particulate air pollution and mortality in china: a time- series … › sites › default › files...
TRANSCRIPT
Confidential: For Review O
nly
Particulate Air Pollution and Mortality in China: A Time-
Series Analysis in the Largest 38 Chinese Cities
Journal: BMJ
Manuscript ID BMJ.2016.033574.R1
Article Type: Research
BMJ Journal: BMJ
Date Submitted by the Author: 20-Oct-2016
Complete List of Authors: Yin, Peng; China CDC He, Guojun; The Hong Kong University of Science and Technology Chiu, Kowk Yan ; HKUST Fan, Maorong; Xiyuan Hospital, China Academy of Chinese Medical Sciences
Liu, Tong; HKUST Mu, Quan; The NatureConservancy Fan, Maoyong; Ball State University, Zhou, Maigeng; CDC,
Keywords: Mortality, Particulate air pollution, Generalized linear model, Heterogeneous effects of air pollution
https://mc.manuscriptcentral.com/bmj
BMJ
Confidential: For Review O
nly
i
Particulate Air Pollution and Mortality in China: A Time-Series
Analysis in the Largest 38 Chinese Cities
Authors:
Peng Yina,*
, Guojun Heb,*
, Maoyong Fanc,*
, Kowk Yan Chiud, Maorong Fan
e, Chang
Liuf, An Xue
g, Tong Liu
d, Yuhang Pan
d, Quan Mu
h, Maigeng Zhou
a,§
Affiliations: a National Center for Chronic and Non-communicable Disease Control and
Prevention, Chinese Center for Disease Control and Prevention, Beijing, China b Division of Social Science, Division of Environment, and Economics Department,
The Hong Kong University of Science and Technology, HK c Department of Economics, Ball State University, Muncie, IN, USA
d The Hong Kong University of Science and Technology, HK
e Xiyuan Hospital, China Academy of Chinese Medical Sciences, Beijing, China
f Scheller College of Business, Georgia Institute of Technology, GA, USA
g Department of Environmental Engineering, Beijing University, Beijing, China
h The Nature Conservancy, Beijing, China
* These authors contribute equally to this manuscript.
§ Corresponding author: Maigeng Zhou National Center for Chronic and Non-
communicable Disease Control and Prevention, Chinese Center for Disease Control
and Prevention, Nanwei Road, Xicheng District, Beijing, 100050, China; Email:
“The Corresponding Authors have the right to grant on behalf of all authors and do
grant on behalf of all authors, a worldwide licence to the Publishers and its licensees
in perpetuity, in all forms, formats and media (whether known now or created in the
future), to i) publish, reproduce, distribute, display and store the Contribution, ii)
translate the Contribution into other languages, create adaptations, reprints, include
within collections and create summaries, extracts and/or, abstracts of the
Contribution, iii) create any other derivative work(s) based on the Contribution, iv) to
exploit all subsidiary rights in the Contribution, v) the inclusion of electronic links
from the Contribution to third party material where-ever it may be located; and, vi)
licence any third party to do any or all of the above.”
Page 1 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
ii
ABSTRACT
Objectives
To estimate the overall effect of particulate air pollution (particulate matter with
aerodynamic diameter <10 µm, or PM10) on mortality and explore the heterogeneity
of air pollution effects in major cities in China.
Design
Generalized linear models with different lag structures using time series data.
Setting
Thirty-eight large cities in 27 provinces of China. These cities have a combined
population of more than 200 million people.
Participants
350,638 deaths (200,912male, 149,726 female) recorded in 38 city districts by the
Disease Surveillance Point System (DSPS) of the Chinese Center for Disease Control
and Prevention (CDC) from January 1st, 2010 through June 29
th, 2013.
Main outcome measure
Daily numbers of deaths from all causes, cardiovascular and respiratory (CVR)
diseases, and non-CVR diseases and among different demographic groups were used
to estimate the associations between particulate air pollutantion and mortality.
Results
A 10-��/�� change in concurrent day PM10 concentrations was associated with 0.44
percent (95 percent CI: 0.30 to 0.58) change in the daily number of deaths. Previous
day and two-day lagged PM10 levels were also statistically significantly associated
with increased mortality at lower magnitudes. The estimate for the effect of PM10 on
CVR deaths was 0.62 percent (95 percent CI: 0.43 to 0.81) per 10 ��/��, compared
to 0.26 percent (95 percent CI: 0.09 to 0.42) for other-cause mortality. Exposure to
PM10 had similar impacts on both males and females. The elderly (≥ 60 years) were
more vulnerable to particulate air pollution than young people at high levels of air
pollution. The PM10 effect varies across cities and decreases as pollution levels rise.
Conclusion
Air pollution has a greater impact on CVR mortality than it does on other-cause
mortality. Older people have a higher risk of death from air pollution than younger
people. The magnitude of the effect varies across cities and decreases as PM10 levels
increase.
Page 2 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
iii
For Print Publication
ABSTRACT
Study question
To estimate the overall effect of particulate air pollution (particulate matter with
aerodynamic diameter <10 µm, or PM10) on mortality and explore the heterogeneity
of air pollution effects in major cities in China.
Methods
Generalized linear models with different lag structures were estimated using time
series data from thirty-eight large cities in 27 provinces of China. There were in total
350,638 deaths (200,912male, 149,726 female) in 38 city districts recorded by the
Disease Surveillance Point System (DSPS) of China CDC from January 1st, 2010
through June 29th
, 2013. Outcome measures are daily numbers of deaths from all
causes, cardiovascular and respiratory (CVR) diseases, and non-CVR diseases. A 10-
��/�� change in concurrent day PM10 concentration was associated with 0.44
percent (95 percent CI: 0.30 to 0.58) change in the daily number of deaths. Previous
day and two-day lagged PM10 levels were also statistically significantly associated
with increased mortality at lower magnitudes. The effects were greater for CVR
mortality than non-CVR mortality. The elderly (≥ 60 years) were more vulnerable to
air pollution than young people. The PM10 effect decreases as pollution levels rise.
Study answer and limitations
The air pollution effects cover a wide range and vary across cities, causes of diseases
and age groups. Other pollutants such as NO2, SO2 and O3 and indoor air pollution
were unexamined due to data limitation.
What this study adds
Acute air pollution effects are city-specific, affected by many local factors, and cannot
be generalized to all cities.
Funding, competing interests, data sharing
The study was funded by the SBI Research Grant from the HKUST and China
National Science and Technology Pillar Program 2013. The authors have no conflicts
of interests to declare. The mortality data can be applied through:
http://www.phsciencedata.cn/Share/edtShare.jsp. Other data are available upon
request.
Page 3 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
iv
Summary Box
What is already known on this topic
Many studies have shown a positive association between daily mortality and
particulate air pollution.
The air pollution effects in developed countries cannot be directly transferred to
China because of differences in air pollution levels, particle compositions, and
population characteristics.
Despite immense interest in the effect of air pollution on a national scale in China,
multi-city analysis is very limited.
What this study adds
Acute air pollution effects are city-specific, affected by many local factors, and
cannot be generalized to all cities.
The air pollution effects are smaller in more polluted cities and are more
homogenous in northern cities than in southern cities in China.
The effect of air pollution is only weakly associated with GDP per capita (the
association is positive but not statistically significant).
Page 4 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
1
Introduction
Air pollution and its negative health consequences are a major public concern
in China nowadays.1-3
According to the Global Burden of Disease Study, a loss of 25
million healthy years and more than 1.2 million premature deaths in China can be
attributed to the outdoor air pollution in 2010.4 OECD (2012) estimated that, world-
wide, up to 3.6 million people annually could die prematurely from air pollution each
year by 2050, with most of the deaths in China and India.5
Many time-series studies conducted in Chinese cities have consistently found
that temporarily elevated air pollution levels were associated with increased
mortality.6 7
A limiting factor in these studies is that the data were from a single city,
or a few cities. Additionally, many of most previous studies focused on heavily
polluted cities where pollution levels were several orders of magnitude greater than
those in cleaner cities. Therefore, these studies might be inadequate for setting
optimal environmental and public health policies at the national level. If air pollution
effects are greater in the most polluted cities and the government policy sets national
pollution control policies based on health risks estimated from heavily polluted cities,
then it might over-regulate pollution which would hinder local economic growth in
cleaner cities. Alternatively, if air pollution effects are smaller in more polluted cities,
then the government may under-assess the health risks and under regulate.
Much of the multi-city analysis has been focused on developed countries
where both health and pollution data are readily available.8-10
Estimates from these
studies have had profound implications for designing environmental regulations to
protect public health in the western world, however, they cannot be directly
transferred to China because of differences in air pollution levels, particle
compositions, and population characteristics. For example, the peak daily PM10
Page 5 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
2
(particulate matter with aerodynamic diameter < 10µm) concentrations in China often
goes beyond 600 ��/�� whereas the typical range of daily PM10 concentrations in
Europe and the United States is from 20 to 80 ��/��.
Due to the lack of multi-city studies in China, researchers have resorted to
meta-analysis to combine estimates from different studies.6 7 11-13
It is noteworthy that
meta-analyses based on published papers often suffer from publication bias and its
validity has been questioned due to incomparability across different studies. First,
“positive” results are more likely to be published than “negative” results, leading to
the censoring of studies with non-significant results.14
Second, heterogeneity in study
samples (e.g. the difference in age structure, ICD codes, and study periods) leads to
different interpretations in the coefficients of differing studies. Third, the differences
in study designs and statistical methods would affect the estimated sizes of air
pollution effects, which make comparisons across different studies difficult or even
impossible.
In this study, we assembled the most up-to-date and most comprehensive daily
mortality and particulate matter pollution data for China. With these data, we then
estimated the associations between PM10 and mortality in the Chinese population.
Daily mortality data and PM10 in China’s largest 38 cities are from Jan 1st, 2010 to
June 29th
, 2013. We used flexible modeling strategies to estimate the relationship
between PM10 and mortality and controlled for potential confounding factors, such as
temperature, dew point, day of week, holiday, and year effects. We estimated a
generalized linear model using daily time-series data for each city. Under this
research design, we were able to: (1) provide city-specific estimates of air pollution
effect and compare the estimates across cities with different levels of PM10
concentrations; and (2) estimate the national air pollution effects using the random
Page 6 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
3
effects meta-analysis and explore the heterogeneity of air pollution effects.
Methods
Study Area
We collected daily data on particulate matter air pollution, mortality, and
weather condition for the largest 38 cities most populated in China. Figure 1 shows
the geographical locations of these cities. They cover 27 provinces and encompass
much of China’s geography. Their combined population totals more than 200 million
people.
Measurement of Air Pollution
Air quality data were collected from the Ministry of Environmental Protection
(MEP). The MEP reported the daily air pollution index (API) and the primary
pollutant in all the major Chinese cities during the study period. The API is based on
the concentrations of three major air pollutants (PM10, SO2, and NO2) and provides an
overall measure of ambient air quality for each city. The higher the API, the worse is
air quality. The method used by the MEP to construct the API allows us to calculate
the daily concentration of PM10. The methods employed for estimating concentrations
of PM10 are available in the Supplementary Material 1 in the Appendix.
Daily Mortality
Daily mortality data are provided by the Disease Surveillance Point System
(DSPS) of the Chinese Center for Disease Control and Prevention (CDC). The DSPS
collects mortality data from certain city districts in each city it covers. Mortality data
include basic demographic characteristics (e.g. sex and age) of the decedent and the
Page 7 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
4
cause of death. The causes of death are coded according to the International
Classification of Diseases 10 (ICD-10). Mortality data are classified by causes of
death: cardiovascular (I00-99) and respiratory (J00-99) diseases, and all other
diseases. For this study we had daily numbers of deaths by age group, gender, and
cause of death for all the DSPS districts in the 38 cities. The data are from January 1st,
2010 through June 29th
, 2013. We discuss the details of the DSPS in the
Supplementary Material 2 in the Appendix.
Measurement of Weather Conditions
We obtained daily weather information from local weather stations from Jan 1,
2010 through June 29th
, 2013, for all 38 cities in the study.
Model
We estimated the associations between PM10 and daily mortality using a set of
generalized linear models (GLMs). For each city, we estimated the following equation
using daily time-series data:
��~��� ����
log���� = � + ������� + ∑ ��� �� , "#$��%�&� + '()� + *+�",�� + -.,/� + #�0�
(1)
where �� is the number of deaths on day 0; �� is assumed to originate from a Poisson
distribution with 12��3 = �� and canonical log-link in the regression. ������ is the
PM10 concentration on day 0 − +, and + is a day lag. � represents the log-relative rate
of mortality associated with air pollution. �� is a set of meteorological factors that are
correlated with air pollution levels, which include temperature and dew point. Also
included are several sets of dummy variables representing different time effects. They
Page 8 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
5
are dummy variables for day of the week ('()�), holidays (*+�",��), and year
(-.,/�). Dummy variables for day of the week capture weekly mortality patters.
Holiday dummy variables were used to capture the effect of holiday related conditions
on mortality such as unusually heavy traffics. Year dummy variables control for
potential discontinuous change in mortality levels in different years due to yearly
changes in policies. We also included a cubic function of time #�0� to control for the
long-term trends.
Meteorological factors � were modeled in the regressions through a set of
natural spline functions ��. The spline functions allow very flexible relationships
between meteorological factors and the outcomes. We chose the degrees of freedom
for each meteorological factor ("#$�) based on its best prediction for air pollution
levels. Using degrees of freedom that best predict air pollution levels is advantageous
because they produce unbiased or asymptotically unbiased estimates of the pollution
log-relative risk.15
The optimal degree of freedom for each natural spline was
obtained via generalized cross-validation method that best predicts PM10
concentrations.16
After controlling for these potential confounding factors, the high
frequency PM10 concentration data should provide a plausible source of exogenous
variation.
Since air pollution may affect mortality in a lagged fashion, we therefore
examined the air pollution effects separately for different lag structures (+ =
0, 1, 2, …). We also explored heterogeneous air pollution effects by examining
different age groups and different genders.
To estimate national air pollution effects, we conducted a heterogeneity test
and used random-effects meta-analysis to synthesize city-specific air pollution effect
estimates. Compared with the fixed effects model, the random-effects approach
Page 9 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
6
allows the true air pollution effects �9 to vary between studies. Such heterogeneity in
treatment effects may be caused by differences in study populations, local
socioeconomic conditions, and baseline population health. In the random-effects
meta-analysis, the air pollution effect is assumed to have a normal distribution around
a mean effect.
Finally, we investigated the patterns of heterogeneous air pollution effects
across cities using a set of linear regressions. The explanatory variables are mean
PM10 concentrations, a geographic (north) indicator, GDP per capita, and share of
workers in construction industry.
Patient involvement
No patients were involved in setting the research question or the outcome
measures, nor were they involved in developing plans for recruitment, design, or
implementation of the study. No patients were asked to advise on interpretation or
writing up of results. There are no plans to disseminate the results of the research
directly to study participants or any specific patient community.
Results
Descriptive Statistics
Table 1 summarizes the descriptive statistics for each city in our sample for
daily mean PM10 concentrations, daily number of all-cause deaths, and cardiovascular
and respiratory (CVR) deaths. Over the sample period (Jan 1st, 2010 to June 29
th,
2013), the daily mean of PM10 concentrations across all locations was 92.9 ��/��,
with a standard deviation of 46.3. The most polluted city in our sample was Urumqi in
Xinjiang Province with an average daily mean PM10 concentrations of 136 ��/��.
The least polluted city was Qinhuangdao in Hebei Province with an average daily
mean PM10 concentrations of 66.9 ��/��. The lowest daily PM10 concentration
Page 10 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
7
observed in our sample was 11 ��/��, while the highest was above 600��/��. On
average, there were 8.6 deaths per day in the sampled city districts, including 4.4 from
CVR diseases. The daily mean temperature and dew point across all cities were
56.3°F and 42.9°F, respectively. More information about these DSPS cities is
provided in Appendix Table SM2.
Mortality and PM10 Associations
We estimated the air pollution effects of concurrent day and lagged (up to 6
days) PM10 concentrations on all-cause deaths by fitting model (1) to the time series
data for each city independently. Figure 2 plots city-specific estimates and their 95
percent confidence intervals (CI) for concurrent day PM10. The estimates were
positive and statistically significant at the 5-percent level for 11 cities. None of the
estimates in the 38 cities was negative and statistically significant.
