Review
Ambient air pollution and depression: A systematic review with metaanalysis
up to 2019
Shu-Jun Fan a
, Joachim Heinrich b,c,d
, Michael S. Bloom e
, Tian-Yu Zhao b,c,f
, Tong-Xing Shi a
, Wen-Ru Feng a
,
Yi Sun a
, Ji-Chuan Shen a
, Zhi-Cong Yang a,⇑
, Bo-Yi Yang g,⇑
, Guang-Hui Dong g,⇑
a
Guangzhou Center for Disease Control and Prevention, Guangzhou 510440, China
b
Institute and Clinic for Occupational, Social and Environmental Medicine, University Hospital, LMU Munich, Ziemssenstrabe 1, 80336 Munich, Germany
c
Comprehensive Pneumology Center Munich, German Center for Lung Research, Ziessenstabe 1, 80336 Munich, Germany
d
Allergy and Lung Health Unit, Melbourne School of Population and Global Health, The University of Melbourne, Melbourne, Australia
e
Departments of Environmental Health Sciences and Epidemiology and Biostatics, University at Albany, State University of New York, Rensselaer, NY 12144, USA
f
Institute of Epidemiology, Helmholtz Zentrum München – German Research Center for Environmental Health, Neuherberg 85764, Germany
g
Guangzhou Key Laboratory of Environmental Pollution and Health Risk Assessment, Guangdong Provincial Engineering Technology Research Center of Environmental and Health
Risk Assessment, Department of Occupational and Environmental Health, School of Public Health, Sun Yat-sen University, Guangzhou 510080, China
h i g h l i g h t s
 A systematic review identified 22
studies from 10 countries of the
world.
 Short-term exposure to NO2 was
associated with an increased odds of
depression.
 Long-term PM2.5, PM10, and NO2
exposure was not associated with
depression.
 Short-term PM2.5, PM10, SO2, and O3
exposure was not associated with
depression.
g r a p h i c a l a b s t r a c t
English databases
Chinese databases
Identification records
(N = 14938)
Screening records
(N = 10896)
Eligibility articles
(N = 19)
Included studies
(N = 22)
No. of
studies
I2 IVhet model,
OR (95% CI)
Long-term
PM2.5 12 51.6 1.12 (0.97, 1.29)
PM10 8 85.7 1.04 (0.88, 1.25)
NO2 7 83.6 1.05 (0.83, 1.34)
Short-term
PM2.5 6 51.6 1.01 (0.99, 1.04)
PM10 5 86.7 1.01 (0.98, 1.04)
NO2 7 65.4 1.02 (1.00, 1.04)
SO2 6 71.2 1.03 (0.99, 1.07)
O3 7 82.2 1.01 (0.99, 1.03)
a r t i c l e i n f o
Article history:
Received 14 July 2019
Received in revised form 27 September
2019
Accepted 27 September 2019
Available online 31 October 2019
Editor: Scott Sheridan
Keywords:
Particulate matter
Gaseous pollutants
Depressive disorder
Random effects model
Inverse Variance Heterogeneity (IVhet)
a b s t r a c t
Although epidemiological studies have evaluated the associations of ambient air pollution with depression,
the results remained mixed. To clarify the nature of the association, we performed a comprehensive
systematic review and meta-analysis with the Inverse Variance Heterogeneity (IVhet) model to estimate
the effect of ambient air pollution on depression. Three English and four Chinese databases were searched
for epidemiologic studies investigating associations of ambient particulate (diameter 2.5 lm (PM2.5),
10 lm (PM10)) and gaseous (nitric oxide (NO), nitrogen dioxide (NO2), carbon monoxide (CO), sulfur
dioxide (SO2) and ozone (O3)) air pollutants with depression. Odds ratios (OR) and corresponding 95%
confidence intervals (CI) were calculated to evaluate the strength of the associations. We identified 22
eligible studies from 10 countries of the world. Under the IVhet model, per 10 mg/m3
increase in longterm
exposure to PM2.5 (OR: 1.12, 95% CI: 0.97–1.29, I2
: 51.6), PM10 (OR: 1.04, 95% CI: 0.88–1.25, I2
:
85.7), and NO2 (OR: 1.05, 95% CI: 0.83–1.34, I2
: 83.6), as well as short-term exposure to PM2.5 (OR:
1.01, 95% CI: 0.99–1.04, I2
: 51.6), PM10 (OR: 1.01, 95% CI: 0.98–1.04, I2
: 86.7), SO2 (OR: 1.03, 95% CI:
0.99–1.07, I2
: 71.2), and O3 (OR: 1.01, 95% CI: 0.99–1.03, I2
: 82.2) was not significantly associated
https://doi.org/10.1016/j.scitotenv.2019.134721
0048-9697/Ó 2019 Elsevier B.V. All rights reserved.
⇑ Corresponding authors at: Guangzhou Center for Disease Control and Prevention, No. 1 Qide Road, Baiyun Area, Guangzhou 510440, China (Z.-C. Yang). Guangzhou Key
Laboratory of Environmental Pollution and Health Risk Assessment, Guangdong Provincial Engineering Technology Research Center of Environmental and Health Risk
Assessment, Department of Occupational and Environmental Health, School of Public Health, Sun Yat-sen University, 74 Zhongshan 2nd Road, Yuexiu District, Guangzhou
510080, China (B.-Y. Yang and G.-H. Dong).
E-mail addresses: yangzc@gzcdc.org.cn (Z.-C. Yang), yangby23@mail.sysu.edu.cn (B.-Y. Yang), donggh5@mail.sysu.edu.cn (G.-H. Dong).
Science of the Total Environment 701 (2020) 134721
Contents lists available at ScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
model with depression. However, we observed significant association between short-term NO2 exposure
(per 10 mg/m3
increase) and depression (OR: 1.02, 95% CI: 1.00–1.04, I2
: 65.4). However, the heterogeneity
was high for all of the pooled estimates, which reduced credibility of the cumulative evidence.
Additionally, publication bias was detected for six of eight meta-estimates. In conclusion, short-term
exposure to NO2, but not other air pollutants, was significantly associated with depression. Given the
limitations, a larger meta-analysis incorporating future well-designed longitudinal studies, and
investigations into potential biologic mechanisms, will be necessary for a more definitive result.
Ó 2019 Elsevier B.V. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1. Search strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2. Selection criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.3. Data extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.4. Methodological quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.5. Risk of bias assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.6. Standardization of data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.7. Meta-analysis methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3. Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.1. Literature retrieval and study characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.2. Study quality and risk of bias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.3. Long-term air pollution exposure and depression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.4. Short-term air pollution exposure and depression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4. Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.1. Principal findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.2. Comparison with other studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.3. Pathophysiological mechanisms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5. Strengths and limitations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
6. Conclusions and future perspectives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Declaration of Competing Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1. Introduction
Ambient or outdoor air pollution (referred to as air pollution
hereafter) has become a major global environmental issue. According
to the recent Global Burden of Diseases report (GBD), air pollution
was responsible for 4.90 million deaths and 1.47 billion
disability-adjusted life-years (DALYs) in 2017, with most of the
burden related to cardiovascular disease, respiratory disease, and
lower respiratory infections (GBD 2017 Risk Factor Collaborators,
2018). More recently, the hazardous effects of air pollution on
mental health, such as depression, have attracted interest and have
global public health implications (Buoli et al., 2018; Gu et al., 2019;
Pun et al., 2017; Kim et al., 2016; Lam et al., 2016).
Depression is characterized by persistent poor mood, diminished
interest in activities, exhaustion, and low energy, and it is
one of the most prevalent mental disorders (American Psychiatric
Association, 2013). Depression has been associated with decreased
work productivity, lower quality of life, and an increased risk of allcause
mortality (Walker et al., 2015; Ferenchick et al., 2019;
Spitzer et al., 1995). The GBD (2018) report estimated more than
43 million years lived with disability (YLDs) attributable to depression
in 2017 (GBD 2017 Disease and Injury Incidence and
Prevalence Collaborators, 2018). Depression prevalence is increasing,
and access to effective treatments remains limited (GBD 2017
Disease and Injury Incidence and Prevalence Collaborators, 2018;
Ramanuj et al., 2019). Therefore, identification of risk factors for
depression and consequent development of prevention strategies
is of public health significance.
Mechanistic studies indicate that air pollutants inhalation can
trigger neuro-inflammation and oxidative stress as well as induce
dopaminergic neurotoxicity (Risom et al., 2005; Ng et al., 2008;
Hurley and Tizabi, 2013; Dantzer et al., 2008). Additionally, previous
studies demonstrated that exposure to higher levels of air pollution
could affect people’s residential satisfaction and selfperceived
health, which have been associated with mental health
(von Lindern et al., 2016; Liu et al., 2018; Nguyen et al., 2017).
Therefore, it has been plausibly hypothesized that air pollution
may contribute to depression pathogenesis. During the past decade,
multiple studies investigated the relationship between longterm
or short-term air pollution exposure and depression
(Szyszkowicz et al., 2009; Lim et al., 2012; Wang et al., 2014;
Cho et al., 2014; Zijlema et al., 2016; Szyszkowicz et al., 2016;
Kim et al., 2016; Vert et al., 2017; Pun et al., 2017; Lin et al.,
2017a,b; Kioumourtzoglou et al., 2017; Kim and Kim, 2017;
Wang et al., 2018; Zock et al., 2018; Roberts et al., 2019; Zhao
et al., 2019). However, the results were inconsistent and contradictory;
some studies detected positive air pollution-depression associations
(Lim et al., 2012; Cho et al., 2014; Kim et al., 2016; Vert
et al., 2017; Roberts et al., 2019) and the others found no association
(Wang et al., 2014; Lin et al., 2017a; Kioumourtzoglou et al.,
2017; Kim and Kim, 2017; Zhao et al., 2019). A recent metaanalysis
reported a higher odds ratio for depression in association
with increasing ambient fine particulate matter (PM2.5) but not
with inhalable particulate matter (PM10) exposure. However, the
meta-analysis did not assess gaseous pollutants, and several key
or recent papers were not included (Gu et al., 2019). Thus, the
2 S.-J. Fan et al. / Science of the Total Environment 701 (2020) 134721
weight of evidence was difficult to interpret, limiting the data
available to provide specific suggestions to policy makers. In addition,
due to high heterogeneity, Gu et al. (2019) applied random
effects model to pool the effect estimates. However, the model
was criticized for overestimating effects and underestimating the
statistical error (Doi et al., 2015).
