Long-term air pollution exposure and Parkinson’s Disease mortality in a large pooled European cohort: an ELAPSE study

Abstract

Background
The link between exposure to ambient air pollution and mortality from cardiorespiratory diseases is well established, while evidence on neurodegenerative disorders including Parkinson’s Disease (PD) remains limited.

Objective
We examined the association between long-term exposure to ambient air pollution and PD mortality in seven European cohorts.

Methods
Within the project ‘Effects of Low-Level Air Pollution: A Study in Europe’ (ELAPSE), we pooled data from seven cohorts among six European countries. Annual mean residential concentrations of fine particulate matter (PM2.5), nitrogen dioxide (NO2), black carbon (BC), and ozone (O3), as well as eight PM2.5 components (copper, iron, potassium, nickel, sulphur, silicon, vanadium, zinc), for 2010 were estimated using Europe-wide hybrid land use regression models. PD mortality was defined as underlying cause of death being either PD, secondary Parkinsonism, or dementia in PD. We applied Cox proportional hazard models to investigate the associations between air pollution and PD mortality, adjusting for potential confounders.

Results
Of 271,720 cohort participants, 381 died from PD during 19.7 years of follow-up. In single-pollutant analyses, we observed positive associations between PD mortality and PM2.5 (hazard ratio per 5 µg/m³: 1.25; 95% confidence interval: 1.01-1.55), NO2 (1.13; 0.95-1.34 per 10 µg/m³), and BC (1.12; 0.94-1.34 per 0.5 x 10⁻⁵m⁻¹), and a negative association with O3 (0.74; 0.58-0.94 per 10 µg/m³). Associations of PM2.5, NO2, and BC with PD mortality were linear without apparent lower thresholds. In two-pollutant models, associations with PM2.5 remained robust when adjusted for NO2 (1.24; 0.95-1.62) or BC (1.28; 0.96-1.71), whereas associations with NO2 or BC attenuated to null. O3 associations remained negative, but no longer statistically significant in models with PM2.5. We detected suggestive positive associations with the potassium component of PM2.5.

Conclusion
Long-term exposure to PM2.5, at levels well below current EU air pollution limit values, may contribute to PD mortality.

Full text

Environment International 171 (2023) 107667
Available online 30 November 2022
0160-4120/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Full length article
Long-term air pollution exposure and Parkinson’s disease mortality in a
large pooled European cohort: An ELAPSE study
Thomas Cole-Hunter
a
, Jiawei Zhang
a
, Rina So
a
, Evangelia Samoli
b
, Shuo Liu
a
, Jie Chen
c
,
Maciej Strak
c
,
d
, Kathrin Wolf
e
, Gudrun Weinmayr
f
, Sophia Rodopolou
b
, Elizabeth Remfry
g
,
Kees de Hoogh
h
,
i
, Tom Bellander
j
,
k
, Jørgen Brandt
l
,
m
, Hans Concin
n
, Emanuel Zitt
n
,
o
,
Daniela Fecht
p
, Francesco Forastiere
q
,
r
, John Gulliver
p
,
s
, Barbara Hoffmann
t
,
Ulla A. Hvidtfeldt
u
, Karl-Heinz J¨
ockel
v
, Laust H. Mortensen
w
,
x
, Matthias Ketzel
l
,
y
,
Diego Yacam´
an M´
endez
z
,
ai
, Karin Leander
j
, Petter Ljungman
j
,
aa
, Elodie Faure
ab
,
Pei-Chen Lee
ab
,
ac
, Alexis Elbaz
ab
, Patrik K.E. Magnusson
ad
, Gabriele Nagel
f
,
G¨
oran Pershagen
j
,
k
, Annette Peters
e
,
ae
, Debora Rizzuto
af
,
ag
, Roel C.H. Vermeulen
c
,
ah
,
Sara Schramm
v
, Massimo Stafoggia
j
,
q
, Klea Katsouyanni
b
,
r
, Bert Brunekreef
c
, Gerard Hoek
c
,
Youn-Hee Lim
a
, Zorana J. Andersen
a
,
*
a
Section of Environmental Health, Department of Public Health, University of Copenhagen, Copenhagen, Denmark
b
Department of Hygiene, Epidemiology and Medical Statistics, Medical School, National and Kapodistrian University of Athens, Athens, Greece
c
Institute for Risk Assessment Sciences, Utrecht University, Utrecht, the Netherlands
d
National Institute for Public Health and the Environment, Bilthoven, the Netherlands
e
Institute of Epidemiology, Helmholtz Zentrum München, Neuherberg, Germany
f
Institute of Epidemiology and Medical Biometry, Ulm University, Ulm, Germany
g
Wolfson Institute of Population Health, Queen Mary University of London, United Kingdom
h
Swiss Tropical and Public Health Institute, Basel, Switzerland
i
University of Basel, Basel, Switzerland
j
Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden
k
Centre for Occupational and Environmental Medicine, Region Stockholm, Stockholm, Sweden
l
Department of Environmental Science, Aarhus University, Roskilde, Denmark
m
iClimate, interdisciplinary Centre for Climate Change, Aarhus University, Roskilde, Denmark
n
Agency for Preventive and Social Medicine (aks), Bregenz, Austria
o
Department of Internal Medicine 3, LKH Feldkirch, Feldkirch, Austria
p
MRC Centre for Environment and Health, School of Public Health, Imperial College London, London, United Kingdom
q
Department of Epidemiology, Lazio Region Health Service / ASL Roma 1, Rome, Italy
r
MRC Centre for Environment and Health, Environmental Research Group, School of Public Health, Imperial College London, London, United Kingdom
s
Centre for Environmental Health and Sustainability & School of Geography, Geology and the Environment, University of Leicester, Leicester, United Kingdom
t
Institute for Occupational, Social and Environmental Medicine, Centre for Health and Society, Medical Faculty, Heinrich Heine University Düsseldorf, Germany
u
Danish Cancer Society Research Center, Copenhagen, Denmark
v
Institute for Medical Informatics, Biometry and Epidemiology, Medical Faculty, University of Duisburg-Essen, Essen, Germany
w
Section of Epidemiology, Department of Public Health, University of Copenhagen, Copenhagen, Denmark
x
Statistics Denmark, Copenhagen, Denmark
y
Global Centre for Clean Air Research (GCARE), University of Surrey, Guildford GU2 7XH, United Kingdom
z
Department of Global Public Health, Karolinska Institutet, Stockholm, Sweden
aa
Department of Cardiology, Danderyd University Hospital, Stockholm, Sweden
ab
University Paris-Saclay, UVSQ, Inserm, Gustave Roussy, “Exposome and Heredity” team, CESP UMR1018, 94805 Villejuif, France
ac
Department of Public Health, National Cheng Kung University, Tainan, Taiwan
ad
Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
ae
Ludwig Maximilians Universit¨
at München, München, Germany
af
Department of Neurobiology, Care Sciences, and Society, Karolinska Institutet and Stockholm University, Stockholm, Sweden
ag
Stockholm Gerontology Research Center, Stockholm, Sweden
ah
Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Utrecht, the Netherlands
ai
Centre for Epidemiology and Community Medicine, Region Stockholm, Stockholm, Sweden
* Corresponding author.
E-mail address: zorana.andersen@sund.ku.dk (Z.J. Andersen).
Contents lists available at ScienceDirect
Environment International
journal homepage: www.elsevier.com/locate/envint
https://doi.org/10.1016/j.envint.2022.107667
Received 29 August 2022; Received in revised form 22 November 2022; Accepted 27 November 2022

