Air pollution and lung cancer incidence in 17 European cohorts : Prospective analyses from the European Study of Cohorts for Air Pollution Effects (ESCAPE)

TLDR: The meta-analyses showed a statistically significant association between risk for lung cancer and PM10 and PM2·5, and no association between lungcancer and nitrogen oxides concentration or traffic intensity on the nearest street.

Abstract: Summary Background Ambient air pollution is suspected to cause lung cancer. We aimed to assess the association between long-term exposure to ambient air pollution and lung cancer incidence in European populations. Methods This prospective analysis of data obtained by the European Study of Cohorts for Air Pollution Eff ects used data from 17 cohort studies based in nine European countries. Baseline addresses were geocoded and we assessed air pollution by land-use regression models for particulate matter (PM) with diameter of less than 10 μm (PM10), less than 2·5 μm (PM2·5), and between 2·5 and 10 μm (PMcoarse), soot (PM2·5absorbance), nitrogen oxides, and two traffi c indicators. We used Cox regression models with adjustment for potential confounders for cohort-specifi c analyses and random eff ects models for meta-analyses. Findings The 312 944 cohort members contributed 4 013 131 person-years at risk. During follow-up (mean 12·8 years), 2095 incident lung cancer cases were diagnosed. The meta-analyses showed a statistically signifi cant association between risk for lung cancer and PM10 (hazard ratio [HR] 1·22 [95% CI 1·03–1·45] per 10 μg/m³). For PM2·5 the HR was 1·18 (0·96–1·46) per 5 μg/m³. The same increments of PM10 and PM2·5 were associated with HRs for adenocarcinomas of the lung of 1·51 (1·10–2·08) and 1·55 (1·05–2·29), respectively. An increase in road traffi c of 4000 vehicle-km per day within 100 m of the residence was associated with an HR for lung cancer of 1·09 (0·99–1·21). The results showed no association between lung cancer and nitrogen oxides concentration (HR 1·01 [0·95–1·07] per 20 μg/m³) or traffi c intensity on the nearest street (HR 1·00 [0·97–1·04] per 5000 vehicles per day).

Figures

Table 4: Associations between PM10 and PM2·5 and risk for lung cancer, according to air pollution thresholds*

Table 3: Associations between PM10 and PM2·5 and risk for lung cancer for all participants and those who did not change residence during follow-up, according to histological cancer subtype

Table 1: Participants, lung cancer cases, mean air pollution concentrations, and traffi c in each cohort

Table 2: Meta-analyses of associations between air pollutants and traffi c indicators and the risk for lung cancer