To help address the pending data gap, we systematically identified
and reviewed epidemiological studies of short- and long-term
exposure to ambient particulate and gaseous air pollutants, including
all epidemiologic study design, up to 2019. Furthermore, we
used the Inverse Variance Heterogeneity (IVhet) model to pool
the data, which outperforms the random effects model in reducing
the estimator mean squared error and maintaining the correct coverage
probability of the confidence interval (CI) (Doi et al., 2015).
Our aim was to provide a more comprehensive and accurate
assessment of the state of the literature to inform policy makers,
investigators, and healthcare professionals on likely magnitude of
effects and to recommend next steps for more definitively addressing
the association between air pollutants and depression.
2. Methods
2.1. Search strategy
We conducted the systematic review and meta-analysis according
to the Preferred Reporting Items for Systematic reviews and
Meta-Analysis (PRISMA) guidelines as shown in Table S1 (Moher
et al., 2009). We did not publish a systematic review protocol
beforehand. A specific research question was formulated according
to the ‘‘Participants”, ‘‘Exposure”, ‘‘Comparator”, and ‘‘Outcomes”
(PECO) framework. The focused question of the present systematic
review and meta-analysis was: ‘‘Is exposure to ambient air pollution
associated with the risk of depression?”. Studies on the relationships
of ambient air pollution and depression (published
before 21 August 2019) indexed in three English databases
(PubMed, Web of Science and Scopus) and four Chinese databases
(China National Knowledge Infrastructure, China Biological Medicine
Database, Chongqing VIP Chinese Science and Technology
Periodical Database, and Wanfang Data), were identified. Our
search strategies were based on combinations of air pollution
terms (‘‘air pollution”, ‘‘particulate matter”, ‘‘air pollutants”, particle,
‘‘sulfur dioxide”, ‘‘nitrogen oxide”, ‘‘nitrogen dioxide”, ozone,
and ‘‘carbon monoxide”) and depression terms (depress*, depression,
depressive, depress, depressed, ‘‘unipolar disorder”, and
‘‘bipolar disorder”). Detailed search strategy is shown in Table S2.
The search was limited to the English and Chinese languages. We
also searched the references lists of eligible articles and grey literature
databases (British National Bibliography for Report Literature,
Social Care Online, System for Information on Grey
Literature in Europe, and National Technical Information Service)
to find additional potentially pertinent studies.
2.2. Selection criteria
The following a priori eligibility criteria were based on the PECO
framework: (P) the study was conducted among humans; (E) the
study focused on short-term (<30 days) or long-term (!30 days)
exposures to ambient air pollution; (C) the study provided quantitative
effect estimates with 95% CIs (or standard errors) by comparing
humans exposed to lower air pollution levels with the
greater exposed humans; (O) depression was assessed and defined
using questionnaire (e.g. Center for Epidemiological Studies
Depression Scale, Diagnostic and Statistical Manual of Mental
Disorders, Hospital Anxiety Depression Scale, and Korean version
of the Geriatric Depression Scale, etc.), clinical assessment (International
Classification of Diseases (ICD), International Classification
of Primary Care (ICPC), and doctor-diagnosed), or
antidepressant medication. If more than one publication was identified
for the same study population, the one with more thorough
reporting or adjustments was included.
After removing duplicates, two (SF and WF) of the co-authors
independently screened titles, abstracts, and full text for eligibility
(Fig. 1). If a disagreement arose, the article was referred to a third
co-author (BY) for adjudication.
2.3. Data extraction
Two (SF and YS) of the co-authors independently extracted the
following information from each included study: authors’ names,
publication year, study period, setting, and design, the size, exposure
assessment strategy and lag pattern, outcome (depression)
definition, and effect estimates including odds ratio (OR), risk ratio
(RR), and hazard ratio (HR) with their corresponding 95% CIs. For
studies reporting sub-stratified risk estimates rather than overall
risk estimates, each stratified risk estimate was considered as an
independent data set (Szyszkowicz et al., 2016). Any conflict was
resolved by discussion with a third co-author (BY).
2.4. Methodological quality assessment
We used the popular Newcastle Ottawa scale (NOS) (Wells
et al., 2010) and the Joanna Briggs Institute (JBI) meta-analysis of
statistics assessment and review instrument (JBI, 2016) to assess
the quality of cohort and cross-sectional studies, respectively. Each
study was assessed an NOS score from 0 to 9 and a JBI score from 0
to 20. An NOS score greater than 7 or a JBI score greater than 16
was considered as ‘‘high quality”; otherwise, the study was considered
as ‘‘low quality”.
To the best of our knowledge, no validated scale is available to
assess the quality of time-series, panel, and case-crossover study
designs. We therefore assessed their quality using a scale suggested
by Mustafic et al. (2012), which was adapted from the validated
NOS scale and the Cochrane risk of bias tool (Higgins et al.,
2011). Mustafic´’s adapted scale included three items, and ‘‘0–3”
points were assigned for the following elements: validation
of depression (0: absence of valid criteria; 1: presence of valid
Records identified through database searching (n = 14,938)
Records after duplicates removed (n = 10,896)
Records screened (n=10,896) Clearly irrelevant based on title and/or
abstract (n = 10,817)
Full-text articles assessed
for eligibility (n = 51)
Full-text articles excluded (n = 32)
(1) Irrelevant (n = 10)
(2) No sufficient data for pooling (n=15)
(3) Duplicate data (n = 1)
(4) Other diseases (n = 6)
Studies (estimates) included
in meta-analysis (n = 22)
Articles included in
qualitative synthesis (n = 19)
Fig. 1. Flowchart of study selection.
S.-J. Fan et al. / Science of the Total Environment 701 (2020) 134721 3
criteria), the quality of air pollutant measurements (0: measurements
were not performed at least daily or >25% data were missing;
1: measurements were performed at least daily and missing
data < 25%), and the extent of confounder adjustments (0: no
adjustment was made for long-term trends, season, and air temperature;
1: only long-term trends, season, and air temperature
were adjusted; 2: adjustment was made either for humidity or
day of week in addition to the corresponding adjustments with a
score of 1; 3: adjustment was made for holidays in addition to
the corresponding adjustments with a score of 2) (Mustafic et al.,
2012). If the maximum score was achieved for all the three items
(i.e., 5 points in total), the study was categorized as ‘‘high quality”;
otherwise, the study was categorized as ‘‘low quality”.
2.5. Risk of bias assessment
We further assessed the risk of bias (ROB) for each study, which
is a new concept in environmental health study and related to but
different from the methodological quality assessment. ROB was
defined as characteristics of a study that can introduce systematic
errors in effect estimate (the magnitude or direction) (Woodruff
and Sutton, 2014). A study conducted with high methodological
standards may still have high ROB, which will ultimately influence
the magnitude or direction of an association. To evaluate ROB for
cohort and cross-sectional studies, we applied the National Institutes
of Environmental Health Sciences National Toxicology Program
Office of Health Assessment and Translation (OHAT) ROB
(OHAT, 2015). In the absence of a validated ROB tool for timeseries
and case-crossover studies, we evaluated the ROB of such
studies according to the OHAT tool as well as using the University
of California at San Francisco (UCSF) Navigation Guide (Lam et al.,
2016; Woodruff and Sutton, 2014).
We used the OHAT tool and UCSF Navigation Guide to assess the
selected studies for key (exposure assessment, outcome assessment,
and confounding bias) and other methodologic criteria
(selection bias, attrition/exclusion bias, selective reporting bias,
conflict of interest, and other sources of bias). We tailored adapted
criteria for the ROB assessment of each study to our specific systematic
review, which is provided in Table S3. For example, in
the ‘‘confounding bias” of the key criteria, we extracted all adjustments
from the individual studies and developed a directed acyclic
graph (DAG) to select the most parsimonious confounders for
adjusting and potential mediators. Then, we ranked the risk of confounding
bias according to the selected covariates and mediators.
The potential confounders selected using DAG included meteorological
factors (e.g. temperature, relatively humidity, barometric
pressure, sunlight hours, and wind speed), age, sex, ethnicity,
household income, smoking, physical activity, day of week, season,
urbanity, population density, region, occupation, domestic fuel
type and ventilation, social-economic status, and the spent outside
and time spent in front of a screen. The potential mediators
included triglycerides levels, health status, social satisfaction, sleep
difficulties, and SBP (Fig. S1). The selected studies were classified as
‘‘low”, ‘‘probably low”, ‘‘probably high”, or ‘‘high” risk levels for
each of these domains. Following OHAT tool guidelines, we
excluded studies from the systematic review that were classified
as ‘‘high” or ‘‘probably high” risk on the key criteria and most of
the other criteria.
2.6. Standardization of data
We summarized measures of association between air pollutant
and depression as ORs. To facilitate a comparison of effect sizes
from the different studies, we standardized the effect estimate
units across studies to a 10 mg/m3
change in air pollutant concentrations
prior to pooling. Other reported quantities were first converted
into mg/m3
as: (a) 1 ppm = 1000 ppb; (b) nitrogen dioxide
(NO2): 1 ppb = 46/22.4 mg/m3
; (c) sulfur dioxide (SO2):
1 ppb = 64/22.4 mg/m3
; and (d) ozone (O3): 1 ppb = 48/22.4 mg/m3
.