Environment International 171 (2023) 107667
2
ARTICLE INFO
Handling Editor: Adrian Covaci
Keywords:
Air pollution
Adults
Parkinson’s Disease
Long-term exposure
Low-level exposure
Pooled-cohort study
ABSTRACT
Background: The link between exposure to ambient air pollution and mortality from cardiorespiratory diseases is
well established, while evidence on neurodegenerative disorders including Parkinson’s Disease (PD) remains
limited.
Objective: We examined the association between long-term exposure to ambient air pollution and PD mortality in
seven European cohorts.
Methods: Within the project ‘Effects of Low-Level Air Pollution: A Study in Europe’ (ELAPSE), we pooled data
from seven cohorts among six European countries. Annual mean residential concentrations of ne particulate
matter (PM
2.5
), nitrogen dioxide (NO
2
), black carbon (BC), and ozone (O
3
), as well as 8 PM
2.5
components
(copper, iron, potassium, nickel, sulphur, silicon, vanadium, zinc), for 2010 were estimated using Europe-wide
hybrid land use regression models. PD mortality was dened as underlying cause of death being either PD,
secondary Parkinsonism, or dementia in PD. We applied Cox proportional hazard models to investigate the as-
sociations between air pollution and PD mortality, adjusting for potential confounders.
Results: Of 271,720 cohort participants, 381 died from PD during 19.7 years of follow-up. In single-pollutant
analyses, we observed positive associations between PD mortality and PM
2.5
(hazard ratio per 5 µg/m
3
: 1.25;
95% condence interval: 1.01–1.55), NO
2
(1.13; 0.95–1.34 per 10 µg/m
3
), and BC (1.12; 0.94–1.34 per 0.5 ×10
-
5
m
-1
), and a negative association with O
3
(0.74; 0.58–0.94 per 10 µg/m
3
). Associations of PM
2.5
, NO
2
, and BC
with PD mortality were linear without apparent lower thresholds. In two-pollutant models, associations with
PM
2.5
remained robust when adjusted for NO
2
(1.24; 0.95–1.62) or BC (1.28; 0.96–1.71), whereas associations
with NO
2
or BC attenuated to null. O
3
associations remained negative, but no longer statistically signicant in
models with PM
2.5
. We detected suggestive positive associations with the potassium component of PM
2.5
.
Conclusion: Long-term exposure to PM
2.5
, at levels well below current EU air pollution limit values, may
contribute to PD mortality.
1. Introduction
The association between exposure to air pollution and cardiovascu-
lar and respiratory diseases is well documented (WHO, 2021). Recently,
attention has been drawn to the potential negative impact of air pollu-
tion on disorders that affect the brain and central nervous system,
including neurodegenerative diseases such as dementia, Alzheimer’s
Disease, and Parkinson’s Disease (PD) (Kim et al., 2020; Raggi and
Leonardi, 2020). PD is the second most prevalent neurodegenerative
disorder, affecting an estimated six million individuals globally, pre-
ceded in prevalence only by Alzheimer’s Disease (Dorsey et al., 2018).
Known risk factors for PD include age, being male, smoking, high con-
sumption of dairy, exposure to pesticides, and genetic predisposition
(Ascherio and Schwarzschild, 2016).
It has been suggested that both gaseous pollutants and ne particles
can directly or indirectly damage the brain by crossing the blood–brain
barrier, causing oxidative stress, neuroinammation and abnormal ag-
gregation of proteins, damaging the olfactory bulb and frontal cortex,
leading to the development of PD (Ascherio and Schwarzschild, 2016).
Despite PD being a progressive disease that leads to death in older age,
epidemiological evidence on long-term exposure to air pollution and PD
consists mainly of studies on the incidence of (rather than mortality
from) PD. These studies on incidence present mixed ndings, with some
studies reporting associations with particulate matter of diameter <10
µm (PM
10
), particulate matter of diameter <2.5 µm (PM
2.5
), nitrogen
dioxide (NO
2
) or ozone (O
3
) (Chen et al., 2017; Kirrane et al., 2015;
Rhew et al., 2021;16(7):e0253253.; Ritz et al., 2016; Shi et al., 2020;
Shin et al., 2018; Yu et al., 2021; Yuchi et al., 2020), while other studies
nd no associations (Cerza et al., 2018; Lee et al., 2016; Liu et al., 2016;
Palacios et al., 2017; Palacios et al., 2014; Toro et al., 2019) (see sup-
plemental Table S1). Only two studies examined long-term exposure to
air pollution with respect to PD mortality: one only examining O
3
,
detecting signicant positive associations (Zhao et al., 2021), and;
another examining PM
2.5
and nding suggestive positive associations,
which were weaker than those detected with incidence (compared to
mortality) of PD in the same study (Rhew et al., 2021). Two studies went
further to examine the association of specic components of PM
2.5
and
PD. Nunez and colleagues examined black carbon (BC), organic matter
(OM), nitrate, sulfate, sea salt, and soil PM
2.5
components, detecting
associations of nitrate and OM with PD hospitalization (Nunez et al.,
2021). Palacios and colleagues suggested an association of mercury with
PD incidence (Palacios et al., 2014). Therefore, further study is war-
ranted to elucidate the long-term effects of specic air pollution com-
ponents, and strengthen the evidence base for potential effects at low
pollutant concentrations, on mortality from neurodegenerative diseases
including PD among large cohorts, such as through the ‘Effects of Low-
Level Air Pollution: A Study in Europe’ (ELAPSE) project. ELAPSE has
recently been able to show that long-term exposure to low levels of air
pollution was associated with premature mortality (Stafoggia et al.,
2022; Strak et al., 2021), as well as incidence of cardiovascular diseases
(Wolf et al., 2021), adult-onset asthma (Liu et al., 2021), chronic
obstructive pulmonary disease (COPD) (Liu et al., 2021), lung cancer
(Hvidtfeldt et al., 2021) and liver cancer (So et al., 2021), as well as
mortality from dementia, psychiatric disorders, and suicide (Andersen
et al., 2022). Here, we aimed to examine the association of long-term
exposure to PM
2.5
, NO
2
, BC, and O
3
, as well as eight PM
2.5
compo-
nents, with mortality from PD among seven cohorts of the ELAPSE
project.
2. Materials and methods
2.1. Study population
Within the ELAPSE project, we analysed pooled data from seven
cohorts (and their sub-cohorts) among six European countries (see
supplemental Figure S1), including information on potential
confounders:
(a) Cardiovascular Effects of Air Pollution and Noise in Stockholm
(CEANS) cohort in Sweden, which combined four sub-cohorts:
Stockholm Diabetes Prevention Program (SDPP) (Eriksson
et al., 2008); Stockholm Cohort of 60-year-olds (SIXTY) (W¨
andell
et al., 2007); Stockholm Screening Across the Lifespan Twin study
(SALT) (Magnusson et al., 2013); Swedish National Study on
Aging and Care in Kungsholmen (SNAC-K) (Lagergren et al.,
2004);
(b) Danish Nurse Cohort (DNC) in Denmark (Hundrup et al., 2012),
including two cohort recruitment rounds in 1993 and 1999;
T. Cole-Hunter et al.

