PLOS MEDICINE

RESEARCH ARTICLE

Obesity and the relation between joint

exposure to ambient air pollutants and
incident type 2 diabetes: A cohort study in UK
Biobank
Xiang Li ID1, Mengying Wang1,2, Yongze Song ID3, Hao Ma1, Tao Zhou1,4, Zhaoxia Liang5,
Lu Qi ID1,6*

1 Department of Epidemiology, School of Public Health and Tropical Medicine, Tulane University, New
a1111111111
Orleans, Louisiana, United States of America, 2 Department of Epidemiology and Biostatistics, School of
a1111111111 Public Health, Peking University, Beijing, China, 3 School of Design and the Built Environment, Curtin
a1111111111 University, Bentley, Perth, Western Australia, Australia, 4 Department of Epidemiology and Biostatistics,
a1111111111 School of Public Health (Shenzhen), Sun Yat-sen University, Guangzhou, Guangdong, China, 5 Department
a1111111111 of Obstetrics, Women’s Hospital, Zhejiang University School of Medicine, Hangzhou, Zhejiang Province,
China, 6 Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, United
States of America

* [email protected]

OPEN ACCESS

Citation: Li X, Wang M, Song Y, Ma H, Zhou T,

Liang Z, et al. (2021) Obesity and the relation
Abstract
between joint exposure to ambient air pollutants
and incident type 2 diabetes: A cohort study in UK
Biobank. PLoS Med 18(8): e1003767. https://doi. Background
org/10.1371/journal.pmed.1003767 Air pollution has been related to incidence of type 2 diabetes (T2D). We assessed the joint
Academic Editor: Yuming Guo, Monash association of various air pollutants with the risk of T2D and examined potential modification
University, AUSTRALIA by obesity status and genetic susceptibility on the relationship.
Received: March 13, 2021

Accepted: August 13, 2021

Methods and findings
A total of 449,006 participants from UK Biobank free of T2D at baseline were included. Of all
Published: August 30, 2021
the study population, 90.9% were white and 45.7% were male. The participants had a mean
Copyright: © 2021 Li et al. This is an open access
age of 56.6 (SD 8.1) years old and a mean body mass index (BMI) of 27.4 (SD 4.8) kg/m2.
article distributed under the terms of the Creative
Commons Attribution License, which permits Ambient air pollutants, including particulate matter (PM) with diameters �2.5 μm (PM2.5),
unrestricted use, distribution, and reproduction in between 2.5 μm and 10 μm (PM2.5–10), nitrogen dioxide (NO2), and nitric oxide (NO) were
any medium, provided the original author and
measured. An air pollution score was created to assess the joint exposure to the 4 air pollut-
source are credited.
ants. During a median of 11 years follow-up, we documented 18,239 incident T2D cases.
Data Availability Statement: Requests to access The air pollution score was significantly associated with a higher risk of T2D. Compared to
the dataset from qualified researchers trained in
human participants confidentiality protocols may
the lowest quintile of air pollution score, the hazard ratio (HR) (95% confidence interval [CI])
be sent to UK Biobank (https://www.ukbiobank.ac. for T2D was 1.05 (0.99 to 1.10, p = 0.11), 1.06 (1.00 to 1.11, p = 0.051), 1.09 (1.03 to 1.15,
uk/researchers/). The current study has been p = 0.002), and 1.12 (1.06 to 1.19, p < 0.001) for the second to fifth quintile, respectively,
conducted using the UK Biobank Resource under
after adjustment for sociodemographic characteristics, lifestyle factors, genetic factors, and
Application 29256.
other covariates. In addition, we found a significant interaction between the air pollution
Funding: LQ received awards from National Heart,
score and obesity status on the risk of T2D (p-interaction < 0.001). The observed associa-
Lung, and Blood Institute (Award # HL071981,
HL034594, HL126024, URL: https://www.nhlbi.nih. tion was more pronounced among overweight and obese participants than in the normal-
gov/); the National Institute of Diabetes and weight people. Genetic risk score (GRS) for T2D or obesity did not modify the relationship

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 1 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

Digestive and Kidney Diseases (Awards # between air pollution and risk of T2D. Key study limitations include unavailable data on
DK115679, DK091718, DK100383, URL: https:// other potential T2D-related air pollutants and single-time measurement on air pollutants.
www.niddk.nih.gov/), the Fogarty International
Center (award # TW010790, URL: https://www.fic.
nih.gov/), and Tulane Research Centers of Conclusions
Excellence Awards (URL: https://sph.tulane.edu/).
We found that various air pollutants PM2.5, PM2.5–10, NO2, and NO, individually or jointly,
XL received award from the American Heart
Associatioion (award #: 19PRE34380036; URL: were associated with an increased risk of T2D in the population. The stratified analyses indi-
https://professional.heart.org/en/research- cate that such associations were more strongly associated with T2D risk among those with
programs/application-information/predoctoral- higher adiposity.
fellowship). The funders have no role in the study
design, data collection and analysis, decision to
publish, or preparation of the manuscript.

Competing interests: The authors have declared

that no competing interests exist. Author summary
Abbreviations: AU
BMI,: body mass index; CI,
Anabbreviationlisthasbeencompiledforthoseusedthroughoutthetext:Pleaseverifythatallentriesarecorrect:
confidence interval; ESCAPE, European Study of Why was this study done?
Cohorts for Air Pollution Effects; GIS, Geographic
Information System; GRS, genetic risk score; HR, • Recent evidence has linked ambient air pollution with type 2 diabetes (T2D). However,
hazard ratio; IDF, International Diabetes Federation; most of them were mainly focused on 1 or 2 air pollutants separately. In addition, it has
IPAQ, International Physical Activity Questionnaire; been reported that the association between air pollution and T2D was augmented
LUR, Land Use Regression; MET, metabolic
among people with increased adiposity.
equivalent task; NO, nitric oxide; NO2, nitrogen
dioxide; PM, particulate matter; SNP, single • Individuals are usually exposed to various air pollutants simultaneously; and the impor-
nucleotide polymorphism; T2D, type 2 diabetes;
tance of assessing multi-air pollutant exposures as a whole has been increasingly
WC, waist circumference.
recognized.
• Thus, in this study, we aimed to analyze the association between the air pollution score
that incorporates various air pollutants with the risk of T2D. We also particularly exam-
ined the potential modification by obesity.

What did the researchers do and find?

• We built an air pollution score, incorporating PM2.5, PM2.5–10, nitrogen dioxide (NO2),
and nitric oxide (NO), and analyzed its association with the risk of T2D among 449,006
participants from UK Biobank. Air pollutants were measured in 2010. During a median
follow-up time of 11 years, 18,239 T2D cases were documented.
• We found that long-term exposure to various air pollutants, individually or jointly, was
associated with an increased risk of T2D. Moreover, such association was more pro-
nounced among people with increased adiposity.

What do these findings mean?

• Our findings highlight the importance of comprehensively assessing various air pollut-
ants as a whole in the prevention of T2D.
• Given the adverse impact of air pollution on T2D, attention should be paid to reducing
and mitigating air pollution.
• Our results also emphasize the merit of obesity management.