Although the majority of estimates were positive, the heterogeneity across the
cities was obvious. We performed a chi-squared test for heterogeneity on all 38 cities.
The chi-squared statistics was 91.6 (p-value=0) indicating considerable between-city
heterogeneity in the effect of PM10. The I-squared statistics showed that 59 percent of
between-city heterogeneity was attributable to variability in the true treatment effect,
rather than sampling variation. Since we used the same model and data period for all
cities, heterogeneity was smaller than that in meta-analysis using different studies.
When we rejected the hypothesis of homogeneity, we took into account the identified
between-cities variation and fitted a DerSimonian-Laird random effects model.17
The
combined estimate of the PM10 effects is at the bottom of figure 2. Overall, we found
that a 10-��/�� increase in PM10 concentration was associated with a 0.44 percent
(95 percent CI: 0.30 to 0.58) increase in total mortality.
Page 11 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
8
For lagged PM10, the estimates for individual cities and combined effects
forest plots are presented in supplementary figures SM1-SM6. The estimates for
individual cities became smaller in magnitude and the heterogeneity across cities was
decreasing as the lag time became longer. Similarly, the combined effects converged
to zero as the lag time became longer. For example, for lag 1 day PM10 pollution, the
combined random effects estimate was 0.26 percent excess mortality (95 percent CI:
0.15 to 0.37) per 10 ��/�� increase in PM10. The overall effect decreased to 0.13
percent (95 percent CI: 0.03 to 0.23) per 10 ��/�� increase in lag 2 days PM10
pollution. The combined effects became statistically insignificant and close to zero for
PM10 lagged more than 2 days. Given the aforementioned results, we also estimated
the air pollution effect using the 3-day moving average (lags 0, 1 and 2).
Supplementary figure SM7 in the Appendix plots the estimates and their 95 percent
confidence intervals. The combined random effects estimate was 0.45 percent excess
mortality (95 percent CI: 0.28 to 0.62) per 10 ��/�� increase in PM10. These
estimates are similar to those using concurrent day PM10.
Heterogeneity by Cause of Death, Gender, and Age Group
We plots the PM10 effects for CVR mortality in figure 3 and non-CVR
mortality in figure 4. We used concurrent day PM10 concentrations for disease
specific analysis and subgroup analysis because the analysis for overall mortality
showed that the estimates for concurrent day pollution were similar to those for 3-day
moving average. For both CVR and non-CVR diseases, the pollution effect showed
significant heterogeneity across cities. For most cities, the PM10 effect for CVR
mortality was more likely to be positive and greater than that for non-CVR mortality.
The combined effect for CVR mortality was more than double that for non-CVR
Page 12 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
9
mortality. For CVR mortality, the combined effect for all cities was: a 10-��/��
change in concurrent day PM10 concentration was associated with 0.62 percent (95
percent CI: 0.43 to 0.81) change in daily number of deaths. The combined estimate on
the effects of PM10 on non-CVR deaths was less than half (0.26 percent per 10
��/�� PM10; 95 percent CI: 0.09 to 0.42).
We then explored the heterogeneous effects of PM10 by examining subgroups
by gender and age. Supplementary figure SM9 plots city-specific estimates and 95
percent confidence intervals for males and females separately. The majority of all
estimates were positive for both males and females. The combined effects for males
were 0.39 percent (95 percent CI: 0.23 to 0.54) and for females 0.51 percent (95
percent CI: 0.34 to 0.68) per 10 ��/��. Air pollution effects were slightly greater for
females than for males, however the difference was not statistically significant as the
95 percent CIs for both estimates substantially overlapped.
Supplementary figure SM10 plots city-specific estimates and 95 percent
confidence intervals for subgroups by age. We divided all deaths into two age groups.
For younger people (< 60 years of age), there were positive and statistically
significant association between PM10 concentrations and mortality only in two cities.
The combined effect for younger people was 0.19 percent per 10 ��/�� and
statistically insignificant (95 percent CI: -0.02 to 0.40). For older people (≥ 60 years
of age), the estimates were positive and statistically significant in thirteen cities. The
combined effect for older people was 0.50 percent per 10 ��/�� and statistically
significant at the 5-percent level (95 percent CI: 0.34 to 0.66). This indicates that
older people are more vulnerable to air pollution than younger people.
Air Pollution Effects and City-specific Characteristics
Page 13 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
10
We explored the relationship between the estimated air pollution effects and
some city-specific characteristics. First, we examined the relationship between
estimated air pollution effects and mean PM10 concentrations. Supplementary figure
SM11 plots this relationship; it shows that air pollution effects were smaller in more
polluted cities. We estimated this correlation using a linear function with the
dependent variable being the estimated air pollution effect in each city. These results
are presented in column 1 of Table 3. The coefficient of PM10 was -0.01 (p-
value=0.07) in the linear regression.
Northern China burns a lot of coal during the winter months to provide central
heating, and its weather patterns are significantly different from Southern China, so
we checked whether air pollution effects differed between the North and the South.
Our division of the North and South followed the Huai River line (typically the
demarcation used in the literature).18
We examine the regional effect-concentration
relationship by regressing pollution effects on mean PM10 concentrations for northern
and southern cities separately. Supplementary figure SM12 compares the relationship
for northern cities with that for southern cities. It shows that the marginal effects of
PM10 was almost constant at different pollution levels for northern cities. The
coefficient of mean PM10 concentrations was 0 (p-value=0.97). In sharp comparison,
the marginal effect of PM10 decreased rapidly as pollution levels increased in southern
cities. The coefficient of mean PM10 concentrations was -0.02 (p-value=0.08).
Finally, we examined the relationship between the size of estimated air
pollution effects and two socio-economic factors: GDP per capita and the share of
workers in the construction industry. Supplementary figure SM13 plots the estimated
effects against those two characteristics. The linear fits showed a weak negative
relationship between pollution effect and GDP per capita and a positive relationship
Page 14 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
11
between pollution effects and the share of workers in the construction industry.
Columns 2-4 of Table 3 presents the corresponding regression results for
supplementary figures SM12 and SM13. The last column of Table 3 shows that even
if all variables were included in the regression at the same time, the observed
relationships in supplementary figures SM12 and SM13 remained; this indicates that
the relationships are unlikely to be spurious.
Discussion
Acquiring internally coherent multi-city estimates of air pollution effects in
fast developing countries like China is valuable to cost-benefit analysis of pollution
abatement strategies and setting optimal air quality standards. Here we estimated the
associations between PM10 concentrations and daily mortality from all causes, CVR
diseases, and non-CVR diseases for the largest 38 cities in China. We employed a set
of flexible generalized linear models to obtain the estimates of air pollution effects.
The specification of the statistical model required a series of analytic choices
including: (1) the specification of lag structure of the air pollution variable, and (2)
how to adjust flexibly for weather conditions.
Principal Findings
The results showed positive associations between daily mortality and PM10
exposure in most sampled cities. Compared with one or more days lagged PM10
pollution, concurrent day PM10 pollution had the largest impact on mortality. For
example, at the national level a 10-��/�� increase in concurrent day PM10, one-day
lagged PM10, and two-day lagged PM10, was respectively associated with 0.44, 0.26,
and 0.13 percent increase in daily number of deaths. We failed to find similar positive
Page 15 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
12
and statistically significant effects for air pollution lagged longer than 2 days. This
suggests that particulate pollution has relatively short lagging acute effect on
mortality.
The results showed that PM10-mortality associations were substantially
heterogeneous across cities. Based on our analysis of the largest 38 cities, we found
that PM10 concentrations were not statistically significantly associated with mortality
for more than half of all cities. Moreover, the effect estimates covered a wide range
and included both negative and positive domains. For example, the estimates of
concurrent day pollution ranged from high of 1.80 percent (95 percent CI: 0.6 to 3)
per 10 ��/�� in Yuxi to low of -0.98 percent (95 percent CI: -2.45 to 0.48) per 10
��/�� PM10 in Panzhihua. These findings show the limitation of past studies that
focused on a particular or a few large cities and were often biased towards positive
results.14
The findings of our study suggest that acute air pollution effects are city-
specific, affected by many local factors, and cannot be generalized to all cities.
In a closer examination of heterogeneity in the PM10 effects on mortality, we
conducted subgroup analyses. We found that air pollution had a much greater impact
on CVR mortality than it did on non-CVR mortality, with the difference being
statistically significant at the 5-percent level. This is consistent with the literature:
people with CVR diseases are more sensitive to short-term air pollution deterioration
than those with non-CVR diseases. Also found was that PM10 concentrations had a
much greater impact on the older people than the younger people. For young people,
we found that the overall effect was not statistically indifferent from zero. In sharp
contrast, the overall effect for older people was 0.50 percent (95 percent CI: 0.34 to
0.66). Positive and statically significant associations between PM10 concentrations
and mortality were found for one third of the cities. This indicates that older people
Page 16 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
13
are more vulnerable to particulate air pollution than younger people at high levels of
air pollution. Our gender subgroup analysis shows that exposure to PM10 had a
slightly greater impact on females than on males but the difference was not
statistically significant.
Interestingly, overall magnitudes of PM10 effects found in our study are
comparable to those found in several large meta-analyses. For example, summarizing
thirty-three time-series and case-crossover studies, Shang and coauthors reported that
a 10 ��/�� increase in PM10 was associated with a 0.32 percent (95 percent CI: 0.28
to 0.35) increase in non-accidental mortality, a 0.44 percent (95 percent CI: 0.33 to
0.54) increase in mortality due to cardiovascular diseases, and a 0.32 percent (95
percent CI: 0.23 to 0.40) increase in mortality due to respiratory diseases.7
Explanation of Heterogeneity
The pattern of the associations between air pollution effects and baseline PM10
level and city-specific characteristics is enlightening. First, air pollution effects were
smaller in more polluted cities. This could be because of the “saturation” effect, in
which underlying biochemical and cellular processes became saturated when
exposed to a very high level of a toxic component.19 20
It is also possible that in
more polluted cities, people adopted more defensive measures, such as reducing
outdoor activities, wearing face masks or installing air filters. As a result,
despite of living in more polluted areas, people’s actual exposure may have
been mitigated by avoidance behaviors. Second, we found that the air pollution
effects were more homogenous in northern cities than in southern cities. This
result may also be related to avoidance behaviors. A recent study showed that
people living in the north are more likely to buy air filters than those in the
Page 17 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
14
south.21
Third, there was only weak evidence that the marginal PM10 effect decreases
as GDP per capita increases (the association was positive but not statistically
significant), indicating that cities with better economic conditions might have better
medical services, and therefore lower marginal effects. The fourth and final
interesting finding was that if a city had a larger share of workers hired in the
construction industry, then the air pollution effects were greater. Possible explanation
for these results may be that construction workers were more likely to be exposed to
air pollution, or, alternatively, that the construction industry generated more
particulates. Much of the above discussion is conjectural because the sample size is
limited in this analysis (38 data points). Future research is warranted on these issues.
Limitations of Study
There are several limitations in this study. First, we were unable to
examine the pollution effects of other pollutants, such as NO2, SO2 and O3 due
to data limitation. Quantification of the health effect of other air pollutants is
also important for setting appropriate air quality standards. Second, we only
focused on urban cities so the estimates of air pollution effects cannot be
generalized to rural areas. Air pollution, including in-door air pollution, might
have a greater impact on rural residents;3 more research focusing on rural areas
is warranted.
Page 18 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
15
Figure 1. Geographical Distribution of 38 cities in the sample
See separate file titled “Figure 1 38 cities”
Note: The figure plots the locations of the 38 cities in our sample. They cover 27 of
31 provinces in China.
Page 19 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
16
Figure 2. Percentage Change in Daily Number of All-Cause Deaths per 10-
��/�� Increase in Concurrent Day PM10 in 38 Chinese Cities
See separate file titled “Figure 2”
PM10 Effect (10 ��/��, Lag=0)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-
specific mean PM10 effects on total mortality, the largest 38 cities in China. The
dependent variable is the percentage change in the number of daily all-cause deaths.
X-axis is percentage change. Each solid square represents an effect size. Horizontal
lines indicate 95 percent CIs. ES=Effect Size.
Page 20 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
17
Figure 3. Percentage Change in Daily Number of Deaths from Cardiovascular
and Respiratory Diseases per 10-��/�� Increase in Concurrent Day PM10 in 38
Chinese Cities
See separate file titled “Figure 3”
PM10 Effect (10 ��/��, Lag=0)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-
specific mean lag PM10 effects on total mortality for cardiorespiratory and non-
cardiorespiratory deaths separately, the largest 38 cities in China. The dependent
variable is the percentage change in the number of daily deaths for cardiorespiratory
and non-cardiorespiratory diseases respectively.
Page 21 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
18
Figure 4. Percentage Change in Daily Number of Deaths from Non-
Cardiovascular and Respiratory Diseases per 10-��/�� Increase in Concurrent
Day PM10 in 38 Chinese Cities
See separate file titled “Figure 4”
PM10 Effect (10 ��/��, Lag=0)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-
specific mean lag PM10 effects on total mortality for cardiorespiratory and non-
cardiorespiratory deaths separately, the largest 38 cities in China. The dependent
variable is the percentage change in the number of daily deaths for cardiorespiratory
and non-cardiorespiratory diseases respectively.
Page 22 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
19
Table 1. Summary Statistics
PM10 (��/��) Daily All-Cause Deaths Daily CVR Deaths
City Mean Std. Dev. IQR Mean Std. Dev. Mean Std. Dev.
Urumqi 136 74.2 64 5.2 2.4 2.6 1.7
Beijing 113.4 71.5 84 21.8 5.4 12 4
Chengdou 109.5 57.2 70 7.1 2.9 3.4 2
Zaozhuang 108.9 53 74 2.4 1.6 1.3 1.2
Zhengzhou 108.5 51.1 58 6.3 3.5 3.2 2.3
Xining 108.3 55.8 66 10 4.8 4 2.7
Nanjing 104.4 53.4 64 9.1 3.3 5 2.6
Anshan 102.7 45.9 54 11.9 3.9 7.3 2.9
Wuhan 101.8 54.3 72 12.2 4.1 6.4 3
Tianjin 101.4 53.9 58 3.7 2.6 2.1 1.7
Tongchuan 100.9 44.6 50 3 1.9 1.6 1.4
Shenyang 100.7 49.6 58 19.4 5.4 10.3 3.6
Harbin 100.2 47.7 58 6.5 2.8 3.7 2
Yinchuan 98.3 49.6 50 4.8 2.6 2.8 2
Panzhihua 97.8 31.1 42 5.5 2.5 2.7 1.8
Maanshan 97.3 37.9 50 3.8 2 1.7 1.4
Xuzhou 97 49.7 52 21.8 5.4 12 4
Chongqing 96.5 49.4 62 3.3 2 1.8 1.4
Hangzhou 92.5 47.1 60 6.5 3 2.9 1.9
Yichang 92.3 42.3 48 6 2.8 2.5 1.7
Taiyuan 92.2 53.4 68 6.5 3.3 3.2 2.1
Changde 92.1 43.7 56 2.5 1.6 1.3 1.1
Changchun 89.9 45.6 48 7.6 2.8 4.2 2.1
Qingdao 89.8 49.3 58 21.8 5.4 12 4
Nanchang 89.8 42 56 3.4 2.2 2.1 1.8
Tangshan 88.7 51.9 52 10.6 3.6 5.1 2.7
Changsha 86 44.4 56 6 2.6 3.1 1.9
Suzhou 84.9 45.7 50 7.3 3.3 3.5 2.2
Zunyi 83.6 30.4 40 7.9 3.3 4.5 2.4
Hohhot 82.2 44.9 59 4 2.7 2.1 1.7
Liuzhou 80 35 42 5 2.5 2 1.5
Qiqihar 75.9 37.2 42 21.8 5.4 12 4
Shanghai 75.1 47.6 52 2.8 1.8 1.7 1.4
Guilin 72.4 38.7 51 1.8 1.4 0.8 0.9
Yantai 71.4 38.9 44 9.5 3.2 4.6 2.2
Yuxi 70.4 24.7 32 7.1 3.1 3.7 2.1
Guangzhou 69.3 31.9 43 19.2 5.6 9.1 3.7
Qinhuangdao 66.9 35.2 32 5.7 2.6 2.7 1.6
All Cities 92.9 46.3 58 8.6 6.9 4.4 4.1
Page 23 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
20
Note: PM10 concentrations are calculated using API reported by the Ministry of
Environmental Protection in China. Mortality Data come from the Chinese Center for Disease
Control and Prevention. See details in the Supplementary Material 1 in the Appendix.