Environment International 171 (2023) 107667
3
(c) Etude Epid´
emiologique aupr`
es de femmes de la Mutuelle
G´
en´
erale de l’Education Nationale (E3N) in France (Clavel-
Chapelon, 2015);
(d) European Prospective Investigation into Cancer and Nutrition-
Netherlands (EPIC-NL) cohort in the Netherlands, which
included two sub-cohorts: Monitoring Project on Risk Factors and
Chronic Diseases in the Netherlands (Morgen) and Prospect
(Beulens et al., 2010);
(e) Heinz Nixdorf Recall study (HNR) in Germany (Schmermund
et al., 2002);
(f) Cooperative Health Research in the Region of Augsburg (KORA)
in Germany (Holle et al., 2005), combining two sub-cohorts from
baseline rounds in 1994–1995 (S3) and 1999–2001 (S4); and;
(g) Vorarlberg Health Monitoring and Prevention Programme
(VHM&PP) in Austria (Ulmer et al., 2007).
The cohorts were recruited in the 1990s or early 2000s, from one or
several large cities and their surrounding towns, except for the two
nationwide cohorts E3N and DNC. Detailed information of each
cohort has been described previously (Chen et al., 2021). All cohorts
were approved by the medical ethics committees in their respective
countries.
2.2. Air pollution exposure assessment
The method for assessment of air pollution exposure via modelling
has previously been described in detail (Hvidtfeldt et al., 2021; Liu et al.,
2021; de Hoogh et al., 2018). Briey, annual mean concentrations of
PM
2.5
, NO
2
, BC, and O
3
(limited to the maximum running 8-hour
average in the boreal warm season, April-September) were estimated
for 2010 at participants’ baseline residential addresses utilising stan-
dardized Europe-wide hybrid land use regression (LUR) models (de
Hoogh et al., 2018). These models were developed based on air pollution
concentration data from routine and research study monitors, satellites,
estimates from chemical transport models, and land use and road vari-
ables as predictors.
PM
2.5
, NO
2
, BC, and O
3
LUR models used a 100 m ×100 m (grid)
spatial scale that performed satisfactorily in vefold hold-out validation,
explaining 66%, 58%, 51and 60%, respectively, of measured spatial
variation. BC was measured by the reectance of PM
2.5
lters from 2009
and 2010, expressed in absorbance units.
In addition, we extrapolated pollutant concentrations for each year
from baseline to the end of follow-up (Hvidtfeldt et al., 2021; Liu et al.,
2021), and incorporated dynamic residential address history during
follow-up. The extrapolation method was applied by utilising estimated
monthly average concentrations, at a spatial resolution of 26 km ×26
km, from the Danish Eulerian Hemispheric Model (DEHM) since 1990
(Brandt et al., 2012). The extrapolation was performed after checking
the agreement between ground measurements versus DEHM predictions
for the four pollutants, as detailed previously to show little impact by
temporal misalignment (Chen et al., 2021). The DEHM provided pre-
dicted modelling exposure data for a complete database to perform
harmonious extrapolation for the four pollutants; ground measurements
for PM
2.5
components were not available back in time. Pollutant con-
centrations were extrapolated for cohorts with available residential
history information, using both a difference method and a ratio method
with 2010 as the reference year (Stafoggia et al., 2022; Strak et al.,
2021).
We estimated exposure to eight components of PM
2.5
for 2010 at the
participants’ baseline residential addresses using Europe-wide LUR
models based on the standardized ESCAPE project monitoring data, with
model details published previously (Chen et al., 2020). The eight com-
ponents were selected to represent major pollution sources: copper (Cu),
iron (Fe), and zinc (Zn) for non-tailpipe trafc emissions; sulfur (S) for
long-range transport of secondary inorganic aerosols; nickel (Ni) and
vanadium (V) for mixed oil burning/industry; silicon (Si) for earth
crustal material; and, potassium (K) for biomass burning. We included
large-scale satellite-model and chemical transport-model estimates of
components to represent background concentrations, in addition to
land-use, road, population, and industrial point source data to model
local spatial variability. We applied two algorithms, supervised linear
regression (SLR) (de Hoogh et al., 2018), and random forest (RF) (Chen
et al., 2019), to develop models for the eight components. The models
explained a moderate-to-large fraction of the measured concentration
variation at the European scale (ranging from 41% to 90% across
components).
2.3. Mortality outcome denition
We dened PD mortality from mortality registries, based solely on
the underlying cause of death (as contributing causes of death were not
available) using the International Classication of Diseases (ICD)-9: 332
(PD); and, ICD-10: G20 (PD), G21 (secondary Parkinsonism), G22
(Parkinsonism in diseases classied elsewhere), and F02.3 (dementia in
PD).
2.4. Statistical analysis
We performed Cox proportional hazards models with age as the
underlying timescale (Thi´
ebaut and B´
enichou, 2004) to examine asso-
ciations between long-term exposure to air pollution and PD, following
the general analytical framework of ELAPSE (Samoli et al., 2021).
Censoring occurred at the time of event of interest, death from other
causes, emigration, loss to follow-up, or the end of follow-up (ranging
from 2011 to 2015 depending on sub-cohort), whichever came rst. The
start of follow-up was the year of enrolment in individual cohorts which
ranged from 1985 to 2001 (supplemental Table S2). Air pollution
exposure was included in models as a linear term. We examined asso-
ciations using three models including a priori dened individual and
area-level covariates, all assessed at the cohort baseline. Model 1
included age (time axis), sex (strata), sub-cohort (strata), and calendar
year of baseline (year of enrolment). Model 2 was additionally adjusted
for smoking status (never, former, current), smoking duration (years) for
current smokers, smoking intensity (linear and squared term: cigarettes/
day) for current smokers, body mass index (BMI, categories: <18.5,
18.5–24.9, 25.0–29.9, and ≥30 kg/m
2
), marital status (married/
cohabiting, divorced/separated, single, widowed), and employment
status (employed/self-employed, other). Model 3 (main model) was
further adjusted for area-level mean annual (2001) income as a proxy for
socio-economic status (SES). Only participants with complete exposure
and covariate information for Model 3 were included in the analyses to
ensure comparability among model results.
For explorative purposes, we performed a subset analysis with Model
3 by excluding participants of exposure levels above certain pre-dened
values (PM
2.5
: 25, 20, 15 µg/m
3
; NO
2
: 40, 30, 20 µg/m
3
; BC: 3, 2.5, 2, 1.5
×10
-5
m
-1
; O
3
: 120, 100 µg/m
3
), based partially on existing EU and US
limit values and previous (2005) WHO guidelines. To assess the shape of
concentration–response functions (associations from Model 3), we
modelled pollutants as natural cubic splines with two degrees of
freedom and tested for deviation of linearity by comparing with linear
models using a likelihood ratio test. Further, for explorative purposes,
we assessed a potential effect modication on these associations by age
(<65 or ≥65 years old), sex (female or male), overweight status (BMI ≥
25 kg/m
2
or not), smoking status (current, former, or never smoker),
and employment status, by including an interaction term into Model 3
tested by the Wald test. In addition, we performed two-pollutant models
based on Model 3 to investigate the contribution of individual
pollutants.
We performed several sensitivity analyses to check the robustness of
our associations. We compared the results of year 2010 exposure in
Model 3 with results of back-extrapolated baseline year exposures and
time-varying annual exposures. Time-varying (exposure) analyses were
T. Cole-Hunter et al.