Full text

www.thelancet.com/neurology Published online July 10, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70279-1
1
Articles
Air pollution and lung cancer incidence in 17 European
cohorts: prospective analyses from the European Study of
Cohorts for Air Pollution Eff ects (ESCAPE)
Ole Raaschou-Nielsen, Zorana J Andersen, Rob Beelen, Evangelia Samoli, Massimo Stafoggia, Gudrun Weinmayr, Barbara Hoff mann, Paul Fischer,
Mark J Nieuwenhuijsen, Bert Brunekreef, Wei W Xun, Klea Katsouyanni, Konstantina Dimakopoulou, Johan Sommar, Bertil Forsberg, Lars Modig,
Anna Oudin, Bente Oftedal, Per E Schwarze, Per Nafstad, Ulf De Faire, Nancy L Pedersen, Claes-Göran Östenson, Laura Fratiglioni, Johanna Penell,
Michal Korek, Göran Pershagen, Kirsten T Eriksen, Mette Sørensen, Anne Tjønneland, Thomas Ellermann, Marloes Eeftens, Petra H Peeters,
Kees Meliefste, Meng Wang, Bas Bueno-de-Mesquita, Timothy J Key, Kees de Hoogh, Hans Concin, Gabriele Nagel, Alice Vilier, Sara Grioni,
Vittorio Krogh, Ming-Yi Tsai, Fulvio Ricceri, Carlotta Sacerdote, Claudia Galassi, Enrica Migliore, Andrea Ranzi, Giulia Cesaroni, Chiara Badaloni,
Francesco Forastiere, Ibon Tamayo, Pilar Amiano, Miren Dorronsoro, Antonia Trichopoulou, Christina Bamia, Paolo Vineis*, Gerard Hoek*
Summary
Background Ambient air pollution is suspected to cause lung cancer. We aimed to assess the association between
long-term exposure to ambient air pollution and lung cancer incidence in European populations.
Methods This prospective analysis of data obtained by the European Study of Cohorts for Air Pollution Eff ects
used
data from 17 cohort studies based in nine European countries. Baseline addresses were geocoded and we assessed air
pollution by land-use regression models for particulate matter (PM) with diameter of less than 10 μm (PM
10
), less than
2·5 μm (PM
2·5
), and between 2·5 and 10 μm (PM
coarse
), soot (PM
2·5absorbance
), nitrogen oxides, and two traffi c indicators.
We used Cox regression models with adjustment for potential confounders for cohort-specifi c analyses and random
eff ects models for meta-analyses.
Findings The 312 944 cohort members contributed 4 013 131 person-years at risk. During follow-up (mean 12·8 years),
2095 incident lung cancer cases were diagnosed. The meta-analyses showed a statistically signifi
cant association between
risk for lung cancer and PM
10
(hazard ratio [HR] 1·22 [95% CI 1·03–1·45] per 10 μg/m³). For PM
2·5
the HR was 1·18
(0·96–1·46) per 5 μg/m³. The same increments of PM
10
and PM
2·5
were associated with HRs for adenocarcinomas of the
lung of 1·51 (1·10–2·08) and 1·55 (1·05–2·29), respectively. An increase in road traffi c of 4000 vehicle-km per day within
100 m of the residence was associated with an HR for lung cancer of 1·09 (0·99–1·21). The results showed no association
between lung cancer and nitrogen oxides concentration (HR 1·01 [0·95–1·07] per 20 μg/m³) or traffi c intensity on the
nearest street (HR 1·00 [0·97–1·04] per 5000 vehicles per day).
Interpretation Particulate matter air pollution contributes to lung cancer incidence in Europe.
Funding European Community’s Seventh Framework Programme.
Introduction