We calculated the standardized OR for each study as (Kim et al.,
2018):
ORStandardized ¼ e
lnðOROriginÞ
IncrementOrigin
ÂIncrementStandardized
 
For short-term exposure studies, authors used different lag patterns
to evaluate immediate and delayed effects of ambient air pollutants
exposure on odds for depression. Some studies provided
multiple estimates for single-day lags (e.g., lag 0, 1, 2 days), while
others provided cumulative lags (e.g., lag 0–7 days). To facilitate
pooling across studies, we selected lags based on the following criteria:
(a) if only one lag estimate was provided, the estimate was
used; or (b) if multiple lags were provided, in order of precedence
we chose the lag that the investigators focused on or stated as a priority,
the lag that was statistically significant, or the lag with the
largest effect estimates (Atkinson et al., 2012; Yang et al., 2018).
Most studies did not perform or report results from multipollutant
models. Therefore, we only extracted and pooled effect
estimates generated from single-pollutant models. When studies
provided results from several nested adjusted models or sensitivity
analyses, we only chose results from the ‘‘main model” designated
by the investigators.
2.7. Meta-analysis methods
We retrieved effect estimates (OR, RR, or HR) and 95% CI for
associations between each air pollutant and depression in the
included studies. Most studies reported ORs, thus we used it as
measure of association across all studies. Since depression was rare
(the prevalence was approximately 4.4%, WHO, 2017), we considered
OR as equivalent to RR and HR (Eze et al., 2015). Betweenstudy
heterogeneity was tested by calculating I2
(I2
= 0–25% represents
no heterogeneity; I2
= 25–50% represents moderate heterogeneity;
I2
= 50–75% represents large heterogeneity; I2
= 75–100%
represents extreme heterogeneity). If I2
< 50%, the MantelHaenszel
fixed effects model was used; otherwise, the IVhet model
(Doi et al., 2015, 2017) was used. We also reported the pooled
effects from the random effects model. However, the IVhet model
outperforms the random effects model, which favours larger studies,
retains a correct coverage probability, and exhibits a lower
observed variance, regardless of heterogeneity (Doi et al., 2015).
We performed sensitivity analysis to test the stability of the overall
estimate by examining the influence of excluding each study. Additionally,
univariate meta-regression was performed to explore the
source of heterogeneity for meta-analysis with !10 studies
included (Higgins and Green, 2011). Potential moderators including
study location (Europe, Asia, North America, and Mixed areas),
study design (cross-sectional and cohort), age (mean/median age
of participants !45 years, and <45 years), background PM2.5 level
(mean/median of background PM2.5 levels !15 mg/m3
, and
<15 mg/m3
), PM2.5 levels measurement method (monitoring station
and models), depression definition (questionnaire, ICD code, ICPC
code, and use antidepressants), gender proportion (male proportion
!50%, and <50%), sample size (studies with !median number
of participants, and <median number), and type of effect estimate
(OR and HR). Publication bias was examined using Doi plot and the
Luis Furuya-Kanamori (LFK) index. LFK index < |1| indicates ‘‘no
asymmetry”, LFK index between |1| and |2| indicates ‘‘minor asymmetry”,
and LFK index > |2| indicates ‘‘major asymmetry” (MetaXL
User Guide, www.epigear.com).
We graded the overall quality of the pooled evidence according
to the Grading of Recommendations Assessment, Development,
and Evaluation (GRADE) Working Group guideline (Morgan et al.,
4 S.-J. Fan et al. / Science of the Total Environment 701 (2020) 134721
2019). The ROB (e.g., bias from exposure assessment, outcome
assessment and confounding, selection bias, attrition/exclusion
bias, selective reporting bias, conflict of interest), inconsistencies
(high heterogeneity and disparate results across studies), indirect
evidence (the evidence cannot directly answer the research question),
imprecision (e.g., small sample size, wide CI), and publication
bias (assessed using LFK index) are considered. All studies were
started at the same ‘‘high quality” rating regardless of study design.
Credibility of evidence was finally classified as ‘‘high”, ‘‘moderate”,
‘‘low”, and ‘‘very low”.
All statistical analyses were performed using the STATA package
version 11.0 software program (StataCorp, College Station, TX, USA)
and MetaXL v.5.3 software (EpiGear International Pty Ltd, Sunrise
Beach, Queensland, Australia, www.epigear.com), and a twotailed
P-value less than 0.05 was defined as statistically significant.
3. Results
3.1. Literature retrieval and study characteristics
After removing duplicates, we were left with 10,896 papers for
screening on titles and abstracts levels (Fig. 1). We retrieved the
full text of 51 relevant papers. Finally, we retained 19 papers
describing the results of 22 studies (or data sets), that met our
inclusion criteria, in the meta-analysis.
Table 1 summarizes the baseline characteristics of the 22 studies
in our retained papers. Seven of the 22 studies were performed in
Asia, five in North America, nine in Europe, and one in multiple countries.
Most studies adopted a cohort design (n = 10), followed by
cross-sectional (n = 7), case-crossover (n = 3), time-series (n = 1),
and panel (n = 1) designs. Thirteen studies investigated the longterm
effects of air pollution on depression, and the sample sizes
for these studies ranged from n = 284 to n = 354,827. Seven studies
investigated the short-term effects and their sample sizes ranged
from n = 537 to n = 69,132. Two studies explored both long-term
and short-term effects. Of the 22 studies, 20 were conducted among
healthy populations (Szyszkowicz et al., 2009; Lim et al., 2012;
Wang et al., 2014; Zijlema et al., 2016; Szyszkowicz et al., 2016;
Kim et al., 2016; Vert et al., 2017; Pun et al., 2017; Lin et al.,
2017a,b; Kioumourtzoglou et al., 2017; Kim and Kim, 2017; Wang
et al., 2018; Roberts et al., 2019; Zhao et al., 2019; Zhang et al.,
2019; Klompmaker et al., 2019), while the remaining two studies
were based on specific populations with cardiovascular diseases,
diabetes mellitus, or asthma (Cho et al., 2014; Zock et al., 2018).
For exposure assessment, nine studies used data from fixed air
monitoring stations to assess air pollution exposure (Szyszkowicz
et al., 2009; Lim et al., 2012; Wang et al., 2014; Cho et al., 2014;
Szyszkowicz et al., 2016; Kim et al., 2016; Lin et al., 2017b; Kim
and Kim, 2017; Wang et al., 2018), twelve studies used air pollution
data predicted by models, such as land use regression (LUR),
and spatiotemporal models (Zijlema et al., 2016; Vert et al.,
2017; Pun et al., 2017; Lin et al., 2017a; Kioumourtzoglou et al.,
2017; Zock et al., 2018; Roberts et al., 2019; Zhang et al., 2019;
Klompmaker et al., 2019), and the remaining one study used monitoring
stations data for short-term exposure and LUR model for
long-term exposure (Zhao et al., 2019). For outcome assessment,
five and one studies used the ICD code and the ICPC code, respectively.
Thirteen studies used depression scales or interviews, and
the remaining three studies used doctor-diagnosed depression
and/or antidepressant medication use.
3.2. Study quality and risk of bias
According to the NOS and JBI scales (Table S4), seven cohort studies
(Wang et al., 2014; Zijlema et al., 2016; Kim et al., 2016; Pun
et al., 2017; Kioumourtzoglou et al., 2017; Roberts et al., 2019;
Zhang et al., 2019), seven cross-sectional studies (Vert et al., 2017;
Lin et al., 2017a,b; Kim and Kim, 2017; Zock et al., 2018; Zhao
et al., 2019; Klompmaker et al., 2019), and one case-crossover study
(Wang et al., 2018) were regarded as ‘‘high quality”, whereas none
of the time-series, panel, or the remaining two case-crossover studies
were ‘‘high quality” according to Mustafic´’s criteria.
With respect to the study ROB assessment, none of the 19 articles
presented a high risk of bias. The detailed account of each
study’s ROB assessment is provided in Table S4.
3.3. Long-term air pollution exposure and depression
Pooled effect estimates examining the associations between
long-term air pollution exposure and depression are illustrated in
Tables 2 and S5. Twelve studies investigated PM2.5 and depression.
Under the IVhet model, exposure to PM2.5 was not significantly
associated with depression (OR = 1.12, 95% CI = 0.97–1.29;
I2
= 51.6%) (Fig. 2). However, the association was significant under
the random effects model (Table S5). Doi plot with LKF index indicates
major asymmetry (Fig. S2). When any single study or some
specific studies were excluded (Zock et al., 2018; Roberts et al.,
2019), the pooled estimates were not materially changed
(Figs. S3 and S4). For example, when Zock et al. (2018) and
Roberts et al. (2019) were excluded at the same time, the pooled
OR for long-term PM2.5 exposure was 1.11 (0.99, 1.24) (Fig. S4).
Meta-regression results showed that study location (P = 0.299),
study design (P = 0.983), age (P = 0.777), background PM2.5 level
(P = 0.938), PM2.5 exposure measurement methods (P = 0.128),
depression definition (P = 0.686), gender ratio (P = 0.420), sample
size (P = 0.234), and type of effect estimate (P = 0.740) were not
the source of the between-study heterogeneity.
