily stroke mortality, air pollutants, and weather conditions in
559
Guangzhou, China
Ac ce p
te
d
M
an
us
cr
ip t
557
28
Page 28 of 34
Percentile Variable
Mean±SD Min
P25
P50
P75
Max
1 January 2007–31 December 2011
560
Daily mortality count
561
5.0±3.0
0
3
5
7
15
Ischemic stroke
2.8±2.0
0
1
3
4
10
Hemorrhagic stroke
2.2±1.7
0
1
2
3
22.5±6.4
5.4
18.0
24.0
71.8±13.2
25.0
64.0
(%)
73.0
82.0
99.0
M
1 January 2009–31 December 2011
33.5
an
Relative humidity
27.7
us
Temperature (°C)
10
cr
Meteorological factors
ip t
Stroke
Air pollution (µg/m3) 56.0±33.5
0.1
29.7
48.8
75.1
198.8
PM2.5-10
14.6±12.7
0.0
5.2
10.7
20.1
74.7
41.4±23.4
23.6
36.6
55.1
134.8
te
d
PM10
0.1
PM1-2.5
4.4±3.7
0.0
2.1
3.2
5.5
39.8
PM1
37.0±21.0
0.1
20.8
33.1
49.6
120.8
NO2
44.1±23.3
8.2
27.2
39.2
56.0
155.2
SO2
37.1±25.7
2.5
19.0
30.0
46.8
176.0
Ac ce p
PM2.5
January 2007–November 2010 PM constituent (µg/m3) Sulfate
17.5±9.8
2.4
9.4
16.1
24.0
60.9
Nitrate
7.7±7.7
0.9
2.3
4.6
11.4
47.5
Ammonium
5.2±3.5
0.1
2.3
4.8
7.4
19.4
OC
10.9±7.1
2.5
6.2
9.2
13.1
53.7
EC
5.6±3.2
1.5
3.1
4.8
7.2
17.5
29
Page 29 of 34
562
Abbreviation: SD, standard deviation; Px, xth percentiles; Min, minimum; Max,
563
maximum.
ip t
564
Table 2
566
Spearman’s correlation coefficients between PM concentrations, gaseous pollutants,
567
and weather conditions in Guangzhou, China
0.87**
1.00
PM2.5
0.97**
0.74**
1.00
an
1.00
PM2.5-10
PM1-2.5
0.83**
0.81**
0.77**
1.00
PM1
0.96**
0.71**
SO2
0.50**
NO2
0.58**
d
PM10
us
PM1-2.5
PM1
SO2
NO2
Temperature
M
PM2.5-10 PM2.5
0.99**
te
PM10
0.71**
1.00
0.28**
0.55**
0.42**
0.56**
0.35**
0.63**
0.41**
-0.64** 0.65** 1.00
Ac ce p
568
Pollutants
cr
565
1.00
Temperature -0.45** -0.27** -0.48** -0.22** -0.50** -0.07*
0.44** 1.00
Humidity
-0.03
-0.22** -0.29** -0.17** -0.13** -0.18** -0.01
0.11**
**p < 0.01, * P< 0.05.
30
Page 30 of 34
Page 31 of 34
ed
ce pt
Ac
M an
cr
us
ip
Page 32 of 34
pt e
ce
Ac d
M
an
cr
us
ip
Page 33 of 34
ed
pt
ce
Ac M
an
cr
us
ip t
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4 Click here to download high resolution image
Page 34 of 34
S1438-4639(15)00152-2 http://dx.doi.org/doi:10.1016/j.ijheh.2015.11.002 IJHEH 12886
To appear in: Received date: Revised date: Accepted date:
4-7-2015 13-11-2015 14-11-2015
Please cite this article as: Lin, H., Tao, J., Du, Y., Liu, T., Qian, Z., Tian, L., Di, Q., Zeng, W., Xiao, J., Guo, L., Li, X., Xu, Y., Ma, W.,Differentiating the effects of characteristics of PM pollution on mortality from ischemic and hemorrhagic strokes, International Journal of Hygiene and Environmental Health (2015), http://dx.doi.org/10.1016/j.ijheh.2015.11.002 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Differentiating the effects of characteristics of PM pollution on mortality from
2
ischemic and hemorrhagic strokes
3
Hualiang Lin a, Jun Tao b, Yaodong Du c, Tao Liu a, Zhengmin Qian d, Linwei Tian e,
5
Qian Di f, Weilin Zeng a, Jianpeng Xiao a, Lingchuan Guo a, Xing Li a, Yanjun Xu g,
6
Wenjun Ma a,*
us
cr
ip t
4
7 8
a
9
Disease Control and Prevention, Guangzhou, 511430, China
an
Guangdong Provincial Institute of Public Health, Guangdong Provincial Center for
10
b
11
Protection, Guangzhou, China
12
c
Guangdong Provincial Weather Center, Guangzhou, China
13
d
Department of Epidemiology, School of Public Health, Saint Louis University, Saint
14
Louis, MO, USA;
16 17 18 19
M
d
te
Ac ce p
15
South China Institute of Environmental Sciences, Ministry of Environmental
e
School of Public Health, The University of Hong Kong, Hong Kong SAR, China
f
Department of Environmental Health, Harvard School of Public Health, Boston,
USA g
Guangdong Provincial Center for Disease Control and Prevention, Guangzhou,
511430, China
20 21
Corresponding author at: Wenjun Ma, Guangdong Provincial Institute of Public
22
Health, Guangdong Provincial Center for Disease Control and Prevention, Qunxian 1
Page 1 of 34
23
Road, 160, Guangzhou, China. E-mail: [email protected]
24
Abbreviations:
26
PM: particulate matter; OC: organic carbon; EC: elemental carbon; PM10: particulate
27
matter with aerodynamic diameter ≤10µm; PM2.5: particulate matter with
28
aerodynamic diameter ≤2.5µm; QA: quality assurance; QC: quality control; ICD-10:
29
International Classification of Diseases, Tenth Revision; SO2: sulfur dioxide; NO2:
30
nitrogen dioxide; GAM: generalized additive model; PH: public holidays; DOW: day
31
of the week; df: degree of freedom; ER: excess risk; CI: confidence interval; IQR:
32
interquartile range.