Environment International 171 (2023) 107667
4
performed for cohorts with available information on residential address
history (CEANS, EPIC-NL, and VHM&PP), with 1-year strata of calendar
time to account for time trends in air pollution and mortality. Finally, we
compared effect estimates in Model 3 using different datasets excluding
one cohort at a time.
The results are presented as hazard ratios (HR) and 95% condence
intervals (CI) for pollutant unit increases of 5 µg/m
3
for PM
2.5
, 10 µg/m
3
for NO
2
, 0.5 ×10
-5
m
-1
for BC, 10 µg/m
3
for O
3
, and IQR for PM
2.5
composition. All statistical analyses were performed in software R
(version 3.4.0).
3. Results
3.1. Population description
Our pooled cohort included 324,728 participants. We excluded
53,008 of those participants with covariate data missing for Model 3. Of
271,720 remaining participants in the nal analyses, 381 died from PD
(380 from PD and 1 from secondary Parkinsonism) during a mean
follow-up time of 19.7 years (Table 1) and a mean age at end of follow-
up of 66.9 years (supplemental Table S2). The large VHM&PP cohort
contributed with the majority (284, or ~ 75%) of the deaths. Baseline
characteristics of participants varied widely across sub-cohorts
(Table 1), supporting the use of strata for sub-cohorts to adjust for dif-
ferences in baseline hazard. Mean age of participants at baseline was
47.1 years, ranging from 42.1 in VHM&PP to 72.9 years in CEANS-
SNACK. The majority of participants (69%) were female, as three
cohorts/sub-cohorts were female-only by design (DNC, E3N, and EPIC-
NL-Prospect). The proportion of current smokers ranged from 13% in
E3N to 37% in DNC-1993. Almost half of our participants (41%) were
overweight, with the lowest proportion (21%) in E3N and the highest
(74%) in HNR (Table 1).
Table 1
Baseline demographic characteristics of participants by the pooled cohort and sub-cohorts.
Cohort/
sub-
cohort
N Deaths,
N
Follow-
up time,
years
Age,
years
Female
(%)
Current
smokers
(%)
Smoking
duration,
years*
Smoking
intensity,
n/day*
Over
weight
(%)ϕ
Married/
cohabiting
(%)
Employed
(%)
Mean
income,
euroy
Pooled
Cohort
271,720 381 19.7 47.1
±
14.0
69 22 21.9 ±12.5 14.7 ±8.9 41 72 68 20.1 ±
6.2
CEANS 20,702 12 13 56.3
±
11.4
58 22 33.6 ±11.0 13.1 ±7.7 51 72 69 25.3 ±
5.6
SDPP 7,727 0 15.9 47.1
±4.9
61 26 27.9 ±8.6 13.5 ±7.4 52 84 91 24.3 ±
4.2
SIXTY 3,969 3 15.5 60.0
±0.0
52 21 36.3 ±9.9 13.4 ±7.6 65 74 68 24.7 ±
6.9
SALT 6,176 1 10.4 57.8
±
10.6
55 21 37.9 ±9.3 12.7 ±8.0 40 68 64 25.3 ±
6.6
SNACK 2,830 8 7.4 72.9
±
10.4
62 14 43.3 ±13.6 11.7 ±8.2 53 46 23 28.7 ±
2.2
DNC 25,171 27 17.3 53.5
±8.3
100 35 30.4 ±9.5 13.7 ±8.0 29 70 78 19.1 ±
2.5
1993 17,043 27 18.7 56.2
±8.4
100 37 31.6 ±9.9 13.9 ±8.2 28 68 70 19.2 ±
2.6
1999 8,128 0 14.4 47.9
±4.2
100 29 27.1 ±7.1 13.3 ±7.3 30 76 95 19.0 ±
2.4
E3N 39,006 29 16.7 53.0
±6.8
100 13 28.6 ±7.6 11.4 ±9.2 21 83 68 11.2 ±
3.0
EPIC-NL 32,872 17 16.7 49.5
±
11.9
75 29 28.9 ±11.2 15.0 ±8.7 52 70 61 12.6 ±
1.6
Morgen 18,302 5 16.8 42.9
±
11.2
55 35 24.8 ±10.6 15.7 ±8.6 50 65 69 12.2 ±
1.6
Prospect 14,570 12 16.4 57.7
±6.1
100 23 36.8 ±7.6 13.7 ±8.7 55 77 51 13.1 ±
1.4
HNR 4,733 8 12 59.7
±7.8
50 24 34.5 ±9.4 18.6 ±12.0 74 75 40 25.2 ±
8.2
KORA 4,853 4 14.3 49.4
±
13.9
51 21 24.7 ±11.8 16.1 ±9.5 68 80 57 37.3 ±
6.0
S3 2,572 2 15.6 49.4
±
13.9
51 20 25.2 ±12.1 16.5 ±9.5 67 80 55 36.7 ±
4.4
S4 2,281 2 12.9 49.3
±
13.8
51 23 24.3 ±11.6 15.7 ±9.5 69 79 59 38.0 ±
7.3
VHM&PP 144,383 284 23.1 42.1
±
15.0
56 20 13.4 ±8.3 15.6 ±8.9 43 69 70 22.9 ±
1.7
Results of participants’ characteristics at baseline are presented as Mean ±SD, Number, or Percentage.
*: Smoking duration and smoking intensity are only for current smokers. We set these variables to zero for never and former smokers.
ϕ: BMI ≥25 kg/m
2
indicates overweight according to the World Health Organization (WHO) categories.
†: Area-level mean year income in euros ×1,000 in the year 2001. The spatial scale of an area varied from neighborhoods and city districts (CEANS, E3N, EPIC-NL, and
HNR) to municipalities (DNC, KORA, and VHM&PP).
Denition of abbreviation: BMI, body mass index; SD, standard deviation.
T. Cole-Hunter et al.

Environment International 171 (2023) 107667
5
3.2. Air pollution description
Fig. 1 represents the distribution of air pollution levels in 2010 by
pooled cohort and sub-cohorts. Exposure distributions varied between
cohorts with the lowest concentrations of PM
2.5
and BC in the Nordic
cohorts (CEANS and DNC). Almost all participants were exposed to
annual PM
2.5
levels below the EU limit value (25
μ
g/m
3
) but above the
US limit value (12 µg/m
3
) and the new WHO guideline (5
μ
g/m
3
).
Similarly, almost all participants were exposed to annual NO
2
levels
below the EU limit value (40
μ
g/m
3
) but above the new WHO guideline
(10
μ
g/m
3
) (Fig. 1). We also observed varying exposure levels across
cohorts and sub-cohorts for the baseline year exposure (supplemental
Figure S2). Comparing with 2010 exposure levels, the concentrations of
PM
2.5
were much higher at baseline, with smaller differences observed
for the other pollutants. Pearson correlations between NO
2
and BC were
moderate to high in sub-cohorts (0.67–0.93) except for in CEANS-
SNACK (0.43; Table S3). PM
2.5
was moderately positively correlated
with exposure to BC and NO
2
in most sub-cohorts, and (warm season) O
3
was moderately negatively correlated with other pollutants in most sub-
cohorts (Table S3).
3.3. Main analyses
In our fully adjusted model (Model 3), signicant positive associa-
tions were observed between PM
2.5
exposure and PD mortality (HR:
1.25; 95% CI: 1.01–1.55), while suggestive positive associations were
evident for NO
2
(1.13; 0.95–1.34) and BC (1.12; 0.94–1.34) (Table 2).
We observed a negative association for O
3
(0.74; 0.58–0.94). Adjust-
ment for individual covariates minimally affected the air pollution effect
estimates (Model 2 versus Model 1). Adjustment for area-level income
mildly decreased the effect estimates (Model 3 versus Model 2).
In two-pollutant models, the association with PM
2.5
was reasonably
robust when including NO
2
or BC although with attenuation towards the
null for either pollutant, and the association with O
3
was mostly un-
changed (Table 2). Two-pollutant results as contributions of NO
2
and BC
individually are difcult to interpret because of the high correlations
between NO
2
and BC in some sub-cohorts (Table S3).
We observed linear associations of PM
2.5
, NO
2
, and BC with PD
mortality, with no evidence of a lower threshold below which air
pollution was not associated with PD mortality (Fig. 2, Table S4). At the
extremes of concentration distribution, the uncertainty of the shape of
Fig. 1. Distribution of annual average concentrations of air pollution for the year 2010 by pooled cohort and sub-cohorts (N ¼271,720). The bold lines in
the middle of the box indicate the median values (the 50th percentile). The lower and upper hinges correspond to the 25th and 75th percentiles. The lower and upper
whiskers extend to the 5th and 95th percentiles. Red dashed lines represent different limited values in EU, U.S., and WHO guidelines (2021 version). For PM
2.5
, they
indicate the annual average limited/guideline values of WHO (5 µg/m
3
), U.S. (12 µg/m
3
), and EU (25 µg/m
3
). For NO
2
, they indicate the annual average limited/
guideline values of WHO (10 µg/m
3
) and EU (40 µg/m
3
). For O
3
, it indicates the 8-hour mean, peak season guideline value of WHO (60 µg/m
3
). O
3
was in the boreal
warm season from April 1st through September 30th. Denition of abbreviation: PM
2.5
, particulate matters with aerodynamic diameters of less than 2.5
μ
m; NO
2
,
nitrogen dioxide; BC, black carbon (measured by the reectance of PM
2.5
lters from 2009 and 2010, expressed in absorbance units); O
3
, ozone. (For interpretation of
the references to colour in this gure legend, the reader is referred to the web version of this article.)
T. Cole-Hunter et al.