Lung cancer is one of the most common cancers and has a
poor prognosis. Active smoking is the main cause, but
occupational exposures, residential radon, and environ-
mental tobacco smoke are also established risk factors.
Furthermore, lower socioeconomic position has been
associated with a higher risk for lung cancer.
1
Ambient air
pollution, specifi cally particulate matter with absorbed
polycyclic aromatic hydrocarbons and other genotoxic
chemicals, is suspected to increase the risk for lung cancer.
Results of several epidemiological studies have shown
higher risks for lung cancer in association with various
measures of air pollution
2–11
and suggested an association
mainly in non-smokers
4,12
and never-smokers
13,14
and in
individuals with low fruit con sumption.
4,13
In developed
countries, overall lung cancer incidence rates have
stabilised during the past few decades, but major shifts
have been recorded in the frequencies of diff erent
histological types of lung cancer, with substantial relative
increases in adenocarcinomas and decreases in squamous-
cell carcinomas.
15
Changes in tobacco blends
15
and ambient
air pollution
16,17
might have contributed to these shifts.
Within the European Study of Cohorts for Air Pollution
Eff ects (ESCAPE), we aimed to analyse data from
17 European cohort studies with a wide range of exposure
levels to investigate the following hypotheses:
that ambient air pollution at the residence (specifi cally
particulate matter) is associated with risk for lung cancer;
that the association between air pollution and risk for
lung cancer is stronger for non-smokers and people with
low fruit intake; and that the association with air pollution
is stronger for adenocarcinomas and squamous-cell
carcinomas than for all lung cancers combined.
Methods
Study design and participants
This study is a prospective analysis of data obtained by
ESCAPE—an investigation into the long-term eff ects of
Published Online
July 10, 2013
http://dx.doi.org/10.1016/
S1470-2045(13)70279-1
See Online/Comment
http://dx.doi.org/10.1016/
S1470-2045(13)70302-4
*Joint last authors
See Online for related
multimedia content
Danish
Cancer Society Research
Center
, Copenhagen, Denmark
(O Raaschou-Nielsen PhD,
Z J Andersen PhD,
K T Eriksen PhD, M Sørensen PhD,
A Tjønneland DMSc); Center for
Epidemiology and Screening,
Department of Public Health,
University of Copenhagen,
Copenhagen, Denmark
(Z J Andersen); Institute for Risk
Assessment Sciences, Utrecht
University, Utrecht,
Netherlands (R Beelen PhD,
Prof B Brunekreef PhD,
M Eeftens MSc, K Meliefste BSc,
M Wang
MSc, G Hoek PhD);
Department of
Hygiene,
Epidemiology and Medical
Statistics, Medical School,
National and Kapodistrian
University of A
thens, Athens,
Greece (E Samoli PhD,
Prof K Katsouyanni PhD,
K Dimakopoulou MSc,
Prof A Trichopoulou MD,
C Bamia PhD); Department of
Epidemiology
, Lazio Regional
Health Service, Local Health
Unit
ASL RME, Rome, Italy
(M Stafoggia MSc,
G Cesaroni MSc, C Badaloni MSc,
F Forastiere PhD); Institute of
Epidemiology and Medical
Biometry, Ulm University, Ulm,
Germany (G Weinmayr PhD,
G Nagel PhD); IUF–Leibniz
Research Institute for
Environmental Medicine,
Düsseldorf, Germany
(G Weinmayr,