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 2 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

Introduction
Type 2 diabetes (T2D) is a persistent public threat worldwide. The prevalence of T2D has been
increasing, and it is set to rise even further. From the most recent International Diabetes Fed-
eration (IDF), it is estimated that there will be more than 700 million adults living with diabe-
tes worldwide by 2045 [1]. While several traditional risk factors, such as poor diet [2,3], low
physical activity [4,5], and poor sleep behaviors have been related to T2D [6,7], recent evidence
suggests that ambient air pollution may also contribute to the development of the disease [8–
10]. However, the relations between air pollution and T2D is a relatively new field, and the pre-
vious findings between air pollution and T2D were inconsistent [11,12]. Notably, the prior
studies are largely limited by cross-sectional design [13–15], and small sample size [16], while
data from large-scale, cohort settings are still scarce. In addition, individuals are usually
exposed to a combination of various air pollutants simultaneously, and the importance of
assessing multi-air pollutant exposures as a whole has been increasingly recognized [17–19].
However, most previous studies were mainly focused on 1 or 2 air pollutants separately, with-
out considering the joint exposure to various air pollutants [15,20–23]. Recently, we have cre-
ated and validated a novel air pollution score incorporating various air pollutants, which has
been shown to be associated with heart failure [24]. To date, no study has jointly investigated
long-term exposure to various air pollutants with the risk of T2D in a cohort study design.
Of note, emerging evidence has also linked air pollutants such as PM2.5, nitrogen dioxide
(NO2), and O3 with increased adiposity [25,26], and it was reported that the association
between air pollutants and T2D was augmented in obese participants [27]. Therefore, we
hypothesized that obesity status might modify the relationship between air pollution and the
risk of T2D. In addition, prior evidence suggests that the genetic factors may modify the envi-
ronment–disease relation, while investigations on the modification effect by the genetic predis-
position on the relation between air pollution and T2D risk are scarce [28].
In the present study, we aimed to analyze the associations of various air pollutants and the
air pollution score, which comprehensively incorporated PM2.5, PM2.5–10, NO2, and nitric
oxide (NO), with the risk of T2D among 449,006 participants from UK Biobank. We particu-
larly examined the potential modification by obesity status (both general and central obesity)
and genetic predisposition to obesity or T2D.

Method
Protocol
This research has been conducted using the UK Biobank Resource under Application Number
29256. Our study did not employ a prospective protocol. Analyses were first planned and per-
formed in January 2021. During the peer review, we added a table of Pearson correlations
between each of the air pollutant and a map of air pollution levels exposed to study participants
at baseline (2010). Minor changes to the manuscript were also made at the request of peer
reviewers.

Study population
The UK Biobank is a prospective cohort based in the United Kingdom aimed to improve the
prevention, diagnosis, and treatment of a wide range of illnesses. Briefly, over 500,000 middle-
aged participants were recruited across 2006 to 2010. Participants provided a wide range of
health-related information through touchscreen questionnaires, physical measurements, and
biological samples at baseline or follow-up assessment. Details of the study design have been

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 3 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

described elsewhere previously [29]. The study was approved by both the National Health Ser-
vice National Research Ethics Service (Ref: 11/NW/0382) and the Institutional Review Board
of Tulane University (2018–1872). All the participants provided written informed consent.
In the current analysis, we excluded participants with T2D (N = 12,185) and those with
missing information on residential air pollution (N = 41,302) at baseline, leaving a total of
449,006 participants for the primary analysis. In the joint association of genetic risk and air
pollution analysis, to avoid heterogeneity, we only included participants of European descent
with complete genotyping data (N = 417,035).

Assessment of air pollution and air pollution score

The annual average air pollution, including PM2.5, PM2.5–10, NO2, and NOx, in 2010, were
modeled for each address using a Land Use Regression (LUR) model developed as part of the
European Study of Cohorts for Air Pollution Effects (ESCAPE, http://www.escapeproject.eu/)
[30,31] and linked to participants’ residential addresses given at baseline visit. The LUR models
calculated the spatial variations of annual average air pollutant concentration at participants’
home addresses using Geographic Information System (GIS)-derived predictors, such as traf-
fic, land use, and topography. The LUR model is based on ESCAPE monitoring done between
January 2010 and January 2011, and air pollution estimates are representative for the year
2010. The ESCAPE estimates for particulates are valid up to 400 km from the monitoring area
[30,31]. The concentration of NO is estimated by subtracting NO2 from NOx following the
previous study [32].
To capture the joint exposure to various air pollutants, we created an air pollution score by
summing up concentrations of 4 ambient air pollutants (PM2.5, PM2.5–10, NO2, and estimated
NO), weighted by the multivariable-adjusted risk estimates (β coefficients) on T2D in the pres-
ent analysis [24]. The β coefficient was from the final model with individual air pollutant as the
independent variable, one at a time. The equation was: air pollution score = (βPM2.5 × PM2.5 +
βPM2.5–10 × PM2.5–10 + βNO2 × NO2 + βNO × NO) × (4 / sum of the β coefficients). The air pollu-
tion score ranges from 31.7 to 140.3, a higher score indicating greater exposure to ambient air
pollution. Since PM2.5–10 was not significantly associated with T2D, we also created a weighted
air pollution score without PM2.5–10.

Assessment of outcomes
Information on incident T2D was collected through February 8, 2020. Incident T2D was
defined by ICD-10 code E11, ascertained by hospital inpatient records containing data on
admissions and diagnoses from the Hospital Episode Statistics for England, Scottish Morbidity
Record data for Scotland, and the Patients Episode Database for Wales.

Genotype data
Genotyping, imputation, and quality control of the genetic data were performed by the UK
Biobank team. The detailed information is available elsewhere (http://www.ukbiobank.ac.uk/
scientists-3/genetic-data/) [33]. A genetic risk score (GRS) for T2D was created using 112
independent single nucleotide polymorphisms (SNPs), which passed quality control out of the
128 SNPs recently identified to be associated with T2D at genome-wide significance [34,35].
Information of the 112 independent SNPs is provided in S1 Table. The GRS for T2D was cal-
culated by the weighted method: GRS = (β1 × SNP1 + β2 × SNP2 + . . . + β112 × SNP112) × (112 /
sum of the β coefficients). Each SNP was recoded as 0, 1, and 2 according to the number of
risk alleles. The β coefficient was obtained from the reported GWAS meta-analysis [35]. The
GRS for T2D in the current analysis ranges from 81.0 to 136.9; a higher score indicates a higher

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 4 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

genetic predisposition to T2D. The BMI-GRS was created in the same way as T2D-GRS, using
the 97 identified body mass index (BMI)-related SNPs [36]. Information of the BMI-related
SNPs is summarized in S2 Table. We determined whether participants were at low, intermedi-
ate, or high genetic risk according to the tertile categories of the GRS.

Assessment of other covariates

Age, sex, ethnicity, and Townsend deprivation index (based on the participant’s postcode,
higher scores indicate a higher degree of deprivation) were obtained at baseline. Weight,
height, and waist circumference (WC) were measured at baseline during the initial assessment
center visit. BMI was calculated as weight divided by height squared (kg/m2) during the initial
Assessment Centre visit. Overall obesity status was defined as follows: normal weight: 18.5 to
less than 25 kg/m2; overweight: 25 to less than 30 kg/m2; obese: 30 kg/m2 and above. Central
obesity was defined using the European cutoff points for WC: �94 cm for men and �80 cm
for women. Physical activity was assessed by the International Physical Activity Questionnaire
(IPAQ) and metabolic equivalent task (MET) score was calculated using the IPAQ guideline
[37,38]. Alcohol intake was assessed by the Touchscreen questionnaire. A healthy diet score
was adapted from the American Heart Association Guidelines and defined as adherence to 4
or 5 components of the following: (1) total fruit intake �4.5 pieces/week; (2) total vegetable
intake �4.5 servings/week (3 tablespoons of vegetable considered as 1 serving); (3) total fish
intake �2 servings/week; (4) processed meat intake less than twice/week; and (5) red meat
intake �5 times/week [39].