The associations between long-term PM10 and NO2 exposure
with depression were estimated by eight and seven studies,
respectively (Table 2). None of the associations was significant
either under the IVhet model (PM10: OR = 1.04, 95% CI = 0.88–
1.25, Fig. S5; NO2: OR = 1.05, 95% CI = 0.83–1.34, Fig. 2) or the random
effects model (PM10: OR = 1.06, 95% CI = 0.94–1.20; NO2:
OR = 1.12, 95% CI = 0.99–1.28; Table S5). The between-study
heterogeneities were extreme for both PM10 (I2
= 85.7%) and NO2
(I2
= 83.6%). The Doi plot showed major asymmetry for PM10 and
NO2 (Figs. S6 and S7). In sensitivity analyses, when any single study
or some specific studies were excluded (Zock et al., 2018; Roberts
et al., 2019), the pooled estimates were not materially changed,
indicating the robustness of the pooled estimates (Figs. S8–S10).
For example, when Zock et al. (2018) and Roberts et al. (2019) were
excluded at the same time, the pooled OR for long-term NO2 exposure
was 1.05 (0.73, 1.51) (Fig. S10). Due to the limited number of
studies, we did not perform meta-regression analyses for them.
Additionally, our confidence in the cumulative evidence was ‘‘very
low” or ‘‘low” for the pooled associations of long-term PM2.5, PM10,
and NO2 with depression based on the GRADE system (Table S6).
The associations of PM2.5absorbance and NOx with depression
were investigated by one study each, and both reported positive
associations (Vert et al., 2017). Two studies looked at O3 and
depression, but the results were mixed (Kioumourtzoglou et al.,
2017; Zhao et al., 2019). Due to limited data, we did not perform
meta-analyses for PM2.5absorbance, NOx, or O3 exposure.
3.4. Short-term air pollution exposure and depression
Pooled effect estimates describing the associations between
short-term air pollution exposure and depression are summarized
in Tables 2 and S5. Six, five, seven, six, and seven studies focused on
short-term exposure to PM2.5, PM10, NO2, SO2, and O3, respectively
(Figs. 3 and S11–S13). The pooled results under the IVhet model
S.-J. Fan et al. / Science of the Total Environment 701 (2020) 134721 5
Table 1
Characteristics of the studies included in the meta-analysis.
Authors (publication
year)
Study Location Study Design Study Participants Studied Pollutants Exposure duration Exposure
assessment
Outcome
definition
Pollutants (increment)
and estimates
Szyszkowicz et al.
(2009)
Canada (North America) Time-series Emergency visits for
depression from six cities in
Canada (n = 27, 047)
PM2.5, PM10, CO, NO2, SO2,
O3
Short-term National Air
Pollution
Surveillance
system
(Monitoring
station)
ICD-9, rubic-296
or 311
% change in RR
PM10 (19.4 mg/m3
):
%RR: 6.4 (3.6, 9.4);
PM2.5 (8.3 mg/m3
):
%RR: 2.3 (-0.2, 4.7);
NO2 (20.1 ppb):
%RR: 10.0 (6.6, 13.6);
SO2 (4.6 ppb):
%RR: 2.6 (-0.1, 5.3);
O3 (18.9 ppb):
%RR: À4.0 (-7.3, À0.6)
Lim et al. (2012) Korea (Asia) Panel Participants who visited a
community center for the
elderly located in Seoul, Korea
(2008À2010), with an
average age of 71 years
(n = 537)
PM10, SO2, NO2, O3 Short-term Monitoring station SGDS-K % change in RR
PM10 (24.2 mg/m3
):
%RR: 17.0 (4.9, 30.5);
NO2 (15 ppb):
%RR: 32.8 (12.6, 56.6);
SO2 (2.3 ppb):
%RR: À20.0 (-36.6, 0.9);
O3 (37.0 ppb):
%RR: 43.7 (11.5, 85.2)
Wang et al. (2014) United States (North
America)
Cohort Participants from the
MOBILIZE Boston Study
(2005–2008) aged greater
than 65 years (n = 732).
PM2.5, NO2, O3, CO, NO Short-term Harvard School of
Public Health
Stationary
ambient
monitoring site
CESD-R PM2.5 (3.40 mg/m3
):
OR: 0.67 (0.46, 0.98);
NO2 (4.07 ppb):
OR: 1.32 (0.99, 1.76);
O3 (13.45 ppb):
OR: 0.71 (0.46, 1.09)
Cho et al. (2014) Korea (Asia) Case-
crossover
Patients (cardiovascular
disease, diabetes mellitus, or
asthma patients) visited the
emergency department for
depression, with a mean age
of 44 years (n = 4985)
PM10, SO2, NO2, O3, CO Short-term Monitoring
stations
ICD-10. F32 PM10 (36.7 mg/m3
):
OR: 1.12 (1.07, 1.18);
NO2 (12.04 ppb):
OR:1.082 (1.03, 1.13);
SO2 (2.33 ppb):
OR: 1.103 (1.043, 1.166);
O3 (10.04 ppb):
OR: 1.059 (0.995, 1.127)
Szyszkowicz et al.
(2016)
Canada (North America) Case-
crossover
People who visited hospital
emergency departments for
depression (n = 118, 602)
PM2.5, SO2, NO2, O3 Short-term Environment
Canada
(Monitoring
stations)
ICD-F32/F33 Males
PM.2.5 (7.12 mg/m3
):
OR: 1.02 (1.00, 1.05);
NO2 (9 ppb):
OR: 1.015 (0.99, 1.04);
SO2 (2.5 ppb):
OR: 1.02 (0.997, 1.04);
O3 (14.5 ppb):
OR: 1.04 (1.01, 1.07);
Females
PM2.5 (7.12 mg/m3
):
OR: 1.01(0.996, 1.03);
NO2 (9 ppb):
OR: 1.03 (1.00, 1.05);
SO2 (2.5 ppb):
OR: 1.02 (1.01, 1.04);
O3 (14.5 ppb):
OR: 1.06 (1.03, 1.08)
Zijlema et al. –
LifeLines (2016)
Netherlands (Europe) Cohort Participants of LifeLines
cohort (2007–2013) with a
PM10, PM2.5, NO2, Long-term LUR model DSM-IV PM10 (10 mg/m3
):
OR: 2.66 (1.63, 4.35);
6S.-J.Fanetal./ScienceoftheTotalEnvironment701(2020)134721
mean age of 43.8 years at
baseline (n = 32, 145)
PM2.5 (5 mg/m3
):
OR: 1.04 (0.32, 3.4);
NO2 (10 mg/m3
):
OR: 1.34 (1.17, 1.53)
Zijlema et al. – KORA
(2016)
Germany (Europe) Cohort Participants of KORA cohort
(2004–2005 and 2006–2008)
with a mean age of 55.3 years
at baseline (n = 5314)
PM10, PM2.5, NO2, Long-term LUR model PHQ-9 PM10 (10 mg/m3
):
OR: 0.74 (0.16, 3.47);
PM2.5 (5 mg/m3
):
OR: 1.06 (0.25, 4.51);
NO2 (10 mg/m3
):
OR: 1.01 (0.67, 1.54)
Zijlema et al. – HUNT
(2016)
Norway (Europe) Cohort Participants of HUNT cohort
(2006–2008) with a mean age
of 54.7 years at baseline
(n = 32, 102)
PM10, PM2.5, NO2, Long-term LUR model HADS-D PM10 (10 mg/m3
):
OR: 0.36 (0.20, 0.66);
NO2 (10 mg/m3
):
OR: 0.79 (0.66, 0.94)
Zijlema et al. –FINRISK
(2016)
Finland (Europe) Cohort Participants of FINRISK cohort
(2007) with a mean age of
51.9 years at baseline
(n = 1367)
PM10, PM2.5, NO2 Long-term LUR model CESD-11 PM2.5 (5 mg/m3
):
OR: 1.39 (0.64, 3.05)
Kim et al. (2016) Korea (Asia) Cohort Participants of NHID cohort
(2002–2010) aged 15–
79 years at baseline
(n = 27,270)
PM2.5 Long-term Monitoring
stations
ICD-10 code
F32.x and
antidepressant
prescription
PM2.5 (10 mg/m3
):
HR: 1.47 (1.14, 1.90)
Vert et al. (2017) Spain (Europe) Cross-
sectional
Participants from the ALFA
cohort (2013–2014) with a
mean age of 56.5 years
(n = 958)
PM10, PM2.5, NO2, PM
coarse, NOx
Long-term LUR model Self-report of
doctor-
diagnosed
depression or
antidepressants
medication use
PM10 (10 mg/m3
):
OR: 6.52 (1.82, 23.35);
PM2.5 (5 mg/m3
):
OR: 4.38 (1.70, 11.3);
NO2 (10 mg/m3
):
OR: 2.00 (1.37, 2.93)
Lin et al. (2017a) China, Ghana, India,
Mexico, Russia, South
Africa
Cross-
sectional
Participants from SAGE cohort
(2007–2010) aged 18 years or
older (n = 41, 785)
PM2.5 Long-term GEOS-Chem
chemical transport
model
WHO WMH-
CIDI
PM2.5 (10 mg/m3
):
OR: 1.10 (1.02, 1.19)
Kioumourtzoglou et al.