Articles
2
www.thelancet.com/neurology Published online July 10, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70279-1
exposure to air pollution on human health in Europe—
which included 36 European areas in which air pollution
was measured, land-use regression models were
developed, and cohort studies were located. The present
study included 17 cohort studies, located in 12 areas, from
which information about incident lung cancer cases and
the most important potential confounders could be
obtained, and where the resources needed for parti cipation
were available. These cohorts were in Sweden (European
Prospective Investigation into Cancer and Nutrition
[EPIC]-Umeå, Swedish National Study on Aging and Care
in Kungsholmen [SNAC-K], Stockholm Screening Across
the Lifespan Twin study and TwinGene [SALT], Stockholm
60 years old and IMPROVE study [Sixty], Stockholm
Diabetes Prevention Program [SDPP]), Norway (Oslo
Health Study [HUBRO]), Denmark (Diet, Cancer and
Health study [DCH]), the Netherlands (EPIC-Monitoring
Project on Risk Factors and Chronic Diseases in the
Netherlands [MORGEN], EPIC-PROSPECT), the UK
(EPIC-Oxford), Austria (Vorarlberg Health Monitoring and
Prevention Programme [VHM&PP]), Italy (EPIC-Varese,
EPIC-Turin, Italian Studies of Respiratory Disorders
in Childhood and Environment [SIDRIA]-Turin,
SIDRIA-Rome), Spain (EPIC-San Sebastian), and Greece
(EPIC-Athens; fi gure 1). The study areas were mostly large
cities and the surrounding suburban or rural communities.
Some of the cohorts covered large regions of the country,
such as EPIC-MORGEN in the Netherlands, EPIC-Oxford
in the UK, and the VHM&PP cohort in Austria. For DCH,
EPIC-Oxford, VHM&PP, and EPIC-Athens, exposure to
air pollution was assessed for part of the original cohort
only, and only those parts were analysed (restrictions are
specifi ed in the appendix pp 8, 11, 12, and 18). The use of
cohort data in ESCAPE was approved by the local ethical
and data protection authorities. Each cohort study followed
the rules for ethics and data protection set up in the
country in which they were based.
Procedures
The association between long-term exposure to air
pollution and incidence of lung cancer was analysed in
each cohort separately at the local centre by common
standardised protocols for exposure assessment,
outcome defi nition, confounder models, and statistical
analyses. Cohort-specifi c eff ect estimates were combined
by meta-analysis at the Danish Cancer Society Research
Center, Copenhagen, Denmark. A pooled analysis of all
cohort data was not possible due to data transfer and
privacy issues.
The main outcome was all cancers of the lung; secondary
analyses addressed adenocarcinomas and squamous-cell
carcinomas of the lung. We included cancers located in
the bronchus and the lung (International Statistical
Classifi cation of Diseases and Related Health Problems,
10th revision [ICD10] and International Classifi cation of
Diseases for Oncology, 3rd edition [ICDO3] C34·0–C34·9).
We only included primary cancers (ie, not metastases).
Each cancer was histologically characterised, and data for
squamous-cell carcinomas (ICDO3 8050–8084; fi fth
digit morphology code 3) and adenocarcinomas
(ICDO3 8140–8384; fi fth digit morphology code 3) in
particular were obtained. Lymphomas in the lung
(ICDO3 morphology codes 9590/3–9729/3) were not
included. The characterisation of histology was based on
routine pathology; this study did not include verifi cation
of tumour histology. The cohort members were followed
up for cancer incidence in national or local cancer
registries, except for EPIC-Athens, in which cancer cases
were identifi ed by questionnaires and telephone
interviews followed by verifi cation of medical records, and
the SIDRIA cohorts, for which hospital discharge and
mortality register data were used.
Exposure assessment
Air pollution concentrations at the baseline residential
addresses of study participants were estimated by
land-use regression models in a three-step, standardised
Prof B Hoff mann MD); Medical
Faculty, Heinrich Heine
University of Düsseldorf,
Düsseldorf, Germany
(Prof B Hoff mann); National
Institute for Public Health and
the Environment, Bilthoven,
Netherlands (P Fischer MSc,
B Bueno-de-Mesquita PhD);
Center for Research in
Environmental Epidemiology,
Parc de Recerca Biomèdica de
Barcelona, Barcelona, Spain
(M J Nieuwenhuijsen PhD); Julius
Center for Health Sciences and
Primary Care, University
Medical Center Utrecht,
Utrecht, Netherlands
(Prof B Brunekreef,
Prof P H Peeters PhD); MRC-HPA
Centre for Environment and
Health, Department of
Epidemiology and
Biostatistics, Imperial College
London, St Mary’s Campus,
Basque country
Rome
Athens
Varese
Turin
Vorarlberg
NetherlandsLondon/Oxford
Umeå
Oslo
Stockholm
Copenhagen
PM, NOX, and NO
2
measured
NOX and NO
2
measured
Figure 1: Areas where cohort members lived, measurements were taken, and land-use regression models for
prediction of air pollution were developed
NO
2
=nitrogen dioxide. NOx=nitrogen oxides (the sum of nitric oxide and nitrogen dioxide). PM=particulate matter.