M
an
us
cr
ip t
25
Ac ce p
te
d
33
2
Page 2 of 34
ABSTRACT
34
Though increasing evidence supports significant association between particulate
35
matter (PM) air pollution and stroke, it remains unclear what characteristics, such as
36
particle size and chemical constituents, are responsible for this association. We
37
investigated the association between particulate matter with different particle sizes,
38
chemical constituents of fine particulate pollution and mortalities from ischemic
39
stroke and hemorrhagic stroke in Guangzhou, China. A time-series model with
40
quasi-Poisson function was applied to assess the association of PM pollution with
41
different particle sizes and chemical constituents with mortalities from ischemic and
42
hemorrhagic strokes in Guangzhou, Chinathe association of PM with different particle
43
sizes and seven major PM2.5 components with mortality from overall stroke, ischemic
44
stroke and hemorrhagic stroke in Guangzhou, we controlled for potential confounding
45
factors in the model, such as temporal trends, day of the week, public holidays,
46
meteorological factors and influenza epidemic.with adjustment for potential
48 49 50
cr
us
an
M
d
te
Ac ce p
47
ip t
33
confounding factors. We found significant association between stroke mortality and various PM fractions, such as PM10, PM2.5 and PM1, with generally larger magnitudes for smaller particles. For the PM2.5 chemical constituents, we found that organic carbon (OC), elemental carbon (EC), sulfate, nitrate, and ammonium were
51
significantly associated with hemorrhagic stroke mortality. The analysis for specific
52
types of stroke suggested that it was hemorrhagic stroke, rather than ischemic stroke,
53
that was significantly associated with PM pollution. Our study shows that various PM
54
pollution fractions are associated with stroke mortality, and constituents primarily 3
Page 3 of 34
from combustion and secondary aerosols might be the harmful components of PM2.5
56
in Guangzhou, and this study suggests that PM pollution is more relevant to
57
hemorrhagic stroke in the study area, however more studies are warranted due to the
58
underlying limitations of this study.
ip t
55
Keywords: Particulate matter air pollution; stroke; particle size; chemical constituents
us
60
cr
59
61
Ac ce p
te
d
M
an
62
4
Page 4 of 34
62 63
Introduction A great number of epidemiological studies have demonstrated a consistent increased risk from stroke with short term exposure to ambient particulate matter (PM)
65
air pollution, which was generally measured as particulate matter with aerodynamic
66
diameter ≤10µm (PM10) or ≤2.5µm (PM2.5) (Dominici et al., 2006, Williams et al.,
67
2014). Previous studies mainly evaluated the relationship of stroke occurrence with
68
total mass concentration of the particles, fewer have examined the stroke effects of
69
different characteristics of the particles, such as particle size and their chemical
70
constituents (Kettunen et al., 2007), and more importantly limited studies have been
71
conducted to differentiate the effects of these PM characteristics on different stroke
72
types, which presented an obstacle to a better understanding of the biological
73
mechanisms of their association.
76 77 78 79
cr
us
an
M
d
te
75
Epidemiological and toxicological studies suggest that smaller particles might be more harmful to human health. Only limited epidemiological studies, however, have
Ac ce p
74
ip t
64
examined the association between stroke and these smaller particles with inconsistent findings. For example, a study examined the relationship between stroke mortality and three size fractions (PM2.5-10, PM1-2.5 and PM1) in Barcelona, Spain, and found that PM1 and PM2.5-10, rather than PM1-2.5, was associated with stroke mortality (Perez
80
et al., 2009). A study from Helsinki, Finland only detected significant stroke effects of
81
PM2.5 and ultrafine particles (PM0.1) in warm season, but not in cold season (Kettunen
82
et al., 2007). And in Copenhagen, Denmark, ultrafine particle, rather than PM10, was
83
found to be associated with ischemic stroke hospitalization without atrial fibrillation 5
Page 5 of 34
84
(Andersen et al., 2010). One recent systematic review suggested that PM2.5 and PM10,
85
rather than PM2.5-10, were associated with stroke mortality (Wang et al., 2014). On the other hand, the current air pollution control guidelines/regulations
87
generally use total mass concentration as the indicator. Although it is important in to
88
protecting human health, more targeted air quality standards need to incorporate PM
89
chemical constituents or emission sources that are more directly related to the health
90
impacts, and this has been viewed as the ultimate goal of air pollution control policies.
91
However, such standards has been hindered by limited information on the toxicity of
92
PM constituents (Dai et al., 2014). PM consists of many chemical components that
93
originate from different sources, such as industrial emission, traffic, biomass burning
94
and coal combustion. Exploring which specific PM component was associated with a
95
given health outcome will also help to explain the underlying mechanism of the PM
96
health effects (Li et al., 2014).
98 99 100 101
cr
us
an
M
d
te
The objective of this study was to investigate the relationship between particulate
Ac ce p
97
ip t
86
matter airPM pollution with different particle sizes (PM10, PM2.5-10, PM2.5, PM1-2.5 and PM1), PM2.5 constituents and mortalities from ischemic stroke and hemorrhagic strokes in Guangzhou, China.
102
Materials and methods
103
Study setting
104
Guangzhou, the capital city of Guangdong Province, is an economic center of
105
south China. Due to the rapid economic development and corresponding rise in 6
Page 6 of 34
energy consumption over the past decades, Guangzhou this city has experienced some
107
of the worstserious air pollution among China’s cities. Guangzhou has a typical
108
subtropical humid-monsoon climate with an average annual temperature of 22 °C and
109
average rainfall of 1500-2000 mm. The residents in urban districts of Guangzhou
110
were selected as the subjects of this study, with a population of about 5.5 million,
111
accounting for 69.7% of the whole population of this city. The urban areas were
112
chosen for two reasons. First, they were close to the air monitoring stations. Second,
113
the mortality data were of high quality because most of the residents living in these
114
districts are permanent residents (Yu et al., 2012).
an
us
cr
ip t
106
116
M
115
Data collection
Daily mortality data from 1 January 2007 to 31 December 2011 were obtained
118
from the death registry system, which included the information of sex, date of death,
119
age at death, residential address, and the underlying cause of death. In Guangzhou, all
121 122 123
te
Ac ce p
120
d
117
deaths were obliged to be reported to the death registry system. Hospital or community doctors were requested to report the cause of death on a death certificate. The government has mandated detailed quality assurance (QA) and quality control (QC) for the death registry. The cause of death was coded using the International
124
Classification of Diseases, Tenth Revision (ICD-10). Mortality from stroke (ICD-10:
125
I60–I66), and sub-categories, including ischemic stroke (ICD-10: I63–I66), and
126
hemorrhagic stroke (ICD-10: I60–I62) were extracted to construct the time series. We
127
also compiled stroke mortality data among the residents near the air monitoring 7
Page 7 of 34
128
station to evaluate the impact of potential exposure misclassification. The air pollution data were collected from two different air monitoring stations.