(2017)
United States (North
America)
Cohort Women from the NHS cohort
(1996–2008) with a mean age
of 66.6 years (n = 41, 844)
PM2.5, O3 Long-term Nationwide
spatiotemporal
model
Self-report of
doctor-
diagnosis/
antidepressant
medication use
PM2.5 (10 mg/m3
):
HR: 1.08 (0.97, 1.20)
Kim and Kim (2017) Korea (Asia) Cross-
sectional
Residents living in 25
communities in Seoul aged
19 years or older (n = 23, 139)
PM10 Long-term National Institute
of Environmental
Research
(Monitoring
stations)
Self-designed
questionnaire
PM10 (10 mg/m3):
OR: 1.01 (0.98, 1.05)
Pun et al. (2017) United States (North
America)
Cohort Participants from the NSHAP
cohort (2005–2006 and
2010–2011) with a mean age
of 69.3 at baseline (n = 4008)
PM2.5 Both Generalized
additive mixed
models
CESD-11 Long-term
PM2.5 (5 mg/m3
):
OR: 1.04 (0.89, 1.22);
Short-term
PM2.5 (5 mg/m3
):
OR: 1.08 (1.00, 1.16)
Lin et al. (2017b) China (Asia) Cross-
sectional
Pregnant women who
regularly visited prenatalcare
Clinics in Shanghai
(2010) with a mean age of
28 years (n = 1931)
PM10, SO2, NO2 Short-term Shanghai
Environmental
Monitoring Center
SCL-90-R PM10 (57 mg/m3
):
OR: 1.04 (0.92, 1.18);
NO2 (12.8 mg/m3
):
OR: 1.21 (0.96, 1.52);
SO2 (14.0 mg/m3
):
OR: 1.22 (1.05, 1. 42).
Wang et al. (2018) China (Asia) Case-
crossover
Hospital admissions related
to depression from 26 cities of
China (n = 19, 646)
PM10, PM2.5 Short-term National Air
Pollution
Monitoring
System
ICD-10. F32/33/
34.1/41.2
% change in RR
PM.10 (76.9 mg/m3
):
%RR:4.36 (2.05, 6.73);
PM.2.5 (47.5 mg/m3
):
%RR:6.21 (3.85, 8.63)
(continued on next page)
S.-J.Fanetal./ScienceoftheTotalEnvironment701(2020)1347217
Table 1 (continued)
Authors (publication
year)
Study Location Study Design Study Participants Studied Pollutants Exposure duration Exposure
assessment
Outcome
definition
Pollutants (increment)
and estimates
Zock et al. (2018) Netherlands (Europe) Cross-
sectional
4450 residents from 135
neighbourhoods in 43 Dutch
municipalities, aged
40.5 years
PM2.5, PM10, NO2 Long-term ESCAPE model ICPC codes PM10 (10 mg/m3
):
OR: 2.33 (0.73, 7.44);
PM2.5 (10 mg/m3
):
OR: 6.42 (1.39, 29.7);
NO2 (10 mg/m3
):
OR: 1.15 (0.95, 1.39).
Roberts et al. (2019) England (Europe) Cohort 284 children living in London
drawn from the
Environmental Risk
Longitudinal Twin study
PM2.5, NO2 Long-term KCLurban model Children‘s
Depression
Inventory
PM2.5 (14.09 mg/m3
):
OR: 1.63 (1.08, 2.46);
NO2 (37.9 mg/m3
):
OR: 1.57 (1.05, 2.35).
Zhao et al. (2019) Germany (Europe) Cross-
sectional
Participants from two
German birth cohorts
(GINIplus and LISA cohorts),
with an average age of
15.20 years (n = 2827)
O3 Both Monitoring
stations for shortterm
exposure;
LUR model for
long-term
exposure
Depression
Screener for
Teenagers
Long-term
O3 (3.2 mg/m3
):
OR: 1.08 (0.94, 1.23);
Short-term
O3 (39.7 g/m3
):
OR: 0.95 (0.82, 1.11).
Zhang et al. (2019) Korea (Asia) Cohort Adults underwent a
comprehensive annual or
biennial health examination
at clinics (n = 123,045)
PM2.5, PM10 Long-term LUR model based
on subjects’
address postal
codes
CESD PM10 (10 mg/m3
):
HR: 1.11 (1.06, 1.16);
PM2.5 (10 mg/m3
):
HR: 1.01 (0.83, 1.22);
Klompmaker et al.
(2019)
Netherlands (Europe) Cross-
sectional
Adults from a national health
survey (n = 354,827)
Long-term LUR model based
on home address
Antidepressant
medication use
PM10 (1.24 mg/m3
):
OR: 0.99 (0.97, 1.01);
PM2.5 (0.83 mg/m3
):
OR: 1.01 (0.99, 1.03);
NO2 (7.85 mg/m3
):
OR: 1.03 (1.00, 1.05).
Abbreviations: ALFA: Alzheimer and Families; CESD-11: 11-item form of the Center for Epidemiological Studies-Depression; CESD-R: Revised Center for Epidemiological Studies Depression Scale; CO: carbon monoxide; DSM-IV:
Diagnostic and Statistical Manual of Mental Disorders; FINRISK: the Finnish National Cardiovascular Risk Factor Survey; GEOS: Geodetic Satellite; HADS-D: depression subscale of the Hospital Anxiety Depression Scale; HR: hazard
ratio; HUNT: Helseundersøkelsen i Nord-Trøndelag; ICD-9: International Classification for Diseases, 9th revision; ICD-10: International Classification of Diseases, 10th revision; ICPC, International Classification of Primary Care;
KORA: Cooperative Health Research in the Region Augsburg; LUR: land use regression; NO2: nitrogen dioxide; NHID: National Health Insurance database; NHS: Nurses’ Health Study; NSHAP: National Social Life, Health and Aging
Project; NO: nitrogen monoxide; NOx: oxynitride; O3: ozone; OR: odd ratio; PHQ-9: depression module of the patient health questionnaire; PM10: particle with aerodynamic diameter 10 mm; PM2.5: particle with aerodynamic
diameter 2.5 mm; PM: particulate matter; RR: relative risk; SAGE: Study on global AGEing and adults health; SCL-90-R: Symptom Checklist-90-Revised Scale; SGDS-K: The Korean version of the Geriatric Depression Scale-Short
Form; SO2: sulfur dioxide; WHO: World Health Organization; WMH-CIDI: World Mental Health Survey version of the Composite International Diagnostic Interview.
8S.-J.Fanetal./ScienceoftheTotalEnvironment701(2020)134721
showed that only NO2 was significantly associated with higher
odds for depression (OR = 1.02, 95% = 1.00–1.04), and no significant
association was found for the remaining four air pollutants (PM2.5:
OR = 1.01, 95% CI = 0.99–1.04; PM10: OR = 1.01, 95% CI = 0.98–1.04;
SO2: OR = 1.03, 95% CI = 0.99–1.07; O3: OR = 1.01, 95% CI = 0.99–
1.03). However, under the conventional random effects model,
the odds for depression were significantly associated with all the
four pollutants (Table S5).
The heterogeneity was large to extreme for all the five pooled
associations (I2
ranged from 51.6% to 86.7%). The Doi plot
showed major asymmetry in PM10, NO2, and O3, and symmetrical
in PM2.5 and SO2 (Figs. S14–S18). Sensitivity analyses excluding
single studies did not materially change the pooled results
(Figs. S19–S23). The credibility of the cumulative evidence was
‘‘low” or ‘‘very low” for short-term exposure to NO2, PM10, and
O3, and was ‘‘moderate” for PM2.5 and SO2 according to GRADE criteria
(Table S6).
One study investigated the association between short-term NO
exposure and depression, and detected non-significant association
(Wang et al., 2014). Two studies focused on short-term CO exposure
and depression, but reported inconsistent results (Wang
et al., 2014; Cho et al., 2014). Due to the limited number of studies,
we did not generate pooled effect estimates for short-term NO or
CO exposure and depression.
Table 2
Meta-analysis results for associations between long-term and short-term air pollution exposure and depression under the IVhet model.
No. of studies Sample size OR (95% CI) Cochran’s Q I2
Pheterogeneity LFK index
Long-term
PM2.5 12 637,297 1.12 (0.97, 1.29) 22.72 51.6 0.019 4.52 (Major asymmetry)
PM10 8 575,980 1.04 (0.88, 1.25) 48.99 85.7 <0.001 2.12 (Major asymmetry)
NO2 7 430,080 1.05 (0.83, 1.34) 36.69 83.6 <0.001 2.67 (Major asymmetry)
Short-term
PM2.5 6 170,027 1.01 (0.99, 1.04) 10.32 51.6 0.067 À0.58 (No asymmetry)
PM10 5 54,146 1.01 (0.98, 1.04) 29.97 86.7 <0.001 6.57 (Major asymmetry)
NO2 7 153,826 1.02 (1.00, 1.04) 17.34 65.4 0.008 6.00 (Major asymmetry)
SO2 6 153,094 1.03 (0.99, 1.07) 17.37 71.2 0.004 À0.13 (No asymmetry)
O3 7 154,722 1.01 (0.99, 1.03) 33.71 82.2 <0.001 À2.34 (Major asymmetry)
Abbreviations: CI: confidence interval; IVhet: Inverse Variance Heterogeneity; LKF: Luis Furuya-Kanamori; NO2: nitrogen dioxide; O3: ozone; OR: odd ratio; PM10: particle
with aerodynamic diameter 10 mm; PM2.5: particle with aerodynamic diameter 2.5 mm; SO2: sulfur dioxide.
Fig. 2. Meta-analysis results for the association between long-term PM2.5 (A) and NO2, (B) exposure and depression.
S.-J. Fan et al. / Science of the Total Environment 701 (2020) 134721 9
4. Discussion
4.1. Principal findings
In this systematic review and meta-analysis, we synthesized 22
studies from 10 countries to comprehensively evaluate the associations
of long-term and short-term ambient air pollutants exposure
with depression. We only observed significant association
between short-term exposure to NO2 and depression under the
advanced IVhet meta-analyses, although more significant associations
were detected under the conventional random effects model.
Due to high between-study heterogeneity, the confidence in the
pooled estimates were ranked as low or moderate, so a future
meta-analysis incorporating additional study populations will be
necessary to draw more definitive conclusions. A handful of studies
explored the impact of other air pollutants, like NO, CO, and O3, and
the results were mixed.