Page 24 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
21
Table 2. Weather Conditions
Variable Mean Std. Dev. Min Max
Temperature (°F) 56.3 21.3 -21.7 96.9
Dew Point (°F) 42.9 23.0 -29.3 83.4
Note: Weather data are from Global Historical Climatology Network. Link:
https://www.ncdc.noaa.gov/data-access/land-based-station-data/land-based-
datasets/global-historical-climatology-network-ghcn.
Page 25 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only22
Table 3. Relationships between Air Pollution Effect and City-specific Factors
(1) (2) (3) (4) (5) (6)
Mean PM10 -0.010*
-0.009*
(0.071) (0.100)
Mean PM10×North -0.000
(0.971)
Mean PM10×South -0.022*
(0.079)
GDP Per Capita -0.034 -0.048
(0.542) (0.308)
Share of Workers in Construction Industry
6.247** 7.491***
(0.025) (0.002)
North Indicator
-0.371**
(0.025)
Constant 1.394** 2.654** 0.671*** 0.542** 0.108 1.213**
(0.018) (0.025) (0.000) (0.036) (0.547) (0.028)
Observations 38 18 38 38 38 38
R-squared 0.075 0.158 0.158 0.008 0.119 0.342
Note: We regress the estimated air pollution effects in Figure 2 on mean PM10 concentrations, a north indicator, GDP per capita (in 10,000
Yuan), and share of workers in construction industry. GDP per capita is collected from city statistical yearbooks. Share of workers in
construction industry is calculated using 2005 micro census data. P-values are in parenthesis. *, **, and *** indicate statistical significance at
10%, 5%, and 1% respectively.
Page 26 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
23
We thank Philip Coelho, Ball State University, for providing comments on the
manuscript.
Contributors: MYF, GH, PY and MZ designed the study. PY and MZ collected and
cleaned the mortality data. MYF, KYC, and GH conducted the analyses. MRF, AX,
and CL reviewed the literature, conducted GIS matching, and contributed to
interpretation of the results. TL, YP, and QM cleaned the pollution data, collected
socioeconomic data, and summarized the results. MYF and GH finished the first draft.
All authors commented on this draft and contributed to the final version. All authors
had full access to all of the data (including statistical reports and tables) in the study
and can take responsibility for the integrity of the data and the accuracy of the data
analysis. MYF, GH, PY and MZ are study guarantors.
Funding: The study was financially supported by the SBI Research Grant
(SBI15HS06) from the Hong Kong University of Science and Technology and China
National Science and Technology Pillar Program 2013 (2013BAI04B02). The funders
were not involved in the research and preparation of the article, including study
design; collection, analysis, and interpretation of data; writing of the article; nor in the
decision to submit it for publication.
Competing interests: All authors have completed the ICMJE uniform disclosure form
at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding
author) and declare: the study was financially supported by the SBI Research Grant
from the Hong Kong University of Science and Technology and China National
Science and Technology Pillar Program 2013; no financial relationships with any
organizations that might have an interest in the submitted work; no other relationships
or activities that could appear to have influenced the submitted work.
Ethical approval: Not required.
Data sharing: The pollution data and weather data are all available from Maoyong Fan
at [email protected]. The mortality data can only be applied through a government data
sharing portal: http://www.phsciencedata.cn/Share/edtShare.jsp.
Page 27 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
24
References
1. Ouyang Y. China wakes up to the crisis of air pollution. The Lancet Respiratory
Medicine 2013;1(1):12.
2. Chan CK, Yao X. Air pollution in mega cities in China. Atmos Environ
2008;42(1):1-42. doi: http://dx.doi.org/10.1016/j.atmosenv.2007.09.003
3. Zhou M, He G, Fan M, et al. Smog episodes, fine particulate pollution and
mortality in China. Environ Res 2015;136(0):396-404. doi:
http://dx.doi.org/10.1016/j.envres.2014.09.038
4. Kassebaum NJ, Bertozzi-Villa A, Coggeshall MS, et al. Global, regional, and
national levels and causes of maternal mortality during 1990–2013: a
systematic analysis for the Global Burden of Disease Study 2013. The
Lancet 2014;384(9947):980-1004.
5. Marchal V, Dellink R, Van Vuuren D, et al. OECD environmental outlook to
2050: Organization for Economic Co-operation and Development, 2011.
6. Lu F, Xu D, Cheng Y, et al. Systematic review and meta-analysis of the adverse
health effects of ambient PM2.5 and PM10 pollution in the Chinese
population. Environ Res 2015;136:196-204. doi:
http://dx.doi.org/10.1016/j.envres.2014.06.029
7. Shang Y, Sun Z, Cao J, et al. Systematic review of Chinese studies of short-term
exposure to air pollution and daily mortality. Environment International
2013;54:100-11. doi: http://dx.doi.org/10.1016/j.envint.2013.01.010
8. Samet JM, Dominici F, Curriero FC, et al. Fine Particulate Air Pollution and
Mortality in 20 U.S. Cities, 1987–1994. N Engl J Med 2000;343(24):1742-
49. doi: doi:10.1056/NEJM200012143432401
9. Dominici F, Samet JM, Zeger SL. Combining evidence on air pollution and daily
mortality from the 20 largest US cities: a hierarchical modelling strategy.
Journal of the Royal Statistical Society: Series A (Statistics in Society)
2000;163(3):263-302. doi: 10.1111/1467-985X.00170
10. Katsouyanni K, Touloumi G, Samoli E, et al. Confounding and effect
modification in the short-term effects of ambient particles on total
mortality: results from 29 European cities within the APHEA2 project.
Epidemiology 2001;12(5):521-31.
11. Aunan K, Pan X-C. Exposure-response functions for health effects of ambient
air pollution applicable for China – a meta-analysis. Sci Total Environ
2004;329(1–3):3-16. doi:
http://dx.doi.org/10.1016/j.scitotenv.2004.03.008
12. Kan H, Chen B, Chen C, et al. Establishment of exposure-response functions of
air particulate matter and adverse health outcomes in China and
worldwide. Biomed Environ Sci 2005;18(3):159-63.
13. Lai H-K, Tsang H, Wong C-M. Meta-analysis of adverse health effects due to
air pollution in Chinese populations. BMC Public Health 2013;13(360) doi:
10.1186/1471-2458-13-360
14. Walker E, Hernandez AV, Kattan MW. Meta-analysis: Its strengths and
limitations. Cleve Clin J Med 2008;75(6):431-39.
15. Dominici F, McDermott A, Hastie TJ. Improved semiparametric time series
models of air pollution and mortality. J Amer Statistical Assoc
2004;99(468):938-48.
16. Gu C. Smoothing spline ANOVA models: Springer 2013.
Page 28 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
25
17. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials
1986;7(3):177-88. doi: http://dx.doi.org/10.1016/0197-2456(86)90046-
2
18. Chen Y, Ebenstein A, Greenstone M, et al. Evidence on the impact of sustained
exposure to air pollution on life expectancy from China’s Huai River
policy. Proceedings of the National Academy of Sciences
2013;110(32):12936-41. doi: 10.1073/pnas.1300018110
19. Ambrose JA, Barua RS. The pathophysiology of cigarette smoking and
cardiovascular diseaseAn update. J Am Coll Cardiol 2004;43(10):1731-37.
doi: 10.1016/j.jacc.2003.12.047
20. Pope CA, Burnett RT, Krewski D, et al. Cardiovascular Mortality and Exposure
to Airborne Fine Particulate Matter and Cigarette Smoke: Shape of the
Exposure-Response Relationship. Circulation 2009;120(11):941-48. doi:
10.1161/circulationaha.109.857888
21. Ito K, Zhang S. Willingness to Pay for Clean Air: Evidence from Air Purifier
Markets in China. NBER Working Paper No. 22367, 2016.
Page 29 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Page 30 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
NOTE: Weights are from random effects analysis
Overall (I-squared = 59.1%, p = 0.000)
Nanchang
Shanghai
Beijing
Wuhan
QiqiharYinchuan
City
Changchun
Hohhot
TianjinChangsha
Shenyang
Tangshan
Changde
Xuzhou
Hangzhou
Yichang
Yuxi
Anshan
Zaozhuang
UrumqiXining
Taiyuan
Harbin
Chengdu
Tongchuan
Nanjing
Qinhuangdao
Suzhou
Guangzhou
Qingdao
Panzhihua
Zunyi
Zhengzhou
Yantai
Guilin
Maanshan
Chongqing
Liuzhou
0.44 (0.30 to 0.58)
0.22 (-0.50 to 0.94)
0.31 (0.00 to 0.61)
0.29 (0.10 to 0.48)
0.77 (0.40 to 1.15)
-0.24 (-1.21 to 0.73)-0.43 (-1.07 to 0.21)
ES (95% CI)
0.64 (0.14 to 1.14)
0.17 (-0.72 to 1.05)
0.54 (0.19 to 0.90)0.54 (-0.05 to 1.13)
0.58 (0.03 to 1.13)
-0.16 (-1.15 to 0.83)
0.71 (0.10 to 1.32)
-0.64 (-1.49 to 0.21)
0.50 (-0.09 to 1.09)
0.47 (-0.60 to 1.54)
1.80 (0.60 to 3.00)
0.12 (-0.53 to 0.77)
1.09 (0.36 to 1.82)
0.03 (-0.43 to 0.49)0.02 (-0.88 to 0.93)
0.76 (0.24 to 1.29)
0.10 (-0.28 to 0.48)
0.29 (-0.20 to 0.79)
0.20 (-1.07 to 1.46)
0.59 (0.16 to 1.02)
0.37 (-0.49 to 1.22)
0.44 (-0.00 to 0.89)
1.65 (1.17 to 2.13)
0.20 (-0.11 to 0.50)
-0.98 (-2.45 to 0.48)
0.45 (-0.46 to 1.37)
0.21 (-0.26 to 0.67)
0.25 (-0.30 to 0.81)
0.42 (-0.76 to 1.60)
1.72 (0.72 to 2.72)
0.48 (0.19 to 0.78)
1.69 (0.88 to 2.50)
100.00
2.23
4.29
4.90
3.90
1.502.53
Weight
3.21
1.71
3.982.76
2.93
1.46
2.68
1.81
2.76
1.30
1.09
2.49
2.20
3.431.66
3.09
3.87
3.22
1.00
3.58
1.79
3.48
3.28
4.30
0.78
1.63
3.39
2.92
1.12
1.43
4.37
1.92
%
0.44 (0.30 to 0.58)
0.22 (-0.50 to 0.94)
0.31 (0.00 to 0.61)
0.29 (0.10 to 0.48)
0.77 (0.40 to 1.15)
-0.24 (-1.21 to 0.73)-0.43 (-1.07 to 0.21)
ES (95% CI)
0.64 (0.14 to 1.14)
0.17 (-0.72 to 1.05)
0.54 (0.19 to 0.90)0.54 (-0.05 to 1.13)
0.58 (0.03 to 1.13)
-0.16 (-1.15 to 0.83)
0.71 (0.10 to 1.32)
-0.64 (-1.49 to 0.21)
0.50 (-0.09 to 1.09)
0.47 (-0.60 to 1.54)
1.80 (0.60 to 3.00)
0.12 (-0.53 to 0.77)
1.09 (0.36 to 1.82)
0.03 (-0.43 to 0.49)0.02 (-0.88 to 0.93)
0.76 (0.24 to 1.29)
0.10 (-0.28 to 0.48)
0.29 (-0.20 to 0.79)
0.20 (-1.07 to 1.46)
0.59 (0.16 to 1.02)
0.37 (-0.49 to 1.22)
0.44 (-0.00 to 0.89)
1.65 (1.17 to 2.13)
0.20 (-0.11 to 0.50)
-0.98 (-2.45 to 0.48)
0.45 (-0.46 to 1.37)
0.21 (-0.26 to 0.67)
0.25 (-0.30 to 0.81)
0.42 (-0.76 to 1.60)
1.72 (0.72 to 2.72)
0.48 (0.19 to 0.78)
1.69 (0.88 to 2.50)
100.00
2.23
4.29
4.90
3.90
1.502.53
Weight
3.21
1.71
3.982.76
2.93
1.46
2.68
1.81
2.76
1.30
1.09
2.49
2.20
3.431.66
3.09
3.87
3.22
1.00
3.58
1.79
3.48
3.28
4.30
0.78
1.63
3.39
2.92
1.12
1.43
4.37
1.92
%
0-4 -2 0 2 4
Page 31 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
NOTE: Weights are from random effects analysis
Overall (I-squared = 59.7%, p = 0.000)
ChengduZunyi
BeijingTianjin
Hangzhou
Liuzhou
Yinchuan
Shanghai
Changde
Shenyang
Wuhan
Zaozhuang
Urumqi
TangshanXuzhouPanzhihua
Guangzhou
Maanshan
Yantai
Suzhou
Xining
Yichang
Nanchang
Qingdao
Qinhuangdao
Nanjing
Yuxi
City
Tongchuan
Hohhot
Chongqing
Guilin
Changsha
Zhengzhou
Qiqihar
Changchun
Harbin
Taiyuan
Anshan
0.62 (0.43 to 0.81)
0.38 (-0.31 to 1.07)0.35 (-0.83 to 1.53)
0.54 (0.28 to 0.80)0.53 (0.09 to 0.97)
0.50 (-0.38 to 1.39)
2.35 (1.13 to 3.57)
-0.23 (-1.11 to 0.66)
0.28 (-0.14 to 0.69)
0.70 (-0.18 to 1.58)
0.68 (-0.01 to 1.38)
1.09 (0.58 to 1.60)
1.52 (0.59 to 2.45)
0.02 (-0.60 to 0.65)
-0.34 (-1.56 to 0.87)-1.09 (-2.28 to 0.10)-1.65 (-3.61 to 0.31)
2.07 (1.39 to 2.75)
2.38 (0.85 to 3.91)
0.90 (0.12 to 1.69)
1.02 (0.37 to 1.67)
0.06 (-1.20 to 1.32)
1.08 (-0.35 to 2.51)
0.51 (-0.47 to 1.49)
0.21 (-0.21 to 0.63)
0.19 (-0.97 to 1.36)
1.10 (0.53 to 1.67)
0.93 (-0.71 to 2.57)
ES (95% CI)
1.52 (-0.15 to 3.19)
-0.01 (-1.14 to 1.13)
0.62 (0.23 to 1.02)
1.98 (0.32 to 3.64)
1.38 (0.58 to 2.18)
0.70 (0.04 to 1.35)
-0.13 (-1.37 to 1.11)
0.20 (-0.49 to 0.90)
0.02 (-0.48 to 0.52)
1.05 (0.30 to 1.80)
0.31 (-0.57 to 1.19)
100.00
3.171.77
4.884.17
2.52
1.70
2.53
4.28
2.53
3.15
3.89
2.39
3.42
1.721.760.82
3.20
1.22
2.83
3.31
1.63
1.35
2.24
4.29
1.81
3.65
1.09
Weight
1.06
1.86
4.37
1.07
2.77
3.31
1.66
3.15
3.94
2.97
2.53
%
0.62 (0.43 to 0.81)
0.38 (-0.31 to 1.07)0.35 (-0.83 to 1.53)
0.54 (0.28 to 0.80)0.53 (0.09 to 0.97)
0.50 (-0.38 to 1.39)
2.35 (1.13 to 3.57)
-0.23 (-1.11 to 0.66)
0.28 (-0.14 to 0.69)
0.70 (-0.18 to 1.58)
0.68 (-0.01 to 1.38)
1.09 (0.58 to 1.60)
1.52 (0.59 to 2.45)
0.02 (-0.60 to 0.65)
-0.34 (-1.56 to 0.87)-1.09 (-2.28 to 0.10)-1.65 (-3.61 to 0.31)
2.07 (1.39 to 2.75)
2.38 (0.85 to 3.91)
0.90 (0.12 to 1.69)
1.02 (0.37 to 1.67)
0.06 (-1.20 to 1.32)