Environment International 171 (2023) 107667
6
the relationship was large (Fig. 2). The likelihood ratio test (results not
shown) indicated no signicant deviations from a linear relationship.
The majority of participants (74% to 100%) in the two Nordic cohorts
(CEANS, DNC) were exposed to PM
2.5
levels mostly below 15
μ
g/m
3
(Fig. 1). This disparity in exposure level across sub-cohorts was less
prevalent for NO
2
and O
3
but not for BC (Fig. 1).
We observed that the association between PD mortality with PM
2.5
was statistically stronger in participants who were normal or under
weight, as compared to overweight participants (p-value for interaction
=0.05). We also note stronger association in males, although no
signicant effect modication was observed (Table S5).
3.4. Sensitivity analyses
A number of sensitivity analyses showed robustness in associations
for all pollutants. Associations with PM
2.5
, NO
2
and BC were (border-
line) signicant when using either of the back-extrapolated (ratio or
difference method) baseline year exposures (Table S6). The HRs and
condence intervals for all three pollutants were substantially smaller
because of the higher exposure levels and variability of baseline
Table 2
Associations between long-term air pollution exposure and Parkinson’s disease mortality (N =271,720).
Model 1 Model 2 Model 3 Model 3 þNO
2
Model 3 þPM
2.5
Model 3 þBC Model 3 þO
3
Parkinson’s disease (381deaths)
NO
2
1.19 (1.01, 1.40) 1.18 (1.01, 1.39) 1.13 (0.95, 1.34) / 1.01 (0.81, 1.26) 1.11 (0.74, 1.68) 0.93 (0.72, 1.20)
PM
2.5
1.30 (1.05, 1.60) 1.30 (1.05, 1.60) 1.25 (1.01, 1.55) 1.24 (0.95, 1.62) / 1.28 (0.96, 1.71) 1.11 (0.85, 1.45)
BC 1.18 (1.00, 1.40) 1.17 (0.99, 1.39) 1.12 (0.94, 1.34) 1.02 (0.67, 1.55) 0.97 (0.76, 1.25) / 0.91 (0.71, 1.18)
O
3
0.70 (0.56, 0.89) 0.70 (0.56, 0.89) 0.74 (0.58, 0.94) 0.68 (0.48, 0.98) 0.79 (0.59, 1.07) 0.67 (0.47, 0.96) /
Results are presented as hazard ratio and 95% condence interval [HR (95%CI)] for the following increases: 10 µg/m
3
for NO
2
, 5 µg/m
3
for PM
2.5
, 0.5 ×10
-5
m
-1
for BC
and 10 µg/m
3
for O
3
. BC was measured by the reectance of PM
2.5
lters from 2009 and 2010, expressed in absorbance units.
Model 1: adjusted for age (time axis), sex (strata), sub-cohort (strata), and calendar year of baseline;
Model 2: additionally adjusted for smoking (status, duration, intensity, and intensity
2
), BMI (category), marital status, and employment status;
Model 3: further adjusted for area-level mean year income.
Fig. 2. Concentration-response curves for the associations between long-term exposure to air pollution and Parkinson’s disease mortality. Natural cubic
splines with two degrees of freedom were t for air pollutants based on Model 3, where the hazard ratios equal to one were for minimum pollutant exposures. Solid
black (horizontal) lined-curves indicate hazard ratio values and dashed black lines (with grey shading) indicate their 95% condence intervals. Solid black (vertical)
lines indicate the 5th and 95th percentiles of air pollutants’ concentrations. Dashed red (vertical) lines represent existing EU, US, and WHO guideline limit values.
The histograms show the distributions of exposures in 2010. X-axes are truncated at 60 µg/m
3
for NO
2
and 3 ×10
-5
m
-1
for BC. Denition of abbreviation: PM
2.5
,
particulate matter with aerodynamic diameter of less than 2.5
μ
m; NO
2
, nitrogen dioxide; BC, black carbon (measured by the reectance of PM
2.5
lters from 2009
and 2010, expressed in absorbance units); O
3
, ozone. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of
this article.)
T. Cole-Hunter et al.

Environment International 171 (2023) 107667
7
exposure compared to 2010 exposure (Figure S2). For time-varying ex-
posures, available in a subset of three cohorts (N =132,952) with
available data (CEANS, EPIC-NL, and VHM&PP), HRs were generally
very similar compared to those with the 2010 exposure (Table S7).
Further, the associations with PD mortality were largely robust to
excluding one sub-cohort at a time, except for NO
2
association when
excluding DNC. Associations for PM
2.5
, NO
2
and BC became very
imprecise and non-signicant when excluding the large VHM&PP which
contributed the majority of PD deaths (Table S8).
PM
2.5
component data were available in a subset of 271,003 par-
ticipants. Exposure to the eight PM
2.5
components estimated by the two
different models, SLR and RF, and their correlation with PM
2.5
mass and
NO
2
can be seen in Supplementary Material (Tables S9-S12). In single
SLR models, Cu, Fe, K, and S were positively associated with PD mor-
tality, and associations were attenuated in two-pollutant models with
PM
2.5
and NO
2
(Fig. 3). Single-pollutant RF models showed signicant
associations between K and PD mortality, and no associations for other
components (Fig. 3).
4. Discussion
In the pooled analysis of 271,720 adults from seven European co-
horts, we found that long-term exposure to PM
2.5
, NO
2
, and BC were
associated with PD mortality, with the strongest and most robust asso-
ciations for PM
2.5
. We observed that associations persisted at low-levels
of pollutant concentration, well below current EU limit values. This
study of a large European dataset on Europeans from seven countries
contributes to the evidence base on air pollution and PD risk, albeit that
this evidence base remains mixed. Air pollution is a complex mix of PM
and gases, with some components able to access the brain through the
circulatory or olfactory system. While further research into the relevant
biological pathways is required, several hypotheses exist including that
air pollution components enter the brain and disrupt proteostatis,
causing injury to mitochondria and inducing inammation.
Of 14 studies investigating the association of long-term exposure to
PM
2.5
with PD development or mortality, the majority reported positive
associations, in agreement with our ndings (Table S1). Seven of these
studies were included in a recent review and meta-analysis of PM
2.5
and
PD incidence, reporting a relative risk (RR) estimate of 1.08 (95% CI:
0.98–1.19) per 10
μ
g/m
3
increase (Han et al., 2020), which is weaker
than our HR estimate of 1.25 (95% CI: 1.01–1.55) per 5
μ
g/m
3
. A similar
comparison can be made for NO
2
, for which Han and colleagues re-
ported a RR of 1.03 (95% CI: 0.99–1.07) per 10
μ
g/m
3
compared to our
HR of 1.13 (95% CI: 0.95–1.34) per 10 µg/m
3
. We found generally
stronger associations than those reported in the literature, which are in
line with generally stronger associations reported in the ELAPSE project
with overall mortality (Stafoggia et al., 2022; Strak et al., 2021)
compared to those in the literature and recent meta-analyses (Chen and
Hoek, 2020). We observed a clear linear PM
2.5
exposure–response
function with PD mortality, with a linear function for NO
2
and BC
exposure especially at lower levels, suggesting that there are no lower
threshold levels below which air pollution is not harmful (Fig. 2),
consistent with observations made for other health outcomes in ELAPSE
and other studies (Stafoggia et al., 2022; Strak et al., 2021).
Our results on long-term exposure to NO
2
and PD mortality are in
accordance with current evidence of a suggestive positive association,
however, not to O
3
, as reported by a 2019 meta-analysis by Kasdagli and
colleagues (Kasdagli et al., 2019). They found a signicant positive as-
sociation with O
3
, while we found a negative association with O
3
in
single pollutant models. In two-pollutant models with PM
2.5
, associa-
tions with O
3
remained negative but no longer statistically signicant.
Our negative association with O
3
may be due to its negative correlation
with PM
2.5
, BC and NO
2
, the small exposure contrasts within each sub-
cohort (Fig. 1), or generally low levels of O
3
exposure in our study. The
inverse relationship between O
3
and PD mortality could also be
explained by the moderate-to-strong negative correlation of O
3
with
NO
2
in our pooled cohort, and strong negative correlations of O
3
with
NO
2
and BC in some of our larger populated sub-cohorts (e.g., VHM&PP)
(Table S3). Alternatively, this negative association between O
3
and PD
mortality could be explained by bias related to the study design and lack
of power due to low numbers of cases. In our study, we exploit exposure
contrasts within relatively small study areas compared to other studies,
making it thus less informative for O
3
which tends to vary on a large
spatial scale (Brunekreef et al., 2021). Moreover, one previous study has
shown that ambient ozone concentration is not a suitable surrogate for
individual exposure assessment and may lead to large misclassication
of exposure (Niu et al., 2018). A large US study has previously suggested
the existence of a threshold at 56 ppb (~110
μ
g/m
3
) for the effect of
warm-season O
3
on all-cause mortality (Jerrett et al., 2009). Of ve
studies on O
3
and PD incidence, three (Shin et al., 2018; Cerza et al.,
2018; Zhao et al., 2021) have detected positive associations, whereas
two have found none (Chen et al., 2017; Lee et al., 2016). We present the
Fig. 3. Associations of PM
2.5
components with
Parkinson’s disease mortality in single-pollutant
and two-pollutant models in SLR and RF ana-
lyses. (N ¼271,003; Parkinson’s mortality ¼381).
Results are presented as hazard ratio and 95% con-
dence interval [HR (95%CI)] for the following in-
crements: 5 µg/m
3
for PM
2.5
, 10 µg/m
3
for NO
2
, and
IQR increments for each PM
2.5
component. The main
model was adjusted for sub-cohort identication, age,
sex, year of enrollment, smoking (status, duration,
intensity, and intensity
2
), BMI categories, marital
status, employment status, and 2001 neighborhood-
level mean income. In two-pollutant models, PM
2.5
and NO
2
exposures were estimated using LUR. De-
nition of abbreviation: BMI, body mass index; CI,
condence interval; Cu, copper; Fe, iron; HR, hazard
ratio; K, potassium; Ni, nickel; PM
2.5
, ne particulate
matter; S, sulfur; Si, silicon; V, vanadium; Zn, zinc.
T. Cole-Hunter et al.

