Environmental Research 180 (2020) 108817

Contents lists available at ScienceDirect

Environmental Research
journal homepage: www.elsevier.com/locate/envres

Review article

Ambient air pollution and diabetes: A systematic review and meta-analysis T

a,b,c,1 d,1 c,e f,g b
Bo-Yi Yang , Shujun Fan , Elisabeth Thiering , Jochen Seissler , Dennis Nowak ,
Guang-Hui Donga, Joachim Heinrichb,h,∗
a
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
b
Institute and Clinic for Occupational, Social and Environmental Medicine, Hospital of the Ludwig-Maximilian University Munich, LMU Munich, Member, German Center
for Lung Research (DZL Munich), CPC (Comprehensive Pneumology Center Munich), Germany
c
Institute of Epidemiology, Helmholtz Zentrum München- German Research Center for Environmental Health, Neuherberg, Germany
d
Guangzhou Center for Disease Control and Prevention, Guangzhu, 510440, China
e
Division of Metabolic and Nutritional Medicine, Dr. von Hauner Children's Hospital, University of Munich Medical Center, Munich, Germany
f
Diabetes Center, Medizinische Klinik und Poliklinik IV – Campus Innenstadt, Ludwig-Maximilians-University, Munich, Germany
g
Clinical Cooperation Group Type 2 Diabetes, Helmholtz Zentrum München, Neuherberg, Germany
h
Allergy and Lung Health Unit, Melbourne School of Population and Global Health, The University of Melbourne, Melbourne, Australia

ARTICLE INFO ABSTRACT

Keywords: Background: Air pollutants are suggested to be related to type 2 diabetes (T2D). Since several high quality papers
Particulate matter on air pollutants and T2D have been published beyond the last reviews, an extended systematic review is highly
Gaseous pollutant warranted. We review epidemiological studies to quantify the association between air pollutants and T2D, and to
Type 2 diabetes answer if diabetes patients are more vulnerable to air pollutants.
Gestational diabetes
Methods: We systematically reviewed the databases of PubMed and Web of Science based on the guidelines of
Mortality
the Preferred Reporting Items for Systematic review and Meta-analysis (PRISMA). We calculated odds ratios
(OR) or hazard ratios (HR) and their 95% confidence intervals (CI) to assess the strength of the associations
between air pollutants [e.g., particulate matter with diameter ≤ 2.5 μm (PM2.5), particulate matter with
diameter ≤ 10 μm (PM10), and nitrogen dioxide (NO2)] and T2D. We evaluated the quality and risk of bias of the
included studies and graded the credibility of the pooled evidence using several recommended tools. We also
performed sensitivity analysis, meta-regression analysis, and publication bias test.
Results: Out of 716 articles identified, 86 were used for this review and meta-analysis. Meta-analyses showed
significant associations of PM2.5 with T2D incidence (11 studies; HR = 1.10, 95% CI = 1.04–1.17 per 10 μg/m3
increment; I2 = 74.4%) and prevalence (11 studies; OR = 1.08; 95% CI = 1.04–1.12 per 10 μg/m3 increment;
I2 = 84.3%), of PM10 with T2D prevalence (6 studies; OR = 1.10; 95% CI = 1.03–1.17 per 10 μg/m3 increment;
I2 = 89.5%) and incidence (6 studies; HR = 1.11; 95% CI = 1.00–1.22 per μg/m3 increment; I2 = 70.6%), and
of NO2 with T2D prevalence (11 studies; OR = 1.07; 95% CI = 1.04–1.11 per 10 μg/m3 increment; I2 = 91.1%).
The majority of studies on glucose-homoeostasis markers also showed increased risks with higher air pollutants
levels, but the studies were too heterogeneous for meta-analysis. Overall, patients with diabetes might be more
vulnerable to PM.
Conclusions: Recent publications strengthened the evidence for adverse effects of ambient air pollutants ex-
posure (especially for PM) on T2D and that diabetic patients might be more vulnerable to air pollutants ex-
posure.

Abbreviations: β, regression coefficient; BC, black carbon; CO, carbon monoxide; CI, confidence interval; GDM, gestational diabetes; GRADE, the Grading of
Recommendations Assessment; Development, and Evaluation; HbA1c, glycated haemoglobin; HOMA, homoeostasis model assessment; HOMA-IR, homoeostasis
model assessment insulin resistance; HR, hazards ratio; JBI, the Joanna Briggs Institute; NOS, Newcastle-Ottawa Scale; NOx, nitric oxides; O3, ozone; OHAT, the
Office of Health Assessment and Translation; OR, odds ratio; PM, particulate matter; PM10, particulate matter with diameter < 10 μm; PM2.5, particulate matter with
diameter < 2.5 μm; ROB, risk of bias; SO2, sulfur dioxide; T2D, Type 2 diabetes mellitus; UFP, ultrafine particle
∗
Corresponding author. Institute and Clinic for Occupational, Social and Environmental Medicine, Hospital of Ludwig Maximilian University, LMU Munich,
Germany Ziemssenstrasse 1, 80336, Muenchen, Germany.
E-mail address: [email protected] (J. Heinrich).
1
The two authors contributed equally to this work.

https://doi.org/10.1016/j.envres.2019.108817
Received 15 June 2019; Received in revised form 10 September 2019; Accepted 8 October 2019
Available online 12 October 2019
0013-9351/ © 2019 Elsevier Inc. All rights reserved.
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