Articles
www.thelancet.com/neurology Published online July 10, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70279-1
3
London, UK (W W Xun MPH,
K de Hoogh PhD,
Prof P Vineis MPH); Division of
Occupational and
Environmental Medicine,
Department of Public Health
and Clinical Medicine, Umeå
University, Umeå, Sweden
(J Sommar MSc,
Prof B Forsberg PhD,
L Modig PhD, A Oudin PhD);
Norwegian Institute of Public
Health, Oslo, Norway
(B Oftedal PhD,
procedure. First, particulate matter with an aerodynamic
diameter of less than 10 μm (PM
10
), particulate matter
with aerodynamic diameter of less than 2·5 μm (PM
2·5
),
blackness of the PM
2·5
exposed fi lter (PM
2·5absorbance
),
determined by measurement of light refl ectance (a
marker for soot and black carbon), nitrogen oxides
(NOx), and nitrogen dioxide (NO
2
) were measured during
diff erent seasons at locations for each cohort population
between October, 2008, and April, 2011.
18,19
PM
coarse
was
calculated as the diff erence between PM
10
and PM
2·5
(ie,
PM with diameter 2·5–10 μm). In three areas, only NO
2
and NOx were measured (fi gure 1). Second, land-use
regression models were developed for each pollutant in
each study area, with the yearly mean concentration as
the dependent variable and an extensive list of
geographical attributes as possible predictors.
20,21
Generally, predictors for PM
10
, PM
2·5
, NOx, and NO
2
were
related to traffi c or roads and population or building
density. Variables related to industry, proximity to a port,
and altitude were also predictors in some models. The
models generally explained a large fraction of measured
spatial variation, the R² from leave-one-out
cross-validation usually falling between 0·60 and 0·80
(appendix p 20). Finally, the models were used to assess
Total
participants
Age at
baseline
(years)
All lung
cancer
Adeno-
carcinoma*
Squamous-
cell
carcinoma*
PM
10
(μg/m
3
)
PM
coarse
(μg/m
3
)
PM
2·5
(μg/m
3
)
PM
2·5absorbance
(10
–5
/m)
NO
2
(μg/m
3
)
NOx
(μg/m
3
)
Traffi c on
nearest
street
(vehicles per
day)
Traffi c load on
major streets
within 100 m
(vehicle-km
per day)
EPIC-Umeå,
Sweden
22 136 46·0
(12·2)
69
(0·31%)
34 (0·15%) 18 (0·08%) NA NA NA NA 5·2 (2·5) 8·7
(5·7)
845 (1530) 102 (417)
HUBRO, Oslo,
Norway
17 640 47·8
(15·0)
75
(0·43%)
25 (0·14%) ·· 13·5
(3·1)
4·0
(2·0)
8·9
(1·3)
1·2 (0·3) 20·9 (8·0) 38·3
(15·5)
2502 (5117) 821 (1840)
SNAC-K,
Stockholm,
Sweden
2384 73·1
(10·7)
18
(0·76%)
13† (0·55%) ·· 16·4
(6·0)
8·6
(4·8)
8·0
(1·3)
0·8 (0·2) 17·5 (4·9) 33·5
(12·6)
3888 (9886) 2298 (3699)
SALT, Stockholm,
Sweden
4732 57·9
(10·2)
29
(0·61%)
12 (0·25%) ·· 14·9
(3·9)
7·3
(3·0)
7·3
(1·3)
0·6 (0·2) 10·9 (4·2) 18·9
(9·4)
1460 (3351) 587 (1623)
Sixty, Stockholm,
Sweden
3813 60·4
(0·1)
38
(1·00%)
22 (0·58%) 5 (0·13%) 15·0
(3·8)
7·3
(2·9)
7·3
(1·3)
0·6 (0·2) 10·7 (4·2) 18·6
(9·4)
1453 (3466) 512 (1446)
SDPP, Stockholm,
Sweden
7116 47·1
(5·0)
35
(0·49%)
22 (0·31%) 5 (0·07) 13·6
(3·2)
6·3
(2·4)
6·6
(1·2)