Environment International 171 (2023) 107667
8
novel result of a positive association between long-term exposure to BC
and PD mortality, with a HR of 1.12 (0.94, 1.34) per 0.5 ×10
-5
m
-1
,
which is in contrast to the only other study on BC and incidence of PD
reporting an inverse association (RR: 0.96; 95%CI: 0.94–0.99) (Yang
et al., 2018). These varied ndings could be partly explained by a lack of
specicity regarding particle component exposure levels across studies,
with components expected to vary in presence and level across study
sites due to varying air pollution sources contributing to the same
component (Nunez et al., 2021).
In our particle component analysis, we detected a suggestive asso-
ciation between the PM
2.5
component of K and mortality from PD
(Fig. 3). This association, however, was attenuated when adjusting for
either PM
2.5
mass or NO
2
. The particle component results shed some
light on the potential local contributing (combustion or otherwise)
sources. K has been seen as a tracer of biomass burning, which in Europe
may be in the form of residential wood re places or municipal water
boilers. K has also been seen in roadside samples of Poland as an element
of salt sprinkles used by local authorities to de-ice road surfaces during
winter, which may suspend as airborne particles outside of winter
(Skorbiłowicz and Skorbiłowicz, 2019). An analysis from the US sug-
gested that 80% of the variability of K can be explained by factors
associated with soil dust, trafc and biomass burning, however more
accurately indicating trafc when considered alongside Fe (for which
we saw a less suggestive association) (Pachon et al., 2013). Our ndings
differ from those of Nunez and colleagues, who detected associations of
trafc-related nitrate and OM with PD hospitalization (Nunez et al.,
2021), as well as of Palacios and colleagues, who found the relevance of
mercury for PD incidence (Palacios et al., 2014). OM is a complex
mixture of compounds including the potentially neurotoxic polycyclic
aromatic hydrocarbons, which are known to evoke an inammatory
response and contribute to degenerative disease-like syndromes in ani-
mal studies (Nunez et al., 2021). While our study was not able to analyse
this specic PM component, it is moderately-to-highly correlated with
PM
2.5
and may explain those associated ndings. Based on the literature,
more studies are needed to determine which PM component is most
biologically relevant for neuropathology related to PD (Nunez et al.,
2021).
We observed signicant effect modication for the associaton of
PM
2.5
with PD mortality by BMI, with the strongest associations in those
who were not overweight (BMI <25 kg/m
2
, as dened by WHO), which
may be indicative of PD-related weight loss with progressive disease
stages due to malnutrition (van der Marck et al., 2012). We hypothesise
that those individuals with more progressed disease and that are closer
to death would have a lower body weight, also as a function of age-
related weight loss – suggested separately by a stronger association of
PM
2.5
with PD mortality among individuals ≥65 years of age.
A strength of our study is the pooling of data from seven European
cohorts within the ELAPSE framework, which allows for a larger sample
size to investigate mortality from PD. Another strength is that we had
harmonized exposure data based on the Europe-wide hybrid LUR
models of a ne spatial scale. Furthermore, we had detailed and
harmonised information available on potential confounders including
sample characteristics at the individual and area level. While we did
have information on an individual’s history of smoking, we did not have
information on history of traumatic brain injury, another major personal
risk factor for PD, which is somewhat controlled for in our models by
controlling for sex: males show increased likelihood to suffer this injury
due to their greater exposure to events such as road trafc accidents or
contact sports; as well as more likely to smoke, as another major per-
sonal risk factor (Rocca, 2018).
The main limitation of our study is related to the denition of PD
based on mortality data. Numerous studies show that PD is not well
ascertained in death certicates and it underestimates the true burden of
PD (Shi and Counsell, 2021; Hobson and Meara, 2018). Furthermore,
many PD patients die from competing illnesses, and it is a weakness that
we could not identify those, as we did not have information on deaths
from PD as a contributing cause. Although the sample size achieved by
pooling seven European cohorts was large, we had limited statistical
power due to a small number of deaths due to PD. Another limitation is
that exposure data were based on the year 2010, and applied at the
baseline of the cohorts recruited in the 1990s and early 2000s. A pre-
vious study reported stable spatial distribution of NO
2
over 10 years in
the Netherlands prior to 2011 (Eeftens et al., 2011). Furthermore, pre-
dictions from our year 2010 model were highly correlated (R
2
>76%)
with year 2000 and 2005 models for NO
2
and O
3
, and the year 2013
model for PM
2.5
(de Hoogh et al., 2018), indicating a limited impact of
temporal misalignment by exposures based on the year 2010. Further-
more, our sensitivity analyses showed robust associations when using
either back-extrapolated baseline year exposure (Table S6) or time-
varying annual exposure in three cohorts with available information
on address history (Table S7). Another limitation is inherent exposure
misclassication due to lack of information on time spent outdoors and
commuting to work (as personal exposure from outdoor sources), other
sources of air pollution indoors, as well as outdoor and indoor sources at
work. Finally, we did not have environmental noise exposure informa-
tion for the VHM&PP cohort, which contains the majority of our cases,
to treat as a potential confounder for the effect of air pollution on PD
mortality.
5. Conclusions
In conclusion, we found that long-term exposure to PM
2.5
, NO
2
and
BC were associated with the risk of dying from PD, with PM
2.5
found to
be the most relevant pollutant for this risk. We also found that associ-
ations persisted at low levels of pollutant concentration, well below
current EU air quality standards. This study based on a large European
population adds strong novel evidence in support of an association be-
tween air pollution and PD.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Data availability
Data will be made available on request.
Acknowledgements
This study was supported by the Health Effects Institute (HEI)
(#4954-RFA14-3/16-5-3) and the Novo Nordisk Foundation Challenge
Programme [NNF17OC0027812]. The HEI is an organization jointly
funded by the United States Environmental Protection Agency (EPA)
(Assistance Award No. R-82811201) and certain motor vehicle and en-
gine manufacturers. The contents of this article do not necessarily reect
the views of HEI, or its sponsors, nor do they necessarily reect the views
and policies of the EPA or motor vehicle and engine manufacturers.
While HEI has reviewed and approved the study design, it was not
involved in data collection and analysis, decision to publish, or prepa-
ration of the manuscript. We give thanks to all participants in the pooled
cohort studies and the respective study teams of the ELAPSE project for
their hard work and effort. Accordingly, we especially thank Marjan
Tewis for conducting data management tasks when creating the pooled
cohort database. In addition, we specically thank the National Institute
for Public Health and the Environment (RIVM), Bilthoven, the
Netherlands, for their contribution to the ELAPSE project. SALT and
TwinGene are sub-studies of The Swedish Twin Registry (STR), which is
managed by Karolinska Institutet and receives additional funding
through the Swedish Research Council (No. 2017-00641). The KORA
study was initiated and nanced by the Helmholtz Zentrum München –
T. Cole-Hunter et al.