130
The two stations were surrounded by residential areas where there were neither major
131
industrial sources nor local fugitive dust sources (Fig. 1). An automatic air monitoring
132
system was installed on the rooftop of Panyu Meteorological Centre (Station 1) to
133
measure daily air pollution from 1 January 2009 through 31 December 2011,
134
including PM10, PM2.5 and PM1, and also gaseous air pollutants including sulfur
135
dioxide (SO2) and nitrogen dioxide (NO2). Measurements of PM10, PM2.5 and PM1
136
were performed using a GRIMM Aerosol Spectrometer (Model 1.108, Grimm Aerosol
137
Technik GmbH, Ainring, Germany). Details about the air pollution collection and
138
quality control have been introduced elsewhere. In brief, the GRIMM model 1.108
139
monitor is designed to measure particle size distribution and the mass concentration
140
based on a light scattering measurement of individual particles in the collected air
141
samples. Thus, the particle density, determined by chemical composition, could affect
143 144 145
cr
us
an
M
d
te
Ac ce p
142
ip t
129
the concentration of PM mass. However, the PM mass concentrations are converted to a mass distribution using a density factor corresponding to the GRIMM established “urban environment” factor. Thus, the measured PM mass should be accurate (Grimm and Eatough 2009). We estimated PM2.5-10 concentrations by subtracting daily PM2.5
146
from PM10, and PM1-2.5 by subtracting PM1 from PM2.5. When no concentration was
147
measured on some observation days, they were treated as missing data; during the
148
measurement period of 2009-2011, the proportion of missing data for the particle
149
concentration was very low (ranging from 1% to 2%). 8
Page 8 of 34
Daily PM2.5 chemical composition data were obtained from another automatic
151
monitoring system, which was located on the rooftop of the South China Institute of
152
Environmental Sciences (Station 2). PM2.5 samples were collected on 47 mm quartz
153
microfiber filters (Whatman International Ltd, Maidstone, England, QMA) using a
154
sampler (BGI Incorporated, Waltham, MA, US, Model PQ200) operating at a flow
155
rate of 16.7 L min-1. For this study, we measured seven major PM2.5 chemical
156
constituents for four months (January, April, July and November) of each year from
157
2007 through 2010. These measures included organic carbon (OC), elemental carbon
158
(EC), and five water-soluble ions (nitrate (NO3–), sulfate (SO42–), ammonium (NH4+),
159
sodium ions (Na+), and chloride ion (Cl–)). Details of the measurement and QA/QC
160
procedure can be found elsewhere (Tao et al., 2012).
M
an
us
cr
ip t
150
We also collected daily meteorological data, including daily mean temperature
162
(°C) and relative humidity (%) from Guangzhou Weather Station. Approval to conduct
163
this study was granted by the Medical Ethics Committee of Guangdong Provincial
165 166 167
te
Ac ce p
164
d
161
Centre for Disease Control and Prevention.
Statistical analysis
Due to different time periods to measure the concentrations of different particle
168
fractions and PM2.5 chemical constituents, we constructed two datasets for the data
169
analyses. The first one involved daily mass concentrations of PM10, PM2.5-10, PM2.5,
170
PM1-2.5, and PM1 for 1 January 2009 through 31 December 2011 and the second
171
included daily concentrations of PM2.5 chemical constituents for January, April, July 9
Page 9 of 34
172
and November of 2007-2010. We examined the short-term association between daily PM pollution and stroke
174
mortality using generalized additive models (GAM) with a quasi-Poisson link
175
function to account for over-dispersion in stroke mortality. In the model, we
176
controlled for public holidays (PH) and day of the week (DOW) using categorical
177
variables. To reduce the issues associated with multiple testing and model
178
specifications, we followed some previous studies to select model specification a
179
priori and the degree of freedom (df) for seasonal patterns and long-term trends and
180
other meteorological variables (Peng et al., 2006). We applied 8 df per year for time
181
trends to filter out the information at time scales of longer than two months, 6 df for
182
mean temperature of the current day (Temp0) and moving average of previous three
183
days (Temp1–3) to account for the potential nonlinearity in the relationship between
184
temperature and mortality, and 3 df for current day’s humidity (Humidity0).
185
Smoothing spline functions were used for all nonlinear time varying variables.
187 188 189 190
cr
us
an
M
d
te
Ac ce p
186
ip t
173
We established the core model for stroke mortality without air pollutants in the
model. We used residual plots to check whether there were discernable patterns for the core model. The core model can be specified as: log[E(Yt)] =α+s(t, df=8/year) + s(Temp0, df =6) + s(Temp1–3, df=6) + s(Humidity0, df=3) + β1*DOW+β2*PH
191
where E(Yt) is the expected stroke mortality count on day t, α is the model intercept,
192
s() indicates a smoother based on penalized splines, df is the degree of freedom, t
193
represents time to adjust for long-term trend and seasonality, Temp0 is the mean 10
Page 10 of 34
temperature on the current day, Temp1–3 means the moving average for the previous 3
195
days’ temperature, Humidity0 presents the humidity on the current day, PH represents
196
a binary variable for the public holiday, DOW is an indicator for day of week, β is the
197
regression coefficient.
ip t
194
After the core model was established, we included the 5 particle size fractions in
199
the models individually to analyze the association between these pollutants and stroke
200
mortality. Because the particulate pollutants were highly inter-correlated (see Table 2),
201
we did not include these pollutants in the same model.
an
us
cr
198
For the association between PM2.5 chemical constituents and stroke mortality, we
203
included month and year, instead of time sequence, in the model to control for
204
seasonal and long-term trends, as we only have data of four months each year. For the
205
chemical constituents which were significantly associated with stroke mortality, we
206
found that they were highly correlated with PM2.5 mass concentration (Table s1), so
207
we did not include PM2.5 mass in the multivariate model in order to avoid collinearity.