0·5 (0·1) 8·4 (1·7) 14·4
(3·3)
861 (1621) 110 (423)
DCH,
Copenhagen,
Denmark
37 447 56·8
(4·4)
638
(1·70%)
236 (0·63%) 106 (0·28%) 17·1
(1·9)
5·7
(1·0)
11·3
(0·9)
1·2 (0·2)
16·3 (7·0) 26·7
(18·4)
2991 (7209) 1221 (2332)
EPIC-MORGEN,
Netherlands
15 993 43·7
(10·7)
92
(0·58%)
32 (0·20%) 24 (0·15%) 25·6
(1·7)
8·6
(1·1)
16·9
(0·6)
1·4 (0·2) 23·8 (7·0) 36·5
(11·8)
1535 (4084) 917 (1979)
EPIC-PROSPECT,
Netherlands
14 630 57·6
(6·0)
112
(0·77%)
43 (0·29%) 16 (0·11%) 25·3
(1·2)
8·5
(0·7)
16·8
(0·5)
1·4 (0·2) 26·7 (4·6) 39·6
(10·5)
1020 (3433) 678 (1513)
EPIC-Oxford, UK 36 832 45·3
(13·6)
78
(0·21%)
19 (0·05%) 9 (0·02%) 16·1
(2·0)
6·4
(0·9)
9·8
(1·1)
1·1 (0·3) 24·5 (8·0) 40·9
(15·6)
1381 (4345) 373 (1287)
VHM&PP,
Vorarlberg, Austria
108 018 42·8
(14·9)
678
(0·63%)
223 (0·21%) 157 (0·15%) 20·7
(2·4)
6·7
(0·9)
13·6
(1·2)
1·7 (0·2) 19·9 (5·5) 40·0
(9·5)
1687 (3582) 294 (991)
EPIC-Varese, Italy 9506 51·6
(8·2)
43
(0·45%)
17 (0·18%) 12 (0·13%) NA NA NA NA 43·8 (17·3) 86·8
(41·9)
NA NA
EPIC-Turin, Italy 7216 50·4
(7·6)
48
(0·67%)
23 (0·32%) ·· 46·6
(4·6)
16·6
(3·0)
30·1
(2·0)
3·1 (0·4) 53·0 (10·9) 96·2
(21·5)
3903 (9164) 465 (912)
SIDRIA-Turin, Italy 4816 44·0
(6·2)
19
(0·39%)
·· ·· 48·1
(4·1)
17·0
(2·5)
31·0
(1·7)
3·2 (0·4) 59·8 (10·6) 107·3
(24·3)
4291 (10 202) 810 (1379)
SIDRIA-Rome,
Italy
9105 44·3
(6·0)
53
(0·58%)
··
·· 36·5
(5·0)
16·7
(3·4)
19·4
(1·8)
2·7 (0·5)
39·1 (9·1) 82·0
(23·9)
2956 (6728) 1392 (2825)
EPIC-San
Sebastian, Spain
7464 49·4
(7·7)
52
(0·70%)
·· ·· NA NA NA NA 23·8 (6·6) 47·1
(12·5)
NA 673 (2614)
EPIC-Athens,
Greece
4096 49·0
(11·7)
18
(0·44%)
6 (0·15%) ·· 45·2
(13·7)
20·8
(2·6)
20·4
(2·7)
2·3 (0·5) 38·0 (13·7) 75·5
(41·0)
9073 (12 512) 11 000
(15 000)
Data are n, mean (SD), and n (%). PM
10
=particulate matter with diameter <10 µm. PM
coarse
=particulate matter with diameter 2·5–10 µm. PM
2·5
=particulate matter with diameter <2·5 µm.
PM
2·5absorbance
=soot.
NO
2
=nitrogen dioxide. NOx=nitrogen oxides (the sum of nitric oxide and nitrogen dioxide). EPIC=European Prospective Investigation into Cancer and Nutrition. NA=not available. HUBRO=Oslo Health Study.
SNAC-K=Swedish National Study on Aging and Care in Kungsholmen. SALT=Screening Across the Lifespan Twin study and TwinGene. Sixty=Stockholm 60 years old and IMPROVE. SDPP=Stockholm Diabetes
Prevention Program. DCH=Diet, Cancer and Health study. MORGEN=Monitoring Project on Risk Factors and Chronic Diseases in the Netherlands. VHM&PP=Vorarlberg Health Monitoring and Prevention
Programme. SIDRIA=Italian Studies of Respiratory Disorders in Childhood and Environment. ··=No data or too few cases for the model to converge. *Of the lung. †Contributed to results for adenocarcinomas of
the lung in participants who lived at the same residence during the whole follow-up, but did not contribute to the results for all participants because the model did not converge.
Table 1: Participants, lung cancer cases, mean air pollution concentrations, and traffi c in each cohort