1. Introduction (PRISMA) (Shamseer et al., 2015). Two electronic databases (i.e.,

PubMed and Web of Science) were searched for peer-reviewed articles
Type 2 diabetes mellitus (T2D), characterized by chronic hy- published before 25 May 2019 and written in English. The combina-
perglycemia, is caused by insulin resistance and beta cell dysfunction. tions of following terms concerning air pollution (i.e., air pollution,
T2D has become a leading contributor to the global burden of disease particulate matter, particles, PM2.5, PM10, PM2.5-10, black smoke, black
and early death (Forouzanfar et al., 2016; Lim et al., 2018). According carbon, NOx, NO2, SO2, CO, and O3) and T2D [i.e., diabetes, type 2
to the International Diabetes Federation (Cho et al., 2018), in 2017 diabetes, T2D, gestational diabetes, glucose, insulin resistance, homo-
approximately 451 million people had diabetes globally, and the pre- eostasis model assessment (HOMA), impaired glucose tolerance, and
valence is still increasing, mostly in African, Pacific, and Asian coun- glycated haemoglobin (HbA1c)] were used for the literature search.
tries (Danaei et al., 2011; Nanditha et al., 2016; NCD risk factors col- Additionally, reference lists of all the included studies and previously
laboration (NCD Risk Factor Collaboration (NCD-RisC), 2016); van published systematic reviews (Balti et al., 2014; Dendup et al., 2018;
Dieren et al., 2010). In parallel, air pollution has also been listed as one Dimakakou et al., 2018; Eze et al., 2015; He et al., 2017; Janghorbani
of the major environmental problems worldwide, especially in devel- et al., 2014; Liu et al., 2019; Park and Wang, 2014; Thiering and
oping nations (Cohen et al., 2017; Guan et al., 2016). Given the ubi- Heinrich, 2015; Wang et al., 2014) were manually searched to find
quitous nature of air pollution, it has affected large proportions of more studies.
people worldwide (Cohen et al., 2017; van Donkelaar et al., 2015).
Ambient air pollution (referred to as air pollution hereafter) consists of 2.2. Study selection and eligibility criteria
particulate matter (PM) of different sizes [e.g., ≤10 μm in diameter
(PM10); ≤2.5 μm (PM2.5); ≤100 nm ultrafine particles (UFP)], and Two authors (BYY and SF) performed study selection, and if there
gaseous compounds [e.g., carbon monoxide (CO), nitric oxides (NOx), was any uncertain then the third author checked (JH). Briefly, we
nitrogen monoxide (NO), nitrogen dioxide (NO2), and sulfur dioxide screened potential studies through the following four steps (Fig. S1).
(SO2), ozone (O3)]. These compounds show variations in their re- First, we downloaded all reference records from the two electronic
activity, dispersion, and toxicity. databases into a reference manager (NoteExpress V3.1, Aegean Soft-
Although accumulating evidence has demonstrated that air pollu- ware, China). Second, duplicates were deleted from the initial reference
tion adversely affects many health outcomes (Brook et al., 2010; Chen records using the software's function. Third, the titles and abstracts of
et al., 2013; Cesaroni et al., 2014; Kampa and Castanas, 2008; Pope the remaining articles were reviewed for eligibility. Fourth, the full
et al., 2015; Raaschou-Nielsen et al., 2013; Shah et al., 2013; Turner texts of the potentially eligible references were further evaluated.
et al., 2011), links between air pollution exposure and T2D have been Studies finally included had to meet the following five criteria: (1)
only recently investigated and the results were still inconsistent. Five assessed the associations of air pollution with T2D or pre-diabetic
recent comprehensive reviews (Dendup et al., 2018; Eze et al., 2015; conditions, such as fasting blood glucose (FBG), insulin resistance, and
Janghorbani et al., 2014; Liu et al., 2019; Thiering and Heinrich, 2015) impaired glucose tolerance; (2) cohort, case-crossover, time series,
summarized epidemiological studies concerning air pollution and T2D cross-sectional, case-control, or panel studies, but not reviews, letters,
and concluded that evidence indicated a positive association between clinical research studies, or animal experiments; (3) focused on outdoor
the two. However, more critical analysis is warranted. First, while most but not indoor air pollution; (4) written in English; (5) for duplicate
studies detected an increase in the risk of T2D (Eze et al., 2015; publication, we only included the study that reported the most recent
Thiering and Heinrich, 2015), other papers published beyond this re- and detailed information.
view failed to detect a significant association (e.g., Coogan et al.,
2016a; Coogan et al., 2016b). Additionally, most studies were con- 2.3. Data extraction
ducted in North America and Western European countries (Eze et al.,
2015; Thiering and Heinrich, 2015). Both the prevalence of T2D and air We extracted the following information from each eligible study:
pollution levels are reported to be much higher in developing countries authors' names, publication year, participants’ age and gender, ex-
(Chen et al., 2011; Rohde and Muller, 2015; Yang et al., 2018a), but the posure assessment of air pollution, exposure period [long-term (≥30
effects of air pollution on T2D have not been well studied before 2015 days) or short-term (< 30 days)], effect estimates [e.g., odds ratio (OR),
in these regions. For instance, several studies were performed in China and hazards ratio (HR)] and their corresponding 95% confidence in-
in recent two years (Lao et al., 2019; Liang et al., 2019; Yang et al., tervals (CI), study setting, and sample size. Two independent authors
2018a, 2018b). The most recent meta-analysis by Liu et al. (2019) only (BYY and SF) extracted this information, and discordances were dis-
pooled data on long-term air pollution exposure and T2D prevalence or cussed with a third author (JH) and resolved by consensus.
incidence, without including short-term studies and those investigating
air pollution and glucose-homoeostasis markers. 2.4. Assessment of study quality and risk of bias
To provide healthcare policy makers and researchers with more
comprehensive evidence on ambient air pollution and T2D and glucose- Two authors (BYY and SF) independently evaluated the quality and
homoeostasis markers, we performed an updated systematic review and risk of bias (ROB) of the included studies, and any conflicts was ad-
meta-analysis of studies that investigated relationships of air pollutants judicated by a third author (JH). We employed Newcastle-Ottawa Scale
with T2D and glucose-homoeostasis markers. We address three funda- (NOS) (Wells et al., 2013) to evaluate the quality of cohort and case-
mental questions: (i) is air pollution exposure associated with increased control studies. Each study was assessed on an NOS score from 0 to 9,
risks of diabetes development and diabetes-related mortality, (ii) is air and the score greater than 7 was classified as “high quality”, otherwise
pollution associated with markers of glucose homeostasis that predis- the study was categorized as “low quality”. We used the scale re-
pose to diabetes, and (iii) are participants with diabetes more vulner- commended by Joanna Briggs Institute (JBI) (http://www.
able to air pollutants exposure than those without diabetes? joannabriggs.edu.au/) to assess the quality of cross-sectional study.
Each study can be assessed on a JBI score from 0 to 20, and the score
2. Methods greater than 16 was considered as “high quality”. For time-series and
case-crossover studies, there is no validated scale to assess their quality.
2.1. Search strategy Thus, we adopted a scale adapted from validated NOS scale and the
Cochrane risk of bias tool to evaluate these studies (Mustafic et al.,
We conducted the study using the standard methods following the 2012). The adapted scale included validation of T2D or its traits (0–1
Preferred Reporting Items for Systematic Review and Meta-Analysis point), the quality of air pollution assessment (0–1 point), and the