209 210 211
d
te
Ac ce p
208
M
202
To examine the temporal relationship of PM pollution with stroke mortality, we
fitted the models with different lag structures from the current day (lag0) up to three lag days (lag3), as previous studies in China showed little evidence of association with a lag beyond 3 days (Yu et al., 2012). We also examined the stroke mortality impacts
212
of multi-day lags (moving averages for the current day and the previous 1, 2 and 3
213
days: lag01, lag02, and lag03, respectively). We reported the estimates at the lag with
214
the most certain results for each PM size and PM2.5 components. To testify the
215
linearity assumption of the relationship between the logarithm of daily stroke 11
Page 11 of 34
216
mortality and PM, we graphically examined the dose-response curve derived from a
217
smoothing function (Kan et al., 2007).
219
ip t
218
Sensitivity analysis
The sensitivity of the key findings was assessed in terms of the degrees of
221
freedom in the smooth function of time trends (6-9 per year) and meteorological
222
variables, including mean temperature (df = 5-9) and relative humidity (df = 3-7). To
223
check the potential exposure misclassification resulting from the pollution data from
224
one individual air monitoring station, we compiled stroke mortality data among the
225
residents near the air monitoring stations, and did a similar analysis by restricting
226
daily mortality data to those residents (with a mean distance of about 10-15 km from
227
the monitoring stations, Fig. 1).
te
d
M
an
us
cr
220
All analyses were conducted using the “mgcv” package in R (version 3.1.0; R
229
Development Core Team, Vienna, Austria). We reported the results as excess risk (ER,
230 231 232 233
Ac ce p
228
with 95% CI) in daily stroke mortality for an interquartile range (IQR) increase of PM pollution. Statistical significance was defined as p< 0.05.
Results
234
From 1 January 2007 to 31 December 2011, we recorded a total of 9,066 stroke
235
deaths, the corresponding annual stroke mortality rate was 33.21 per 100,000
236
population. oOf whichthem, 5113 were ischemic stroke and 3953 were hemorrhagic
237
stroke. On average, there were 5.0 stroke deaths per day (Table 1). 12
Page 12 of 34
During the period of 2009-2011, daily concentrations of PM10, PM2.5-10, PM2.5,
239
PM1-2.5 and PM1 in Guangzhou were 56.0, 14.6, 41.4, 4.4 and 37.0 µg/m3, respectively
240
(Table 1). PM2.5 accounted for a substantive fraction of PM10 concentration: the ratio
241
of PM2.5 to PM10 ranged from 40.5% to 98.1%, with an average of 76.0%; and PM1
242
also accounted for a substantial fraction of PM2.5 (89.3% on average, ranging from
243
35.5% to 99.3%). Daily mean concentrations of NO2 and SO2 were 44.1 and 37.1
244
µg/m3, respectively (Table 1). Various PMs were strongly correlated with each other
245
(Table 2). PM10 was strongly associated with PM2.5 (correlation coefficient, r = 0.97)
246
and with PM1 (r = 0.96). PM2.5 and PM1 were highly correlated (r = 0.99). Correlation
247
coefficients for PM with gaseous pollutants and weather factors ranged from low to
248
moderate.
M
an
us
cr
ip t
238
Over the 4 years (2007-2010), the daily averaged concentrations were 10.9 µg/m3
250
for OC, 5.6 µg/m3 for EC, 7.7 µg/m3 for nitrate, and 17.5 µg/m3 for sulfate (Table 1).
251
The other contributors to PM2.5 were ammonium (5.2 µg/m3), chloride (1.8 µg/m3)
253 254 255 256
te
Ac ce p
252
d
249
and sodium (1.8 µg/m3).
Generally, moderate to high correlations (r = 0.21–0.94) were observed for PM2.5
with the PM2.5 chemical constituents (supplemental Table s1). A significant correlation was observed among the PM2.5 components with the exception of sodium and EC.
257
Fig. 2 presents the risk estimates for different particle fractions. We found PM10,
258
PM2.5 and PM1 were associated with stroke mortality. The excess risk (ER) for per
259
IQR increase in lag 2 day’s PM10, PM2.5, and PM1 were 7.18% (95% CI: 1.62%, 13
Page 13 of 34
13.05%), 6.62% (95% CI: 1.28%, 12.25%), and 8.74% (95% CI: 1.33%, 16.68%),
261
respectively. We observed significant risk estimates for various PM fractions on
262
hemorrhagic stroke mortality; however, this analysis did not find any association
263
between PM pollutants and ischemic stroke mortality.
ip t
260
Fig. 3 illustrates the percentage change in overall stroke mortality for per IQR
265
increase in each of the PM2.5 chemical constituents, obtained from single-pollutant
266
and multi-pollutant models. There was a positive and statistically significant
267
association between overall stroke mortality and sulfate, nitrate, ammonium, OC, and
268
EC in both single-pollutant and multi-pollutant models. We found most of the
269
components had the highest impact at lag 3 day in single-day effects. In the
270
single-pollutant model, an IQR increases in the 3-day lagged concentration of sulfate
271
(14.6 ug/m3), nitrate (8.9 ug/m3), ammonium (5.1 ug/m3), OC (6.9 ug/m3), and EC
272
(4.1 ug/m3) corresponded to a 1.39% (95% CI: 0.34%, 2.45%), 1.34% (95% CI:
273
0.12%, 2.58%), 2.59% (95% CI: 0.96%, 4.25%), 1.67% (95% CI: 0.72%, 2.63%), and
275 276 277
us
an
M
d
te
Ac ce p
274
cr
264
2.71% (95% CI: 1.05%, 4.39%) increase in stroke mortality, respectively. In multi-pollutant models with SO2 or NO2 being adjusting for, we found similar effect estimates for various constituents. Consistent results were observed for hemorrhagic stroke, while no association was detected for ischemic stroke.