4.2. Comparison with other studies
We systematically searched seven databases in two languages
and were aware of one recently published meta-analysis on air pollution
and depression (Gu et al., 2019, Table S7). In that metaanalysis,
the authors investigated short-term effects of PM10 and
PM2.5 as well as long-term effects of PM2.5. They included seven
studies (n = 255,181) on PM2.5 and PM10 exposure and depression,
and reported that both long-term (OR = 1.25, 95% CI = 1.07–1.45)
and short-term PM2.5 (OR = 1.18, 95% CI = 1.04–1.34) exposure
were associated with higher odds for depression. No significant
association was observed for short-term PM10 exposure. By comparison,
we not only focused on PM10 and PM2.5, but also NO2,
SO2, O3, CO, and NO. Thus, our pooled estimates were based on triple
the number of studies and a substantially larger study population,
which provided a more precise estimate of the association
between ambient air pollution and depression. In addition, while
Fig. 3. Meta-analysis results for the association between short-term PM2.5 (A) and NO2, (B) exposure and depression.
10 S.-J. Fan et al. / Science of the Total Environment 701 (2020) 134721
the prior meta-analysis only adopted the random effects model,
our current analysis applied both the random effects model and
the IVhet model. Under the conventional random effects model,
our pooled effects for both long- and short-term PM2.5 exposure
with depression were consistent with those from Gu et al.’s
meta-analysis. However, the significant relationships disappear
under the IVhet model. As is aforementioned in the methods section,
the IVhet model outperforms the random effects model (Doi
et al., 2015), thus our current pooled estimates would be more
robust. Moreover, apart from study quality assessment, we also
did ROB assessment for each study, performed meta-regression
analyses to explore the sources of heterogeneity, and graded the
credibility of the cumulative evidence. Collectively, our systematic
review and meta-analysis builds on the prior meta-analysis by covering
more air pollutants, including more studies and participants,
and performing more thorough assessment for the included studies.
As such, the evidence from our study might be more comprehensive
and precise.
We excluded five studies concerning air pollution and depression
that did not provide sufficient quantitative data for pooling
in the current meta-analysis (Wang and Yang, 2018; Wang et al.,
2019; Qiu et al., 2019; Sheffield et al., 2018; Shin et al., 2018).
Specifically, two cross-sectional studies in China observed that
greater levels of PM2.5, SO2, and total suspended particulate emission
intensities were consistently associated with a higher prevalence
of depression (Wang and Yang, 2018; Wang et al., 2019). A
time-series study of 10,947 participants showed that short-term
PM2.5 exposure contributed to 12.07% of total hospital admissions
for depression (Qiu et al., 2019). A cohort study of 557 pregnant U.
S. women observed greater mid-pregnancy PM2.5 exposures to be
associated with a greater frequency of depressive symptoms
(Sheffield et al., 2018). Finally, Shin et al. also reported higher odds
for depression in association with greater quartiles of ambient CO,
NO2, and PM10 levels in more than 120,000 Koreans (Shin et al.,
2018). The results from these studies collectively supported a positive
association between air pollution and depression, which were
consistent with our meta-analytical estimates under the conventional
random effects model, but were contrary with the estimates
from the IVhet model.
Our results also indicate that the exposure period may modify
associations between air pollution exposure and depression. More
specifically, we found a significant association for depression with
short-term exposure to NO2, yet non-significant association with
long-term exposure for NO2. Still, most included short-term exposure
studies did not adjust for long-term air pollution exposure,
and we were thus unable to disentangle mutually confounded
short- and long-term NO2 effects. Participants’ characteristics like
age, sex, and income were also hypothesized to modify the association
between air pollution and health outcomes. However, most
included studies did not report stratified results, and thus we were
unable to conduct subgroup analyses to assess the potential modification
by these variables.
4.3. Pathophysiological mechanisms
The pathophysiological mechanisms underlying the association
between ambient air pollution exposure and depression remain
unclear, but several pathways have been proposed (Block and
Calderón-Garcidueñas, 2009; MohanKumar et al., 2008). It has
been well-documented that inhalation of air pollutants can cause
oxidative stress and systemic inflammation (Risom et al., 2005;
Araujo, 2010; Møller and Loft, 2010). Mechanistic evidence indicates
that ambient air pollutants might affect psychological status
by causing oxidative stress and neuro-inflammation (Ng et al.,
2008; Hurley and Tizabi, 2013). Oxidative stress can damage
dopaminergic neurons; depletion of central nervous system dopamine
is likely involved in the neuropathology of depression
(Hasler, 2010; Block et al., 2004). Additionally, ambient air pollutions
exposure may cause cerebrovascular damage and neurodegeneration
by increasing expression of inflammatory mediators
(e.g., hippocampal pro-inflammatory cytokines), upregulating
expression of innate immunity, promoting autoantibodies to cell
junction and neural proteins production, and activating neuroinflammation
responses (Fonken et al., 2011; Sama et al., 2007;
Calderón-Garcidueñas et al., 2003). Moreover, many neuroimaging
studies also found that air pollutants can hurt brain tissues (e.g.,
white matter, cortical gray matter, and basal ganglia) and thus
cause cognitive disorders in humans (de Prado Bert et al., 2018;
Wilker et al., 2015). Additionally, previous researches demonstrated
that air pollution exposure impaired people’s residential
satisfaction and self-perceived health, which have been associated
with mental health (von Lindern et al., 2016; Liu et al., 2018;
Nguyen et al., 2017). Furthermore, exposure to ambient air pollutions
has been linked to cardiovascular diseases, asthma, and
chronic obstructive pulmonary diseases, and cancers (Franklin
et al., 2015; Guarnieri and Balmes, 2014; Zhang et al., 2016;
Hamra et al., 2014), which are also important predictors of depression
(Ng et al., 2008; Maurer et al., 2008; Scott et al., 2007; Ossola
et al., 2018; Sotelo et al., 2014). The positive associations between
NO2 exposure and depression in our meta-analysis are consistent
with these hypothesized mechanisms.
5. Strengths and limitations
The current study has several strengths. First, we provide the
most comprehensive evidence on the relationship between depression
and ambient air pollution exposure to date. The total number
of participants was large, we not only estimated airborne PM, but
also airborne gaseous pollutants. Second, we assessed the quality
and ROB for each included study according to validated or widely
accepted scales and determined confidence of our pooled estimates
according to the GRADE system. Thus, our pooled results may be
valuable for researchers in this area to identify research gaps and
to improve future study designs. Third, we conducted multiple
sensitivity analyses by excluding any single study or some specific
studies (e.g., two studies that included children and teenagers
<18 years of age as participants (Zock et al., 2018; Roberts et al.,
2019)), and found that the effects estimates were consistent with
those from the overall analyses. This indicates that these pooled
results were reliable.
However, our study also has some limitations. First, there was
high between-study heterogeneity for all of the pooled air
pollutants-depression associations, although we adopted the
robust IVhet model to generate pooled estimates. With a limited
number of studies, we were unable to perform sub-group and
meta-regression analyses to identify the sources of the heterogeneity
except for the association between long-term PM2.5 exposure
and depression. However, we found that none of the detected
potential moderators contributed to the heterogeneity for the association
of long-term PM2.5 exposure and depression. This indicates
that other unmeasured or unreported variables would be more
responsible for the heterogeneity. Second, according to the GRADE
system, the credibility of the cumulative evidence was ‘‘low” or
‘‘very low” for all meta-analyses except that for short-term exposure
to PM2.5 and SO2 (moderate). The main causes for degrading
the quality of the cumulative evidence included high heterogeneity
and inconsistent results across studies as well as publication bias.
The strength of recommendation in healthy and environmental
policy-making guidelines was therefore greatly compromised.
Third, while the included cohort and cross-sectional studies were
ranked as ‘‘high” quality, those with a time-series, panel, or caseS.-J.
Fan et al. / Science of the Total Environment 701 (2020) 134721 11
crossover design has low quality. The main reason for the low quality
was that such short-term studies did not control for long-term
air pollution exposure, which preclude us to disentangle between
short-term triggering effects of air pollution and long-term effects.
In addition, about half of the included studies assessed exposure
using data from fixed air monitoring stations, which might have
caused exposure misclassification. Further, many studies defined
depression using questionnaires, but not clinically diagnosed, thus
outcome misclassification is also possible. Fourth, the number of
studies (or data sets) for some air pollutants (e.g., NO and CO)
was small (n 2) (Wang et al., 2014; Cho et al., 2014), which precluded
meta-analysis. Fifth, the included studies were not representative
of global populations, as nearly all included were from
China, Korea, North America, and Europe, and so our results may
not be generalizable to other areas. Sixth, most studies only
reported estimates from single-pollutant model, thus we were
unable to explore potential synergistic or additive effects of correlated
exposures in multi-pollutant mixtures (Billionnet et al.,
2012). Seventh, most studies used a linear model to fit the air
pollutant-depression associations, but associations may be nonlinear.
Eighth, nearly all of our included studies controlled for
potential confounders, yet confounding factors varied among the
studies, and some important confounding factors such as noise
and greenspace were not considered. Thus, we cannot rule out
the possibility for residual and unmeasured confounding in our
pooled estimates. Ninth, we did not convert HR and RR to OR, but
pooled them directly, this might have biased the pooled results.
Finally, the approach used to select lag for short-term exposure
studies might have biased to a greater pooled effect estimate.
6. Conclusions and future perspectives
In summary, the present systematic review and meta-analysis
indicates an association for short-term exposure to NO2 and
depression, but not for the other air pollutants. However, due to
high between-study heterogeneity and small sample sizes for
some pollutants, it is difficult to draw a robust conclusion on the
plausibility of an air pollution-depression association. In the future,
research focus should be extended to other geographic areas, especially
those with high ambient air pollution levels like Africa and
India. Advanced methods should be applied to assess individual
air pollution exposure, and more strict definition and diagnosis
for depression should be adopted. More sophisticated statistical
analyses including multi-pollutant models, non-linear association
investigation, and effect modification assessment, should be
employed. Some important confounding variables, including meteorologic
factors, noise, and green space, should be collected and
adjusted. To facilitate future quantitative synthesis, authors should
improve results importing, providing numerical estimates and
describing bias concerns. Finally, mechanistic studies remain
needed to clearly elucidate the biological pathways underlying
associations between air pollution and depression.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared
to influence the work reported in this paper.