Statistical analysis
Follow-up time was calculated from the recruitment date to the first diagnosis of T2D, lost to
follow-up, death, or end of the current follow-up, whichever came first. Cox proportional haz-
ard models were used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs)
for incident T2D associated with the individual air pollutant or the air pollution score. The
proportional hazards assumption was tested by creating a time-dependent variable, and no
violation was found. Cox regression models were adjusted for age, sex, and ethnicity in model
1. Model 2 was further adjusted for Townsend deprivation index, assessment center, alcohol
intake, smoking status, physical activity, sedentary hours, and healthy diet score. In model 3,
we additionally adjusted for BMI, systolic blood pressure, antihypertension medication use,
high cholesterol, and T2D-GRS. We examined the dose–response relationship between the air
pollution score and incident T2D using the restricted cubic spline analysis. We particularly
tested whether the association between the air pollution score and risk of T2D was modified
by obesity status (overall obesity and central obesity) or genetic predisposition to T2D or obe-
sity, by including an interaction term between the air pollution score and the continuous BMI
or WC. To disentangle the temporal concerns regarding the relationship between air pollution
and T2D, we further conducted a sensitivity analysis by excluding T2D cases that occurred in
the first 2 years of follow-up. In addition, we performed the analyses among participants who
have been living in the current address for at least 5 years to assess the long-term effect of air
pollution on T2D.
Statistical analyses were performed with SAS version 9.4 (SASAU Institute,
: PerPLOSstyle; All p-
Cary, NC).donotuseInc:excep
values were two-sided, and p < 0.05 was considered statistically significant.
The study was reported as per the Strengthening the Reporting of Observational Studies in
Epidemiology (STROBE) guidelines (S1 STROBE checklist).

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 5 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

Results
Among the 449,006 participants, a total of 18,239 incident cases of T2D were recorded during
a median follow-up time of 11 years. Baseline characteristics of the study participants accord-
ing to the quintiles of the air pollution score are presented in Table 1. Participants with higher
exposure to ambient air pollution were younger, with a greater Townsend deprivation index
(indicating a greater degree of deprivation), more likely to be current smokers, and less likely
to be physically active. The Pearson correlation coefficients among the individual air pollutants
were shown in S3 Table. Air pollution levels in areas where participants lived in 2010 were
shown in Fig 1. The distribution patterns of air pollution levels in the current study are compa-
rable to the spatial distribution of air pollutants from the public UK Air Information Resources
(https://uk-air.defra.gov.uk/data/pcm-data). The administrative boundary data of UK is
sourced from UK Government Open Data portal (https://data.gov.uk/dataset/3fd8d2d2-b591-
42ff-b333-c53a6a513e96/countries-december-2017-full-clipped-boundaries-in-great-britain).
These data are UK government–released open data.

Table 1. Baseline characteristics of the UK Biobank participants according to the quintiles of air pollution score (N = 449,006).
Air Pollution Score
Q1 Q2 Q3 Q4 Q5
Age, year 57.2 (7.8) 56.9 (8.0) 56.7 (8.1) 56.1 (8.2) 55.4 (8.3)
White 85,814 (95.2) 83,849 (93.6) 82,394 (92.2) 79,749 (89.7) 74,540 (84.1)
Male 40,524 (45.1) 40,443 (45.0) 40,483 (45.1) 40,514 (45.1) 41,078 (45.7)
PM2.5, μg/m3 8.7 (0.4) 9.5 (0.3) 9.9 (0.3) 10.4 (0.4) 11.5 (0.9)
PM2.5–10, μg/m3 6.2 (0.8) 6.2 (0.8) 6.3 (0.8) 6.5 (0.9) 6.9 (1.0)
NO2, μg/m3 17.3 (2.8) 22.8 (2.7) 26.3 (2.8) 29.8 (2.9) 37.0 (6.4)
NO, μg/m3 9.4 (2.8) 13.4 (3.6) 16.0 (3.7) 18.9 (3.9) 29.1 (11.9)
BMI, kg/m2 27.0 (4.4) 27.3 (4.6) 27.4 (4.7) 27.5 (4.8) 27.5 (5.0)
WC, cm 89.2 (13.0) 89.9 (13.1) 90.3 (13.3) 90.5 (13.4) 90.4 (13.6)
Systolic blood pressure, mm Hg 138.8 (18.5) 138.2 (18.5) 138.1 (18.4) 137.4 (18.4) 136.2 (18.3)
Townsend deprivation index −3.2 [−4.2–−2.1] −2.9 [−4.1–−1.3] −2.4 [−3.7–−0.30] −1.4 [−3.0–1.0] 0.9 [−1.6–3.6]
Total MET 2,466.0 [1,048.5–3,155.0] 2,435 [996.0–2,979.0] 2,493.0 [1,010.0–2,946.0] 2,559.0 [1,022.0–2,932.0] 2,506.5 [1,032.0–2,882.5]
Healthy diet score 2.1 (0.9) 2.1 (0.9) 2.1 (0.9) 2.1 (0.9) 2.1 (0.9)
Sedentary time 4.0 [3.0–6.0] 4.0 [3.0–6.0] 4.0 [3.0–6.0] 4.0 [3.0–6.0] 4.0 [3.0–6.0]
Smoking status
Never 52,165 (58.3) 50,860 (56.9) 49,428 (55.4) 48,503 (54.4) 44,740 (50.2)
Previous 30,835 (34.5) 30,926 (34.6) 30,944 (34.7) 30,458 (34.1) 31,140 (35.0)
Current 6,501 (7.3) 7,597 (8.5) 8,902 (10.0) 10,251 (11.5) 13,200 (14.8)
Alcohol intake
Daily or almost daily 22,284 (24.8) 18,132 (20.2) 17,114 (19.1) 16,624 (18.6) 18,330 (20.5)
3–4 times/week 23,211 (25.9) 21,628 (24.1) 20,492 (22.9) 19,924 (22.3) 18,832 (21.1)
1–2 times/week 22,442 (25.0) 23,985 (26.8) 24,009 (26.8) 23,404 (26.2) 21,626 (24.2)
1–3 times/month 8,929 (10.0) 10,198 (11.4) 10,379 (11.6) 10,413 (11.6) 9,728 (10.9)
Special occasions only 7,932 (8.8) 9,650 (10.8) 10,612 (11.9) 11,280 (12.6) 11,520 (12.9)
Never 4,905 (5.5) 6,035 (6.7) 6,967 (7.8) 7,867 (8.8) 9,268 (10.4)
Hypertension, yes 47,228 (52.6) 47,418 (52.8) 47,391 (52.8) 45,971 (51.2) 43,542 (48.5)
Antihypertension meds, yes 16,572 (18.5) 17,633 (19.8) 18,015 (20.2) 17,939 (20.2) 17,245 (19.4)
Cholesterol lowering meds, yes 12,826 (14.4) 13,935 (15.6) 14,648 (16.4) 14,894 (16.7) 14,602 (16.5)

Data are mean (SD), Median [IQR], or N (%).