1.08 (-0.35 to 2.51)
0.51 (-0.47 to 1.49)
0.21 (-0.21 to 0.63)
0.19 (-0.97 to 1.36)
1.10 (0.53 to 1.67)
0.93 (-0.71 to 2.57)
ES (95% CI)
1.52 (-0.15 to 3.19)
-0.01 (-1.14 to 1.13)
0.62 (0.23 to 1.02)
1.98 (0.32 to 3.64)
1.38 (0.58 to 2.18)
0.70 (0.04 to 1.35)
-0.13 (-1.37 to 1.11)
0.20 (-0.49 to 0.90)
0.02 (-0.48 to 0.52)
1.05 (0.30 to 1.80)
0.31 (-0.57 to 1.19)
100.00
3.171.77
4.884.17
2.52
1.70
2.53
4.28
2.53
3.15
3.89
2.39
3.42
1.721.760.82
3.20
1.22
2.83
3.31
1.63
1.35
2.24
4.29
1.81
3.65
1.09
Weight
1.06
1.86
4.37
1.07
2.77
3.31
1.66
3.15
3.94
2.97
2.53
%
0-4 -2 0 2 4
Page 32 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
NOTE: Weights are from random effects analysis
Overall (I-squared = 38.4%, p = 0.010)
City
Tongchuan
Zunyi
Changchun
Urumqi
Yantai
Tangshan
Hangzhou
TaiyuanYichang
Shenyang
Wuhan
Xuzhou
Qinhuangdao
Qiqihar
Anshan
Chongqing
Hohhot
Suzhou
Liuzhou
Yinchuan
Chengdu
Harbin
Tianjin
Changsha
Yuxi
Nanchang
Guilin
Nanjing
Panzhihua
Qingdao
Zhengzhou
Xining
Shanghai
Beijing
Guangzhou
Changde
Maanshan
Zaozhuang
0.26 (0.09 to 0.42)
ES (95% CI)
-1.48 (-3.31 to 0.35)
0.58 (-0.77 to 1.94)
1.16 (0.43 to 1.89)
0.04 (-0.58 to 0.66)
-0.40 (-1.18 to 0.37)
0.21 (-1.34 to 1.75)
0.50 (-0.24 to 1.24)
0.48 (-0.21 to 1.18)-0.20 (-1.77 to 1.37)
0.43 (-0.44 to 1.31)
0.40 (-0.13 to 0.94)
-0.06 (-1.28 to 1.16)
0.53 (-0.64 to 1.70)
-0.41 (-1.92 to 1.10)
-0.11 (-1.10 to 0.87)
0.32 (-0.12 to 0.75)
0.36 (-0.86 to 1.58)
-0.12 (-0.73 to 0.50)
1.20 (0.12 to 2.28)
-0.64 (-1.58 to 0.31)
0.23 (-0.32 to 0.77)
0.19 (-0.36 to 0.73)
0.57 (0.03 to 1.11)
-0.39 (-1.25 to 0.48)
2.72 (1.09 to 4.35)
-0.07 (-1.05 to 0.91)
-1.03 (-2.64 to 0.58)
-0.06 (-0.68 to 0.57)
-0.25 (-2.43 to 1.93)
0.19 (-0.25 to 0.63)
-0.27 (-0.92 to 0.39)
-0.02 (-1.37 to 1.33)
0.35 (-0.10 to 0.79)
-0.03 (-0.31 to 0.25)
1.26 (0.62 to 1.90)
0.72 (-0.06 to 1.49)
1.17 (-0.16 to 2.50)
0.64 (-0.30 to 1.59)
100.00
%Weight
0.70
1.20
3.04
3.65
2.82
0.95
2.98
3.210.93
2.38
4.24
1.43
1.53
0.99
2.00
5.12
1.42
3.73
1.74
2.13
4.20
4.17
4.20
2.43
0.86
2.01
0.89
3.64
0.51
5.07
3.44
1.21
5.00
6.50
3.51
2.82
1.23
2.13
0.26 (0.09 to 0.42)
ES (95% CI)
-1.48 (-3.31 to 0.35)
0.58 (-0.77 to 1.94)
1.16 (0.43 to 1.89)
0.04 (-0.58 to 0.66)
-0.40 (-1.18 to 0.37)
0.21 (-1.34 to 1.75)
0.50 (-0.24 to 1.24)
0.48 (-0.21 to 1.18)-0.20 (-1.77 to 1.37)
0.43 (-0.44 to 1.31)
0.40 (-0.13 to 0.94)
-0.06 (-1.28 to 1.16)
0.53 (-0.64 to 1.70)
-0.41 (-1.92 to 1.10)
-0.11 (-1.10 to 0.87)
0.32 (-0.12 to 0.75)
0.36 (-0.86 to 1.58)
-0.12 (-0.73 to 0.50)
1.20 (0.12 to 2.28)
-0.64 (-1.58 to 0.31)
0.23 (-0.32 to 0.77)
0.19 (-0.36 to 0.73)
0.57 (0.03 to 1.11)
-0.39 (-1.25 to 0.48)
2.72 (1.09 to 4.35)
-0.07 (-1.05 to 0.91)
-1.03 (-2.64 to 0.58)
-0.06 (-0.68 to 0.57)
-0.25 (-2.43 to 1.93)
0.19 (-0.25 to 0.63)
-0.27 (-0.92 to 0.39)
-0.02 (-1.37 to 1.33)
0.35 (-0.10 to 0.79)
-0.03 (-0.31 to 0.25)
1.26 (0.62 to 1.90)
0.72 (-0.06 to 1.49)
1.17 (-0.16 to 2.50)
0.64 (-0.30 to 1.59)
100.00
%Weight
0.70
1.20
3.04
3.65
2.82
0.95
2.98
3.210.93
2.38
4.24
1.43
1.53
0.99
2.00
5.12
1.42
3.73
1.74
2.13
4.20
4.17
4.20
2.43
0.86
2.01
0.89
3.64
0.51
5.07
3.44
1.21
5.00
6.50
3.51
2.82
1.23
2.13
0-4 -2 0 2 4
Page 33 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
1
Supplementary Materials to
Particulate Air Pollution and Mortality in China: A Time-Series Analysis in the Largest 38 Chinese Cities
Authors: Peng Yina,*, Guojun Heb,*, Maoyong Fanc,*, Kowk Yan Chiud, Maorong Fane, Chang Liuf, An Xueg, Tong Liud, Yuhang Pand, Quan Muh, Maigeng Zhoua,§
Affiliations: a National Center for Chronic and Non-communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention, Beijing, China b Division of Social Science, Division of Environment, and Economics Department, The Hong Kong University of Science and Technology, HK c Department of Economics, Ball State University, Muncie, IN, USA d The Hong Kong University of Science and Technology, HK e Xiyuan Hospital, China Academy of Chinese Medical Sciences, Beijing, China f Scheller College of Business, Georgia Institute of Technology, GA, USA g Department of Environmental Engineering, Beijing University, Beijing, China h The Nature Conservancy, Beijing, China * These authors contribute equally to this manuscript. § Corresponding author: Maigeng Zhou National Center for Chronic and Non-communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention, Nanwei Road, Xicheng District, Beijing, 100050, China; Email: [email protected].
Page 34 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
2
Supplementary Material 1: PM10 calculation
The API is constructed based on the concentrations of 3 atmospheric pollutants, namely sulfur dioxide (𝑆𝑆𝑂𝑂2), nitrogen dioxide (𝑁𝑁𝑂𝑂2), and suspended particulates of 10 micrometers or less (𝑃𝑃𝑀𝑀10) measured at the monitoring stations throughout each city. It is a proxy measure of the ambient air quality. The API indicates the maximum concentration of the three pollutants. Table SM1 shows the relationship between the API and the concentration of the three air pollutants.
Table SM1. The Relationship between the API and Air Pollutant Concentrations
API 𝑆𝑆𝑂𝑂2 𝑁𝑁𝑂𝑂2 𝑃𝑃𝑀𝑀10 Air Quality Levels 0-50 0-0.05 0-0.08 0-0.05 Excellent 50-100 0.05-0.15 0.08-0.12 0.05-0.15 Good 100-200 0.15-0.8 0.12-0.28 0.15-0.35 Slightly Polluted 200-300 0.8-1.6 0.28-0.565 0.35-0.42 Moderately Polluted 300-400 1.6-2.1 0.565-0.75 0.42-0.5 Severely Polluted 400-500 2.1-2.62 0.75-0.94 0.5-0.6 Severely Polluted Note: Pollutant concentration is measured by 𝑚𝑚𝑚𝑚/𝑚𝑚3. The last column is the official air quality description based on the API
The construction of the API takes four steps. First, measure the daily average concentration of each pollutant. Second, for each pollutant, find out its corresponding concentration interval in Table SM1. Third, calculate the pollution index (PI) of each pollutant linearly. Finally, take the maximum of all pollution indices and define it as the API. For example, assume the concentrations of the three pollutants are: 𝐶𝐶𝑆𝑆𝑂𝑂2 =0.07𝑚𝑚𝑚𝑚/𝑚𝑚3, 𝐶𝐶𝑁𝑁𝑂𝑂2 = 0.10𝑚𝑚𝑚𝑚/𝑚𝑚3, and 𝐶𝐶𝑃𝑃𝑀𝑀10 = 0.30𝑚𝑚𝑚𝑚/𝑚𝑚3, then use Table SM1 we find that the concentrations of 𝑆𝑆𝑂𝑂2, and 𝑁𝑁𝑂𝑂2 are in the interval [50,100] while the 𝑃𝑃𝑀𝑀10 concentration falls into the interval [100,200]. Within each interval we can calculate pollution index of each pollutant linearly:
𝑃𝑃𝐼𝐼𝑆𝑆𝑂𝑂2 =100 − 50
0.15 − 0.05∗ (0.07 − 0.05) + 50 = 60
𝑃𝑃𝐼𝐼𝑁𝑁𝑂𝑂2 =100 − 50
0.12 − 0.08∗ (0.10 − 0.08) + 50 = 75
𝑃𝑃𝐼𝐼𝑃𝑃𝑀𝑀10 =200 − 1000.35 − 0.15
∗ (0.30 − 0.15) + 100 = 175 Then the 𝐴𝐴𝑃𝑃𝐼𝐼 = max[𝑃𝑃𝐼𝐼𝑆𝑆𝑂𝑂2 ,𝑃𝑃𝐼𝐼𝑁𝑁𝑂𝑂2 ,𝑃𝑃𝐼𝐼𝑃𝑃𝑀𝑀10} = 175 and PM10 is called the
primary pollutant. Reverse this process, we can recover the concentrations of the primary pollutant. We use daily API to recover daily 𝑃𝑃𝑀𝑀10 concentrations because the Chinese government did not provide daily individual pollution concentrations to the public. In our daily API data, PM10 is the primary pollutant for more than 90% of the days. So the reverse calculation can provide us accurate PM10 concentration data for 90% of the time. To deal with missing values in time series data, we use two different strategies to interpolate PM10 concentrations for the rest less than 10% sample: (1) treat those days as if PM10 is the primary pollutant, and (2) use linear interpolate for the missing values. We tried both methods and it turned out that both methods generated quantitatively similar empirical results. We also tried dropping the days
Page 35 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
3
with missing PM10 and re-estimated all the equations, and the results remained the same. The main results reported in the paper used PM10 concentrations from method (1), and the results using alterative PM10 measures are available upon request.
Page 36 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
4
Supplementary Material 2: Disease Surveillance Point System
Daily mortality data come from the Disease Surveillance Point System (DSPS) of the Chinese Center for Disease Control and Prevention (CDC). The DSPS was established by the Chinese government to provide timely information on the causes and number of deaths in 1978. To represent national population and mortality trends, the DSPS adopted a multi-stage cluster population probability sampling method. The main objectives of the DSPS are to: (1) identify the number of deaths related to each disease category and provide basic mortality information about the deceased for public health officials; and (2) provide feedback to evaluate the impacts of the public health interventions. The DSPS initially covered 71 counties and city-districts in 29 provinces; this was expanded to 145 counties and city-districts in 31 provinces in 1990. The DSPS was overhauled following the Severe Acute Respiratory Syndrome (SARS) outbreak in 2003; 161 city districts and counties were in the system from 2004 to the present. Currently 81.5 million people, or roughly 6 percent of the Chinese population live in those DSPS city districts and counties.
In the event of a death, the doctor or decedent’s family is required to fill out a death certificate and submit it to the DSPS. The mortality data include basic demographic characteristics of the decedent and the cause of death. The causes of death are coded in the International Classification of Diseases 10 (ICD-10). Total mortality is classified by causes of death: cardiovascular (I00-99) and respiratory (J00-99) diseases, and all other diseases. For this study, we had daily numbers of deaths by age group, gender, and cause of death for all the DSPS districts in the largest 38 cities. The data period is from Jan 1st, 2010 to June 29th, 2013. Table SM2 presents more details about the cities covered by the DSPS.
Page 37 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
5
Table SM2. Characteristics of DSPS Cities
City DSPS Districts City Area
(KM2) City Pop.
(1000) District Pop.
Share of Female
Population
Share of the Elderly
(>60)
GDP Per Capita
(10,000)
Share of Construction
Workers Anshan Qianshan 9254 3512 211410 0.49 0.15 6.6 0.08 Beijing Dongcheng,Tongzhou 16411 20166 1206066 0.49 0.12 7.1 0.08 Changchun Nanguan 20604 7592 550888 0.5 0.12 7.0 0.04 Changde Wuling 18670 5908 530319 0.5 0.12 5.4 0.06 Changsha Tianxin 11816 6920 472816 0.5 0.13 10.4 0.07 Chengdu Qingyang 12125 13247 823355 0.5 0.15 4.2 0.05 Chongqing Dazu 82677 29162 704640 0.49 0.18 2.6 0.02 Guangzhou Yuexiu 7323 12767 1163069 0.5 0.15 14.6 0.04 Guilin Xiufeng 27809 5145 155925 0.5 0.13 4.6 0.01 Hangzhou Xiacheng 16588 8748 506795 0.49 0.16 11.1 0.02 Harbin Nangang 53068 9929 1313002 0.49 0.13 5.2 0.03 Hohhot Huimin 17292 2719 394146 0.5 0.11 10.9 0.05 Liuzhou Liubei 18617 3677 425676 0.49 0.14 8.8 0.03 Maanshan Yushan 3259 1921 295972 0.48 0.15 7.9 0.18 Nanchang Donghu 7402 5081 575977 0.5 0.14 7.2 0.02 Nanjing Pukou 6587 7531 719366 0.48 0.13 9.1 0.04 Panzhihua Renhe 7427 1188 261717 0.48 0.15 6.1 0.01 Qingdao Shibei 11079 8056 512573 0.5 0.12 10.3 0.04 Qinhuangdao Haigang 7709 2970 610139 0.5 0.13 6.0 0.04 Qiqihar Meilisi 42469 5648 165790 0.49 0.13 3.3 0.01 Shanghai Luwan 6340 23435 1553413 0.48 0.09 11.9 0.06 Shenyang Shenbei 12980 7224 347655 0.49 0.15 8.0 0.04
Page 38 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
6
Suzhou Wuzhong 8488 9148 1165065 0.49 0.11 14.5 0.11 Taiyuan Xinghualingqu 6972 4049 650546 0.5 0.14 5.8 0.04 Tangshan Kaiping 13472 7549 270456 0.48 0.15 7.1 0.05 Tianjin Hongqiao 11865 13557 550101 0.48 0.19 11.2 0.12 Tongchuan WangYi 3882 838 200762 0.5 0.17 2.5 0.10 Urumqi Tianshan 13788 2501 711443 0.5 0.12 6.3 0.02 Wuhan Jiang'an 8494 9502 902296 0.5 0.15 9.5 0.05 Xining Chengzhong 7662 2218 298421 0.5 0.13 4.8 0.05 Xuzhou Yunlong 11259 8955 320533 0.5 0.14 6.1 0.04 Yantai Zhifu 13746 6671 828652 0.5 0.13 10.2 0.04 Yichang Wujiagang 21084 4047 213884 0.49 0.14 7.2 0.05 Yinchuan Xingqing 8975 2026 679976 0.47 0.06 5.4 0.04 Yuxi Hongta 15285 2318 494672 0.49 0.13 10.7 0.06 Zaozhuang Xuecheng 4563 3875 391613 0.48 0.14 3.1 0.06 Zhengzhou Zhongyuan 7446 8849 620825 0.49 0.11 4.1 0.05 Zunyi Qianshan 30762 6117 627764 0.49 0.11 3.5 0.03
Note: Data are collected from city statistical yearbooks and 2005 small census.