2
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

extent of covariates adjustments (0–3 points). If the maximum score 3. Results

was achieved for all the three items (i.e., 5 points), the study was
classified as “high quality”, otherwise “low quality”. 3.1. Literature retrieval and study characteristics
We further employed the Office of Health Assessment and
Translation (OHAT, 2015) to assess the ROB for cohort and cross-sec- The systematic literature searches yielded a total of 716 hits (Fig.
tional studies. As there is no validated tool for assessing ROB of time S1). After excluding duplicates and other publications according to the
series and case-crossover studies, we thus employed the tool re- eligibility assessment by abstract and full text screens, 86 articles were
commended by the National Institute of Environmental Health Sciences identified as eligible for this review. The basic characteristics of the 86
National Toxicology Program (OHAT, 2015) and the Navigation Guide selected publications are shown in Tables S1–S3 of the supplement,
by the University of California for these studies (Lam et al., 2016). The which were grouped by study design and outcomes [diabetes mortality,
included studies were assessed for key criteria (i.e., assessment of ex- diabetes incidence and prevalence, diabetes-related traits, and gesta-
posure and outcome, and confounding bias) and other criteria (i.e., tional diabetes (GDM)], and were ordered by date of publication. The
attrition/exclusion bias, selection bias, reporting bias, conflict of in- included studies were from 19 countries, most of which are western
terest, and other bias). Each of the criteria was assessed as “low”, developed nations such as North America and European countries.
“probably low”, “probably high”, or “high” risk. If a study for which the Cohort and cross-sectional are the most commonly used study designs.
key criteria and most of the other criteria are categorized as “high” or About two thirds of the included studies assessed air pollution exposure
“probably high”, then the study should be removed from the review. using models (e.g., land use regression model, hybrid spatial-temporal
model, satellite-based model), and the remaining studies used data from
fixed air monitoring stations. Twenty studies adopted multi-pollutant
2.5. Statistical analysis models, but the pattern of the associations were mixed; while some
studies found that the effect estimates in the single-pollutant models
Due to apparent differences in characteristics (e.g., study design, were attenuated by additionally adjusted for other pollutant(s), the
participants’ health status, exposure assessment, and outcomes) for others observed that the effects were strengthened or not materially
studies investigating diabetic mortality in general population, all-cause affected (Tables S1–S3). Concerning the study quality assessment, 57
mortality in diabetic patients, and other diabetic traits (e.g., FBG, in- studies were considered as having “high quality” and the remaining
sulin, insulin resistance), we only performed meta-analyses for air studies were evaluated as “low quality” (Table S4). With respect to the
pollutants with T2D prevalence or incidence. In addition, we only fo- ROB assessment, none of the included studies was excluded because of
cused on PM and NO2, as the number of studies on other air pollutants high ROB (Table S4).
was limited.
For air pollutants with T2D prevalence or incidence available from 3.2. Is exposure to air pollution associated with an increased risk of
more than four studies, we extracted point effect estimates (OR or HR diabetes?
with their corresponding 95% CI) from individual studies and then
pooled them using the inverse variance method. The Z test was used to To answer this question, we will make use of study findings on
test the significance of the pooled estimates. The between-study het- mortality caused by diabetes in the general population as well as in-
erogeneity was evaluated using both I2 statistics and Q-test (Higgins cidence and prevalence of T2D in prospective cohorts and cross-sec-
and Thompson, 2011). If I2 ≥ 50% or P value for Q-test was < 0.01, tional studies. Detailed information of the included studies is sum-
then the heterogeneity was considered as “high” and the Der Simonian marized in Tables S1–S3 of the supplement.
and Laird random effects model was used to pool effect estimates.
Otherwise, the heterogeneity was considered as “moderate or low” and 3.2.1. Effects of air pollution on diabetes-related mortality in the general
the Mantel-Haenszel fixed effects model was used to pool effect esti- population
mates (Der Simonian and Laird, 1986). For meta-analysis with 10 data Eleven population-based studies have investigated the relationships
sets or more (Higgins and Green, 2011) and with high between-study for short-term exposure to air pollution and diabetes-related mortality
heterogeneity, we performed univariate meta-regression to explore (Fig. 1 and Table S1). All studies, except one (Zuniga et al., 2016),
potential sources of the heterogeneity. The covariates included pub- suggested an increase in daily diabetes-caused mortality after exposure
lication year, study settings, study design, exposure assessment, sample to days of high PM or black carbon (BC), and six out of the 11 studies
size, participants' age and sex, background air pollutants levels, and reported statistically significant increased mortality risks (e.g., a 10-μg/
diagnosed method of T2D. We also performed sensitivity analyses by m3 increment in PM2.5 was associated with 6–49% increase in diabetes-
excluding each study. Egger's regression test was used to evaluate po- related mortality). Due to high heterogeneity in design (time series,
tential publication bias (Egger et al., 1997). case-crossover) and the limited number of such studies, a meta-analysis
As only a small part of studies applied multi-pollutant model, we was not performed.
extracted and pooled effect estimates from single-pollutant model. In Four cohort studies explored long-term exposure to air pollution and
addition, most published studies used a generalized linear model, thus T2D related mortality. Higher levels of PM2.5 and NO2 were associated
we assumed a linear association and pooled all effects estimates for a with greater T2D mortality in all the four studies, but no significant
fixed 10 μg/m3 increase in air pollutant concentration. Other reported association was observed for O3 (Table S1).
units or scales were harmonized as published elsewhere (Yang et al.,
2018c). 3.2.2. Long-term effects of air pollution on T2D incidence
Finally, we employed the Grading of Recommendations Assessment, Prospective studies follow population at risk (i.e., participants
Development, and Evaluation (GRADE) Working Group to grade the without diabetes) for a certain time and investigated relationships of air
credibility of the pooled evidence. pollution exposure at baseline or during the follow-up period (annual
All the above statistical analyses were performed using STATA averages levels of air pollutants were often used as surrogates of long-
package version 12.1 program (StatCorp, College Station, TX, USA). A term exposure) with the incidence of T2D cases. Fig. 2 illustrates the
P-value < 0.05 was considered as statistically significant. summary findings of overall increased incidence of T2D with increased

3
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

Fig. 1. Overview of studies examining

diabetes mortality in the general popu-
lation and all-cause mortality in diabetes
patients. The direction of the effect is de-
noted by “+” for positive association, “o”
for no association, and “(+)” for non-sig-
nificant associations or associations that are
restricted to subgroups. For short-term ex-
posure studies that reported multiple lags,
we showed the results from the lag that the
authors stated as a priority, the lag was
statistically significant, or the lag that re-
ported the largest effect estimates. PM10
indicates particulate matter with
diameter < 10 μm; PM2.5, particulate matter
with diameter < 2.5 μm; NO2, nitrogen di-
oxide; NOx, nitric oxides; O3, ozone; BC,
black carbon; CO, carbon monoxide; SO2,
sulfur dioxide; SO4, sulfate.