BMIAU
, body
: AbbreviationlistshavebeencompiledforthoseusedinTables1
mass index; MET, metabolic equivalent task; NO, nitric oxide; NO2, 3:Pleaseverifythatallentriesarecorrect:
nitrogen dioxide; PM, particulate matter; WC, waist circumference.

https://doi.org/10.1371/journal.pmed.1003767.t001

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 6 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

Fig 1. Map of air pollutions (PM2.5, PM2.5–10, NO2, and NO) of areas where participants lived in 2010. (a) PM2.5, (b) PM2.5–10, (c) NO2, and (d) NO. The
administrative boundary data of the UK are sourced from UK Government Open Data portal (https://data.gov.uk/dataset/3fd8d2d2-b591-42ff-
b333-c53a6a513e96/countries-december-2017-full-clipped-boundaries-in-great-britain). These data are UK government–released open data. NOAU , nitric
: Abbreviationlistshav
oxide;
NO2, nitrogen dioxide; PM2.5, particulate matter with aerodynamic diameter �2.5 μm; PM2.5–10, particulate matter with an aerodynamic diameter between 2.5
and 10 μm.
https://doi.org/10.1371/journal.pmed.1003767.g001

The associations between individual air pollutants and the risk of T2D are shown in
Table 2. We found that PM2.5, NO2, and NO were each associated with an increased risk of
T2D in the multivariate-adjusted models, while a positive and marginal association was found
for PM2.5–10. Compared to individuals exposed to the lowest quintile of air pollutant, the HR
for T2D for those exposed to the highest quintile was 1.12 (95% CI: 1.05 to 1.18, p = 0.002) for
PM2.5, 1.11 (1.04 to 1.18, p < 0.001) for NO2, and 1.12 (1.06 to 1.18, p < 0.001) for NO,
respectively.
When jointly considering the 4 air pollutants by the air pollution score, we found that par-
ticipants exposed to higher levels of the score were significantly associated with a higher risk of
T2D (Table 3). In the age, ethnicity, and sex-adjusted model (model 1), individuals within the

Table 2. HRs of T2D by individual air pollutant concentration among 461,191 UK Biobank participants.
HR (95% CI) per SD increase Air pollutant concentration in quintiles P for trend
Q1 Q2 Q3 Q4 Q5
PM2.5 1.04 (1.02, 1.05) Ref. 1.08 (1.02, 1.14) 1.06 (1.01, 1.12) 1.06 (1.01, 1.12) 1.12 (1.05, 1.18) 0.002
PM2.5–10 1.01 (0.99, 1.03) Ref. 1.02 (0.97, 1.07) 1.04 (0.98, 1.09) 1.01 (0.96, 1.06) 1.06 (1.00, 1.11) 0.081
NO2 1.05 (1.02, 1.07) Ref. 1.04 (0.99, 1.10) 1.07 (1.01, 1.13) 1.11 (1.05, 1.17) 1.11 (1.04, 1.18) <0.001
NO 1.04 (1.02, 1.05) Ref. 1.06 (1.00, 1.11) 1.06 (1.01, 1.12) 1.12 (1.06, 1.18) 1.12 (1.06, 1.18) <0.001

CI, confidence interval; GRS, genetic risk score; HR, hazard ratio; NO, estimated nitric oxide; NO2, nitrogen dioxide; PM2.5, particulate matter with aerodynamic
diameter �2.5 μm; PM2.5–10, particulate matter with an aerodynamic diameter between 2.5 and 10 μm; T2D, type 2 diabetes.
Models were adjusted for age, ethnicity, sex, Townsend deprivation index, center, alcohol intake, smoking status, physical activity, sedentary hour, healthy diet score,
BMI, systolic blood pressure, antihypertension meds, high cholesterol, and T2D GRS.

https://doi.org/10.1371/journal.pmed.1003767.t002

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 7 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

Table 3. Associations between air pollution score and incident T2D among 461,191 UK Biobank participants.
Model 1 Model 2 Model 3
HR (95% CI) p-value HR (95% CI) p-value HR (95% CI) p-value
Air pollution score per SD 1.17 (1.16, 1.19) <0.001 1.05 (1.03, 1.07) <0.001 1.04 (1.02, 1.06) <0.001
Q1 ref - ref - ref -
Q2 1.20 (1.14, 1.27) <0.001 1.07 (1.02, 1.13) 0.007 1.05 (0.99, 1.10) 0.11
Q3 1.34 (1.28, 1.41) <0.001 1.11 (1.05, 1.16) <0.001 1.06 (1.00, 1.11) 0.051
Q4 1.51 (1.44, 1.58) <0.001 1.15 (1.09, 1.21) <0.001 1.09 (1.03, 1.15) 0.002
Q5 1.71 (1.63, 1.80) <0.001 1.16 (1.10, 1.22) <0.001 1.12 (1.06, 1.19) <0.001

Model 1: adjusted for age, ethnicity, and sex.

Model 2: Model 1+ Townsend deprivation index, center, alcohol intake, smoking status, physical activity, sedentary hour, and healthy diet score.
Model 3: Model 2+ BMI, SBP, antihypertension meds, high cholesterol, and T2D-GRS.
BMI, body mass index; CI, confidence interval; GRS, genetic risk score; HR, hazard ratio; SBP, systolic blood pressure; T2D, type 2 diabetes.

https://doi.org/10.1371/journal.pmed.1003767.t003

fifth quintile (Q5) of air pollution score had a 71% increased risk of developing T2D, compared to
those in Q1 (p < 0.001). After further adjustment for assessment center, Townsend deprivation
index, alcohol intake, smoking status, total physical activity, sedentary hours, and healthy diet
score, the association between air pollution score and risk of T2D attenuated but remained signifi-
cant (HR for extreme quintiles was 1.16 (1.10 to 1.22, p < 0.001) in model 2). Additional adjust-
ment for BMI, systolic blood pressure, antihypertension medication use, high cholesterol, and
GRS of T2D did not appreciably change the result (HR for extreme quintiles was 1.12 (1.06 to
1.109, p < 0.001) in model 3). Also, the results were stable when we excluded PM2.5–10 in the air
pollution score (S4 Table). Sensitivity analysis excluding T2D cases developed within the first 2
years of follow-up yielded similar results (S5 Table). Moreover, the results were robust when only
participants living in the current address for at least 5 years were included (S6 Table).
Intriguingly, we found significant interactions between the air pollution score and BMI on
the risk of T2D (S7 Table; p-interaction < 0.001). In the stratified analysis according to the
obesity status (normal-weight, overweight, and obesity), the associations between air pollution
score and T2D risk appeared to be stronger among overweight and obese participants, while
no association was found among those with normal-weight (Fig 2). The spline analysis showed

Fig 2. Dose-responsive relationship of air pollution score with T2D incidence according to obesity status. Dashed lines represent the 95% CIs of the HR.
Multivariable models were adjusted for age, sex, Townsend deprivation index, center, alcohol intake, smoking status, physical activity, sedentary hours, healthy
diet score, systolic blood pressure, antihypertension meds, high cholesterol, and T2D GRS. Sample size for normal weight, overweight, and obese subgroup
were 149,945, 193,492, and 105,569, respectively. CI, confidence interval; GRS, genetic risk score; HR, hazard ratio; T2D, type 2 diabetes.
https://doi.org/10.1371/journal.pmed.1003767.g002

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 8 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

a significant linear relationship between air pollution score and the risk of T2D among over-
weight and obese participants (p for curvature was 0.19 and 0.42, p for linearity was 0.02 and
0.0001, respectively, in the overweight and obese group). A similar interaction pattern was
observed between air pollution score and central obesity status (S1 Fig and S7 Table; p-inter-
action = 0.015). We also further tested the interaction between air pollution score and obesity
GRS on the risk of T2D. However, we did not find a significant modification effect by the obe-
sity GRS, though the relationship was statistically significant among participants with a high
genetic risk of obesity (S2 Fig).
We did not observe a significant interaction between the air pollution score and the genetic
risk of T2D, suggesting that such association was consistent regardless of participants’ genetic
predisposition to T2D (S3 Fig). We noted that the relationship between air pollution score and
T2D were stronger among those with intermediate/high genetic predisposition to T2D, though
the test for interaction was not significant.