Page 39 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
7
Figure SM1. Percentage Change in Daily Number of All-Cause Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in PM10 in 38 Chinese Cities (Lag 1 Day)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean PM10 effects on total mortality, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily all-cause deaths. X-axis is percentage change. Each solid square represents an effect size. Horizontal lines indicate 95 percent CIs. ES=Effect Size.
NOTE: Weights are from random effects analysis
Overall (I-squared = 35.6%, p = 0.017)
Zaozhuang
Tangshan
Taiyuan
Liuzhou
Urumqi
Chengdu
Changde
Beijing
Guangzhou
Anshan
Maanshan
Xuzhou
Wuhan
Hangzhou
Qinhuangdao
Qiqihar
Tongchuan
Xining
Shenyang
Nanchang
City
Tianjin
Qingdao
Guilin
Suzhou
Harbin
Nanjing
Zhengzhou
Chongqing
Shanghai
Changsha
Yantai
Zunyi
Changchun
Yinchuan
Panzhihua
Yuxi
Yichang
Hohhot
0.26 (0.15 to 0.37)
0.72 (-0.01 to 1.45)
-0.79 (-1.74 to 0.16)
0.21 (-0.31 to 0.73)
1.46 (0.66 to 2.26)
-0.05 (-0.51 to 0.41)
-0.19 (-0.69 to 0.31)
0.23 (-0.38 to 0.84)
0.14 (-0.05 to 0.33)
1.07 (0.60 to 1.54)
0.20 (-0.45 to 0.84)
1.08 (0.08 to 2.08)
-0.16 (-0.97 to 0.65)
0.50 (0.14 to 0.86)
0.17 (-0.41 to 0.76)
0.00 (-0.85 to 0.86)
-0.21 (-1.20 to 0.77)
0.20 (-1.04 to 1.43)
-0.36 (-1.26 to 0.54)
0.58 (0.05 to 1.12)
0.14 (-0.55 to 0.83)
ES (95% CI)
0.36 (0.01 to 0.72)
0.06 (-0.24 to 0.36)
0.49 (-0.68 to 1.65)
0.33 (-0.10 to 0.76)
0.20 (-0.17 to 0.57)
0.47 (0.05 to 0.90)
0.11 (-0.35 to 0.58)
0.33 (0.03 to 0.62)
0.19 (-0.10 to 0.49)
0.63 (0.06 to 1.20)
0.19 (-0.36 to 0.74)
0.61 (-0.30 to 1.52)
0.06 (-0.46 to 0.57)
-0.36 (-0.99 to 0.28)
-0.37 (-1.84 to 1.10)
1.32 (0.14 to 2.50)
0.48 (-0.61 to 1.57)
0.11 (-0.76 to 0.99)
100.00
1.79
1.16
2.94
1.54
3.45
%
3.12
2.36
7.05
3.36
2.15
1.05
1.51
4.51
2.49
1.38
1.08
0.72
1.27
2.84
1.97
Weight
4.59
5.35
0.80
3.73
4.37
3.83
3.43
5.43
5.38
2.60
2.74
1.25
3.00
2.22
0.52
0.79
0.90
1.33
0.26 (0.15 to 0.37)
0.72 (-0.01 to 1.45)
-0.79 (-1.74 to 0.16)
0.21 (-0.31 to 0.73)
1.46 (0.66 to 2.26)
-0.05 (-0.51 to 0.41)
-0.19 (-0.69 to 0.31)
0.23 (-0.38 to 0.84)
0.14 (-0.05 to 0.33)
1.07 (0.60 to 1.54)
0.20 (-0.45 to 0.84)
1.08 (0.08 to 2.08)
-0.16 (-0.97 to 0.65)
0.50 (0.14 to 0.86)
0.17 (-0.41 to 0.76)
0.00 (-0.85 to 0.86)
-0.21 (-1.20 to 0.77)
0.20 (-1.04 to 1.43)
-0.36 (-1.26 to 0.54)
0.58 (0.05 to 1.12)
0.14 (-0.55 to 0.83)
ES (95% CI)
0.36 (0.01 to 0.72)
0.06 (-0.24 to 0.36)
0.49 (-0.68 to 1.65)
0.33 (-0.10 to 0.76)
0.20 (-0.17 to 0.57)
0.47 (0.05 to 0.90)
0.11 (-0.35 to 0.58)
0.33 (0.03 to 0.62)
0.19 (-0.10 to 0.49)
0.63 (0.06 to 1.20)
0.19 (-0.36 to 0.74)
0.61 (-0.30 to 1.52)
0.06 (-0.46 to 0.57)
-0.36 (-0.99 to 0.28)
-0.37 (-1.84 to 1.10)
1.32 (0.14 to 2.50)
0.48 (-0.61 to 1.57)
0.11 (-0.76 to 0.99)
100.00
1.79
1.16
2.94
1.54
3.45
%
3.12
2.36
7.05
3.36
2.15
1.05
1.51
4.51
2.49
1.38
1.08
0.72
1.27
2.84
1.97
Weight
4.59
5.35
0.80
3.73
4.37
3.83
3.43
5.43
5.38
2.60
2.74
1.25
3.00
2.22
0.52
0.79
0.90
1.33
0-4 -2 0 2 4
PM10 Effect (10μg/m3, Lag=1)
Page 40 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
8
Figure SM2. Percentage Change in Daily Number of All-Cause Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in PM10 in 38 Chinese Cities (Lag 2 Days)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean PM10 effects on total mortality, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily all-cause deaths. X-axis is percentage change. Each solid square represents an effect size. Horizontal lines indicate 95 percent CIs. ES=Effect Size.
NOTE: Weights are from random effects analysis
Overall (I-squared = 21.3%, p = 0.125)
Yuxi
Taiyuan
QiqiharQinhuangdao
Shenyang
Wuhan
Changsha
Zunyi
ChongqingYantai
Changchun
Panzhihua
Beijing
Hangzhou
Harbin
Nanchang
Changde
GuangzhouLiuzhou
Qingdao
Nanjing
Suzhou
City
HohhotZhengzhou
YichangZaozhuang
Urumqi
Xining
Xuzhou
Tongchuan
ShanghaiGuilin
Tianjin
Anshan
Yinchuan
Maanshan
TangshanChengdu
0.13 (0.03 to 0.23)
1.26 (0.11 to 2.41)
0.27 (-0.26 to 0.79)
-0.43 (-1.42 to 0.55)-0.48 (-1.36 to 0.39)
0.31 (-0.24 to 0.87)
-0.04 (-0.40 to 0.32)
0.15 (-0.42 to 0.73)
0.64 (-0.27 to 1.54)
0.18 (-0.11 to 0.48)0.17 (-0.38 to 0.72)
0.24 (-0.27 to 0.74)
-1.13 (-2.60 to 0.34)
-0.03 (-0.22 to 0.16)
0.50 (-0.08 to 1.07)
0.07 (-0.31 to 0.44)
0.20 (-0.47 to 0.88)
0.06 (-0.55 to 0.67)
0.68 (0.22 to 1.14)1.22 (0.42 to 2.02)
0.20 (-0.10 to 0.50)
0.32 (-0.11 to 0.74)
0.30 (-0.13 to 0.73)
ES (95% CI)
-0.08 (-0.96 to 0.79)-0.12 (-0.59 to 0.35)
0.64 (-0.46 to 1.74)0.54 (-0.21 to 1.29)
-0.17 (-0.64 to 0.29)
0.34 (-0.53 to 1.20)
0.06 (-0.74 to 0.87)
-0.00 (-1.25 to 1.24)
0.10 (-0.19 to 0.38)0.07 (-1.09 to 1.24)
-0.02 (-0.38 to 0.34)
0.25 (-0.40 to 0.89)
-0.19 (-0.82 to 0.43)
0.50 (-0.52 to 1.52)
-0.55 (-1.50 to 0.40)-0.50 (-1.00 to 0.00)
100.00
0.67
2.73
0.901.13
2.50
4.80
2.36
1.05
6.172.54
2.90
0.42
9.27
2.36
4.53
1.78
2.15
3.361.33
6.08
3.83
3.76
Weight
1.133.27
0.731.49
3.33
1.14
1.31
0.58
6.260.66
4.73
1.93
2.07
0.85
0.962.96
%
0.13 (0.03 to 0.23)
1.26 (0.11 to 2.41)
0.27 (-0.26 to 0.79)
-0.43 (-1.42 to 0.55)-0.48 (-1.36 to 0.39)
0.31 (-0.24 to 0.87)
-0.04 (-0.40 to 0.32)
0.15 (-0.42 to 0.73)
0.64 (-0.27 to 1.54)
0.18 (-0.11 to 0.48)0.17 (-0.38 to 0.72)
0.24 (-0.27 to 0.74)
-1.13 (-2.60 to 0.34)
-0.03 (-0.22 to 0.16)
0.50 (-0.08 to 1.07)
0.07 (-0.31 to 0.44)
0.20 (-0.47 to 0.88)
0.06 (-0.55 to 0.67)
0.68 (0.22 to 1.14)1.22 (0.42 to 2.02)
0.20 (-0.10 to 0.50)
0.32 (-0.11 to 0.74)
0.30 (-0.13 to 0.73)
ES (95% CI)
-0.08 (-0.96 to 0.79)-0.12 (-0.59 to 0.35)
0.64 (-0.46 to 1.74)0.54 (-0.21 to 1.29)
-0.17 (-0.64 to 0.29)
0.34 (-0.53 to 1.20)
0.06 (-0.74 to 0.87)
-0.00 (-1.25 to 1.24)
0.10 (-0.19 to 0.38)0.07 (-1.09 to 1.24)
-0.02 (-0.38 to 0.34)
0.25 (-0.40 to 0.89)
-0.19 (-0.82 to 0.43)
0.50 (-0.52 to 1.52)
-0.55 (-1.50 to 0.40)-0.50 (-1.00 to 0.00)
100.00
0.67
2.73
0.901.13
2.50
4.80
2.36
1.05
6.172.54
2.90
0.42
9.27
2.36
4.53
1.78
2.15
3.361.33
6.08
3.83
3.76
Weight
1.133.27
0.731.49
3.33
1.14
1.31
0.58
6.260.66
4.73
1.93
2.07
0.85
0.962.96
%
0-4 -2 0 2 4
PM10 Effect (10μg/m3, Lag=2)
Page 41 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
9
Figure SM3. Percentage Change in Daily Number of All-Cause Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in PM10 in 38 Chinese Cities (Lag 3 Days)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean PM10 effects on total mortality, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily all-cause deaths. X-axis is percentage change. Each solid square represents an effect size. Horizontal lines indicate 95 percent CIs. ES=Effect Size.
NOTE: Weights are from random effects analysis
Overall (I-squared = 29.2%, p = 0.049)
Beijing
Qinhuangdao
Nanchang
Zaozhuang
Yinchuan
Changde
Xuzhou
Qingdao
Yuxi
Taiyuan
Guilin
Shanghai
Zunyi
SuzhouAnshan
Changsha
Wuhan
Urumqi
Hohhot
Tongchuan
Shenyang
Tangshan
Chongqing
Nanjing
Panzhihua
Qiqihar
Yichang
Liuzhou
Chengdu
Xining
Maanshan
Zhengzhou
Changchun
Harbin
City
Hangzhou
TianjinYantai
Guangzhou
0.09 (-0.01 to 0.19)
-0.12 (-0.30 to 0.07)
0.05 (-0.84 to 0.93)
-0.24 (-0.91 to 0.43)
0.44 (-0.31 to 1.18)
0.08 (-0.53 to 0.69)
0.02 (-0.59 to 0.63)
-0.19 (-1.00 to 0.62)
0.25 (-0.05 to 0.55)
1.13 (-0.02 to 2.28)
0.33 (-0.20 to 0.86)
-0.84 (-2.02 to 0.34)
0.13 (-0.16 to 0.42)
0.40 (-0.50 to 1.31)
0.31 (-0.12 to 0.73)0.27 (-0.38 to 0.91)
-0.04 (-0.62 to 0.54)
0.05 (-0.31 to 0.41)
0.22 (-0.24 to 0.68)
-0.49 (-1.37 to 0.39)
-0.04 (-1.28 to 1.21)
0.07 (-0.49 to 0.63)
0.58 (-0.37 to 1.53)
0.10 (-0.19 to 0.39)
0.37 (-0.04 to 0.79)
-0.22 (-1.67 to 1.24)
-0.57 (-1.57 to 0.42)
-0.12 (-1.24 to 0.99)
0.65 (-0.15 to 1.45)
-0.72 (-1.22 to -0.21)
0.25 (-0.62 to 1.12)
-0.11 (-1.14 to 0.91)
-0.18 (-0.65 to 0.29)
0.25 (-0.26 to 0.75)
-0.06 (-0.44 to 0.32)
ES (95% CI)
1.00 (0.44 to 1.56)
-0.17 (-0.53 to 0.20)-0.15 (-0.70 to 0.40)
0.44 (-0.03 to 0.90)
100.00
7.96
1.22
1.93
1.62
2.24
2.26
1.41
5.69
0.75
2.83
0.72
5.87
1.16
3.822.07
2.46
4.64
3.42
1.23
0.65
2.56
1.06
5.72
3.92
0.48
0.98
0.80
1.44
3.02
1.26
0.93
3.34
2.97
4.39
Weight
2.58
4.582.63
3.40
%
0.09 (-0.01 to 0.19)
-0.12 (-0.30 to 0.07)
0.05 (-0.84 to 0.93)
-0.24 (-0.91 to 0.43)
0.44 (-0.31 to 1.18)
0.08 (-0.53 to 0.69)
0.02 (-0.59 to 0.63)
-0.19 (-1.00 to 0.62)
0.25 (-0.05 to 0.55)
1.13 (-0.02 to 2.28)
0.33 (-0.20 to 0.86)
-0.84 (-2.02 to 0.34)
0.13 (-0.16 to 0.42)
0.40 (-0.50 to 1.31)
0.31 (-0.12 to 0.73)0.27 (-0.38 to 0.91)
-0.04 (-0.62 to 0.54)
0.05 (-0.31 to 0.41)
0.22 (-0.24 to 0.68)
-0.49 (-1.37 to 0.39)
-0.04 (-1.28 to 1.21)
0.07 (-0.49 to 0.63)
0.58 (-0.37 to 1.53)
0.10 (-0.19 to 0.39)
0.37 (-0.04 to 0.79)
-0.22 (-1.67 to 1.24)
-0.57 (-1.57 to 0.42)
-0.12 (-1.24 to 0.99)
0.65 (-0.15 to 1.45)
-0.72 (-1.22 to -0.21)
0.25 (-0.62 to 1.12)
-0.11 (-1.14 to 0.91)
-0.18 (-0.65 to 0.29)
0.25 (-0.26 to 0.75)
-0.06 (-0.44 to 0.32)
ES (95% CI)
1.00 (0.44 to 1.56)
-0.17 (-0.53 to 0.20)-0.15 (-0.70 to 0.40)
0.44 (-0.03 to 0.90)
100.00
7.96
1.22
1.93
1.62
2.24
2.26
1.41
5.69
0.75
2.83
0.72
5.87
1.16
3.822.07
2.46
4.64
3.42
1.23
0.65
2.56
1.06
5.72
3.92
0.48
0.98
0.80
1.44
3.02
1.26
0.93
3.34
2.97
4.39
Weight
2.58
4.582.63
3.40
%
0-4 -2 0 2 4
PM10 Effect (10μg/m3, Lag=3)
Page 42 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
10
Figure SM4. Percentage Change in Daily Number of All-Cause Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in PM10 in 38 Chinese Cities (Lag 4 Days)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean PM10 effects on total mortality, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily all-cause deaths. X-axis is percentage change. Each solid square represents an effect size. Horizontal lines indicate 95 percent CIs. ES=Effect Size.