exposure to at least one air pollutant. Fifteen out of 21 studies reported method of T2D) was the potential source of the heterogeneity (data now
statistically significant increased risks with air pollutants. shown). Based on the GRADE system, the credibility of all the pooled
In meta-analysis, the pooled effect for T2D incidence was 1.10 (95% evidence was “low” or “very low” (Table S6).
CI 1.04–1.17) and 1.11 (95% CI 1.00–1.22) for an increment of 10 μg/
m3 in PM2.5 and PM10, respectively, based on 11 and six studies (Fig. 3
3.3. Is exposure to air pollution associated with markers of glucose
and Table S5). We also meta-analyzed seven studies investigating the
homeostasis?
risk for the development of T2D in relation to NO2 and observed that
pooled effect was 1.01 (95% CI 0.99–1.02) (Fig. 3 and Table S5). High
The results of studies on air pollution and markers of glucose
between-study heterogeneity was observed for all the above three meta-
homeostasis are shown in Fig. 5 (for study details see Table S3 of the
analyses (Table S5). Meta-regression analysis for PM2.5 and DM in-
supplement). Out of 17 studies on fasting blood glucose, 14 reported
cidence showed that none of the investigated covariates was a sub-
increased risk for at least one of the investigated pollutants. For ex-
stantial source of heterogeneity (data not shown). Sensitivity analyses
ample, 14 studies investigated PM2.5 and reported that a 10 μg/m3 in-
by excluding each study did not change the overall estimates (Figs.
crement in PM2.5 was significantly associated with 0.05–24.7 mg/dL
S2–S4). Publication bias was detected for the meta-analyses for PM2.5
higher levels of fasting blood glucose. Also, HbA1c showed increased
and PM10 with T2D incidence (Table S5; Figs. S5–S7). Based on the
risk in seven out of eight studies. Consistent with these findings,
GRADE system, confidence in the pooled evidence was “very low” for
homeostatic model assessment of insulin resistance (HOMA-IR) was
both PM2.5 and PM10, and moderate for NO2 (Table S5).
increased with increased exposure to air pollution in eight out of 10
Two prospective studies investigated ozone and T2D, and both re-
studies. The results of few studies on 2h-blood glucose and insulin levels
ported increased risks (Jerrett et al., 2017; Renzi et al., 2018) (Fig. 2).
are mixed. Evidence has shown that women experiencing gestational
diabetes are more likely to have subsequent T2D (Kwak et al., 2013). To
date, six out of eight studies reported positive associations between
3.2.3. Long-term effects of air pollution on T2D prevalence
gestational diabetes and exposure to air pollution. For example, in a
A total of 25 cross-sectional studies were performed to evaluate
cross-sectional analysis of 410,267 American women, Hu et al. (2015)
relationships for air pollution and T2D prevalence and results are illu-
found that the risk of gestational diabetes significantly increased 20%
strated in Fig. 2 and Table S2. The vast majority of these studies (18 out
per 5 μg/m3 increase in PM2.5 (Table S3; Fig. 5).
of 25) showed increased risks for T2D prevalence for at least one of the
investigated pollutants. For meta-analysis, we extracted 11 studies on
PM2.5, six on PM10, and 11 on NO2 with T2D prevalence and found 3.4. Are subjects with diabetes more vulnerable to air pollution exposure
increased risk (Fig. 4 and Table S6). The overall effect estimate was than subjects without diabetes?
increased by 1.08 (95% CI 1.04–1.12), 1.10 (95% CI 1.03–1.17), and
1.07 (95% CI 1.04–1.11) for an increment of 10 μg/m3 of PM2.5, PM10, Case-crossover and time-series studies were performed to compare
and NO2, respectively (Fig. 4 and Table S6). Sensitivity analyses by the effects of short-term air pollution exposure on all-cause mortality
excluding each study did not change the overall estimates (Figs. among T2D patients and participants without diabetes. As shown in
S8–S10), and evidence of publication bias was detected for all the three Table S1 and Fig. 1, although not all of the estimates were statistically
air pollutants (Table S6; Figs. S11–S13). We detected high between- significant, the cumulative evidence seems to indicate a positive asso-
study heterogeneity for all the three meta-analyses (Table S6). Meta- ciation between short-term air pollution exposure and the daily all-
regression analysis showed that none of the investigated covariates (e.g. cause mortality risk in diabetic subjects. In addition, most of these
study setting, exposure assessment, sample size, age, and diagnosed studies reported a greater PM-associated mortality risk in diabetic

4
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

Fig. 2. Overview of studies ex-

amining ambient air pollution ex-
posure and its association with
diabetes development. The studies
investigate associations between am-
bient air pollution exposure with
diabetes incidence and prevalence.
The direction of the effect is denoted
by “+” for positive association, “o”
for no association, and “(+)” for non-
significant associations or associa-
tions that are restricted to subgroups.
PM10 indicates particulate matter
with diameter < 10 μm; PM2.5, parti-
culate matter with diameter
< 2.5 μm; PM2.5-10, particulate
matter with diameter ranging from
2.5 to 10 μm; PM1, particulate matter
with diameter < 1.0 μm; NO2, ni-
trogen dioxide; NOx, nitric oxides;
O3, ozone; BC, black carbon; CO,
carbon monoxide; SO2, sulfur dioxide.

5
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

Fig. 3. Meta-analysis of type 2 diabetes incidence and exposure to air pollutants for increments of 10 μg/m3. PM10 indicates particulate matter with
diameter < 10 μm; PM2.5, particulate matter with diameter < 2.5 μm; NO2, nitrogen dioxide.

patients than in non-diabetic population (Table S1). 4. Discussion

For example, two studies investigated the associations between O3
exposure and all-cause mortality among diabetes patients (Zanobetti 4.1. Key findings
and Schwartz, 2011; Stafoggia et al., 2010), and both reported sig-
nificant and positive associations (Table S1; Fig. 1). Stafoggia et al. In the current systematic review and meta-analysis of 86 epide-
(2010) also observed that the association of O3 with mortality was miological studies conducted in 19 countries, we performed a com-
stronger in diabetic participants (5.5% (95% CI = 1.4–9.8%) per 10 μg/ prehensive evaluation of available data on ambient air pollution and
m3 increment) than in general population (1.5% (95% CI = 0.9–2.1%) T2D. Most short-term exposure studies reported a positive association
per 10 μg/m3 increment). between higher air pollution levels and greater diabetes-caused mor-
A multi-center study on the effects of short-term NO2 exposure re- tality. Meta-analytical results showed significant associations of PM2.5,
ported an increased risk of all-cause mortality risk in diabetic patients PM10, and NO2 with T2D prevalence, and of PM2.5 and PM10 with T2D
[3.61% (95% CI = 1.73–5.53%) per 10 μg/m3 increment], which was incidence. However, the between-study heterogeneity for these meta-
greater than those observed in participants without diabetes [2.03% analyses was high, which significantly reduced the credibility of the
(95% CI = 0.87–3.21% per 10 μg/m3 increment) (Chiusolo et al., 2011) pooled evidence. Significant association was also found for air pollu-
(Table S1; Fig. 1). tants and glucose homeostasis markers (especially for fasting blood
In summary, compared to non-diabetic subjects, the observed up to glucose, HbA1c, and HOMA-IR) and gestational diabetes, but the stu-
2-fold increased risk of mortality suggest that diabetic patients might be dies were varied in study design, outcome definition, exposure assess-
more vulnerable to the adverse effects of short-term air pollution ex- ment, and the number of studies for some pollutants were limited.
posure (Table S1). Further, several short-term studies indicated higher mortality rates
among diabetic patients than those in non-diabetic individuals.

6
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

Fig. 4. Meta-analysis of type 2 diabetes prevalence and exposure to air pollutants for increments of 10 μg/m3. PM10 indicates particulate matter with
diameter < 10 μm; PM2.5, particulate matter with diameter < 2.5 μm; NO2, nitrogen dioxide.