Discussion
In this large-scale cohort, we found that long-term exposure to various ambient air pollutants
including PM2.5, PM2.5–10, NO2, and NO, individually or jointly as an air pollution score, was
significantly associated with an increased risk of T2D in the whole population in a dose-
responsive fashion, independent of the traditional risk factors. In addition, we found that the
association was significantly modified by obesity status, with more pronounced associations
observed among overweight/obese individuals than in normal-weight individuals.
The relationship between air pollution and T2D has only been recently investigated. Our
findings on the relations between individual air pollutants and T2D are in line with the results
from several cohort studies assessing the long-term exposure to air pollutants [16,20–
23,27,40]. A study conducted in Hong Kong (N = 57,053) among an elder population (aged 65
+) with a mean follow-up of 9.8 years demonstrated that PM2.5 was associated with an
increased risk of T2D [40]. NO2 and NOx were also found to be associated with an increased
risk of T2D [16,20,22]. The association between PM2.5–10 and T2D was less studied, and the
results were inconsistent [14,23,41]. The conflicting findings from the limited body of previous
studies are partly accounted for by the varies in study design, population, sample size, air pol-
lutants measurement, as well as exposure duration.
We observed significant associations of the joint exposure to various air pollutants, evalu-
ated by an air pollution score, with the risk of T2D. Evaluating the health impact of multiple
air pollutants has been recognized as a priority in the US Environmental Protection Agency’s
integrated, cross-disciplinary research planning [17,19]. The recent assessment of public health
consequences of air pollution has been encouraged to move toward a multipollutant approach
[42]. Indeed, humans are simultaneously exposed to a complex mixture of air pollutants. The
air pollution score could reflect a more comprehensive exposure to various air pollutants and
acknowledge the importance of assessing the health burden from simultaneous exposure to
multiple air pollutants. This simple algorithm is also easy to interpret and facilitate public
health protection from air pollution. Our study indicates that various air pollutants, when
exposed together, may jointly influence the T2D risk.
Although the underlying mechanisms linking various air pollutants and increased risk of
T2D are still unclear, several mechanisms have been proposed. PM is among the most studied
air pollutant on the risk of T2D. A main working hypothesis is that PM may induce oxidative
stress and subsequent visceral adipose tissue inflammation, which further leads to insulin resis-
tance [43–46]. Other possible mechanisms explaining the association between air pollution
and T2D include the disturbed autonomic nervous system [47], epigenetic changes [48,49],

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 9 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

mitochondrial dysfunction [50], as well as alterations in the composition and function of the
human gut microbiome [51].
Interestingly, we found that being overweight/obese significantly amplified the association
between air pollution score and the risk of T2D, while no association was observed among nor-
mal-weight participants. Previous evidence in the Danish Nurse Cohort Study also showed that
the association between PM2.5 and T2D was augmented among obese women [27]. Notably, the
increased susceptibility of obesity to the adverse effect of air pollution has also been seen with
regard to other diseases, including cardiovascular events [52], chronic obstructive pulmonary dis-
ease [53], and hypertension [54], lending support to the modification effects of metabolic status
on the relations between air pollution and T2D. Our finding on the interaction between air pollu-
tion and obesity status is biologically plausible. A previous study has shown that obesity may
enhance associations between air pollution and systemic inflammation [55]. As obesity is a proin-
flammatory state, being overweight/obese may predispose to a heightened inflammatory response
and oxidative stress. WhenAU both: PleaseconfirmthattheedittoWhenbothconditionsðairpollutionandobesityÞare
conditions (air pollution and obesity) are present, there is a mul-
tiplicative effect [53,55]. Furthermore, several hormones such as adiponectin, leptin, and resistin,
which are inversely related to adiposity, may accentuate the adverse effect of air pollution
[25,56,57]. Moreover, changes in respiratory physiology among obese individuals may also
explain the enhanced susceptibility to air pollution. Previous particle deposition studies and a
weight loss trial have shown an association between obesity and enlarged tidal volume, which lead
to a greater dose of inhaled particles among obese participants [58–60].
We did not observe the modification effects by the genetic predisposition to obesity or dia-
betes on the relationship between air pollution and risk of T2D. The null results were partly
due to the small variation of BMI or T2D risk explained by the discovered genetic loci [36].
Further studies with a large sample size and prospective design are warranted to replicate our
findings.
To the best of our knowledge, this is the first study to investigate the joint associations of
various ambient air pollutants with the risk of T2D. The major strengths of our study include a
large sample size, prospective study design of UK Biobank, and comprehensive incorporation
of various air pollutants in the air pollution score. However, we acknowledge the current anal-
ysis has several potential limitations. First, although we have comprehensively considered vari-
ous air pollutants, some of the previously reported air pollutants were not available in UK
Biobank, such as O3, sulfur dioxides, and carbon monoxide. Second, the participants of UK
Biobank are mainly of European descent. Whether the observed association could be applied
to other ethnic groups and areas warrants further investigation. Third, the observational
nature of the study precludes conclusions about causality. Fourth, although we have compre-
hensively considered typical risk factors and potential confounders, because of the observa-
tional study design, residual confounding is inevitable. Lastly, we do not have air pollution
data during follow-up.
In conclusion, we found that long-term exposure to various air pollutants, individually or
jointly, was associated with an increased risk of T2D, and that the association was more pro-
nounced among overweight/obese participants and central obese participants. Our findings
highlight the importance to comprehensively assess the various air pollutants and body weight
management in the prevention of T2D.

Supporting information
S1 STROBE checklist. Checklist of items that should be included in reports of cohort stud-
ies.
(DOCX)

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 10 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

S1 Table. Information of genetic variants associated with T2D in the UK Biobank study.
EA, effect allele; NEA, noneffect allele; OR, odds ratio; SE, standard error; SNP, single nucleo-
tide polymorphism; T2D, type 2 diabetes.
(DOCX)
S2 Table. Information of genetic variants associated with BMI in the UK Biobank study. β,
beta coefficient; BMI, body mass index; Chr, chromosome; EA, effect allele; NEA, noneffect
allele; SE, standard error; SNP, single nucleotide polymorphism.
(DOCX)
S3 Table. Pearson correlations between the individual air pollutant incorporated in the air
pollution score. � Indicates p < 0.001.
(DOCX)
S4 Table. Associations between modified air pollution score (by removing PM2.5–10) and
incident T2D among 461,191 UK Biobank participants. Model 1: adjusted for age, ethnicity,
and sex; Model 2: Model 1+ Townsend deprivation index, center, alcohol intake, smoking sta-
tus, physical activity, sedentary hour, healthy diet score; Model 3: Model 2+ BMI, SBP, antihy-
pertension meds, high cholesterol, and T2D-GRS. BMI, body mass index; GRS, genetic risk
score; SBP, systolic blood pressure; T2D, type 2 diabetes.
(DOCX)
S5 Table. Associations between air pollution score and incident T2D by excluding T2D
cases occurred in the first 2 years of follow-up. Model 1: adjusted for age, ethnicity, and sex;
Model 2: Model 1+ Townsend deprivation index, center, alcohol intake, smoking status, physi-
cal activity, sedentary hour, healthy diet score; Model 3: Model 2+ BMI, SBP, antihypertension
meds, high cholesterol, and T2D-GRS. BMI, body mass index; GRS, genetic risk score; SBP,
systolic blood pressure; T2D, type 2 diabetes.
(DOCX)
S6 Table. Associations between air pollution score and incident T2D by excluding partici-
pants who live in the current address for less than 5 years. Model 1: adjusted for age, ethnic-
ity, and sex; Model 2: Model 1+ Townsend deprivation index, center, alcohol intake, smoking
status, physical activity, sedentary hour, healthy diet score; Model 3: Model 2+ BMI, SBP, anti-
hypertension meds, high cholesterol, and T2D-GRS. BMI, body mass index; GRS, genetic risk
score; SBP, systolic blood pressure; T2D, type 2 diabetes.
(DOCX)
S7 Table. Association between the air pollution score and T2D risk according to general
obesity and central obesity. Multivariable models were adjusted for age, sex, Townsend depri-
vation index, center, alcohol intake, smoking status, physical activity, sedentary hours, healthy
diet score, systolic blood pressure, antihypertension meds, high cholesterol, and T2D GRS.
GRS, genetic risk score; T2D, type 2 diabetes.
(DOCX)
S1 Fig. Dose–response relationship of air pollution score with T2D incidence according to
central obesity status. Dashed lines represent the 95% CIs of the HR. Multivariable models
were adjusted for age, sex, Townsend deprivation index, center, alcohol intake, smoking status,
physical activity, sedentary hours, healthy diet score, systolic blood pressure, antihypertension
meds, high cholesterol, and T2D GRS. Sample size for noncentral obese and central obese sub-
group were 177,791 and 271,215, respectively. CIAU , confidence
: Abbreviationlistshavebeencompiledforthoseusedin
interval; GRS, genetic risk score;