NOTE: Weights are from random effects analysis
Overall (I-squared = 0.5%, p = 0.460)
Taiyuan
Urumqi
Changde
Zunyi
Qingdao
Xining
Changchun
Qinhuangdao
Zaozhuang
Qiqihar
Shanghai
Guilin
Maanshan
Chengdu
Zhengzhou
Suzhou
Beijing
City
Shenyang
Guangzhou
Hohhot
Xuzhou
Tongchuan
Panzhihua
Tangshan
Wuhan
Chongqing
AnshanYantai
Hangzhou
Yinchuan
Liuzhou
Nanchang
Nanjing
Yichang
Changsha
Tianjin
Harbin
Yuxi
0.08 (-0.00 to 0.16)
-0.28 (-0.82 to 0.26)
0.13 (-0.33 to 0.59)
0.08 (-0.53 to 0.69)
0.16 (-0.74 to 1.06)
0.20 (-0.09 to 0.50)
-0.74 (-1.63 to 0.15)
0.04 (-0.48 to 0.56)
-0.16 (-1.08 to 0.77)
0.90 (0.15 to 1.64)
-0.26 (-1.25 to 0.73)
0.25 (-0.03 to 0.53)
-1.28 (-2.47 to -0.09)
0.52 (-0.49 to 1.53)
-0.10 (-0.60 to 0.40)
0.04 (-0.43 to 0.50)
0.13 (-0.29 to 0.56)
-0.12 (-0.31 to 0.07)
ES (95% CI)
0.10 (-0.47 to 0.67)
0.30 (-0.16 to 0.76)
0.02 (-0.85 to 0.90)
0.09 (-0.71 to 0.89)
0.30 (-0.96 to 1.56)
0.14 (-1.31 to 1.59)
1.04 (0.07 to 2.01)
0.05 (-0.31 to 0.42)
0.18 (-0.11 to 0.48)
-0.00 (-0.65 to 0.64)0.02 (-0.53 to 0.57)
0.48 (-0.09 to 1.05)
0.21 (-0.40 to 0.82)
0.66 (-0.14 to 1.47)
-0.22 (-0.90 to 0.46)
0.16 (-0.26 to 0.57)
-0.30 (-1.43 to 0.83)
-0.06 (-0.64 to 0.52)
-0.19 (-0.56 to 0.18)
0.25 (-0.13 to 0.62)
0.48 (-0.67 to 1.63)
100.00
2.13
2.93
1.68
0.78
7.09
0.80
2.32
0.74
1.14
0.64
7.79
0.45
0.62
2.51
2.88
3.46
17.18
Weight
1.93
2.93
0.82
0.98
0.40
0.30
0.67
4.75
7.00
1.512.07
1.93
1.68
0.97
1.36
3.62
0.50
1.88
4.60
4.50
0.48
%
0.08 (-0.00 to 0.16)
-0.28 (-0.82 to 0.26)
0.13 (-0.33 to 0.59)
0.08 (-0.53 to 0.69)
0.16 (-0.74 to 1.06)
0.20 (-0.09 to 0.50)
-0.74 (-1.63 to 0.15)
0.04 (-0.48 to 0.56)
-0.16 (-1.08 to 0.77)
0.90 (0.15 to 1.64)
-0.26 (-1.25 to 0.73)
0.25 (-0.03 to 0.53)
-1.28 (-2.47 to -0.09)
0.52 (-0.49 to 1.53)
-0.10 (-0.60 to 0.40)
0.04 (-0.43 to 0.50)
0.13 (-0.29 to 0.56)
-0.12 (-0.31 to 0.07)
ES (95% CI)
0.10 (-0.47 to 0.67)
0.30 (-0.16 to 0.76)
0.02 (-0.85 to 0.90)
0.09 (-0.71 to 0.89)
0.30 (-0.96 to 1.56)
0.14 (-1.31 to 1.59)
1.04 (0.07 to 2.01)
0.05 (-0.31 to 0.42)
0.18 (-0.11 to 0.48)
-0.00 (-0.65 to 0.64)0.02 (-0.53 to 0.57)
0.48 (-0.09 to 1.05)
0.21 (-0.40 to 0.82)
0.66 (-0.14 to 1.47)
-0.22 (-0.90 to 0.46)
0.16 (-0.26 to 0.57)
-0.30 (-1.43 to 0.83)
-0.06 (-0.64 to 0.52)
-0.19 (-0.56 to 0.18)
0.25 (-0.13 to 0.62)
0.48 (-0.67 to 1.63)
100.00
2.13
2.93
1.68
0.78
7.09
0.80
2.32
0.74
1.14
0.64
7.79
0.45
0.62
2.51
2.88
3.46
17.18
Weight
1.93
2.93
0.82
0.98
0.40
0.30
0.67
4.75
7.00
1.512.07
1.93
1.68
0.97
1.36
3.62
0.50
1.88
4.60
4.50
0.48
%
0-4 -2 0 2 4
PM10 Effect (10μg/m3, Lag=4)
Page 43 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
11
Figure SM5. Percentage Change in Daily Number of All-Cause Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in PM10 in 38 Chinese Cities (Lag 5 Days)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean PM10 effects on total mortality, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily all-cause deaths. X-axis is percentage change. Each solid square represents an effect size. Horizontal lines indicate 95 percent CIs. ES=Effect Size.
NOTE: Weights are from random effects analysis
Overall (I-squared = 0.0%, p = 0.566)
Suzhou
Changsha
Wuhan
Chongqing
Taiyuan
Urumqi
Nanchang
Hangzhou
Zaozhuang
Changde
Zhengzhou
Yuxi
Changchun
Qingdao
Liuzhou
Yichang
Hohhot
Yantai
Beijing
Guangzhou
Shenyang
Xining
Zunyi
Qiqihar
Tongchuan
Nanjing
Guilin
Tianjin
Maanshan
City
Harbin
Xuzhou
Chengdu
Anshan
Qinhuangdao
Shanghai
Yinchuan
PanzhihuaTangshan
0.07 (-0.01 to 0.15)
0.22 (-0.20 to 0.65)
-0.17 (-0.76 to 0.41)
0.02 (-0.35 to 0.38)
0.19 (-0.11 to 0.49)
-0.44 (-0.98 to 0.10)
0.27 (-0.20 to 0.74)
-0.31 (-0.99 to 0.37)
0.10 (-0.47 to 0.67)
0.50 (-0.25 to 1.24)
-0.27 (-0.88 to 0.34)
0.16 (-0.31 to 0.62)
-0.12 (-1.27 to 1.03)
-0.38 (-0.90 to 0.15)
0.13 (-0.17 to 0.43)
0.34 (-0.47 to 1.15)
-0.10 (-1.23 to 1.02)
0.16 (-0.71 to 1.02)
-0.10 (-0.65 to 0.46)
-0.02 (-0.21 to 0.17)
0.41 (-0.05 to 0.87)
0.18 (-0.38 to 0.75)
-0.18 (-1.02 to 0.67)
0.07 (-0.83 to 0.98)
-0.65 (-1.64 to 0.35)
-0.05 (-1.30 to 1.20)
0.07 (-0.35 to 0.48)
-0.68 (-1.84 to 0.49)
-0.06 (-0.42 to 0.31)
-0.01 (-1.03 to 1.01)
ES (95% CI)
0.04 (-0.33 to 0.42)
0.19 (-0.61 to 0.98)
-0.18 (-0.69 to 0.32)
0.30 (-0.34 to 0.94)
0.26 (-0.66 to 1.17)
0.38 (0.10 to 0.66)
0.00 (-0.61 to 0.61)
0.58 (-0.87 to 2.02)1.37 (0.39 to 2.35)
100.00
3.48
1.84
4.65
6.94
2.12
2.85
1.34
1.89
1.13
1.66
2.85
0.47
2.25
7.03
0.95
0.49
0.83
2.03
17.64
2.94
1.95
0.87
0.76
0.63
0.40
3.62
0.46
4.60
0.60
Weight
4.46
1.00
2.46
1.52
0.75
7.95
1.67
0.300.65
%
0.07 (-0.01 to 0.15)
0.22 (-0.20 to 0.65)
-0.17 (-0.76 to 0.41)
0.02 (-0.35 to 0.38)
0.19 (-0.11 to 0.49)
-0.44 (-0.98 to 0.10)
0.27 (-0.20 to 0.74)
-0.31 (-0.99 to 0.37)
0.10 (-0.47 to 0.67)
0.50 (-0.25 to 1.24)
-0.27 (-0.88 to 0.34)
0.16 (-0.31 to 0.62)
-0.12 (-1.27 to 1.03)
-0.38 (-0.90 to 0.15)
0.13 (-0.17 to 0.43)
0.34 (-0.47 to 1.15)
-0.10 (-1.23 to 1.02)
0.16 (-0.71 to 1.02)
-0.10 (-0.65 to 0.46)
-0.02 (-0.21 to 0.17)
0.41 (-0.05 to 0.87)
0.18 (-0.38 to 0.75)
-0.18 (-1.02 to 0.67)
0.07 (-0.83 to 0.98)
-0.65 (-1.64 to 0.35)
-0.05 (-1.30 to 1.20)
0.07 (-0.35 to 0.48)
-0.68 (-1.84 to 0.49)
-0.06 (-0.42 to 0.31)
-0.01 (-1.03 to 1.01)
ES (95% CI)
0.04 (-0.33 to 0.42)
0.19 (-0.61 to 0.98)
-0.18 (-0.69 to 0.32)
0.30 (-0.34 to 0.94)
0.26 (-0.66 to 1.17)
0.38 (0.10 to 0.66)
0.00 (-0.61 to 0.61)
0.58 (-0.87 to 2.02)1.37 (0.39 to 2.35)
100.00
3.48
1.84
4.65
6.94
2.12
2.85
1.34
1.89
1.13
1.66
2.85
0.47
2.25
7.03
0.95
0.49
0.83
2.03
17.64
2.94
1.95
0.87
0.76
0.63
0.40
3.62
0.46
4.60
0.60
Weight
4.46
1.00
2.46
1.52
0.75
7.95
1.67
0.300.65
%
0-4 -2 0 2 4
PM10 Effect (10μg/m3, Lag=5)
Page 44 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
12
Figure SM6. Percentage Change in Daily Number of All-Cause Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in PM10 in 38 Chinese Cities (Lag 6 Days)
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean PM10 effects on total mortality, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily all-cause deaths. X-axis is percentage change. Each solid square represents an effect size. Horizontal lines indicate 95 percent CIs. ES=Effect Size.
NOTE: Weights are from random effects analysis
Overall (I-squared = 14.5%, p = 0.221)
Anshan
Yuxi
Urumqi
Shenyang
Changchun
Taiyuan
Qingdao
ChengduQiqihar
Wuhan
Yinchuan
Chongqing
Tongchuan
Yantai
Harbin
Xuzhou
Hangzhou
Zhengzhou
Suzhou
Nanchang
Xining
Liuzhou
Beijing
Maanshan
City
Tangshan
Tianjin
Hohhot
Guilin
Changsha
Shanghai
Changde
Qinhuangdao
Panzhihua
Zaozhuang
Guangzhou
Zunyi
Nanjing
Yichang
0.05 (-0.04 to 0.14)
0.16 (-0.48 to 0.80)
-0.25 (-1.41 to 0.90)
0.25 (-0.23 to 0.72)
-0.08 (-0.65 to 0.50)
0.03 (-0.49 to 0.54)
-0.28 (-0.83 to 0.26)
0.20 (-0.10 to 0.49)
-0.47 (-0.98 to 0.05)-0.59 (-1.58 to 0.41)
-0.13 (-0.50 to 0.24)
0.37 (-0.24 to 0.98)
0.13 (-0.18 to 0.43)
0.23 (-1.02 to 1.49)
0.12 (-0.43 to 0.67)
-0.04 (-0.42 to 0.33)
-0.62 (-1.44 to 0.21)
0.03 (-0.55 to 0.61)
-0.01 (-0.47 to 0.46)
0.14 (-0.29 to 0.56)
-0.31 (-1.01 to 0.38)
-0.01 (-0.86 to 0.85)
0.04 (-0.77 to 0.86)
-0.11 (-0.30 to 0.08)
-0.21 (-1.23 to 0.82)
ES (95% CI)
1.75 (0.78 to 2.72)
0.09 (-0.28 to 0.45)
0.73 (-0.11 to 1.57)
0.09 (-1.06 to 1.24)
-0.01 (-0.60 to 0.58)
0.27 (-0.01 to 0.56)
-0.00 (-0.62 to 0.61)
-0.37 (-1.29 to 0.55)
0.30 (-1.15 to 1.75)
-0.36 (-1.12 to 0.40)
0.57 (0.11 to 1.03)
-0.28 (-1.20 to 0.65)
0.12 (-0.30 to 0.53)
0.53 (-0.59 to 1.65)
100.00
1.87
0.61
3.17
2.26
2.73
2.48
6.52
2.780.82
4.75
2.07
6.29
0.52
2.45
4.64
1.17
2.25
3.23
3.79
1.62
1.09
1.20
10.91
0.77
Weight
0.85
4.79
1.13
0.62
2.19
6.88
2.02
0.95
0.39
%
1.36
3.33
0.95
3.88
0.66
0.05 (-0.04 to 0.14)
0.16 (-0.48 to 0.80)
-0.25 (-1.41 to 0.90)
0.25 (-0.23 to 0.72)
-0.08 (-0.65 to 0.50)
0.03 (-0.49 to 0.54)
-0.28 (-0.83 to 0.26)
0.20 (-0.10 to 0.49)
-0.47 (-0.98 to 0.05)-0.59 (-1.58 to 0.41)
-0.13 (-0.50 to 0.24)
0.37 (-0.24 to 0.98)
0.13 (-0.18 to 0.43)
0.23 (-1.02 to 1.49)
0.12 (-0.43 to 0.67)
-0.04 (-0.42 to 0.33)
-0.62 (-1.44 to 0.21)
0.03 (-0.55 to 0.61)
-0.01 (-0.47 to 0.46)
0.14 (-0.29 to 0.56)
-0.31 (-1.01 to 0.38)
-0.01 (-0.86 to 0.85)
0.04 (-0.77 to 0.86)
-0.11 (-0.30 to 0.08)
-0.21 (-1.23 to 0.82)
ES (95% CI)
1.75 (0.78 to 2.72)
0.09 (-0.28 to 0.45)
0.73 (-0.11 to 1.57)
0.09 (-1.06 to 1.24)
-0.01 (-0.60 to 0.58)
0.27 (-0.01 to 0.56)
-0.00 (-0.62 to 0.61)
-0.37 (-1.29 to 0.55)
0.30 (-1.15 to 1.75)
-0.36 (-1.12 to 0.40)
0.57 (0.11 to 1.03)
-0.28 (-1.20 to 0.65)
0.12 (-0.30 to 0.53)
0.53 (-0.59 to 1.65)
100.00
1.87
0.61
3.17
2.26
2.73
2.48
6.52
2.780.82
4.75
2.07
6.29
0.52
2.45
4.64
1.17
2.25
3.23
3.79
1.62
1.09
1.20
10.91
0.77
Weight
0.85
4.79
1.13
0.62
2.19
6.88
2.02
0.95
0.39
%
1.36
3.33
0.95
3.88
0.66
0-4 -2 0 2 4
PM10 Effect (10μg/m3, Lag=6)
Page 45 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
13
Figure SM7. Percentage Change in Daily Number of All-Cause Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in 3-day Moving Average PM10 (lags 0, 1, and 2) in 38 Chinese Cities
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean lag PM10 effects on total mortality, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily all-cause deaths. X-axis is percentage change. Each solid square represents an effect size. Horizontal lines indicate 95 percent CIs. ES=Effect Size.