278
Supplemental Fig. s1 and s2 illustrate the effects of various PM2.5 chemical
279
constituents on mortalities from ischemic and hemorrhagic strokes. Consistent results
280
were observed for hemorrhagic stroke, while no association was detected for ischemic
281
stroke. 14
Page 14 of 34
Fig. 4 shows the exposure-response relationships of PM10, PM2.5 and PM1 with
283
stroke mortality. An approximately linear relationship was observed, with no clear
284
evidence of obvious threshold concentrations below which PM concentration had no
285
significant impact on stroke mortality.
ip t
282
Sensitivity analyses found that the estimates obtained from this study were
287
relatively robust to the degrees of freedom of the smoothing adjustment for time trend
288
and weather variables. And when we only used the mortality from residents
289
geographically close to the air monitoring station, we found comparable estimates
290
(supplemental Fig. s3 and s4).
an
us
cr
286
292
M
291
Discussion
There has been limited information on the association between characteristics of
294
PM pollution and stroke mortality (Kettunen et al., 2007). To our knowledge, this
295
might be the first study to simultaneously quantify the effects of particle size and
297 298 299
te
Ac ce p
296
d
293
chemical constituents of PM pollution on stroke mortality. We found a significant association between stroke mortality and PM10, PM2.5, and PM1, and we found that PM2.5 chemical constituents, mainly combustion-related and secondary particle components, exhibited associations with stroke mortality; and our analysis found
300
there was significant association of PM pollution with hemorrhagic stroke mortality,
301
but not ischemic stroke mortality.
302
The finding of this study was generally consistent with previous studies (Perez et
303
al., 2009, Andersen et al., 2010, Liu et al., 2013). For example, a case-crossover study 15
Page 15 of 34
in Barcelona, Spain, also found that PM1 was associated with cardiovascular mortality,
305
with ER for a 10 µg/m3 increase in PM1 on lag day 1 being 2.8% (95% CI: 0.0%-5.8%)
306
(Perez et al., 2009). Two recent studies from Beijing and Shenyang, China also
307
demonstrated significant cardiovascular impacts of PM0.1 and PM0.5, respectively (Liu
308
et al., 2013, Meng et al., 2013). Significant mortality effects of smaller particles were
309
also observed in other areas (Andersen et al., 2010, Belleudi et al., 2010) and in one
310
recent systematic review (Wang et al., 2014). On the other hand, some studies
311
reported that coarse and fine particles, but not ultrafine particles, might account for
312
the association between PM pollution and human health (Andersen et al., 2008), and
313
some studies did not find any significant effect of PM pollution on occurrence of
314
stroke (Villeneuve et al., 2012).
M
an
us
cr
ip t
304
This study found that ambient levels of EC and OC were among the toxic
316
components with respect to stroke mortality. This is in agreement with a recent
317
meta-analysis of short-term associations between ambient EC and cardiovascular
319 320 321
te
Ac ce p
318
d
315
mortality (Smith et al., 2009). Several other studies have also reported significant associations between cardiovascular health and OC/EC. A study from California found that an increase in daily ambient OC and EC were associated with higher cardiovascular mortality risk (Ostro et al., 2007), and further analysis suggested
322
people with lower education levels were the most vulnerable population (Ostro et al.,
323
2008). Significant cardiovascular effects have also been reported in Atlanta, US and
324
Xi’an City, China (Metzger et al., 2004, Cao et al., 2012). Several possible biological
325
pathways have been suggested to explain this association. For example, one study 16
Page 16 of 34
from Germany found higher levels of OC or EC could lead to changes in myocardial
327
repolarization, which potentially increase the risk of sudden death (Henneberger et al.,
328
2005). The ambient EC exposure has also been linked with ST-segment depression
329
among the elderly (Gold et al., 2005).
ip t
326
Our analysis found a significant association between stroke mortality and
331
secondary aerosol components of PM2.5, including sulfate, nitrate, and ammonium.
332
Sulfates have long been blamed as the main toxic component in fine particulates.
333
Consistent with our study, Mar et al. found increased cardiovascular mortality
334
associated with sulfate in Phoenix, US (Mar et al., 2000). On the other hand, a few
335
other studies did not find significant association between sulfate and human health
336
(Fairley 1999, Cao et al., 2012). Further examination of this issue is warranted.
337
Nitrate was positively associated with stroke mortality in our study. Consistent
338
findings were also reported in California, US (Fairley 2003), the Netherlands (Hoek et
339
al., 2000), and Xi’an, China (Cao et al., 2012). However, Klemm and co-authors did
341 342 343
us
an
M
d
te
Ac ce p
340
cr
330
not identify a significant association between nitrate and mortality in Atlanta, US (Klemm et al., 2004). The significant relationship between ammonium and stroke mortality observed in this study was in accordance with the findings from several recent studies. Peng’s study including 119 US communities observed a significant
344
association in single-pollutant models, and the association was not statistical
345
significant after adjusting for co-pollutants (Peng et al., 2009). In Xi’an, higher
346
mortality was also associated with ammonium, but the association was not statistically
347
significant (Huang et al., 2012). 17
Page 17 of 34
Several mechanisms have been proposed for the relationship between particulate
349
pollution and stroke mortality. The effects on ischemic stroke might be through
350
increased plasma viscosity (Yorifuji et al., 2011). While a few pathways have been
351
proposed for their effects on hemorrhagic stroke. First, when inhaled in the pulmonary
352
tract, the particles could cause pulmonary inflammation, oxidative stress, heart rate
353
variability, and alterations in cardiovascular function; local inflammation may further
354
contribute to a systemic inflammatory state, which in turn was able to activate
355
hemorrhagic pathways, deteriorate vascular function, and possibly promote rupture of
356
atherosclerotic plaques (Brook et al., 2004); second, direct ischemic damage to blood
357
vessels induced by PM pollution might result in brain hemorrhage; third, PM
358
pollution has been linked with atherosclerosis and hypertension, which may cause
359
rupture of brain vessels and hemorrhagic stroke. Furthermore, the stronger association
360
for hemorrhagic stroke than ischemic stroke might be due to existence of other
361
environmental and genetic risk factors, such as hypertension, dietary pattern, or
363 364 365
cr
us
an
M
d
te
Ac ce p
362
ip t
348
physical activity, which might be associated with hemorrhagic stroke in Guangzhou. We found significant effects of PM pollution on hemorrhagic stroke, but not
ischemic stroke, and this finding was consistent in the results of particle sizes and chemical constituents. Previous findings about the association of air pollution with
366
hemorrhagic and ischemic strokes remained inconsistent among different countries.