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 11 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

HR, hazard ratio; T2D, type 2 diabetes.

(TIF)
S2 Fig. Adjusted HR for T2D according to obesity GRS. Stratified analyses were performed
by tertiles of obesity GRS. Multivariable models were adjusted for age, sex, Townsend depriva-
tion index, center, alcohol intake, smoking status, physical activity, sedentary hours, healthy
diet score, BMI, systolic blood pressure, antihypertension meds, and high cholesterol. BMI,
body mass index; CI, confidence interval; GRS, genetic risk score; HR, hazard ratio; T2D, type
2 diabetes.
(TIF)
S3 Fig. Adjusted HR for T2D according to T2D GRS. Stratified analyses were performed by
tertiles of T2D GRS. Multivariable models were adjusted for age, sex, Townsend deprivation
index, center, alcohol intake, smoking status, physical activity, sedentary hours, healthy diet
score, BMI, systolic blood pressure, antihypertension meds, and high cholesterol. BMI, body
mass index; CI, confidence interval; GRS, genetic risk score; HR, hazard ratio; T2D, type 2 dia-
betes.
(TIF)

Acknowledgments
The authors appreciate the participants in UK Biobank for their participation and contribution
to the research. The study has been conducted using the UK Biobank Resource under Applica-
tion 29256.

Author Contributions
Conceptualization: Xiang Li, Lu Qi.
Data curation: Xiang Li.
Formal analysis: Xiang Li, Mengying Wang, Yongze Song.
Funding acquisition: Lu Qi.
Investigation: Xiang Li.
Methodology: Xiang Li.
Project administration: Lu Qi.
Resources: Lu Qi.
Software: Yongze Song.
Supervision: Lu Qi.
Validation: Mengying Wang.
Visualization: Yongze Song.
Writing – original draft: Xiang Li.
Writing – review & editing: Mengying Wang, Yongze Song, Hao Ma, Tao Zhou, Zhaoxia
Liang, Lu Qi.

References
1. International Diabetes Federation. IDF Diabetes Atlas, Ninth Edition 2019 (IDF, 2019). 2019.

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 12 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

2. Hu Y, Ding M, Sampson L, Willett WC, Manson JE, Wang M, et al. Intake of whole grain foods and risk
of type 2 diabetes: results from three prospective cohort studies. BMJ. 2020; 370:m2206. https://doi.
org/10.1136/bmj.m2206 PMID: 32641435
3. Qian F, Liu G, Hu FB, Bhupathiraju SN, Sun Q. Association Between Plant-Based Dietary Patterns and
Risk of Type 2 Diabetes: A Systematic Review and Meta-analysis. JAMA Intern Med. 2019; 179:1335–
44. https://doi.org/10.1001/jamainternmed.2019.2195 PMID: 31329220
4. Umpierre D, Ribeiro PAB, Kramer CK, Leitao CB, Zucatti ATN, Azevedo MJ, et al. Physical activity
advice only or structured exercise training and association with HbA1c levels in type 2 diabetes: a sys-
tematic review and meta-analysis. JAMA. 2011; 305:1790–9. https://doi.org/10.1001/jama.2011.576
PMID: 21540423
5. Avery L, Flynn D, Van Wersch A, Sniehotta FF, Trenell MI. Changing physical activity behavior in type 2
diabetes: a systematic review and meta-analysis of behavioral interventions. Diabetes Care. 2012;
35:2681–9. https://doi.org/10.2337/dc11-2452 PMID: 23173137
6. Shan Z, Ma H, Xie M, Yan P, Guo Y, Bao W, et al. Sleep duration and risk of type 2 diabetes: a meta-
analysis of prospective studies. Diabetes Care. 2015; 38:529–37. https://doi.org/10.2337/dc14-2073
PMID: 25715415
7. Wang M, Zhou T, Li X, Ma H, Liang Z, Fonseca VA, et al. Baseline Vitamin D Status, Sleep Patterns,
and the Risk of Incident Type 2 Diabetes in Data From the UK Biobank Study. Diabetes Care. 2020; 43:
2776 LP– 2784. https://doi.org/10.2337/dc20-1109 PMID: 32847829
8. Yang B-Y, Fan S, Thiering E, Seissler J, Nowak D, Dong G-H, et al. Ambient air pollution and diabetes:
A systematic review and meta-analysis. Environ Res. 2020; 180:108817. https://doi.org/10.1016/j.
envres.2019.108817 PMID: 31627156
9. Li Y, Xu L, Shan Z, Teng W, Han C. Association between air pollution and type 2 diabetes: an updated
review of the literature. Ther Adv Endocrinol Metab. 2019; 10:2042018819897046. https://doi.org/10.
1177/2042018819897046 PMID: 31903180
10. Rajagopalan S, Brook RD. Air pollution and type 2 diabetes: mechanistic insights. Diabetes. 2012;
61:3037–45. https://doi.org/10.2337/db12-0190 PMID: 23172950
11. Coogan PF, White LF, Yu J, Burnett RT, Seto E, Brook RD, et al. PM2. 5 and diabetes and hypertension
incidence in the Black Women’s Health Study. Epidemiology. 2016; 27:202. https://doi.org/10.1097/
EDE.0000000000000418 PMID: 26595125
12. Coogan PF, White LF, Yu J, Burnett RT, Marshall JD, Seto E, et al. Long term exposure to NO2 and dia-
betes incidence in the Black Women’s Health Study. Environ Res. 2016; 148:360–6. https://doi.org/10.
1016/j.envres.2016.04.021 PMID: 27124624
13. Li X, Tang K, Jin X-R, Xiang Y, Xu J, Yang L-L, et al. Short-term air pollution exposure is associated
with hospital length of stay and hospitalization costs among inpatients with type 2 diabetes: a hospital-
based study. J Toxicol Environ Heal Part A. 2018; 81:819–29. https://doi.org/10.1080/15287394.2018.
1491912 PMID: 30015599
14. Strak M, Janssen N, Beelen R, Schmitz O, Vaartjes I, Karssenberg D, et al. Long-term exposure to par-
ticulate matter, NO(2) and the oxidative potential of particulates and diabetes prevalence in a large
national health survey. Environ Int. 2017; 108:228–36. https://doi.org/10.1016/j.envint.2017.08.017
PMID: 28886416
15. Pearson JF, Bachireddy C, Shyamprasad S, Goldfine AB, Brownstein JS. Association between fine par-
ticulate matter and diabetes prevalence in the US. Diabetes Care. 2010; 33:2196–201. https://doi.org/
10.2337/dc10-0698 PMID: 20628090
16. Krämer U, Herder C, Sugiri D, Strassburger K, Schikowski T, Ranft U, et al. Traffic-related air pollution
and incident type 2 diabetes: results from the SALIA cohort study. Environ Health Perspect. 2010;
118:1273–9. https://doi.org/10.1289/ehp.0901689 PMID: 20504758
17. U.S. EPA. Air, Climate, and Energy Strategic Research Action Plan 2012–2016. 2012; 601/R–12/00.
Available from: www.epa.gov/research
18. U.S. EPA. Air, Climate, and Energy: Strategic Research Action Plan 2016–2019. 2015.
19. U.S. EPA. Air and Energy Strategic Research Action Plan FY 2019–2022. 2020.
20. Coogan PF, White LF, Jerrett M, Brook RD, Su JG, Seto E, et al. Air Pollution and Incidence of Hyper-
tension and Diabetes Mellitus in Black Women Living in Los Angeles. Circulation. 2012; 125:767–72.
https://doi.org/10.1161/CIRCULATIONAHA.111.052753 PMID: 22219348
21. Chen H, Burnett RT, Kwong JC, Villeneuve PJ, Goldberg MS, Brook RD, et al. Risk of incident diabetes
in relation to long-term exposure to fine particulate matter in Ontario. Canada Environ Health Perspect.
2013; 121:804–10. https://doi.org/10.1289/ehp.1205958 PMID: 23632126