4.2. Comparison with prior meta-analyses and interpretations analysis (Dimakakou et al., 2018; He et al., 2017). More specifically, the
systematic review by Dimakakou et al. (2018) only included 18 studies
Studies for air pollutants with T2D morbidity and mortality pub- investigating air pollution and T2D and diabetes traits. The meta-ana-
lished before 2015 have been partly analyzed in five recent meta-ana- lysis by He et al. (2017) included 11 cohort studies on PM2.5 and T2D
lyses (Balti et al., 2014; Eze et al., 2015; Janghorbani et al., 2014; Park incidence and reported that a 10 μg/m3 increase in PM2.5 was asso-
et al., 2014; Wang et al., 2014). All of these published meta-analyses ciated with a 25% higher risk of T2D. The most recent meta-analysis by
documented a positive association for long-term exposure to air pol- Liu et al. (2019) meta-analyzed 30 studies published before 2018, and
lutants and T2D, but the included studies, the methodology, and the reported significant associations for PM10, PM2.5, and NO2 with T2D
results of these reviews and meta-analyses are different (Thiering and prevalence, and for PM2.5 with T2D incidence. By comparison, the re-
Heinrich, 2015). In addition, the conducted studies have limitations in sults from these published meta-analyses were not exactly the same as
correlation with other unmeasured air pollutants, uncontrolled con- those from our current study, which might be caused by the different
founding (e.g., physical activity, smoking), other environmental ex- number of the included studies. We updated the literature search up to
posures (e.g., indoor, occupational, smoking, greenspace), and co- May 25, 2019, which generated more studies and participants (Table
morbidities (Eze et al., 2014). A few more recent systematic review S7). Thus, we believe our estimates would be more comprehensive.
articles also reported statistically significant associations between air Despite this, it is notable that the between-study heterogeneity in our
pollutants and T2D, but were incomplete or did not perform meta- meta-analyses was high, which substantially reduced the credibility of

7
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

Fig. 5. Overview of studies examining ambient air pollu-

tion exposure and its association with predisposing con-
ditions for diabetes. The studies on ambient air pollution
exposure in relation to fasting blood glucose (FBG), fasting
insulin, 2h-blood glucose (2h-BG) and insulin (2h-insulin),
glycated haemoglobin (HbA1c), homeostatic model assessment
of insulin resistance (HOMA-IR), and gestational diabetes
(GDM). The direction of the effect is denoted by “+” for po-
sitive association, “o” for no association, and “(+)” for non-
significant associations or associations that are restricted to
subgroups. For short-term exposure studies that reported mul-
tiple lags, we showed the results from the lag that the authors
stated as a priority, the lag was statistically significant, or the
lag that reported the largest effect estimates. PM10 indicates
particulate matter with diameter < 10 μm; PM2.5, particulate
matter with diameter < 2.5 μm; PM2.5-10, particulate matter
with diameter ranging from 2.5 to 10 μm; PM1, particulate
matter with diameter < 1.0 μm; NO2, nitrogen dioxide; NOx,
nitric oxides; O3, ozone; BC, black carbon; CO, carbon mon-
oxide; SO2, sulfur dioxide.

8
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

the accumulative evidence. We incorporated a set of covariates into and gaps in the existing evidence.
meta-regression analysis to detect potential sources of the hetero- However, some limitations should be acknowledged. First, the in-
geneity, but found that none of the study-level variables was re- cluded studies differed markedly in outcome assessment (e.g., self-re-
sponsible for the heterogeneity. Therefore, other unmeasured or un- ported versus diagnosed diabetes), exposure assessment (e.g., mea-
reported variables would be more important in explaining the surements from air-monitoring stations versus spatial-statistical model),
heterogeneity. and exposure period (e.g., lag days for short-term exposure and ex-
For studies investigating short-term effects of air pollution on dia- posure window of long-term exposure), which might have biased effect
betic caused mortality, a cautious interpretation is needed, although estimates. In meta-analysis, we also detected high between-study het-
half of the effect estimates are significantly increased. These studies erogeneity, which reduced the credibility of the cumulative evidence,
cannot disentangle between short-term triggering effects of PM and and thus healthcare policy makers and researcher should be cautious in
long-term effects, because it is mostly not known whether exposed using and interpreting the evidence. We also performed meta-regression
subjects developed T2D and further exposure is just triggering mortality by considering a set of covariates, but did not detect any major source
with underlying diabetes. However, evidence from prospective cohort of heterogeneity. The unmeasured or unreported variables such as po-
studies has indicated a positive association between long-term exposure pulation sensitivity and other potential sources and constituents of air
to air pollution and T2D mortality, albeit the number of such cohort pollutants might be more responsible for the heterogeneity. Second, we
studies was small. employed the random effects model to pool these data; however, the
Studies on markers of glucose homeostasis, which might be changed method also has limitations in overestimating effects and under-
in subjects with a higher risk of diabetes, can help shed light on the estimating the statistical error (Doi et al., 2015). Third, although sev-
association of air pollution exposure with diabetes and the potential eral recent studies have been performed in developing countries such as
biological mechanisms underlying the association. Two very recent China, most of the included studies were from developed countries.
systematic reviews and meta-analyses (Dang et al., 2018; Elshahidi, Thus, the generalizability of the current evidence to other highly pol-
2019) have pooled the results on air pollution with HOMA-IR and ge- luted areas is limited, especially considering possible non-linear re-
stational diabetes by including 6 and 8 studies, respectively. Dang et al. lationship between air pollution and health outcomes (Burnett et al.,
(2018) observed that long-term exposure to NO2 and PM10, but not 2014). Fourth, while most studies focused on PM and NO2, the diabetic
PM2.5, was associated with higher risk of insulin resistance. However, effects of other air pollutants (e.g., O3) were less studied. Fifth, one
they found serious publication bias for the included studies, which re- third of the included studies used air pollutant monitoring data but not
duced the credibility of the pooled evidence. Elshahidi (2019) in- personal exposure measurements. In addition, while some studies ob-
vestigated the associations between seven air pollutants (CO, NO, NO2, jectively diagnosed diabetes, the others defined diabetes using a self-
NOx, O3, SO2, PM10, and PM2.5) and gestational diabetes, and found report of doctor diagnosis or registered data. These assessments might
significant associations for NO and SO2. Likewise, the included studies have misclassified actual exposure and outcomes and thus under-
in that meta-analysis have high heterogeneity in study design, exposure estimated the effect estimates. Sixth, the number of covariates differed
assessment, and outcome definition, and the number of studies for each among studies and several important covariates (e.g., noise and
air pollutant was limited, which restricts the ability to obtain a more greenspace) were not controlled in most studies. Thus, residual con-
precise estimate with minimal heterogeneity. Thus, more well-designed founding is still possible. Seventh, due to high correlation between air
longitudinal studies remain warranted to explore these topics. pollutants, investigators usually adopted single-pollutant model to
generate effect estimates, while it is likely that the diabetic effects of
4.3. Potential biological mechanisms one pollutant is masked or dominated by the other pollutant(s). Eighth,
we mainly focused on the studies conducted in adults and did not ex-
Although the accurate pathways underlying air pollution and dia- plore the health effects of air pollution on T2D during different stages of
betes remain unclear, several mechanisms have been proposed. the life course as Lim and Thurston (2019) published recently.
Evidence has well-demonstrated that PM exposure is associated with
elevated systemic inflammation and oxidative stress, endoplasmatic
reticulum stress, impaired endothelial function, cardiac autonomic 4.5. Concluding remarks and future perspectives
nervous system dysfunction, and mitochondrial dysfunction (Brook
et al., 2010; Rajagopalan and Brook, 2012; Rückerl et al., 2014; Liu The evidence we reviewed indicates an association of long-term
et al., 2013, 2016). Additionally, evidence also showed that exposure to exposure to air pollution with T2D development and diabetes-related
air pollutants can cause epigenetic changes, which may lead to changes mortality in adults. The results from studies assessing biomarkers of
in markers of coagulation, inflammation, and endothelial function glucose homeostasis in healthy individuals provide more support for a
(Bind et al., 2012; Carmona et al., 2014). Moreover, a few animal potential association. However, because of limited studies available, a
studies suggested a possible role of PM2.5 exposure and insulin re- robust conclusion on the association of air pollution exposure and
sistance (Sun et al., 2009; Liu et al., 2014; Xu et al., 2011). Finally, the diabetes-related traits and GDM cannot be presently drawn. For short
effect modification of air pollution and T2D by a genetic risk score (er)-term air pollution effects in diabetic patients, the recent overall
provided an additional argument for a causal interrelationship between results strengthened the evidence for an adverse triggering effect of air
air pollution and T2D (Eze et al., 2016). pollution on acute mortality. However, these effects can be partly due
to long-term exposure and T2D development, especially considering
4.4. Strengths and limitations that those short-term studies did not adjust for long-term air pollution
exposure. Therefore, results of prospective cohort studies are more
A major strength of our review is that we provided the most sys- valid for the assessment of air pollution effects on mortality cause by
tematic assessment of the current epidemiological evidence on ambient T2D.
air pollution and T2D. We covered many air pollutants (e.g., PM2.5, In conclusion, overall and in particular the recently published stu-
PM10, NO2, SO2, O3, CO) and a rich set of T2D metrics (e.g., prevalence, dies strengthened substantially the evidence for adverse effects of air
incidence, mortality, FBG, HOMA-IR, HbA1C), which otherwise would pollution exposure on T2D. Future studies are also needed to clarify the
be difficult to be obtained from individual studies or isolated reviews or role of air pollution exposure at other places than outdoors and T2D and
meta-analyses. In addition, we assessed the quality and ROB of each to investigate how and to what extent air pollution control strategies
study and graded the credibility of the pooled evidence, which might be can reduce the burden of diabetes-related diseases.
useful for researchers and policy makers to understand the limitations