Environment International 171 (2023) 107667
9
German Research Center for Environmental Health, which is funded by
the German Federal Ministry of Education and Research (BMBF) and by
the State of Bavaria. The Novo Nordisk Foundation, the Swedish
Research Council, the German Federal Ministry of Education and
Research, and the State of Bavaria were not involved in the study design,
data collection and analysis, decision to publish, or preparation of the
manuscript.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.envint.2022.107667.
References
Andersen, Z.J., Zhang, J., Jørgensen, J.T., Samoli, E., Liu, S., Chen, J., et al., 2022. Long-
term exposure to air pollution and mortality from dementia, psychiatric disorders,
and suicide in a large pooled European cohort: ELAPSE study. Environ. Int. 170.
Ascherio, A., Schwarzschild, M.A., 2016. The epidemiology of Parkinson’s disease: risk
factors and prevention. Lancet Neurol. 15 (12), 1257–1272.
Beulens, J.W., Monninkhof, E.M., Verschuren, W.M., van der Schouw, Y.T., Smit, J.,
Ocke, M.C., et al., 2010. Cohort prole: the EPIC-NL study. Int. J. Epidemiol. 39 (5),
1170–1178.
Brandt, J., Silver, J.D., Frohn, L.M., Geels, C., Gross, A., Hansen, A.B., et al., 2012. An
integrated model study for Europe and North America using the Danish Eulerian
Hemispheric Model with focus on intercontinental transport of air pollution. Atmos.
Environ. 53, 156–176.
Brunekreef, B., Strak, M., Chen, J., Andersen, Z.J., Atkinson, R., Bauwelinck, M., et al.,
2021. Mortality and Morbidity Effects of Long-Term Exposure to Low-Level PM
2.5
,
BC, NO
2
, and O
3
: An Analysis of European Cohorts in the ELAPSE Project. Res Rep
Health Eff Inst. 208, 1–127.
Cerza, F., Renzi, M., Agabiti, N., Marino, C., Gariazzo, C., Davoli, M., et al., 2018.
Residential exposure to air pollution and incidence of Parkinson’s disease in a large
metropolitan cohort. Environ. Epidemiol. 2 (3).
Chen, J., de Hoogh, K., Gulliver, J., Hoffmann, B., Hertel, O., Ketzel, M., et al., 2019.
A comparison of linear regression, regularization, and machine learning algorithms
to develop Europe-wide spatial models of ne particles and nitrogen dioxide.
Environ. Int. 130, 104934.
Chen, J., Hoek, G., 2020. Long-term exposure to PM and all-cause and cause-specic
mortality: A systematic review and meta-analysis. Environ. Int. 143, 105974.
Chen, J., de Hoogh, K., Gulliver, J., Hoffmann, B., Hertel, O., Ketzel, M., et al., 2020.
Development of Europe-Wide Models for Particle Elemental Composition Using
Supervised Linear Regression and Random Forest. Environ. Sci. Technol. 54 (24),
15698–15709.
Chen, C.Y., Hung, H.J., Chang, K.H., Hsu, C.Y., Muo, C.H., Tsai, C.H., et al., 2017. Long-
term exposure to air pollution and the incidence of Parkinson’s disease: A nested
case-control study. PLoS One 12 (8), e0182834.
Chen, J., Rodopoulou, S., de Hoogh, K., Strak, M., Andersen, Z.J., Atkinson, R., et al.,
2021. Long-Term Exposure to Fine Particle Elemental Components and Natural and
Cause-Specic Mortality-a Pooled Analysis of Eight European Cohorts within the
ELAPSE Project. Environ. Health Perspect. 129 (4), 47009.
Clavel-Chapelon, F., 2015. Cohort Prole: The French E3N Cohort Study. Int. J.
Epidemiol. 44 (3), 801–809.
de Hoogh, K., Chen, J., Gulliver, J., Hoffmann, B., Hertel, O., Ketzel, M., et al., 2018.
Spatial PM
2.5
, NO
2
, O
3
and BC models for Western Europe - Evaluation of
spatiotemporal stability. Environ. Int. 120, 81–92.
de Hoogh, K., Chen, J., Gulliver, J., Hoffmann, B., Hertel, O., Ketzel, M., et al., 2018.
Spatial PM
2.5
, NO
2
, O
3
and BC models for Western Europe - Evaluation of
spatiotemporal stability. Environ. Int. 120, 81–92.
Dorsey, E.R., Elbaz, A., Nichols, E., Abbasi, N., Abd-Allah, F., Abdelalim, A., et al., 2018.
Global, regional, and national burden of Parkinson’s disease, 1990–2016: a
systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. 17
(11), 939–953.
Eeftens, M., Beelen, R., Fischer, P., Brunekreef, B., Meliefste, K., Hoek, G., 2011. Stability
of measured and modelled spatial contrasts in NO
2
over time. Occup. Environ. Med.
68 (10), 765.
Eriksson, A.K., Ekbom, A., Granath, F., Hilding, A., Efendic, S., Ostenson, C.G., 2008.
Psychological distress and risk of pre-diabetes and Type 2 diabetes in a prospective
study of Swedish middle-aged men and women. Diabet. Med. : J. British Diabet.
Assoc. 25 (7), 834–842.
Han, C., Lu, Y., Cheng, H., Wang, C., Chan, P., 2020. The impact of long-term exposure to
ambient air pollution and second-hand smoke on the onset of Parkinson disease: a
review and meta-analysis. Publ. Health 179, 100–110.
Hobson, P., Meara, J., 2018. Mortality and quality of death certication in a cohort of
patients with Parkinson’s disease and matched controls in North Wales, UK at 18
years: a community-based cohort study. BMJ Open 8 (2), e018969.
Holle, R., Happich, M., L¨
owel, H., Wichmann, H.E., 2005. KORA–a research platform for
population based health research. Gesundheitswesen. 67 (Suppl 1), S19–S25.
Hundrup, Y.A., Simonsen, M.K., Jørgensen, T., Obel, E.B., 2012. Cohort prole: the
Danish nurse cohort. Int. J. Epidemiol. 41 (5), 1241–1247.
Hvidtfeldt, U.A., Severi, G., Andersen, Z.J., Atkinson, R., Bauwelinck, M., Bellander, T.,
et al., 2021. Long-term low-level ambient air pollution exposure and risk of lung
cancer - A pooled analysis of 7 European cohorts. Environ. Int. 146, 106249.
Jerrett, M., Burnett, R.T., Pope 3rd, C.A., Ito, K., Thurston, G., Krewski, D., et al., 2009.
Long-term ozone exposure and mortality. N. Engl. J. Med. 360 (11), 1085–1095.
Kasdagli, M.I., Katsouyanni, K., Dimakopoulou, K., Samoli, E., 2019. Air pollution and
Parkinson’s disease: A systematic review and meta-analysis up to 2018. Int. J. Hyg.
Environ. Health. 222 (3), 402–409.
Kim, H., Kim, W.H., Kim, Y.Y., Park, H.Y., 2020. Air Pollution and Central Nervous
System Disease: A Review of the Impact of Fine Particulate Matter on Neurological
Disorders. Front. Publ. Health. 8, 575330.
Kirrane, E.F., Bowman, C., Davis, J.A., Hoppin, J.A., Blair, A., Chen, H., et al., 2015.
Associations of Ozone and PM
2.5
Concentrations With Parkinson’s Disease Among
Participants in the Agricultural Health Study. J. Occup. Environ. Med. 57 (5),
509–517.
Lagergren, M., Fratiglioni, L., Hallberg, I.R., Berglund, J., Elmståhl, S., Hagberg, B., et al.,
2004. A longitudinal study integrating population, care and social services data. The
Swedish National study on Aging and Care (SNAC). Aging Clin. Exp. Res. 16 (2),
158–168.
Lee, P.C., Liu, L.L., Sun, Y., Chen, Y.A., Liu, C.C., Li, C.Y., et al., 2016. Trafc-related air
pollution increased the risk of Parkinson’s disease in Taiwan: A nationwide study.
Environ. Int. 96, 75–81.
Liu, S., Jorgensen, J.T., Ljungman, P., Pershagen, G., Bellander, T., Leander, K., et al.,
2021. Long-term exposure to low-level air pollution and incidence of chronic
obstructive pulmonary disease: The ELAPSE project. Environ. Int. 146, 106267.
Liu, S., Jørgensen, J.T., Ljungman, P., Pershagen, G., Bellander, T., Leander, K., et al.,
2021. Long-term exposure to low-level air pollution and incidence of asthma: the
ELAPSE project. Eur. Respir. J. 57 (6).
Liu, S., Jørgensen, J.T., Ljungman, P., Pershagen, G., Bellander, T., Leander, K., et al.,
2021. Long-term exposure to low-level air pollution and incidence of chronic
obstructive pulmonary disease: The ELAPSE project. Environ. Int. 146, 106267.
Liu, R., Young, M.T., Chen, J.C., Kaufman, J.D., Chen, H., 2016. Ambient Air Pollution
Exposures and Risk of Parkinson Disease. Environ. Health Perspect. 124 (11),
1759–1765.
Magnusson, P.K., Almqvist, C., Rahman, I., Ganna, A., Viktorin, A., Walum, H., et al.,
2013. The Swedish Twin Registry: establishment of a biobank and other recent
developments. Twin Res. Hum. Genet.: Ofc. J. Int. Soc. Twin Stud. 16 (1), 317–329.
Niu, Y., Cai, J., Xia, Y., Yu, H., Chen, R., Lin, Z., et al., 2018. Estimation of personal ozone
exposure using ambient concentrations and inuencing factors. Environ. Int. 117,
237–242.
Nunez, Y., Boehme, A.K., Li, M., Goldsmith, J., Weisskopf, M.G., Re, D.B., et al., 2021.
Parkinson’s disease aggravation in association with ne particle components in New
York State. Environ. Res. 201, 111554.
Pachon, J.E., Weber, R.J., Zhang, X., Mulholland, J.A., Russell, A.G., 2013. Revising the
use of potassium (K) in the source apportionment of PM
2.5
. Atmospheric. Pollut. Res.
4 (1), 14–21.
Palacios, N., Fitzgerald, K.C., Hart, J.E., Weisskopf, M.G., Schwarzschild, M.A.,
Ascherio, A., et al., 2014. Particulate matter and risk of Parkinson disease in a large
prospective study of women. Environ. Health. 13, 80.
Palacios, N., Fitzgerald, K.C., Hart, J.E., Weisskopf, M., Schwarzschild, M.A.,
Ascherio, A., et al., 2017. Air Pollution and Risk of Parkinson’s Disease in a Large
Prospective Study of Men. Environ. Health Perspect. 125 (8), 087011.
Raggi, A., Leonardi, M., 2020. Burden of brain disorders in Europe in 2017 and
comparison with other non-communicable disease groups. J. Neurol. Neurosurg.
Psychiatry 91 (1), 104–105.
Rhew, S.H., Kravchenko, J., Lyerly, H.K., 2021. Exposure to low-dose ambient ne
particulate matter PM
2.5
and Alzheimer’s disease, non-Alzheimer’s dementia, and
Parkinson’s disease in North Carolina. PLoS One 16 (7) e0253253.
Ritz, B., Lee, P.C., Hansen, J., Lassen, C.F., Ketzel, M., Sorensen, M., et al., 2016. Trafc-
Related Air Pollution and Parkinson’s Disease in Denmark: A Case-Control Study.
Environ. Health Perspect. 124 (3), 351–356.
Rocca, W.A., 2018. The burden of Parkinson’s disease: a worldwide perspective. Lancet
Neurol. 17 (11), 928–929.
Samoli, E., Rodopoulou, S., Hvidtfeldt, U.A., Wolf, K., Stafoggia, M., Brunekreef, B.,
et al., 2021. Modeling multi-level survival data in multi-center epidemiological
cohort studies: Applications from the ELAPSE project. Environ. Int. 147, 106371.
Schmermund, A., M¨
ohlenkamp, S., Stang, A., Gr¨
onemeyer, D., Seibel, R., Hirche, H.,
et al., 2002. Assessment of clinically silent atherosclerotic disease and established
and novel risk factors for predicting myocardial infarction and cardiac death in
healthy middle-aged subjects: rationale and design of the Heinz Nixdorf RECALL
Study. Risk Factors, Evaluation of Coronary Calcium and Lifestyle. Am. Heart J. 144
(2), 212–218.
Shi, H., Counsell, C., 2021. Accuracy of death certicates for recording parkinsonian
syndromes and associated dementia. J. Neurol. 268 (1), 140–146.
Shi, L., Wu, X., Danesh Yazdi, M., Braun, D., Abu Awad, Y., Wei, Y., et al., 2020. Long-
term effects of PM2.5 on neurological disorders in the American Medicare
population: a longitudinal cohort study. Lancet Planet Health 4 (12) e557-e65.
Shin, S., Burnett, R.T., Kwong, J.C., Hystad, P., van Donkelaar, A., Brook, J.R., et al.,
2018. Effects of ambient air pollution on incident Parkinson’s disease in Ontario,
2001 to 2013: a population-based cohort study. Int. J. Epidemiol. 47 (6),
2038–2048.
Skorbiłowicz, M., Skorbiłowicz, E., 2019. Content of calcium, magnesium, sodium and
potassium in the street dust from the area of Białystok (Poland). J. Ecol. Eng. 20 (10),
125–131.
T. Cole-Hunter et al.