NOTE: Weights are from random effects analysis
Overall (I-squared = 53.3%, p = 0.000)
Nanchang
Hangzhou
Qinhuangdao
Chongqing
Chengdu
Qingdao
Changsha
Tangshan
Taiyuan
Hohhot
Suzhou
Panzhihua
Liuzhou
Zhengzhou
Qiqihar
Wuhan
Harbin
Tongchuan
Urumqi
ShanghaiAnshan
Yantai
Changde
Shenyang
Tianjin
MaanshanGuangzhou
Beijing
City
Zunyi
Xining
YichangNanjing
Zaozhuang
Yuxi
Yinchuan
Changchun
Xuzhou
Guilin
0.45 (0.28 to 0.62)
0.29 (-0.60 to 1.17)
0.60 (-0.17 to 1.37)
-0.04 (-1.21 to 1.13)
0.45 (0.12 to 0.78)
-0.19 (-0.81 to 0.42)
0.20 (-0.19 to 0.58)
0.63 (-0.08 to 1.35)
-0.84 (-2.17 to 0.50)
0.66 (-0.02 to 1.34)
-0.02 (-1.19 to 1.15)
0.62 (0.04 to 1.20)
-1.39 (-3.35 to 0.57)
2.14 (1.16 to 3.12)
0.14 (-0.46 to 0.74)
-0.35 (-1.56 to 0.85)
0.58 (0.14 to 1.03)
0.26 (-0.26 to 0.79)
0.11 (-1.52 to 1.73)
-0.19 (-0.78 to 0.41)
0.36 (-0.05 to 0.78)0.34 (-0.52 to 1.19)
0.37 (-0.36 to 1.11)
0.50 (-0.24 to 1.23)
0.74 (0.03 to 1.45)
0.64 (0.13 to 1.14)
1.75 (0.45 to 3.05)1.66 (1.08 to 2.24)
0.25 (-0.02 to 0.51)
ES (95% CI)
0.98 (-0.15 to 2.11)
-0.04 (-1.25 to 1.18)
0.79 (-0.55 to 2.13)0.80 (0.26 to 1.34)
1.33 (0.39 to 2.27)
2.10 (0.68 to 3.52)
-0.53 (-1.38 to 0.33)
0.57 (-0.11 to 1.26)
-0.60 (-1.71 to 0.50)
0.50 (-0.91 to 1.91)
100.00
2.28
2.67
1.55
4.84
3.33
4.54
2.88
1.26
3.04
1.54
3.51
0.66
1.99
3.39
1.48
4.24
3.79
0.92
3.44
4.382.36
2.80
2.80
2.92
3.89
1.333.49
5.18
Weight
1.62
1.47
1.263.70
2.09
1.15
2.35
3.02
1.68
1.16
%
0.45 (0.28 to 0.62)
0.29 (-0.60 to 1.17)
0.60 (-0.17 to 1.37)
-0.04 (-1.21 to 1.13)
0.45 (0.12 to 0.78)
-0.19 (-0.81 to 0.42)
0.20 (-0.19 to 0.58)
0.63 (-0.08 to 1.35)
-0.84 (-2.17 to 0.50)
0.66 (-0.02 to 1.34)
-0.02 (-1.19 to 1.15)
0.62 (0.04 to 1.20)
-1.39 (-3.35 to 0.57)
2.14 (1.16 to 3.12)
0.14 (-0.46 to 0.74)
-0.35 (-1.56 to 0.85)
0.58 (0.14 to 1.03)
0.26 (-0.26 to 0.79)
0.11 (-1.52 to 1.73)
-0.19 (-0.78 to 0.41)
0.36 (-0.05 to 0.78)0.34 (-0.52 to 1.19)
0.37 (-0.36 to 1.11)
0.50 (-0.24 to 1.23)
0.74 (0.03 to 1.45)
0.64 (0.13 to 1.14)
1.75 (0.45 to 3.05)1.66 (1.08 to 2.24)
0.25 (-0.02 to 0.51)
ES (95% CI)
0.98 (-0.15 to 2.11)
-0.04 (-1.25 to 1.18)
0.79 (-0.55 to 2.13)0.80 (0.26 to 1.34)
1.33 (0.39 to 2.27)
2.10 (0.68 to 3.52)
-0.53 (-1.38 to 0.33)
0.57 (-0.11 to 1.26)
-0.60 (-1.71 to 0.50)
0.50 (-0.91 to 1.91)
100.00
2.28
2.67
1.55
4.84
3.33
4.54
2.88
1.26
3.04
1.54
3.51
0.66
1.99
3.39
1.48
4.24
3.79
0.92
3.44
4.382.36
2.80
2.80
2.92
3.89
1.333.49
5.18
Weight
1.62
1.47
1.263.70
2.09
1.15
2.35
3.02
1.68
1.16
%
0-4 -2 0 2 4
All Causes
PM10 Effect (10μg/m3, 3-day Moving Average)
Page 46 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
14
Figure SM8. Percentage Change in Daily Number of All-Cause Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in 4-day (Lag 3-6 Days) Moving Average of PM10
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean lag PM10 effects on total mortality, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily all-cause deaths. X-axis is percentage change. Each solid square represents an effect size. Horizontal lines indicate 95 percent CIs. ES=Effect Size.
NOTE: Weights are from random effects analysis
Overall (I-squared = 21.0%, p = 0.128)
Maanshan
Yantai
Liuzhou
Shenyang
Panzhihua
Changchun
Anshan
Zaozhuang
Zhengzhou
City
Guangzhou
Chengdu
Yinchuan
NanchangZunyi
Harbin
Shanghai
Guilin
TangshanYichang
Chongqing
Beijing
Changde
Yuxi
Qingdao
XiningXuzhou
Tianjin
Suzhou
Urumqi
Tongchuan
Hangzhou
Qinhuangdao
Taiyuan
Qiqihar
Changsha
Hohhot
Wuhan
Nanjing
0.19 (-0.02 to 0.40)
1.62 (-0.97 to 4.21)
-0.36 (-1.53 to 0.81)
0.65 (-0.91 to 2.21)
0.23 (-0.84 to 1.30)
-0.67 (-3.44 to 2.09)
-0.09 (-1.19 to 1.00)
-0.49 (-1.80 to 0.82)
0.80 (-0.45 to 2.04)
0.26 (-0.68 to 1.20)
ES (95% CI)
0.90 (-0.23 to 2.04)
0.42 (-0.50 to 1.33)
-1.58 (-2.89 to -0.27)
0.30 (-1.19 to 1.79)0.34 (-1.38 to 2.06)
0.04 (-0.70 to 0.78)
-0.06 (-0.77 to 0.66)
1.55 (-0.82 to 3.92)
1.20 (-0.66 to 3.06)0.95 (-1.38 to 3.29)
-0.34 (-1.04 to 0.35)
0.28 (-0.16 to 0.72)
-0.22 (-1.36 to 0.92)
3.16 (0.69 to 5.63)
0.11 (-0.60 to 0.81)
-1.22 (-3.34 to 0.90)-1.57 (-3.48 to 0.34)
0.58 (-0.22 to 1.37)
-0.94 (-1.99 to 0.11)
0.48 (-0.39 to 1.35)
0.16 (-2.17 to 2.49)
0.02 (-1.58 to 1.62)
1.34 (-0.31 to 2.99)
0.08 (-1.00 to 1.16)
0.44 (-1.18 to 2.07)
0.62 (-0.65 to 1.88)
2.22 (0.74 to 3.70)
0.01 (-0.83 to 0.85)
0.56 (-0.45 to 1.56)
100.00
0.62
2.60
1.58
2.98
0.55
2.87
2.14
2.33
3.64
Weight
2.71
3.79
2.14
1.711.33
5.02
5.27
0.73
1.150.75
5.45
8.53
2.69
0.68
5.38
0.911.10
4.61
3.07
4.04
0.75
%
1.51
1.44
2.94
1.47
2.27
1.74
4.27
3.26
0.19 (-0.02 to 0.40)
1.62 (-0.97 to 4.21)
-0.36 (-1.53 to 0.81)
0.65 (-0.91 to 2.21)
0.23 (-0.84 to 1.30)
-0.67 (-3.44 to 2.09)
-0.09 (-1.19 to 1.00)
-0.49 (-1.80 to 0.82)
0.80 (-0.45 to 2.04)
0.26 (-0.68 to 1.20)
ES (95% CI)
0.90 (-0.23 to 2.04)
0.42 (-0.50 to 1.33)
-1.58 (-2.89 to -0.27)
0.30 (-1.19 to 1.79)0.34 (-1.38 to 2.06)
0.04 (-0.70 to 0.78)
-0.06 (-0.77 to 0.66)
1.55 (-0.82 to 3.92)
1.20 (-0.66 to 3.06)0.95 (-1.38 to 3.29)
-0.34 (-1.04 to 0.35)
0.28 (-0.16 to 0.72)
-0.22 (-1.36 to 0.92)
3.16 (0.69 to 5.63)
0.11 (-0.60 to 0.81)
-1.22 (-3.34 to 0.90)-1.57 (-3.48 to 0.34)
0.58 (-0.22 to 1.37)
-0.94 (-1.99 to 0.11)
0.48 (-0.39 to 1.35)
0.16 (-2.17 to 2.49)
0.02 (-1.58 to 1.62)
1.34 (-0.31 to 2.99)
0.08 (-1.00 to 1.16)
0.44 (-1.18 to 2.07)
0.62 (-0.65 to 1.88)
2.22 (0.74 to 3.70)
0.01 (-0.83 to 0.85)
0.56 (-0.45 to 1.56)
100.00
0.62
2.60
1.58
2.98
0.55
2.87
2.14
2.33
3.64
Weight
2.71
3.79
2.14
1.711.33
5.02
5.27
0.73
1.150.75
5.45
8.53
2.69
0.68
5.38
0.911.10
4.61
3.07
4.04
0.75
%
1.51
1.44
2.94
1.47
2.27
1.74
4.27
3.26
0-4 -2 0 2 4
PM10 Effect (10μg/m3, Lag 3-6 Days Moving Average)
Page 47 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Figure SM9. Percentage Change in Daily Number of Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in Concurrent Day PM10 in 38 Chinese Cities: Males vs. Females
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean lag PM10 effects on total mortality for cardiorespiratory and non-cardiorespiratory deaths separately, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily deaths for males and females separately.
NOTE: Weights are from random effects analysis
Overall (I-squared = 45.8%, p = 0.001)
Nanjing
TaiyuanShanghai
Zunyi
Maanshan
Wuhan
Tongchuan
Tianjin
Xuzhou
Chengdu
City
Yuxi
Yichang
Harbin
Xining
Zhengzhou
Shenyang
Chongqing
Qingdao
Yinchuan
Guangzhou
Qiqihar
Urumqi
Hangzhou
Qinhuangdao
Changsha
Tangshan
Nanchang
Beijing
Liuzhou
Panzhihua
Yantai
Guilin
Zaozhuang
ChangdeChangchun
Hohhot
Suzhou
Anshan
0.39 (0.23 to 0.54)
0.64 (0.06 to 1.22)
0.42 (-0.24 to 1.08)0.43 (0.02 to 0.83)
0.11 (-1.06 to 1.29)
1.48 (0.18 to 2.78)
0.50 (0.01 to 0.99)
-0.21 (-1.71 to 1.29)
0.80 (0.33 to 1.27)
-1.12 (-2.24 to -0.00)
0.20 (-0.38 to 0.78)
ES (95% CI)
2.19 (0.62 to 3.76)
0.09 (-1.35 to 1.53)
-0.04 (-0.54 to 0.45)
-0.36 (-1.51 to 0.79)
-0.13 (-0.72 to 0.47)
0.60 (-0.11 to 1.32)
0.47 (0.08 to 0.86)
0.27 (-0.14 to 0.67)
-0.12 (-0.93 to 0.69)
1.71 (1.08 to 2.34)
-0.01 (-1.25 to 1.24)
-0.10 (-0.69 to 0.48)
0.65 (-0.11 to 1.40)
0.10 (-0.99 to 1.18)
0.60 (-0.19 to 1.38)
-0.10 (-1.37 to 1.17)
0.41 (-0.46 to 1.28)
0.21 (-0.05 to 0.47)
1.81 (0.78 to 2.84)
-1.55 (-3.40 to 0.30)
0.40 (-0.34 to 1.14)
0.82 (-0.69 to 2.32)
0.61 (-0.26 to 1.49)
0.64 (-0.11 to 1.39)0.64 (-0.04 to 1.31)
0.32 (-0.70 to 1.35)
-0.05 (-0.67 to 0.57)
0.32 (-0.50 to 1.15)
100.00
3.56
3.124.70
1.43
1.21
4.13
0.96
4.27
1.55
3.56
Weight
0.88
1.03
4.08
1.49
3.48
%
2.86
4.80
4.72
2.44
3.29
1.31
3.54
2.66
1.61
2.56
1.26
2.22
5.79
1.74
0.66
2.75
0.95
2.20
2.683.04
1.77
3.35
2.39
0.39 (0.23 to 0.54)
0.64 (0.06 to 1.22)
0.42 (-0.24 to 1.08)0.43 (0.02 to 0.83)
0.11 (-1.06 to 1.29)
1.48 (0.18 to 2.78)
0.50 (0.01 to 0.99)
-0.21 (-1.71 to 1.29)
0.80 (0.33 to 1.27)
-1.12 (-2.24 to -0.00)
0.20 (-0.38 to 0.78)
ES (95% CI)
2.19 (0.62 to 3.76)
0.09 (-1.35 to 1.53)
-0.04 (-0.54 to 0.45)
-0.36 (-1.51 to 0.79)
-0.13 (-0.72 to 0.47)
0.60 (-0.11 to 1.32)
0.47 (0.08 to 0.86)
0.27 (-0.14 to 0.67)
-0.12 (-0.93 to 0.69)
1.71 (1.08 to 2.34)
-0.01 (-1.25 to 1.24)
-0.10 (-0.69 to 0.48)
0.65 (-0.11 to 1.40)
0.10 (-0.99 to 1.18)
0.60 (-0.19 to 1.38)
-0.10 (-1.37 to 1.17)
0.41 (-0.46 to 1.28)
0.21 (-0.05 to 0.47)
1.81 (0.78 to 2.84)
-1.55 (-3.40 to 0.30)
0.40 (-0.34 to 1.14)
0.82 (-0.69 to 2.32)
0.61 (-0.26 to 1.49)
0.64 (-0.11 to 1.39)0.64 (-0.04 to 1.31)
0.32 (-0.70 to 1.35)
-0.05 (-0.67 to 0.57)
0.32 (-0.50 to 1.15)
100.00
3.56
3.124.70
1.43
1.21
4.13
0.96
4.27
1.55
3.56
Weight
0.88
1.03
4.08
1.49
3.48
%
2.86
4.80
4.72
2.44
3.29
1.31
3.54
2.66
1.61
2.56
1.26
2.22
5.79
1.74
0.66
2.75
0.95
2.20
2.683.04
1.77
3.35
2.39
0-4 -2 0 2 4
Males
PM10 Effect (10μg/m3, Lag=0)
NOTE: Weights are from random effects analysis
Overall (I-squared = 38.2%, p = 0.010)
City
Yinchuan
Nanchang
Harbin
Chongqing
Hangzhou
Shenyang
Panzhihua
TangshanSuzhou
Changsha
Zhengzhou
Changde
Anshan
Guilin
Wuhan
Xining
Qingdao
Yuxi
Xuzhou
Guangzhou
Taiyuan
Yantai
Qiqihar
Tongchuan
Qinhuangdao
Liuzhou
Changchun
Zunyi
Zaozhuang
Tianjin
Nanjing
Shanghai
Urumqi
Yichang
Maanshan
Hohhot
ChengduBeijing
0.51 (0.34 to 0.68)
ES (95% CI)
-0.92 (-1.95 to 0.10)
-0.08 (-1.17 to 1.02)
0.29 (-0.27 to 0.85)
0.50 (0.07 to 0.94)
0.32 (-0.56 to 1.20)
0.55 (-0.28 to 1.37)
-0.16 (-2.45 to 2.13)
-0.23 (-1.71 to 1.25)0.99 (0.35 to 1.64)
0.47 (-0.41 to 1.35)
0.67 (-0.04 to 1.37)
0.81 (-0.07 to 1.69)
-0.17 (-1.18 to 0.84)
-0.21 (-2.08 to 1.67)
1.12 (0.58 to 1.66)
0.69 (-0.82 to 2.19)
0.11 (-0.34 to 0.57)
1.28 (-0.47 to 3.03)
-0.05 (-1.27 to 1.17)
1.58 (0.87 to 2.29)
1.30 (0.47 to 2.13)
0.05 (-0.77 to 0.88)
-0.61 (-2.13 to 0.91)
0.92 (-1.16 to 2.99)
0.75 (-0.58 to 2.07)
1.48 (0.24 to 2.72)
0.64 (-0.12 to 1.40)
0.97 (-0.48 to 2.41)
1.80 (0.82 to 2.78)
0.22 (-0.29 to 0.72)
0.53 (-0.08 to 1.14)
0.17 (-0.29 to 0.62)
0.24 (-0.46 to 0.94)
1.01 (-0.61 to 2.63)
2.06 (0.45 to 3.67)
-0.14 (-1.52 to 1.24)
0.43 (-0.24 to 1.09)0.39 (0.10 to 0.67)
100.00
Weight
2.04
1.85
4.37
5.39
2.55
2.79
0.51
1.133.77
2.56
3.41
2.55
2.11
0.74
4.50
1.09
5.22
0.84
1.56
3.40
2.75
2.79
1.08
0.61
1.37
1.53
3.10
1.18
2.19
4.79
4.00
5.18
3.44
0.97
0.98
1.27
3.666.73
%
0.51 (0.34 to 0.68)
ES (95% CI)
-0.92 (-1.95 to 0.10)
-0.08 (-1.17 to 1.02)
0.29 (-0.27 to 0.85)
0.50 (0.07 to 0.94)
0.32 (-0.56 to 1.20)
0.55 (-0.28 to 1.37)
-0.16 (-2.45 to 2.13)
-0.23 (-1.71 to 1.25)0.99 (0.35 to 1.64)
0.47 (-0.41 to 1.35)
0.67 (-0.04 to 1.37)
0.81 (-0.07 to 1.69)
-0.17 (-1.18 to 0.84)
-0.21 (-2.08 to 1.67)
1.12 (0.58 to 1.66)
0.69 (-0.82 to 2.19)
0.11 (-0.34 to 0.57)
1.28 (-0.47 to 3.03)
-0.05 (-1.27 to 1.17)
1.58 (0.87 to 2.29)
1.30 (0.47 to 2.13)
0.05 (-0.77 to 0.88)
-0.61 (-2.13 to 0.91)
0.92 (-1.16 to 2.99)
0.75 (-0.58 to 2.07)
1.48 (0.24 to 2.72)
0.64 (-0.12 to 1.40)
0.97 (-0.48 to 2.41)
1.80 (0.82 to 2.78)
0.22 (-0.29 to 0.72)
0.53 (-0.08 to 1.14)
0.17 (-0.29 to 0.62)
0.24 (-0.46 to 0.94)
1.01 (-0.61 to 2.63)
2.06 (0.45 to 3.67)
-0.14 (-1.52 to 1.24)
0.43 (-0.24 to 1.09)0.39 (0.10 to 0.67)
100.00
Weight
2.04
1.85
4.37
5.39
2.55
2.79
0.51
1.133.77
2.56
3.41
2.55
2.11
0.74
4.50
1.09
5.22
0.84
1.56
3.40
2.75
2.79
1.08
0.61
1.37
1.53
3.10
1.18
2.19
4.79
4.00
5.18
3.44
0.97
0.98
1.27
3.666.73
%
0-4 -2 0 2 4
Females
PM10 Effect (10μg/m3, Lag=0)
Page 48 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
16
Figure SM10. Percentage Change in Daily Number of Deaths per 10- 𝜇𝜇𝑚𝑚/𝑚𝑚3 Increase in Concurrent Day PM10 in 38 Chinese Cities: Elderly vs. Young
Note: Maximum likelihood estimates and 95 percent confidence intervals of the city-specific mean lag PM10 effects on total mortality for cardiorespiratory and non-cardiorespiratory deaths separately, the largest 38 cities in China. The dependent variable is the percentage change in the number of daily deaths for elderly and young people separately.