9
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

Funding sources (9785), 31–40.

Dang, J., Yang, M., Zhang, X., Ruan, X., Qin, G., Fu, J., et al., 2018. Association of ex-
posure to air pollution with insulin resistance: a systematic review and meta-analysis.
The work was funded by the National Natural Science Foundation of Int. J. Environ. Res. Public Health 15 (11), e2593.
China (No. 81872582; No. 81703179; No.81803196). Dendup, T., Feng, X., Clingan, S., Astell-Burt, T., 2018. Environmental risk factors for
developing type 2 diabetes mellitus: a systematic review. Int. J. Environ. Res. Public
Health 15 (1), e78.
Ethics statement Der Simonian, R., Laird, N., 1986. Meta-analysis in clinical trials. Contr. Clin. Trials 7 (3),
177–188.
The study was conducted with the principles stipulated by the Dimakakou, E., Johnston, H.J., Streftaris, G., Cherrie, J.W., 2018. Exposure to environ-
mental and occupational particulate air pollution as a potential contributor to neu-
Declaration of Helsinki and all procedures were approved by the ethics rodegeneration and diabetes: a systematic review of epidemiological research. Int. J.
review committee of Sun Yat-Sen University. Environ. Res. Public Health 15 (8), e1704 2018.
Doi, S.A., Barenregt, J.J., Khan, S., Thlib, L., Williams, G.M., 2015. Simulation compar-
ison of the quality effects and random effects methods of meta-analysis.
Declaration of competing interest
Epidemiology 26 (4), e42–e44.
Egger, M., Davery, S.G., Schneider, M., Minder, C., 1997. Bias in meta-analysis detected
We declare no competing interests. by simple, graphical test. BMJ 315 (7109), 629–634.
Elshahidi, M.H., 2019. Outdoor air pollution and gestational diabetes mellitus: a sys-
tematic review and meta-analysis. Iran. J. Public Health 48 (1), 9–19.
Acknowledgements 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
We thank Dr Carlana Ramlochansingh (Institute and Outpatient North America: systematic review and meta-analysis. Environ. Health Perspect. 123
(5), 381–389.
Clinic for Occupational, Social and Environmental Medicine, University Eze, I.C., Schaffner, E., Fischer, E., Schikowski, T., Adam, M., Imboden, M., et al., 2014.
Hospital of Ludwig Maximilians University, Munich, Germany) for as- Long-term air pollution exposure and diabetes in a population-based Swiss cohort.
sistance in the literature search. Environ. Int. 70, 95–105.
Eze, I.C., Imboden, M., Kumar, A., von Eckardstein, A., Stolz, D., Gerbase, M.W., et al.,
2016. Air pollution and diabetes association: modification by type 2 diabetes genetic
Appendix A. Supplementary data risk score. Environ. Int. 94, 263–271.
Forouzanfar, M.H., Afshin, A., Alexander, L.T., Anderson, H.R., Bhutta, Z.A., Biryukov, S.,
et al., 2016. Global, regional, and national comparative risk assessment of 79 beha-
Supplementary data to this article can be found online at https:// vioural, environmental and occupational, and metabolic risks or clusters of risks,
doi.org/10.1016/j.envres.2019.108817. 1990-2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet
388 (10053), 1659–1724.
Guan, W.J., Zheng, X.Y., Chungm, K.F., Zhong, N.S., 2016. Impact of air pollution on the
References
burden of chronic respiratory diseases in China: time for urgent action. Lancet 388,
1939–1951.
Balti, E.V., Echouffo-Tcheugui, J.B., Yako, Y.Y., Kengne, A.P., 2014. Air pollution and risk He, D., Wu, S., Zhao, H., Qiu, H., Fu, Y., Li, X., et al., 2017. Association between parti-
of type 2 diabetes mellitus: a systematic review and meta-analysis. Diabetes Res. Clin. culate matter 2.5 and diabetes mellitus: a meta-analysis of cohort studies. J. Diabetes
Pract. 106 (2), 161–172. Investig. 8 (5), 687–696.
Bind, M.A., Baccarelli, A., Zanobetti, A., Tarantini, L., Suh, H., Vokonas, P., et al., 2012. Higgins, J.P.T., Green, S., 2011. Selection of study characteristics for subgroup Analyses
Air pollution and markers of coagulation, inflammation, and endothelial function: and meta-regression. Cochran Handbook for Systematic Reviews of Interventions.
associations and epigene-environment interactions in an elderly cohort. Version 5.1.0. http://www.handbook.cochran.org, Accessed date: 1 January 2019.
Epidemiology 23 (2), 332–340. Higgins, J.P., Thompson, S.G., 2011. Quantifying heterogeneity in a meta-analysis. Stat.
Brook, R.D., Rajagopalan, S., Pope 3rd, C.A., Brook, J.R., Bhatnagar, A., Diez-Roux, A.V., Med. 21 (11), 1539–1558.
et al., 2010. Particulate matter air pollution and cardiovascular disease: an update to Hu, H., Ha, S., Henderson, B.H., et al., 2015. Association of atmospheric particulate
the scientific statement from the American Heart Association. Circulation 121 (21), matter and ozone with gestational diabetes mellitus. Environ. Health Perspect. 123
2331–2378. (9), 853–859.
Burnett, r.t., Pope 3rd, C.A., Ezzati, M., Olives, C., Lim, S.S., et al., 2014. An integrated Janghorbani, M., Momeni, F., Mansourian, M., 2014. Systematic review and meta-ana-
risk function for estimating the global burden of disease attributable to ambient fine lysis of air pollution exposure and risk of diabetes. Eur. J. Epidemiol. 29 (4),