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 13 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

22. Andersen ZJ, Raaschou-Nielsen O, Ketzel M, Jensen SS, Hvidberg M, Loft S, et al. Diabetes Incidence
and Long-Term Exposure to Air Pollution. Diabetes Care. 2012; 35:92LP–98. https://doi.org/10.2337/
dc11-1155 PMID: 22074722
23. Renzi M, Cerza F, Gariazzo C, Agabiti N, Cascini S, Di Domenicantonio R, et al. Air pollution and occur-
rence of type 2 diabetes in a large cohort study. Environ Int. 2018; 112:68–76. https://doi.org/10.1016/j.
envint.2017.12.007 PMID: 29253730
24. Wang M, Zhou T, Song Y, Li X, Ma H, Hu Y, et al. Joint exposure to various ambient air pollutants and
incident heart failure: a prospective analysis in UK Biobank. Eur Heart J. 2021. https://doi.org/10.1093/
eurheartj/ehaa1031 PMID: 33527989
25. Li W, Dorans KS, Wilker EH, Rice MB, Schwartz J, Coull BA, et al. Residential proximity to major road-
ways, fine particulate matter, and adiposity: The framingham heart study. Obesity. 2016; 24:2593–9.
https://doi.org/10.1002/oby.21630 PMID: 27804220
26. Li M, Qian Z, Vaughn M, Boutwell B, Ward P, Lu T, et al. Sex-specific difference of the association
between ambient air pollution and the prevalence of obesity in Chinese adults from a high pollution
range area: 33 communities Chinese health study. Atmos Environ. 2015; 117:227–33.
27. Hansen AB, Ravnskjær L, Loft S, Andersen KK, Bräuner EV, Baastrup R, et al. Long-term exposure to
fine particulate matter and incidence of diabetes in the Danish Nurse Cohort. Environ Int. 2016; 91:243–
50. https://doi.org/10.1016/j.envint.2016.02.036 PMID: 26989812
28. Eze IC, Imboden M, Kumar A, von Eckardstein A, Stolz D, Gerbase MW, et al. Air pollution and diabetes
association: Modification by type 2 diabetes genetic risk score. Environ Int. 2016; 94:263–71. https://
doi.org/10.1016/j.envint.2016.04.032 PMID: 27281273
29. Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access
resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS
Med. 2015; 12:e1001779–9. https://doi.org/10.1371/journal.pmed.1001779 PMID: 25826379
30. Beelen R, Hoek G, Vienneau D, Eeftens M, Dimakopoulou K, Pedeli X, et al. Development of NO2 and
NOx land use regression models for estimating air pollution exposure in 36 study areas in Europe–The
ESCAPE project. Atmos Environ. 2013; 72:10–23. https://doi.org/10.1016/j.atmosenv.2013.02.037
31. Eeftens M, Beelen R, de Hoogh K, Bellander T, Cesaroni G, Cirach M, et al. Development of Land Use
Regression models for PM(2.5), PM(2.5) absorbance, PM(10) and PM(coarse) in 20 European study
areas; results of the ESCAPE project. Environ Sci Technol. 2012; 46:11195–205. https://doi.org/10.
1021/es301948k PMID: 22963366
32. Cyrys J, Eeftens M, Heinrich J, Ampe C, Armengaud A, Beelen R, et al. Variation of NO2 and NOx con-
centrations between and within 36 European study areas: Results from the ESCAPE study. Atmos
Environ. 2012; 62:374–90. https://doi.org/10.1016/j.atmosenv.2012.07.080
33. Bycroft C, Freeman C, Petkova D, Band G, Elliott LT, Sharp K, et al. Genome-wide genetic data on
~500,000 UK Biobank participants. bioRxiv. 2017. https://doi.org/10.1101/166298 [preprint]
34. Vetter C, Dashti HS, Lane JM, Anderson SG, Schernhammer ES, Rutter MK, et al. Night shift work,
genetic risk, and type 2 diabetes in the UK biobank. Diabetes Care. 2018; 41:762–9. https://doi.org/10.
2337/dc17-1933 PMID: 29440150
35. Scott RA, Scott LJ, Mägi R, Marullo L, Gaulton KJ, Kaakinen M, et al. An Expanded Genome-Wide
Association Study of Type 2 Diabetes in Europeans. Diabetes. 2017; 66:2888–902. https://doi.org/10.
2337/db16-1253 PMID: 28566273
36. Locke A, Kahali B, Berndt S, Justice A, Pers T. Genetic studies of body mass index yield new insights
for obesity biology. Nature. 2015; 518:197–206. https://doi.org/10.1038/nature14177 PMID: 25673413
37. IPAQ Research Committee. Guidelines for data processing and analysis of the International Physical
Activity Questionnaire (IPAQ)-short and long forms. 2005. Available from: http//wwwipaqkise/scoring
pdf.
38. Craig CL, Marshall AL, Sjöström M, Bauman AE, Booth ML, Ainsworth BE, et al. International physical
activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003; 35:1381–95.
https://doi.org/10.1249/01.MSS.0000078924.61453.FB PMID: 12900694
39. Lloyd-Jones DM, Hong Y, Labarthe D, Mozaffarian D, Appel LJ, Van Horn L, et al. Defining and setting
national goals for cardiovascular health promotion and disease reduction: The american heart associa-
tion’s strategic impact goal through 2020 and beyond. Circulation. 2010; 121:586–613. https://doi.org/
10.1161/CIRCULATIONAHA.109.192703 PMID: 20089546
40. Qiu H, Schooling CM, Sun S, Tsang H, Yang Y, Lee RS-Y, et al. Long-term exposure to fine particulate
matter air pollution and type 2 diabetes mellitus in elderly: a cohort study in Hong Kong. Environ Int.
2018; 113:350–6. https://doi.org/10.1016/j.envint.2018.01.008 PMID: 29357993