NOTE: Weights are from random effects analysis
Overall (I-squared = 60.8%, p = 0.000)
Zunyi
Guangzhou
Panzhihua
Tangshan
Shanghai
Beijing
Xuzhou
City
Yinchuan
Maanshan
GuilinHarbinZhengzhou
Changchun
Qingdao
Taiyuan
Chengdu
Shenyang
AnshanYichang
Hohhot
Qinhuangdao
Hangzhou
Xining
Yantai
Qiqihar
Tianjin
Urumqi
Changde
Chongqing
Nanchang
Yuxi
Tongchuan
Zaozhuang
Changsha
Liuzhou
Nanjing
Suzhou
Wuhan
0.50 (0.34 to 0.66)
0.49 (-0.56 to 1.55)
1.80 (1.27 to 2.33)
-1.09 (-2.81 to 0.63)
-0.57 (-1.69 to 0.55)
0.39 (0.05 to 0.73)
0.29 (0.08 to 0.50)
-0.42 (-1.35 to 0.52)
ES (95% CI)
-0.01 (-0.75 to 0.74)
1.73 (0.63 to 2.83)
0.06 (-1.30 to 1.41)0.12 (-0.32 to 0.55)0.19 (-0.34 to 0.73)
0.84 (0.27 to 1.42)
0.22 (-0.12 to 0.56)
0.96 (0.35 to 1.57)
0.27 (-0.28 to 0.81)
0.69 (0.06 to 1.32)
0.34 (-0.41 to 1.10)0.35 (-0.88 to 1.57)
-0.52 (-1.51 to 0.47)
0.06 (-0.92 to 1.03)
0.58 (-0.05 to 1.20)
0.35 (-0.65 to 1.36)
0.44 (-0.19 to 1.07)
-0.61 (-1.82 to 0.61)
0.54 (0.14 to 0.93)
-0.13 (-0.66 to 0.40)
0.99 (0.31 to 1.67)
0.66 (0.33 to 0.99)
0.19 (-0.62 to 1.01)
1.44 (0.11 to 2.77)
0.26 (-1.27 to 1.79)
1.19 (0.39 to 1.99)
0.52 (-0.16 to 1.19)
2.06 (1.12 to 3.00)
0.60 (0.12 to 1.07)
0.72 (0.23 to 1.21)
0.95 (0.54 to 1.36)
100.00
1.62
3.34
0.75
1.49
4.26
4.82
1.90
Weight
2.47
1.53
1.123.793.33
3.16
4.27
3.00
3.28
2.93
2.451.30
%
1.76
1.80
2.93
1.72
2.93
1.32
4.00
3.35
2.72
4.33
2.25
1.15
0.92
2.28
2.74
1.89
3.62
3.54
3.93
0.50 (0.34 to 0.66)
0.49 (-0.56 to 1.55)
1.80 (1.27 to 2.33)
-1.09 (-2.81 to 0.63)
-0.57 (-1.69 to 0.55)
0.39 (0.05 to 0.73)
0.29 (0.08 to 0.50)
-0.42 (-1.35 to 0.52)
ES (95% CI)
-0.01 (-0.75 to 0.74)
1.73 (0.63 to 2.83)
0.06 (-1.30 to 1.41)0.12 (-0.32 to 0.55)0.19 (-0.34 to 0.73)
0.84 (0.27 to 1.42)
0.22 (-0.12 to 0.56)
0.96 (0.35 to 1.57)
0.27 (-0.28 to 0.81)
0.69 (0.06 to 1.32)
0.34 (-0.41 to 1.10)0.35 (-0.88 to 1.57)
-0.52 (-1.51 to 0.47)
0.06 (-0.92 to 1.03)
0.58 (-0.05 to 1.20)
0.35 (-0.65 to 1.36)
0.44 (-0.19 to 1.07)
-0.61 (-1.82 to 0.61)
0.54 (0.14 to 0.93)
-0.13 (-0.66 to 0.40)
0.99 (0.31 to 1.67)
0.66 (0.33 to 0.99)
0.19 (-0.62 to 1.01)
1.44 (0.11 to 2.77)
0.26 (-1.27 to 1.79)
1.19 (0.39 to 1.99)
0.52 (-0.16 to 1.19)
2.06 (1.12 to 3.00)
0.60 (0.12 to 1.07)
0.72 (0.23 to 1.21)
0.95 (0.54 to 1.36)
100.00
1.62
3.34
0.75
1.49
4.26
4.82
1.90
Weight
2.47
1.53
1.123.793.33
3.16
4.27
3.00
3.28
2.93
2.451.30
%
1.76
1.80
2.93
1.72
2.93
1.32
4.00
3.35
2.72
4.33
2.25
1.15
0.92
2.28
2.74
1.89
3.62
3.54
3.93
0-4 -2 0 2 4
Elderly (>=60 years)
PM10 Effect (10μg/m3, Lag=0)
NOTE: Weights are from random effects analysis
Overall (I-squared = 21.0%, p = 0.128)
City
Zhengzhou
Liuzhou
Urumqi
Qiqihar
Nanchang
Zunyi
Tianjin
Qingdao
Harbin
Changchun
Tangshan
Yuxi
Beijing
Wuhan
Chongqing
YichangAnshan
Panzhihua
Hohhot
Taiyuan
Maanshan
Qinhuangdao
Shenyang
Yantai
HangzhouNanjing
Xining
Changsha
ChengduTongchuan
Zaozhuang
Guilin
Changde
Suzhou
Xuzhou
Guangzhou
Yinchuan
Shanghai
0.19 (-0.02 to 0.40)
ES (95% CI)
0.26 (-0.68 to 1.20)
0.65 (-0.91 to 2.21)
0.48 (-0.39 to 1.35)
0.44 (-1.18 to 2.07)
0.30 (-1.19 to 1.79)
0.34 (-1.38 to 2.06)
0.58 (-0.22 to 1.37)
0.11 (-0.60 to 0.81)
0.04 (-0.70 to 0.78)
-0.09 (-1.19 to 1.00)
1.20 (-0.66 to 3.06)
3.16 (0.69 to 5.63)
0.28 (-0.16 to 0.72)
0.01 (-0.83 to 0.85)
-0.34 (-1.04 to 0.35)
0.95 (-1.38 to 3.29)-0.49 (-1.80 to 0.82)
-0.67 (-3.44 to 2.09)
2.22 (0.74 to 3.70)
0.08 (-1.00 to 1.16)
1.62 (-0.97 to 4.21)
1.34 (-0.31 to 2.99)
0.23 (-0.84 to 1.30)
-0.36 (-1.53 to 0.81)
0.02 (-1.58 to 1.62)0.56 (-0.45 to 1.56)
-1.22 (-3.34 to 0.90)
0.62 (-0.65 to 1.88)
0.42 (-0.50 to 1.33)0.16 (-2.17 to 2.49)
0.80 (-0.45 to 2.04)
1.55 (-0.82 to 3.92)
-0.22 (-1.36 to 0.92)
-0.94 (-1.99 to 0.11)
-1.57 (-3.48 to 0.34)
0.90 (-0.23 to 2.04)
-1.58 (-2.89 to -0.27)
-0.06 (-0.77 to 0.66)
100.00
Weight
3.64
1.58
4.04
1.47
1.71
1.33
4.61
5.38
5.02
2.87
1.15
0.68
8.53
4.27
5.45
0.752.14
0.55
1.74
2.94
0.62
1.44
2.98
2.60
1.513.26
0.91
2.27
3.790.75
2.33
0.73
2.69
3.07
1.10
%
2.71
2.14
5.27
0.19 (-0.02 to 0.40)
ES (95% CI)
0.26 (-0.68 to 1.20)
0.65 (-0.91 to 2.21)
0.48 (-0.39 to 1.35)
0.44 (-1.18 to 2.07)
0.30 (-1.19 to 1.79)
0.34 (-1.38 to 2.06)
0.58 (-0.22 to 1.37)
0.11 (-0.60 to 0.81)
0.04 (-0.70 to 0.78)
-0.09 (-1.19 to 1.00)
1.20 (-0.66 to 3.06)
3.16 (0.69 to 5.63)
0.28 (-0.16 to 0.72)
0.01 (-0.83 to 0.85)
-0.34 (-1.04 to 0.35)
0.95 (-1.38 to 3.29)-0.49 (-1.80 to 0.82)
-0.67 (-3.44 to 2.09)
2.22 (0.74 to 3.70)
0.08 (-1.00 to 1.16)
1.62 (-0.97 to 4.21)
1.34 (-0.31 to 2.99)
0.23 (-0.84 to 1.30)
-0.36 (-1.53 to 0.81)
0.02 (-1.58 to 1.62)0.56 (-0.45 to 1.56)
-1.22 (-3.34 to 0.90)
0.62 (-0.65 to 1.88)
0.42 (-0.50 to 1.33)0.16 (-2.17 to 2.49)
0.80 (-0.45 to 2.04)
1.55 (-0.82 to 3.92)
-0.22 (-1.36 to 0.92)
-0.94 (-1.99 to 0.11)
-1.57 (-3.48 to 0.34)
0.90 (-0.23 to 2.04)
-1.58 (-2.89 to -0.27)
-0.06 (-0.77 to 0.66)
100.00
Weight
3.64
1.58
4.04
1.47
1.71
1.33
4.61
5.38
5.02
2.87
1.15
0.68
8.53
4.27
5.45
0.752.14
0.55
1.74
2.94
0.62
1.44
2.98
2.60
1.513.26
0.91
2.27
3.790.75
2.33
0.73
2.69
3.07
1.10
%
2.71
2.14
5.27
0-4 -2 0 2 4
Young (<60 years)
PM10 Effect (10μg/m3, Lag=0)
Page 49 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nlyFigure SM11. City-Specific Air Pollution Estimates and Average PM10 Concentrations
Note: The figure plots estimated effect size over the mean PM10 concentrations. The solid line is a linear regression.
Anshan
Beijing
Changchun ChangdeChangsha
Chengdu
Chongqing
Guangzhou
Guilin
Hangzhou
Harbin
Hohhot
Liuzhou Maanshan
Nanchang
Nanjing
Panzhihua
Qingdao
Qinhuangdao
Qiqihar
Shanghai
Shenyang
Suzhou
Taiyuan
Tangshan
Tianjin
Tongchuan
Urumqi
Wuhan
Xining
Xuzhou
Yantai
Yichang
Yinchuan
Yuxi
Zaozhuang
Zhengzhou
Zunyi
-1
-.5
0
.5
1
1.5
2
Estim
ated
Coe
ffici
ent
70 80 90 100 110 120 130 140 150PM10 Mean Concentration (μg/m3)
Estimated Effect Fitted Line 95% CI
Page 50 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
18
Figure SM12. City-Specific Air Pollution Estimates and Average PM10 Concentrations: North vs. South
Note: The figure plots estimated effect size over the mean PM10 concentrations. The solid dots are northern cities. The hollow triangles are southern cities. The solid line is a linear regression for northern cities. The dashed line is for southern cities.
Anshan
Beijing
Changchun
Harbin
HohhotQingdao
Qinhuangdao
Qiqihar
ShenyangTaiyuan
Tangshan
Tianjin
Tongchuan
UrumqiXining
Xuzhou
Yantai
Yinchuan
Zaozhuang
Zhengzhou
ChangdeChangsha
Chengdu
Chongqing
Guangzhou
Guilin
Hangzhou
Liuzhou Maanshan
Nanchang
Nanjing
Panzhihua
Shanghai Suzhou
Wuhan
Yichang
Yuxi
Zunyi
-1
-.5
0
.5
1
1.5
2
Estim
ated
Coe
ffici
ent
70 80 90 100 110 120 130 140 150PM10 Mean Concentration
Estimated Effect, North Linear Fit, NorthEstimated Effect, South Linear Fit, South
Page 51 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review O
nly
19
Figure SM13. Socio-economic Factors and Marginal Pollution Effects
-1
-.5
0
.5
1
1.5
2
Estim
ated
Coe
ffici
ent
1 2 3 4 5 6 7 8GDP Per Capita (10,000 Yuan)
Estimated Effect Fitted Line 95% CI
-1
-.5
0
.5
1
1.5
2
Estim
ated
Coe
ffici
ent
0 .02 .04 .06 .08 .1 .12 .14 .16 .18Share of Workers Employed in Construction Industry
Estimated Effect Fitted Line 95% CI
Page 52 of 52
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960