367
Most such studies from western countries have showed significant association with
368
ischemic but not hemorrhagic stroke, while studies from Asian countries have
369
suggested an association between air pollution and hemorrhagic stroke (Yorifuji et al., 18
Page 18 of 34
2014). For example, a Japanese study also reported a stronger effect of PM pollution
371
on hemorrhagic stroke than ischemic stroke (Yorifuji et al., 2011), while studies from
372
US and European countries suggested that PM pollution was associated with ischemic
373
stroke, rather than hemorrhagic stroke (Le Tertre et al., 2002, Wellenius et al., 2005).
374
Our study, from a more in-depth analysis based on different particle size fractions and
375
chemical compositions, supported the previous observations.
cr
us
376
ip t
370
A few limitations should be noted. First, we calculated PM2.5-10 and PM1-2.5 by subtracting PM2.5 from PM10, and PM1 from PM2.5, which was subjected to two
378
sources of measurement errors. Second, the study design was ecological in nature,
379
which did not allow us to explore individual-based association and control for
380
potential confounder at the individual level, such as smoking, hypertension,
381
occupational air pollution exposure, etc. Third, we used the data from one single
382
monitoring station, which may introduce exposure misclassification though the
383
sensitivity analysis by restricting the analysis to residents close to the monitoring
385 386 387
M
d
te
Ac ce p
384
an
377
station indicated the robustness of risk estimates. Fourth, the stroke mortality rate in the study area was relatively lower than that in other industrialized cities in China (Chen et al., 2013), indicating the possibility of under-reporting of the mortality data. Furthermore, we did not have the information of other pollutants, such as ultrafine
388
particles and ozone, which might also relate to stroke mortality. The onset date of
389
stroke symptom likely preceded date of death by a few days or longer, therefore the
390
analysis using stroke mortality may under-estimate the association (Wellenius et al.,
391
2005). 19
Page 19 of 34
392
394
Conclusion In summary, the present study suggests particle size and chemical composition
ip t
393
might be important determinant of the association between PM pollution and stroke
396
mortality, and this study suggests that PM pollution might be more relevant to
397
hemorrhagic stroke in the study area, however more studies are warranted due to the
398
underlying limitations of this study.
us
cr
395
an
399
Acknowledgments We sincerely thank those who participated in data collection and
401
management of this study. This study was supported by Guangdong Medical Research
402
Foundation (B2013077).
406 407 408 409
d None.
Ac ce p
405
Conflict of interest
te
403 404
M
400
References
Andersen, Z.J., Olsen, T.S., Andersen, K.K., Loft, S., Ketzel, M., Raaschou-Nielsen, O., 2010. Association between short-term exposure to ultrafine particles and hospital admissions for stroke in Copenhagen, Denmark. European Heart Journal 31(16):
410
2034-2040.
411
Andersen, Z.J., Wahlin, P., Raaschou-Nielsen, O., Ketzel, M., Scheike, T., Loft, S.,
412
2008. Size distribution and total number concentration of ultrafine and accumulation
413
mode particles and hospital admissions in children and the elderly in Copenhagen, 20
Page 20 of 34
Denmark. Occupational and Environmental Medicine 65(7): 458-466.
415
Belleudi, V., Faustini, A., Stafoggia, M., Cattani, G., Marconi, A., Perucci, C.A.,
416
Forastiere, F., 2010. Impact of Fine and Ultrafine Particles on Emergency Hospital
417
Admissions for Cardiac and Respiratory Diseases. Epidemiology 21(3): 414-423.
418
Brook, R.D., Franklin, B., Cascio, W., Hong, Y., Howard, G., Lipsett, M., Luepker, R.,
419
Mittleman, M., Samet, J., Smith, S.C., 2004. Air pollution and cardiovascular disease:
420
A Statement for Healthcare Professionals From the Expert Panel on Population and
421
Prevention Science of the American Heart Association. Circulation 109(21):
422
2655-2671.
423
Cao, J., Xu, H., Xu, Q., Chen, B., Kan, H., 2012. Fine Particulate Matter Constituents
424
and Cardiopulmonary Mortality in a Heavily Polluted Chinese City. Environmental
425
Health Perspectives 120(3): 373-378.
426
Chen, R., Zhang, Y., Yang, C., Zhao, Z., Xu, X., Kan, H., 2013. Acute Effect of
427
Ambient Air Pollution on Stroke Mortality in the China Air Pollution and Health
429 430 431
cr
us
an
M
d
te
Ac ce p
428
ip t
414
Effects Study. Stroke.
Dai, L., Zanobetti, A., Koutrakis, P., Schwartz, J.D., 2014. Associations of Fine Particulate Matter Species with Mortality in the United States: A Multicity Time-Series Analysis. Environ Health Perspect 122(8): 837-842.
432
Dominici, F., Peng, R.D., Bell, M.L., Pham, L., McDermott, A., Zeger, S.L., Samet,
433
J.M., 2006. Fine particulate air pollution and hospital admission for cardiovascular
434
and respiratory diseases. JAMA: The Journal of the American Medical Association
435
295(10): 1127-1134. 21
Page 21 of 34
Fairley, D., 1999. Daily mortality and air pollution in Santa Clara County, California:
437
1989-1996. Environmental Health Perspectives 107(8): 637-641.
438
Fairley, D., 2003. Mortality and air pollution for Santa Clara County, California,
439
1989–1996. Revised analyses of time-series studies of air pollution and health.
440
Special report. Boston: Health Effects Institute 97: 106.
441
Gold, D.R., Litonjua, A.A., Zanobetti, A., Coull, B.A., Schwartz, J., MacCallum, G.,
442
Verrier, R.L., Nearing, B.D., Canner, M.J., Suh, H., 2005. Air pollution and
443
ST-segment depression in elderly subjects. Environmental Health Perspectives 113(7):
444
883- 887.
445
Grimm, H., Eatough, D.J., 2009. Aerosol measurement: the use of optical light
446
scattering for the determination of particulate size distribution, and particulate mass,
447
including the semi-volatile fraction. Journal of the Air & Waste Management
448
Association 59(1): 101-107.
449
Henneberger, A., Zareba, W., Ibald-Mulli, A., Rückerl, R., Cyrys, J., Couderc, J.-P.,
451 452 453
cr
us
an
M
d
te
Ac ce p
450
ip t
436
Mykins, B., Woelke, G., Wichmann, H.-E., Peters, A., 2005. Repolarization changes induced by air pollution in ischemic heart disease patients. Environmental Health Perspectives 113(4): 440- 446.