Acknowledgements
This work was funded by the National Natural Science Foundation
of China (No. 81803196) and the China Scholarship Council,
China (No. 201708120056).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.scitotenv.2019.134721.
References
American Psychiatric Association, 2013. Diagnostic and Statistical Manual of Mental
Disorders, Fifth Edition (DSM-5). American Psychiatric Association Publishing,
Washington, DC.
Araujo, J.A., 2010. Particulate air pollution, systemic oxidative stress, inflammation,
and atherosclerosis. Air Qual. Atmos. Health 4, 79–93.
Atkinson, R.W., Cohen, A., Mehta, S., Anderson, H.R., 2012. Systematic review and
meta-analysis of epidemiological time-series studies on outdoor air pollution
and health in Asia. Air Qual. Atmos. Health 5 (4), 383–391.
Billionnet, C., Sherrill, D., Annesi-Maesano, I., 2012. Estimating the health effects of
exposure to multi-pollutant mixture. Ann. Epidemiol. 22, 126–141.
Block, M.L., Calderón-Garcidueñas, L., 2009. Air pollution: mechanisms of
neuroinflammation and CNS disease. Trends Neurosci. 32, 506–516.
Block, M.L., Wu, X., Pei, Z., Li, G., Wang, T., Qin, L., et al., 2004. Nanometer size diesel
exhaust particles are selectively toxic to dopaminergic neurons: the role of
microglia, phagocytosis, and NADPH oxidase. FASEB J. 18, 1618–1620.
Buoli, M., Grassi, S., Caldiroli, A., Carnevali, G.S., Mucci, F., Iodice, S., et al., 2018. Is
there a link between air pollution and mental disorders? Environ. Int. 118, 154–
168.
Calderón-Garcidueñas, L., Maronpot, R.R., Torres-Jardon, R., Henríquez-Roldán, C.,
Schoonhoven, R., Acuña-Ayala, H., et al., 2003. DNA damage in nasal and brain
tissues of canines exposed to air pollutants is associated with evidence of
chronic brain inflammation and neurodegeneration. Toxicol. Pathol. 31, 524–
538.
Cho, J., Choi, Y.J., Suh, M., Sohn, J., Kim, H., Cho, S.K., et al., 2014. Air pollution as a
risk factor for depressive episode in patients with cardiovascular disease,
diabetes mellitus, or asthma. J. Affect. Disord. 157, 45–51.
Dantzer, R., O’Connor, J.C., Freund, G.G., Johnson, R.W., Kelley, K.W., 2008. From
inflammation to sickness and depression: when the immune system subjugates
the brain. Nat. Rev. Neurosci. 9, 46–56.
de Prado Bert, P., Mercader, E.M.H., Pujol, J., Sunyer, J., Mortamais, M., 2018. The
effects of air pollution on the brain: a review of studies interfacing
environmental epidemiology and neuroimaging. Curr. Environ. Health Rep. 5,
351–364.
Doi, S.A., Barendregt, J.J., Khan, S., Thalib, L., Williams, G.M., 2015. Advances in the
meta-analysis of heterogeneous clinical trials I: the inverse variance
heterogeneity model. Contemp. Clin. Trials 45, 130–138.
Doi, S.A.R., Furuya-Kanamori, L., Thalib, L., Barendregt, J.J., 2017. Meta-analysis in
evidence-based healthcare: a paradigm shift away from random effects is
overdue. Int. J. Evid. Based Healthc. 15, 152–160.
Eze, I.C., Hemkens, L.G., Bucher, H.C., Hoffmann, B., Schindler, C., Künzli, N., et al.,
2015. Association between ambient air pollution and diabetes mellitus in
Europe and North America: systematic review and meta-analysis. Environ.
Health Perspect. 123, 381–389.
Ferenchick, E.K., Ramanuj, P., Pincus, H.A., 2019. Depression in primary care: part 1screening
and diagnosis. BMJ 365,. https://doi.org/10.1136/bmj.l794 l794.
Fonken, L.K., Xu, X., Weil, Z.M., Chen, G., Sun, Q., Rajagopalan, S., et al., 2011. Air
pollution impairs cognition, provokes depressive-like behaviors and alters
hippocampal cytokine expression and morphology. Mol. Psychiatry 16 (987–
995), 973.
Franklin, B.A., Brook, R., Arden Pope 3rd, C., 2015. Air pollution and cardiovascular
disease. Curr. Probl. Cardiol. 40, 207–238.
GBD 2017 Disease and Injury Incidence and Prevalence Collaborators, 2018. Global,
regional, and national incidence, prevalence, and years lived with disability for
354 diseases and injuries for 195 countries and territories, 1990–2017: a
systematic analysis for the Global Burden of Disease Study 2017. Lancet 392,
1789–1858.
GBD 2017 Risk Factor Collaborators, 2018. Global, regional, and national
comparative risk assessment of 84 behavioural, environmental and
occupational, and metabolic risks or clusters of risks for 195 countries and
territories, 1990–2017: a systematic analysis for the Global Burden of Disease
Study 2017. Lancet 392, 1923–1994.
Gu, X., Liu, Q., Deng, F., Wang, X., Lin, H., Guo, X., et al., 2019. Association between
particulate matter air pollution and risk of depression and suicide: systematic
review and meta-analysis. Br. J. Psychiatry 1–12. https://doi.org/10.1192/
bjp.2018.295.
Guarnieri, M., Balmes, J.R., 2014. Outdoor air pollution and asthma. Lancet 383,
1581–1592.
Hamra, G.B., Guha, N., Cohen, A., Laden, F., Raaschou-Nielsen, O., Samet, J.M., et al.,
2014. Outdoor particulate matter exposure and lung cancer: a systematic
review and meta-analysis. Environ. Health Perspect. 122, 906–911.
Hasler, G., 2010. Pathophysiology of depression: do we have any solid evidence of
interest to clinicians? World Psychiatry 9, 155–161.
Higgins, J.P.T., Green, S., 2011. Selection of Study Characteristics for Subgroup
Analyses and Meta-regression. Cochran Handbook for Systematic Reviews of
Interventions, Version 5.1.0 (accessed 1 January 2017) http://www.
handbook.cochran.org..
12 S.-J. Fan et al. / Science of the Total Environment 701 (2020) 134721
Higgins, J.P.T., Altman, D.G., Gotzsche, P.C., et al., 2011. Cochrane Bias Methods
Group; Cochrane Statistical Methods Group. The Cochrane Collaboration’s tool
for assessing risk of bias in randomised trials. BMJ 343. https://doi.org/10.1136/
bmj.d5928.
Hurley, L.L., Tizabi, Y., 2013. Neuroinflammation, neurodegeneration, and
depression. Neurotoxicol. Res. 23, 131–144.
Kim, H.B., Shim, J.Y., Park, B., Lee, Y.J., 2018. Long-term exposure to air pollutants
and cancer mortality: a meta-analysis of cohort studies. Int. J. Environ. Res.
Public Health 15, 2608.
Kim, J., Kim, H., 2017.Demographic and environmentalfactors associated withmental
health: a cross-sectional study. Int. J. Environ. Res. Public Health 14, pii: E431.
Kim, K.N., Lim, Y.H., Bae, H.J., Kim, M., Jung, K., Hong, Y.C., 2016. Long-term fine
particulate matter exposure and major depressive disorder in a communitybased
urban cohort. Environ. Health Perspect. 124, 1547–1553.
Kioumourtzoglou, M.A., Power, M.C., Hart, J.E., Okereke, O.I., Coull, B.A., Laden, F.,
et al., 2017. The association between air pollution and onset of depression
among middle-aged and older women. Am. J. Epidemiol. 185, 801–809.
Klompmaker, J.O., Hoek, G., Bloemsma, L.D., Wijga, A.H., van den Brink, C.,
Brunekreef, B., et al., 2019. Associations of combined exposures to
surrounding green, air pollution and traffic noise on mental health. Environ.
Int. 129, 525–537.
Lam, J., Sutton, P., Kalkbrenner, A., Windham, G., Halladay, A., Koustas, E., et al.,
2016. A systematic review and meta-analysis of multiple airborne pollutants
and autism spectrum disorder. PLoS One 11, e0161851.
Lim, Y.H., Kim, H., Kim, J.H., Bae, S., Park, H.Y., Hong, Y.C., 2012. Air pollution and
symptoms of depression in elderly adults. Environ. Health Perspect. 120, 1023–
1028.
Lin, H., Guo, Y., Kowal, P., Airhihenbuwa, C.O., Di, Q., Zheng, Y., et al., 2017a. Exposure
to air pollution and tobacco smoking and their combined effects on depression in
six low- and middle-income countries. Br. J. Psychiatry 211, 157–162.
Lin, Y., Zhou, L., Xu, J., Luo, Z., Kan, H., Zhang, J., et al., 2017b. The impacts of air
pollution on maternal stress during pregnancy. Sci. Rep. 7, 40956.
Liu, S., Ouyang, Z., Chong, A.M., Wang, H., 2018. Neighborhood environment,
residential satisfaction, and depressive symptoms among older adults in
residential care homes. Int. J. Aging Hum. Dev. 87 (3), 268–288.
Maurer, J., Rebbapragada, V., Borson, S., Goldstein, R., Kunik, M.E., Yohannes, A.M.,
et al., 2008. Anxiety and depression in COPD: current understanding,
unanswered questions, and research needs. Chest 134, 43S–56S.