particulate matter exposure. Environ. Health Perspect. 122 (4), 397–403. 231–242.
Carmona, J.J., Sofer, T., Hutchinson, J., Cantone, L., Coull, B., Maity, A., et al., 2014. Jerrett, M., Brook, R., White, L.F., Burnett, R.T., Yu, J., Su, J., et al., 2017. Ambient ozone
Short-term airborne particulate matter exposure alters the epigenetic landscape of and incident diabetes: a prospective analysis in a large cohort of African American
human genes associated with the mitogen-activated protein kinase network: a cross- women. Environ. Int. 102, 42–47.
sectional study. Environ. Health 13, 94. Kampa, M., Castanas, E., 2008. Human health effects of air pollution. Environ. Pollut. 151
Cesaroni, G., Forastiere, F., Stafoggia, M., Andersen, Z.J., Badaloni, C., Beelen, R., et al., (2), 362–367.
2014. Long term exposure to ambient air pollution and incidence of acute coronary Kwak, S.H., Choi, S.H., Jung, H.S., Cho, Y.M., Lim, S., Cho, N.H., et al., 2013. Clinical and
events: prospective cohort study and meta-analysis in 11 European cohorts from the genetic risk factors for type 2 diabetes at early or late postpartum after gestational
ESCAPE Project. BMJ 348, f7412. diabetes mellitus. J. Clin. Endocrinol. Metab. 98 (4), E744–E752.
Chen, H., Goldberg, M.S., Burnett, R.T., Jerrett, M., Wheeler, A.J., Villeneuve, P.J., 2013. Lam, J., Sutton, P., Kalkbrenner, A., Windham, G., Halladay, A., Koustas, E., et al., 2016.
Long-term exposure to traffic-related air pollution and cardiovascular mortality. A systematic review and meta-analysis of multiple airborne pollutants and autism
Epidemiology 24 (1), 35–43. spectrum disorder. PLoS One 11 (9), e0161851.
Chen, L., Magliano, D.J., Zimmet, P.Z., 2011. The worldwide epidemiology of type 2 Lao, X.Q., Guo, C., Chang, L.Y., Bo, Y., Zhang, Z., Chuang, Y.C., et al., 2019. Long-term
diabetes mellitus—present and future perspectives. Nat. Rev. Endocrinol. 8 (4), exposure to ambient fine particulate matter (PM2.5) and incidence type 2 diabetes: a
228–236. longitudinal cohort study. Diabetologia. https://doi.org/10.1007/s00125-019-
Chiusolo, M., Cadum, E., Stafoggia, M., Galassi, C., Berti, G., Faustini, A., et al., 2011. 4825-1.
Short-term effects of nitrogen dioxide on mortality and susceptibility factors in 10 Liang, F., Yang, X., Liu, F., Li, J., Xiao, Q., Chen, J., et al., 2019. Long-term exposure to
Italian cities: the EpiAir Study. Environ. Health Perspect. 119 (9), 1233–1238. ambient fine particulate matter and incidence of diabetes in China: a cohort study.
Cho, N.H., Shaw, J.E., Karuranga, S., Huang, Y., da Rocha Fernandes, J.D., Ohlrogge, Environ. Int. 126, 568–575.
A.W., et al., 2018. IDF diabetes atlas: global estimates of diabetes prevalence for 2017 Lim, C.C., Thurston, G.G., 2019. Air pollution, oxidative stress, and diabetes: a life course
and projections for 2045. Diabetes Res. Clin. Pract. 138, 271–281. epidemiologic perspective. Curr. Diabetes Rep. 19 (8), 58.
Cohen, A.J., Brauer, M., Burnett, R., Anderson, H.R., Frostad, J., Estep, K., et al., 2017. Lim, C.C., Hayes, R.B., Ahn, J., Shao, Y., Silverman, D.T., Jones, R.R., et al., 2018.
Estimates and 25-year trends of the global burden of disease attributable to ambient Association between long-term exposure to ambient air pollution and diabetes mor-
air pollution: an analysis of data from the Global Burden of Diseases Study 2015. tality in the US. Environ. Res. 165, 330–336.
Lancet 389 (10082), 1907–1918. Liu, F., Chen, G., Huo, W., Wang, C., Liu, S., Li, N., et al., 2019. Associations between
Coogan, P.F., White, L.F., Yu, J., Burnett, R.T., Seto, E., Brook, R.D., et al., 2016a. Long long-term exposure to ambient air pollution and risk of type 2 diabetes mellitus: a
term exposure to NO2 and diabetes incidence in the Black Women's Health Study. systematic review and meta-analysis. Environ. Pollut. 252 (Pt B), 1235–1245.
Environ. Res. 148, 360–366. Liu, C., Ying, Z., Harkenma, J., Sun, Q., Rajagopalan, S., et al., 2013. Epidemiological and
Coogan, P.F., White, L.F., Yu, J., Burnett, R.T., Seto, E., Brook, R.D., et al., 2016b. PM2.5 experimental links between air pollution and type 2 diabetes. Toxicol. Pathol. 41 (2),
and diabetes and hypertension incidence in the Black Women's Health Study. 361–373.
Epidemiology 10, 202–210. Liu, C., Yang, C., Zhao, Y., Ma, Z., Bi, J., Liu, Y., et al., 2016. Associations between long-
Danaei, G., Finucane, M.M., Lu, Y., Singh, G.M., Cowan, M.J., Paciorek, C.J., et al., 2011. term exposure to ambient air pollution and type 2 diabetes prevalence, blood glucose
National, regional, and global trends in fasting plasma glucose and diabetes pre- and glycosylated hemoglobin levels in China. Environ. Int. 92–93, 416–421.
valence since 1980: systematic analysis of health examination surveys and epide- Liu, C., Xu, X., Bai, Y., Wang, T.Y., Rao, X., Wang, A., et al., 2014. Air pollution-mediated
miological studies with 370 country-years and 2.7 million participants. Lancet 378 susceptibility to inflammation and insulin resistance: influence of CCR2 pathways in