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 14 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

41. Puett RC, Hart JE, Schwartz J, Hu FB, Liese AD, Laden F. Are particulate matter exposures associated
with risk of type 2 diabetes? Environ Health Perspect. 2011; 119:384–9. https://doi.org/10.1289/ehp.
1002344 PMID: 21118784
42. Dominici F, Peng RD, Barr CD, Bell ML. Protecting human health from air pollution: shifting from a sin-
gle-pollutant to a multipollutant approach. Epidemiology. 2010; 21:187–94. https://doi.org/10.1097/
EDE.0b013e3181cc86e8 PMID: 20160561
43. Sun Q, Yue P, Deiuliis JA, Lumeng CN, Kampfrath T, Mikolaj MB, et al. Ambient air pollution exagger-
ates adipose inflammation and insulin resistance in a mouse model of diet-induced obesity. Circulation.
2009; 119:538–46. https://doi.org/10.1161/CIRCULATIONAHA.108.799015 PMID: 19153269
44. Brook RD, Rajagopalan S, Pope CA III, Brook JR, Bhatnagar A, Diez-Roux AV, et al. Particulate matter
air pollution and cardiovascular disease: an update to the scientific statement from the American Heart
Association. Circulation. 2010; 121:2331–78. https://doi.org/10.1161/CIR.0b013e3181dbece1 PMID:
20458016
45. Hoffmann B, Moebus S, Dragano N, Stang A, Möhlenkamp S, Schmermund A, et al. Chronic residential
exposure to particulate matter air pollution and systemic inflammatory markers. Environ Health Per-
spect. 2009; 117:1302–8. https://doi.org/10.1289/ehp.0800362 PMID: 19672412
46. Hennig F, Fuks K, Moebus S, Weinmayr G, Memmesheimer M, Jakobs H, et al. Association between
source-specific particulate matter air pollution and hs-CRP: local traffic and industrial emissions. Envi-
ron Health Perspect. 2014; 122:703–10. https://doi.org/10.1289/ehp.1307081 PMID: 24755038
47. Chen J-C, Stone PH, Verrier RL, Nearing BD, MacCallum G, Kim J-Y, et al. Personal Coronary Risk
Profiles Modify Autonomic Nervous System Responses to Air Pollution. J Occup Environ Med 2006; 48.
Available from: https://journals.lww.com/joem/Fulltext/2006/11000/Personal_Coronary_Risk_Profiles_
Modify_Autonomic.7.aspx https://doi.org/10.1097/01.jom.0000245675.85924.7e PMID: 17099449
48. Bind M-A, Baccarelli A, Zanobetti A, Tarantini L, Suh H, Vokonas P, et al. Air pollution and markers of
coagulation, inflammation and endothelial function: Associations and epigene-environment interactions
in an elderly cohort. Epidemiology. 2012; 23:332. https://doi.org/10.1097/EDE.0b013e31824523f0
PMID: 22237295
49. Carmona JJ, Sofer T, Hutchinson J, Cantone L, Coull B, Maity A, et al. Short-term airborne particulate
matter exposure alters the epigenetic landscape of human genes associated with the mitogen-activated
protein kinase network: a cross-sectional study. Environ Health. 2014; 13:94. https://doi.org/10.1186/
1476-069X-13-94 PMID: 25395096
50. Xu X, Liu C, Xu Z, Tzan K, Zhong M, Wang A, et al. Long-term exposure to ambient fine particulate pol-
lution induces insulin resistance and mitochondrial alteration in adipose tissue. Toxicol Sci. 2011;
124:88–98. https://doi.org/10.1093/toxsci/kfr211 PMID: 21873646
51. Fouladi F, Bailey MJ, Patterson WB, Sioda M, Blakley IC, Fodor AA, et al. Air pollution exposure is asso-
ciated with the gut microbiome as revealed by shotgun metagenomic sequencing. Environ Int. 2020;
138:105604. https://doi.org/10.1016/j.envint.2020.105604 PMID: 32135388
52. Miller KA, Siscovick DS, Sheppard L, Shepherd K, Sullivan JH, Anderson GL, et al. Long-Term Expo-
sure to Air Pollution and Incidence of Cardiovascular Events in Women. N Engl J Med. 2007; 356:447–
58. https://doi.org/10.1056/NEJMoa054409 PMID: 17267905
53. McCormack MC, Belli AJ, Kaji DA, Matsui EC, Brigham EP, Peng RD, et al. Obesity as a susceptibility
factor to indoor particulate matter health effects in COPD. Eur Respir J. 2015; 45:1248 LP–1257.
https://doi.org/10.1183/09031936.00081414 PMID: 25573407
54. Dong G-H, Wang J, Zeng X-W, Chen L, Qin X-D, Zhou Y, et al. Interactions Between Air Pollution and
Obesity on Blood Pressure and Hypertension in Chinese Children. Epidemiology 2015; 26. Available
from: https://journals.lww.com/epidem/Fulltext/2015/09000/Interactions_Between_Air_Pollution_and_
Obesity_on.18.aspx https://doi.org/10.1097/EDE.0000000000000336 PMID: 26133026
55. Dubowsky SD, Suh H, Schwartz J, Coull BA, Gold DR. Diabetes, Obesity, and Hypertension May
Enhance Associations between Air Pollution and Markers of Systemic Inflammation. Environ Health
Perspect. 2006; 114:992–8. https://doi.org/10.1289/ehp.8469 PMID: 16835049
56. Chen W, Han Y, Wang Y, Chen X, Qiu X, Li W, et al. Associations between changes in adipokines and
exposure to fine and ultrafine particulate matter in ambient air in Beijing residents with and without pre-
diabetes. BMJ Open Diabetes Res Care. 2020; 8:e001215. https://doi.org/10.1136/bmjdrc-2020-
001215 PMID: 33361362
57. Wolf K, Popp A, Schneider A, Breitner S, Hampel R, Rathmann W, et al. Association Between Long-
term Exposure to Air Pollution and Biomarkers Related to Insulin Resistance, Subclinical Inflammation,
and Adipokines. Diabetes. 2016; 65:3314 LP–3326. https://doi.org/10.2337/db15-1567 PMID:
27605624
58. Graham DR, Chamberlain MJ, Hutton L, King M, Morgan WK. Inhaled particle deposition and body hab-
itus. Br J Ind Med. 1990; 47:38–43. https://doi.org/10.1136/oem.47.1.38 PMID: 2310705

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 15 / 16

PLOS MEDICINE Air pollution, obesity, and type 2 diabetes

59. Bennett WD, Zeman KL. Effect of body size on breathing pattern and fine-particle deposition in children.
J Appl Physiol. 2004; 97:821–6. https://doi.org/10.1152/japplphysiol.01403.2003 PMID: 15107416
60. Matos CMP, Moraes KS, França DC, Tomich GM, Farah MW, Dias RC, et al. Changes in breathing pat-
tern and thoracoabdominal motion after bariatric surgery: a longitudinal study. Respir Physiol Neurobiol.
2012; 181:143–7. https://doi.org/10.1016/j.resp.2012.02.009 PMID: 22415065

PLOS Medicine | https://doi.org/10.1371/journal.pmed.1003767 August 30, 2021 16 / 16