MohanKumar, S.M., Campbell, A., Block, M., Veronesi, B., 2008. Particulate matter,
oxidative stress and neurotoxicity. Neurotoxicology 29, 479–488.
Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., PRISMA Group, 2009. Preferred
reporting items for systematic reviews and meta-analyses: the PRISMA
statement. Ann. Intern. Med. 151 (264–269), W64.
Møller, P., Loft, S., 2010. Oxidative damage to DNA and lipids as biomarkers of
exposure to air pollution. Environ. Health Perspect. 118, 1126–1136.
Morgan, R.L., Thayer, K.A., Santesso, N., Holloway, A.C., Blain, R., Eftim, S.E., et al.,
2019. A risk of bias instrument for non-randomized studies of exposures: a users’
guide to its application in the context of GRADE. Environ. Int. 122, 168–184.
Mustafic, H., Jabre, P., Caussin, C., Murad, M.H., Escolano, S., Tafflet, M., et al., 2012.
Main air pollutants and myocardial infarction: a systematic review and metaanalysis.
JAMA 307, 713–721.
Ng, F., Berk, M., Dean, O., Bush, A.I., 2008. Oxidative stress in psychiatric disorders:
evidence base and therapeutic implications. Int. J. Neuropsychopharmacol. 11,
851–876.
Nguyen, H.A., Anderson, C.A.M., Miracle, C.M., Rifkin, D.E., 2017. The association
between depression, perceived health status, and quality of life among
individuals with chronic kidney disease: an analysis of the National Health
and Nutrition Examination Survey 2011–2012. Nephron 136 (2), 127–135.
OHAT, 2015. Translation (OHAT) Division of the National Toxicology Program
National Institute of Environmental Health Sciences.
Ossola, P., Gerra, M.L., De Panfilis, C., Tonna, M., Marchesi, C., 2018. Anxiety,
depression, and cardiac outcomes after a first diagnosis of acute coronary
syndrome. Health Psychol. 37, 1115–1122.
Pun, V.C., Manjourides, J., Suh, H., 2017. Association of ambient air pollution with
depressive and anxiety symptoms in older adults: results from the NSHAP
study. Environ. Health Perspect. 125, 342–348.
Qiu, H., Zhu, X., Wang, L., Pan, J., Pu, X., Zeng, X., et al., 2019. Attributable risk of
hospital admissions for overall and specific mental disorders due to particulate
matter pollution: a time-series study in Chengdu, China. Environ. Res. 170, 230–
237.
Ramanuj, P., Ferenchick, E.K., Pincus, H.A., 2019. Depression in primary care: part 2management.
BMJ 365,. https://doi.org/10.1136/bmj.l835 l835.
Risom, L., Møller, P., Loft, S., 2005. Oxidative stress-induced DNA damage by
particulate air pollution. Mutat. Res. 592, 119–137.
Roberts, S., Arseneault, L., Barratt, B., Beevers, S., Danese, A., Odgers, C.L., et al., 2019.
Exploration of NO2 and PM2.5 air pollution and mental health problems using
high-resolution data in London-based children from a UK longitudinal cohort
study. Psychiatry Res. 272, 8–17.
Sama, P., Long, T.C., Hester, S., Tajuba, J., Parker, J., Chen, L.C., et al., 2007. The cellular
and genomic response of an immortalized microglia cell line (BV2) to
concentrated ambient particulate matter. Inhal. Toxicol. 19, 1079–1087.
Scott, K.M., Von Korff, M., Ormel, J., Zhang, M.Y., Bruffaerts, R., Alonso, J., et al., 2007.
Mental disorders among adults with asthma: results from the World Mental
Health Survey. Gen. Hosp. Psychiatry 29, 123–133.
Sheffield, P.E., Speranza, R., Chiu, Y.M., Hsu, H.L., Curtin, P.C., Renzetti, S., et al., 2018.
Association between particulate air pollution exposure during pregnancy and
postpartum maternal psychological functioning. PLoS One 13, e0195267.
Shin, J., Park, J.Y., Choi, J., 2018. Long-term exposure to ambient air pollutants and
mental health status: a nationwide population-based cross-sectional study.
PLoS One 13, e0195607.
Sotelo, J.L., Musselman, D., Nemeroff, C., 2014. The biology of depression in cancer
and the relationship between depression and cancer progression. Int. Rev.
Psychiatry 26, 16–30.
Spitzer, R.L., Kroenke, K., Linzer, M., Hahn, S.R., Williams, J.B., deGruy, F.V., 3rd, et al.,
1995. Health-related quality of life in primary care patients with mental
disorders. Results from the PRIME-MD 1000 Study. JAMA 274, 1511–
1517.
Szyszkowicz, M., Rowe, B.H., Colman, I., 2009. Air pollution and daily emergency
department visits for depression. Int. J. Occup. Med. Environ. Health 22, 355–
362.
Szyszkowicz, M., Kousha, T., Kingsbury, M., Colman, I., 2016. Air pollution and
emergency department visits for depression: a multicity case-crossover study.
Environ. Health Insights 10, 155–161.
The Joanna Briggs Institute Critical Appraisal tools for use in JBI Systematic
Reviews: Checklist for Analytical Cross Sectional Studies, 2016. Accessed at:
http://joannabriggs.org/research/critical-appraisaltools.html (accessed date:
January 21, 2018).
Vert, C., Sánchez-Benavides, G., Martínez, D., Gotsens, X., Gramunt, N., Cirach, M.,
et al., 2017. Effect of long-term exposure to air pollution on anxiety and
depression in adults: a cross-sectional study. Int. J. Hyg. Environ. Health 220,
1074–1080.
Von Lindern, E., Hartig, T., Lercher, P., 2016. Traffic-related exposures, constrained
restoration, and health in the residential context. Health Place 39, 92–100.
Walker, E.R., McGee, R.E., Druss, B.G., 2015. Mortality in mental disorders and global
disease burden implications: a systematic review and meta-analysis. JAMA
Psychiatry 72, 334–341.
Wang, F., Liu, H., Li, H., Liu, J., Guo, X., Yuan, J., et al., 2018. Ambient concentrations
of particulate matter and hospitalization for depression in 26 Chinese cities: a
case-crossover study. Environ. Int. 114, 115–122.
Wang, Q., Yang, Z., 2018. Does chronic disease influence susceptibility to the effects
of air pollution on depressive symptoms in China? Int. J. Ment. Health Syst. 12,
33.
Wang, R., Liu, Y., Xue, D., Yao, Y., Liu, P., Helbich, M., 2019. Cross-sectional
associations between long-term exposure to particulate matter and depression
in China: the mediating effects of sunlight, physical activity, and neighborly
reciprocity. J. Affect. Disord. 249, 8–14.
Wang, Y., Eliot, M.N., Koutrakis, P., Gryparis, A., Schwartz, J.D., Coull, B.A., et al.,
2014. Ambient air pollution and depressive symptoms in older adults: results
from the MOBILIZE Boston study. Environ. Health Perspect. 122, 553–558.
Wells, G.A., Shea, B., O’Connell, D., Peterson, J., Welch, V., Losos, M., et al., 2010. The
Newcastle-Ottawa Scale (NOS) for Assessing the Quality of Nonrandomised
Studies in Meta-analyses. http://www.ohri.ca/programs/clinical_epidemiology/
oxford.asp. 50(4), pp. 1088–1101.
Wilker, E.H., Preis, S.R., Beiser, A.S., Wolf, P.A., Au, R., Kloog, I., et al., 2015. Long-term
exposure to fine particulate matter, residential proximity to major roads and
measures of brain structure. Stroke 46, 1161–1166.
Woodruff, T.J., Sutton, P., 2014. The Navigation Guide systematic review
methodology: a rigorous and transparent method for translating
environmental health science into better health outcomes. Environ. Health
Perspect. 122, 1007–1014.
World Health Organization, 2017. Depression and Other Common Mental
Disorders: Global Health Estimates. Available at: https://www.who.
int/mental_health/management/depression/prevalence_global_heath_estima
tes/en/ (accessed 22 March 2018).
Yang, B.Y., Qian, Z., Howard, S.W., Vaughn, M.G., Fan, S.J., Liu, K.K., et al., 2018. Global
association between ambient air pollution and blood pressure: a systematic
review and meta-analysis. Environ. Pollut. 235, 576–588.
Zhang, S., Li, G., Tian, L., Guo, Q., Pan, X., 2016. Short-term exposure to air pollution
and morbidity of COPD and asthma in East Asian area: a systematic review and
meta-analysis. Environ. Res. 148, 15–23.
Zhang, Z., Zhao, D., Hong, Y.S., Chang, Y., Ryu, S., Kang, D., et al., 2019. Long-term
particulate matter exposure and onset of depression in middle-aged men and
women. Environ. Health Perspect. 127, 77001.
Zhao, T., Markevych, I., Standl, M., Schulte-Körne, G., Schikowski, T., Berdel, D., et al.,
2019. Ambient ozone exposure and depressive symptoms in adolescents:
results of the GINIplus and LISA birth cohorts. Environ. Res. 170, 73–81.
Zijlema, W.L., Wolf, K., Emeny, R., Ladwig, K.H., Peters, A., Kongsgård, H., et al., 2016.
The association of air pollution and depressed mood in 70,928 individuals from
four European cohorts. Int. J. Hyg. Environ. Health 219, 212–219.
Zock, J.P., Verheij, R., Helbich, M., Volker, B., Spreeuwenberg, P., Strak, M., et al.,
2018. The impact of social capital, land use, air pollution and noise on individual
morbidity in Dutch neighbourhoods. Environ. Int. 121, 453–460.
S.-J. Fan et al. / Science of the Total Environment 701 (2020) 134721 13