Hoek, G., Brunekreef, B., Verhoeff, A., Wijnen, J.v., Fischer, P., 2000. Daily mortality
454
and air pollution in the Netherlands. Journal of the Air & Waste Management
455
Association 50(8): 1380-1389.
456
Huang, W., Cao, J., Tao, Y., Dai, L., Lu, S.-E., Hou, B., Wang, Z., Zhu, T., 2012.
457
Seasonal Variation of Chemical Species Associated With Short-Term Mortality 22
Page 22 of 34
Effects of PM2.5 in Xi’an, a Central City in China. American Journal of
459
Epidemiology 175(6): 556-566.
460
Kan, H., London, S.J., Chen, G., Zhang, Y., Song, G., Zhao, N., Jiang, L., Chen, B.,
461
2007. Differentiating the effects of fine and coarse particles on daily mortality in
462
Shanghai, China. Environ Int. 33(3): 376-384.
463
Kettunen, J., Lanki, T., Tiittanen, P., Aalto, P.P., Koskentalo, T., Kulmala, M., Salomaa,
464
V., Pekkanen, J., 2007. Associations of fine and ultrafine particulate air pollution with
465
stroke mortality in an area of low air pollution levels. Stroke 38(3): 918-922.
466
Klemm, R., Lipfert, F., Wyzga, R., Gust, C., 2004. Daily mortality and air pollution in
467
Atlanta: two years of data from ARIES. Inhalation toxicology 16(s1): 131-141.
468
Le Tertre, A., Medina, S., Samoli, E., Forsberg, B., Michelozzi, P., Boumghar, A.,
469
Vonk, J., Bellini, A., Atkinson, R., Ayres, J., 2002. Short-term effects of particulate air
470
pollution on cardiovascular diseases in eight European cities. Journal of epidemiology
471
and community health 56(10): 773-779.
473 474 475
cr
us
an
M
d
te
Ac ce p
472
ip t
458
Li, P., Xin, J., Wang, Y., Li, G., Pan, X., Wang, S., Cheng, M., Wen, T., Wang, G., Liu, Z., 2014. Association between particulate matter and its chemical constituents of urban air pollution and daily mortality or morbidity in Beijing City. Environmental Science and Pollution Research 22(1): 358-368.
476
Liu, L., Breitner, S., Schneider, A., Cyrys, J., Brüske, I., Franck, U., Schlink, U.,
477
Marian Leitte, A., Herbarth, O., Wiedensohler, A., 2013. Size-fractioned particulate
478
air pollution and cardiovascular emergency room visits in Beijing, China.
479
Environmental Research 121: 52-63. 23
Page 23 of 34
Mar, T.F., Norris, G.A., Koenig, J.Q., Larson, T.V., 2000. Associations between air
481
pollution and mortality in Phoenix, 1995-1997. Environmental Health Perspectives
482
108(4): 347-353.
483
Meng, X., Ma, Y., Chen, R., Zhou, Z., Chen, B., Kan, H., 2013. Size-Fractionated
484
Particle Number Concentrations and Daily Mortality in a Chinese City.
485
Environmental Health Perspectives 121(10): 1174-1178.
486
Metzger, K.B., Tolbert, P.E., Klein, M., Peel, J.L., Flanders, W.D., Todd, K.,
487
Mulholland, J.A., Ryan, P.B., Frumkin, H., 2004. Ambient air pollution and
488
cardiovascular emergency department visits. Epidemiology 15(1): 46-56.
489
Ostro, B., Feng, W.-Y., Broadwin, R., Green, S., Lipsett, M., 2007. The effects of
490
components of fine particulate air pollution on mortality in California: results from
491
CALFINE. Environmental Health Perspectives 115(1): 13-19.
492
Ostro, B., Feng, W.-Y., Broadwin, R., Malig, B., Green, R., Lipsett, M., 2008. The
493
impact of components of fine particulate matter on cardiovascular mortality in
495 496 497
cr
us
an
M
d
te
Ac ce p
494
ip t
480
susceptible subpopulations. Occupational and environmental medicine 65(11): 750-756.
Peng, R.D., Bell, M.L., Geyh, A.S., McDermott, A., Zeger, S.L., Samet, J.M., Dominici, F., 2009. Emergency Admissions for Cardiovascular and Respiratory
498
Diseases and the Chemical Composition of Fine Particle Air Pollution. Environ
499
Health Perspect. 117(6): 957-963.
500
Peng, R.D., Dominici, F., Louis, T.A., 2006. Model choice in time series studies of air
501
pollution and mortality. J. Roy. Statist. Soc. Ser. A 169(2): 179-203. 24
Page 24 of 34
Perez, L., Medina-Ramon, M., Kunzli, N., Alastuey, A., Pey, J., Perez, N., Garcia, R.,
503
Tobias, A., Querol, X., Sunyer, J., 2009. Size fractionate particulate matter, vehicle
504
traffic, and case-specific daily mortality in Barcelona, Spain. Environmental Science
505
& Technology 43(13): 4707-4714.
506
Smith, K.R., Jerrett, M., Anderson, H.R., Burnett, R.T., Stone, V., Derwent, R.,
507
Atkinson, R.W., Cohen, A., Shonkoff, S.B., Krewski, D., Pope, C.A., Thun, M.J.,
508
Thurston, G., 2009. Public health benefits of strategies to reduce greenhouse-gas
509
emissions: health implications of short-lived greenhouse pollutants. The Lancet
510
374(9707): 2091-2103.
511
Tao, J., Shen, Z., Zhu, C., Yue, J., Cao, J., Liu, S., Zhu, L., Zhang, R., 2012. Seasonal
512
variations and chemical characteristics of sub-micrometer particles (PM1) in
513
Guangzhou, China. Atmospheric Research 118: 222-231.
514
Villeneuve, P.J., Johnson, J.Y., Pasichnyk, D., Lowes, J., Kirkland, S., Rowe, B.H.,
515
2012. Short-term effects of ambient air pollution on stroke: who is most vulnerable?
517 518 519
cr
us
an
M
d
te
Ac ce p
516
ip t
502
Science of the Total Environment 430: 193-201. Wang, Y., Eliot, M.N., Wellenius, G.A., 2014. Short‐term Changes in Ambient Particulate Matter and Risk of Stroke: A Systematic Review and Meta‐analysis. Journal of the American Heart Association 3(4): e000983.