10
B.-Y. Yang, et al. Environmental Research 180 (2020) 108817

mice. Environ. Health Perspect. 122, 17–26. mouse model of diet-induced obesity. Circulation 114 (4), 538–546.
Mustafic, H., Jabre, P., Caussin, C., Murad, M.H., Escolano, S., Tafflet, M., et al., 2012. The Office of Health Assessment and Translation (OHAT), 2015. OHAT Devision of the
Main air pollutants and myocardial infarction: a systematic review and meta-analysis. National Toxicology Program. National Institute of Environmental Health Sciences
J. Am. Med. Assoc. 307 (7), 713–721. 2015.
Nanditha, A., Ma, R.C., Ramachandran, A., Snehalatha, C., Chan, J.C., Chia, K.S., et al., Thiering, E., Heinrich, J., 2015. Epidemiology of air pollution and diabetes. Trends
2016. Diabetes in Asia and the Pacific: implications for the global epidemic. Diabetes Endocrinol. Metab. 26 (7), 384–394.
Care 39, 472–485. Turner, M.C., Krewski, D., Pope 3rd, C.A., Chen, Y., Gapstur, S.M., Thun, M.j., 2011. Long-
NCD Risk Factor Collaboration (NCD-RisC), 2016. Worldwide trends in diabetes since term ambient fine particulate matter air pollution and lung cancer in a large cohort of
1980: a pooled analysis of 751 population-based studies with 4.4 million participants. never-smokers. Am. J. Respir. Crit. Care Med. 184 (12), 1374–1381.
Lancet 9 (387), 1513–1530. van Dieren, S., Beulens, J.W., van der Schouw, Y.T., Grobbee, D.F., Neal, B., 2010. The
Park, S.K., Wang, W., 2014. Ambient air pollution and type 2 diabetes: a systematic re- global burden of diabetes and its complications: an emerging pandemic. Eur. J.
view of epidemiologic research. Curr. Environ. Health Rep. 1 (3), 275–286. Cardiovasc. Prev. Rehabil. 17 (Suppl. 1), S3–S8.
Pope, C.A., Turner, M.C., Burnett, R.T., Jerrett, M., Gapstur, S.M., Diver, W.R., et al., van Donkelaar, A., Martin, R.V., Brauer, M., Boys, B.L., 2015. Use of satellite observations
2015. Relationships between fine particulate air pollution, cardiometabolic disorders, for long-term exposure assessment of global concentrations of fine particulate matter.
and cardiovascular mortality. Circ. Res. 116, 108–115. Environ. Health Perspect. 123 (2), 135–143.
Raaschou-Nielsen, O., Andersen, Z.J., Beelen, R., Samoli, E., Stafoggia, M., et al., 2013. Wang, B., Xu, D., Jing, Z., Liu, D., Yan, S., Wang, Y., 2014. Effect of long-term exposure to
Air pollution and lung cancer incidence in 17 European cohorts: prospective analyses air pollution on type 2 diabetes mellitus risk: a systemic review and meta-analysis of
from the European study of cohorts for air pollution effects (ESCAPE). Lancet Oncol. cohort studies. Eur. J. Endocrinol. 171 (5), R173–R182.
14 (9), 813–822. Wells, G., O'Connell, S.B., Peterson, D., Welch, J., Losos, V., Tugwell, M.P., 2013. The
Rajagopalan, S., Brook, R.D., 2012. Air pollution and type 2 diabetes: mechanistic in- Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in
sights. Diabetes 61 (12), 3037–3045. meta-analyses. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp,
Renzi, M., Cerza, F., Gariazzo, C., Agabiti, N., Cascini, S., Di Domenicantonio, R., et al., Accessed date: 7 March 2019.
2018. Air pollution and occurrence of type 2 diabetes in a large cohort study. Xu, X., Liu, C., Xu, Z., Tzan, K., Zhong, M., Wang, A., et al., 2011. Long-term exposure to
Environ. Int. 112, 68–76. ambient fine particulate pollution induces insulin resistance and mitochondrial al-
Rohde, R.A., Muller, R.A., 2015. Air pollution in China: mapping of concentrations and teration in adipose tissue. Toxicol. Sci. 124, 88–98.
sources. PLoS One 10, e0135749. Yang, B.Y., Qian, Z.M., Li, S., Chen, G., Bloom, M.S., Elliott, M., et al., 2018a. Ambient air
Rückerl, R., Hampel, R., Breitner, S., Cyrys, J., Krau, U., Carter, J., et al., 2014. pollution in relation to diabetes and glucose homoeostasis markers in China: a cross-
Associations between ambient air pollution and blood markers of inflammation and sectional study with findings from the 33 Communities Chinese Health Study. Lancet
coagulation/fibrinolysis in susceptible populations. Environ. Int. 70, 32–49. Planet. Health 2 (2), e64–e73.
Shah, A.S., Langrish, J.P., Nair, H., McAllister, D.A., Hunter, A.L., Donaldson, K., et al., Yang, Y., Guo, Y., Qian, Z.M., Ruan, Z., Zheng, Y., Woodward, A., et al., 2018b. Ambient
2013. Global association of air pollution and heart failure: a systematic review and fine particulate pollution associated with diabetes mellitus among the elderly aged 50
meta-analysis. Lancet 382 (9897), 1039–1048. years and older in China. Environ. Pollut. 243 (Pt B), 815–823.
Shamseer, L., Moher, D., Clarke, M., Ghersi, D., Liberati, A., Petticrew, M., et al., 2015. Yang, B.Y., Qian, Z., Howard, S.W., Vaughn, M.G., Fan, S.J., Liu, K.K., et al., 2018c.
Preferred reporting items for systematic review and meta-analysis protocols Global association between ambient air pollution and blood pressure: a systematic
(PRISMA-P) 2015: elaboration and explanation. BMJ 350, g7647. review and meta-analysis. Environ. Pollut. 235, 576–588.
Stafoggia, M., Forastiere, F., Faustini, A., Biggeri, A., Bisanti, L., Cadum, E., et al., 2010. Zanobetti, A., Schwartz, J., 2011. Ozone and survival in four cohorts with potentially
Susceptibility factors to ozone-related mortality: a population-based case-crossover predisposing diseases. Am. J. Respir. Crit. Care Med. 184 (7), 836–841.
analysis. Am. J. Respir. Crit. Care Med. 182 (3), 376–384. Zuniga, J., Tarajia, M., Herrera, V., Urriola, W., Gomez, B., Motta, J., 2016. Assessment of
Sun, Q., Yue, P., Deiuliis, J.A., Lumeng, C.N., Kampfrath, T., Mikolaj, M.B., et al., 2009. the possible association of air pollutants PM10, O3, NO2 with an increase in cardio-
Ambient air pollution exaggerates adipose inflammation and insulin resistance in a vascular, respiratory, and diabetes mortality in Panama city. Medicine 95 (2), e2464.

11