Environment International 171 (2023) 107667
10
So, R., Chen, J., Mehta, A.J., Liu, S., Strak, M., Wolf, K., et al., 2021. Long-term exposure
to air pollution and liver cancer incidence in six European cohorts. Int. J. Cancer.
149 (11), 1887–1897.
Stafoggia, M., Oftedal, B., Chen, J., Rodopoulou, S., Renzi, M., Atkinson, R.W., et al.,
2022. Long-term exposure to low ambient air pollution concentrations and mortality
among 28 million people: results from seven large European cohorts within the
ELAPSE project. Lancet Planet. Health. 6 (1), e9–e18.
Strak, M., Weinmayr, G., Rodopoulou, S., Chen, J., de Hoogh, K., Andersen, Z.J., et al.,
2021. Long term exposure to low level air pollution and mortality in eight European
cohorts within the ELAPSE project: pooled analysis. BMJ 374, n1904.
Thi´
ebaut, A.C., B´
enichou, J., 2004. Choice of time-scale in Cox’s model analysis of
epidemiologic cohort data: a simulation study. Stat. Med. 23 (24), 3803–3820.
Toro, R., Downward, G.S., van der Mark, M., Brouwer, M., Huss, A., Peters, S., et al.,
2019. Parkinson’s disease and long-term exposure to outdoor air pollution: A
matched case-control study in the Netherlands. Environ. Int. 129, 28–34.
Ulmer, H., Kelleher, C.C., Fitz-Simon, N., Diem, G., Concin, H., 2007. Secular trends in
cardiovascular risk factors: an age-period cohort analysis of 698,954 health
examinations in 181,350 Austrian men and women. J. Intern. Med. 261 (6),
566–576.
van der Marck, M.A., Dicke, H.C., Uc, E.Y., Kentin, Z.H., Borm, G.F., Bloem, B.R., et al.,
2012. Body mass index in Parkinson’s disease: a meta-analysis. Parkinsonism Relat.
Disord. 18 (3), 263–267.
W¨
andell, P.E., Wajngot, A., de Faire, U., Hell´
enius, M.L., 2007. Increased prevalence of
diabetes among immigrants from non-European countries in 60-year-old men and
women in Sweden. Diabetes Metab. 33 (1), 30–36.
WHO, 2021. WHO global air quality guidelines: particulate matter (PM2.5 and PM10),
ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. World Health
Organization.
Wolf, K., Hoffmann, B., Andersen, Z.J., Atkinson, R.W., Bauwelinck, M., Bellander, T.,
et al., 2021. Long-term exposure to low-level ambient air pollution and incidence of
stroke and coronary heart disease: a pooled analysis of six European cohorts within
the ELAPSE project. Lancet Planet. Health. 5 (9), e620–e632.
Yang, Y., Tang, R., Qiu, H., Lai, P.C., Wong, P., Thach, T.Q., et al., 2018. Long term
exposure to air pollution and mortality in an elderly cohort in Hong Kong. Environ.
Int. 117, 99–106.
Yu, Z., Wei, F., Zhang, X., Wu, M., Lin, H., Shui, L., et al., 2021. Air pollution,
surrounding green, road proximity and Parkinson’s disease: A prospective cohort
study. Environ. Res. 197, 111170.
Yuchi, W., Sbihi, H., Davies, H., Tamburic, L., Brauer, M., 2020. Road proximity, air
pollution, noise, green space and neurologic disease incidence: a population-based
cohort study. Environ. Health. 19 (1), 8.
Zhao, N., Pinault, L., Toyib, O., Vanos, J., Tjepkema, M., Cakmak, S., 2021. Long-term
ozone exposure and mortality from neurological diseases in Canada. Environ. Int.
157, 106817.
T. Cole-Hunter et al.