520
Wellenius, G.A., Schwartz, J., Mittleman, M.A., 2005. Air pollution and hospital
521
admissions for ischemic and hemorrhagic stroke among medicare beneficiaries.
522
Stroke 36(12): 2549-2553.
523
Williams, M.L., Atkinson, R.W., Anderson, H.R., Kelly, F.J., 2014. Associations 25
Page 25 of 34
between daily mortality in London and combined oxidant capacity, ozone and
525
nitrogen dioxide. Air Quality, Atmosphere & Health 7(4): 407-414.
526
Yorifuji, T., Kawachi, I., Sakamoto, T., Doi, H., 2011. Associations of outdoor air
527
pollution with hemorrhagic stroke mortality. Journal of occupational and
528
environmental medicine 53(2): 124-126.
529
Yorifuji, T., Suzuki, E., Kashima, S., 2014. Cardiovascular Emergency Hospital Visits
530
and Hourly Changes in Air Pollution. Stroke 45(5): 1264-1268.
531
Yu, I.T.S., Zhang, Y.H., San Tam, W.W., Yan, Q.H., Xu, Y.J., Xun, X.J., Wu, W., Ma,
532
W.J., Tian, L.W., Tse, L.A., Lao, X.Q., 2012. Effect of ambient air pollution on daily
533
mortality rates in Guangzhou, China. Atmos Environ 46: 528-535.
cr
us
an
M
Ac ce p
te
d
534 535
ip t
524
26
Page 26 of 34
535
Figure legends:
536
Fig. 1. Geographical distribution of air pollution monitoring stations in Guangzhou,
538
China.
ip t
537
cr
539
Fig. 2. Percent increase (ERR with 95% CI) in stroke mortality for per IQR increase
541
in particulate pollutants with different lag days (single lags for the current day (lag0)
542
to 3 days before the current day (lag3) and multiday lags for the current day and prior
543
1 day before (lag01), 2 days (lag02), or 3 days (lag03)).
an
us
540
M
544
Fig. 3. Percent increase (ERR with 95% CI) in overall stroke mortality for an
546
interquartile increase in PM2.5 chemical constituents with different lag days (single
547
lags for the current day (lag0) to 3 days before the current day (lag3) and multiday
548
lags for the current day and prior 1 day before (lag01), 2 days (lag02), or 3 days
550 551 552
te
Ac ce p
549
d
545
(lag03)).
Fig. 4. Smoothing plots of daily PM concentration against stroke mortality. X-axis is the PM concentration (1 µg/m3). Confounding factors included DOW, public holidays,
553
time trend, influenza epidemic, the current day’s mean temperature, previous three
554
days’ moving average of mean temperature, relative humidity, SO2 and NO2.
555
27
Page 27 of 34
555
Tables
556
Table 1
558
Summary statistics of daily stroke mortality, air pollutants, and weather conditions in
559
Guangzhou, China
Ac ce p
te
d
M
an
us
cr
ip t
557
28
Page 28 of 34
Percentile Variable
Mean±SD Min
P25
P50
P75
Max
1 January 2007–31 December 2011
560
Daily mortality count
561
5.0±3.0
0
3
5
7
15
Ischemic stroke
2.8±2.0
0
1
3
4
10
Hemorrhagic stroke
2.2±1.7
0
1
2
3
22.5±6.4
5.4
18.0
24.0
71.8±13.2
25.0
64.0
(%)
73.0
82.0
99.0
M
1 January 2009–31 December 2011
33.5
an
Relative humidity
27.7
us
Temperature (°C)
10
cr
Meteorological factors
ip t
Stroke
Air pollution (µg/m3) 56.0±33.5
0.1
29.7
48.8
75.1
198.8
PM2.5-10
14.6±12.7
0.0
5.2
10.7
20.1
74.7
41.4±23.4
23.6
36.6
55.1
134.8
te
d
PM10
0.1
PM1-2.5
4.4±3.7
0.0
2.1
3.2
5.5
39.8
PM1
37.0±21.0
0.1
20.8
33.1
49.6
120.8
NO2
44.1±23.3
8.2
27.2
39.2
56.0
155.2
SO2
37.1±25.7
2.5
19.0
30.0
46.8
176.0
Ac ce p
PM2.5
January 2007–November 2010 PM constituent (µg/m3) Sulfate
17.5±9.8
2.4
9.4
16.1
24.0
60.9
Nitrate
7.7±7.7
0.9
2.3
4.6
11.4
47.5
Ammonium
5.2±3.5
0.1
2.3
4.8
7.4
19.4
OC
10.9±7.1
2.5
6.2
9.2
13.1
53.7
EC
5.6±3.2
1.5
3.1
4.8
7.2
17.5
29
Page 29 of 34
562
Abbreviation: SD, standard deviation; Px, xth percentiles; Min, minimum; Max,
563
maximum.
ip t
564
Table 2
566
Spearman’s correlation coefficients between PM concentrations, gaseous pollutants,
567
and weather conditions in Guangzhou, China
0.87**
1.00
PM2.5
0.97**
0.74**
1.00
an
1.00
PM2.5-10
PM1-2.5
0.83**
0.81**
0.77**
1.00
PM1
0.96**
0.71**
SO2
0.50**
NO2
0.58**
d
PM10
us
PM1-2.5
PM1
SO2
NO2
Temperature
M
PM2.5-10 PM2.5
0.99**
te
PM10
0.71**
1.00
0.28**
0.55**
0.42**
0.56**
0.35**
0.63**
0.41**
-0.64** 0.65** 1.00
Ac ce p
568
Pollutants
cr
565
1.00
Temperature -0.45** -0.27** -0.48** -0.22** -0.50** -0.07*
0.44** 1.00
Humidity
-0.03
-0.22** -0.29** -0.17** -0.13** -0.18** -0.01
0.11**
**p < 0.01, * P< 0.05.
30
Page 30 of 34
Page 31 of 34
ed
ce pt
Ac
M an
cr
us
ip
Page 32 of 34
pt e
ce
Ac d
M
an
cr
us
ip
Page 33 of 34
ed
pt
ce
Ac M
an
cr
us
ip t
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4 Click here to download high resolution image
Page 34 of 34