Articles
4
www.thelancet.com/neurology Published online July 10, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70279-1
P E Schwarze PhD,
Prof P Nafstad MD); Institute of
Health and Society, University
of Oslo, Oslo, Norway
(Prof P Nafstad); Institute of
Environmental Medicine
(Prof U De Faire PhD,
J Penell PhD, M Korek MSc,
Prof G Pershagen PhD),
Department of Medical
Epidemiology and Biostatistics
(Prof N L Pedersen PhD),
Department of Molecular
Medicine and Surgery
(Prof C-G Östenson PhD), and
Aging Research Center,
Department of Neurobiology,
exposure at the baseline address of each cohort member.
We also collected information on two indicators of traffi c
at the residence: traffi c intensity (vehicles per day) on the
nearest street and total traffi c load (vehicle-km driven per
day) on all major roads within 100 m.
Statistical analyses
Proportional hazards Cox regression models were fi tted
for each cohort, with age as the underlying timescale.
Participants were followed up for lung cancer from
enrolment until the time of a lung cancer diagnosis or
censoring. Participants with a cancer (except non-
melanoma skin cancer) before enrolment were excluded.
Censoring was done at the time of death, a diagnosis of
any other cancer (except non-melanoma skin cancer),
emigration, disappearance, loss to follow-up for other
reasons, or end of follow-up, whichever came fi rst. For
the analyses of histological subtypes of lung cancer, cases
of diff erent histological subtypes were censored.
Air pollution exposure was analysed as a linear
variable in three a-priori specifi ed confounder models.
Model 1 included sex, calendar time (year of enrolment;
linear), and age (time axis). Model 2 additionally
adjusted for smoking status (never, former, or current),
smoking intensity, square of smoking intensity,
smoking duration, time since quitting smoking,
environmental tobacco smoke, occupation, fruit intake,
marital status, level of education, and employment
status (all referring to baseline). We entered a squared
term of smoking intensity because we expected a
non-linear association with lung cancer. Model 3 (the
main model) further adjusted for area-level
socioeconomic status. A cohort was included only if
information about age, sex, calendar time, smoking
status, smoking intensity, and smoking duration were
available.
We assessed individual characteristics as a-priori
potential eff ect modifi ers: age (<65 years or ≥65 years),
sex, level of education, smoking status, fruit intake
(<150 g, 150–300 g, or ≥300 g per day). Age was analysed
time dependently. For a few cohorts (HUBRO, Sixty,
SDPP) for which there was information about fruit
intake in categories such as “a few times per week”,
“daily”, and “several times per day”, the lowest category
was analysed as less than 150 g per day, the medium
category as 150–300 g per day, and the highest category as
300 g per day or greater.
We undertook several sensitivity analyses and model
checks for each cohort, all with confounder model 3.
First, we restricted the analyses to participants who had
lived at the baseline address throughout follow-up to
minimise misclassifi cation of long-term exposure
relevant to the development of lung cancer. Second, we
added an indicator of extent of urbanisation to model 3.
Third, we tested the linear assumption in the relation
between each air pollutant and lung cancer by replacing
the linear term with a natural cubic spline with three
equally spaced inner knots, and compared the model fi t
of the linear and the spline models by the likelihood-ratio
test. Fourth, to investigate if an association between air
pollution and risk for lung cancer was detectable below
a-priori defi ned thresholds, we ran models including
only participants exposed to air pollution concentrations
below those thresholds.
In the meta-analysis, we used random-eff ects models
to pool the results for cohorts.
22
I² statistics
23
and p values
for the χ² test from Cochran’s Q were calculated to
investigate the heterogeneity among cohort-specifi c
eff ect estimates. Eff ect modifi cation was tested by
meta-analysing the pooled estimates from the diff erent
strata with the χ² test of heterogeneity. We assessed the
HUBRO
SNAC-K
SALT
Sixty
SDPP
DCH
EPIC-MORGEN
EPIC-PROSPECT
EPIC-Oxford
VHM&PP
EPIC-Turin
SIDRIA-Turin
SIDRIA-Rome
EPIC-Athens
HUBRO
SNAC-K
SALT
Sixty
SDPP
DCH
EPIC-MORGEN
EPIC-PROSPECT
EPIC-Oxford
VHM&PP
EPIC-Turin
SIDRIA-Turin
SIDRIA-Rome
EPIC-Athens
A
B
0
PM
10
concentration (μg/m
3
)
4020 60 10080
0
PM
2·5
concentration (μg/m
3
)
2010 30 401552535
Figure 2: Distribution of particulate matter air pollution at participant addresses in each cohort
PM
10
concentration (A) and PM
2·5
concentration (B) in each of the cohort studies. Pink boxes show median (central
vertical line) and 25th and 75th percentiles (ends of box); lines extending from the left of each box show the
concentration range from the 10th to the 25th percentile; lines extending from the right of each box show the
concentration range from the 75th to the 90th percentile. The black circles show each concentration below the
10th percentile and above the 90th percentile. PM
10
=particulate matter with diameter <10 µm. PM
2·5
=particulate
matter with diameter <2·5 µm.

Articles
www.thelancet.com/neurology Published online July 10, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70279-1
5
Care Sciences and Society
(L Fratiglioni PhD), Karolinska
Institute, Stockholm, Sweden;
Department of Environmental
Science, Aarhus University,
Roskilde, Denmark
(T Ellermann PhD); Cancer
Epidemiology Unit, Nuffi eld
Department of Clinical
Medicine, University of Oxford,
Oxford, UK (Prof T J Key DPhil);
Agency for Preventive and
Social Medicine, Bregenz,
Austria (H Concin MD, G Nagel);
INSERM, Centre for Research in
Epidemiology and Population
Health, U 1018, Nutrition,
Hormones and Women’s
Health Team, Villejuif, France
(A Vilier MSc); University Paris
Sud, UMRS 1018, Villejuif,
France (A Vilier); Institut
Gustave-Roussy, Villejuif,
France (A Vilier); Epidemiology
and Prevention Unit,
Fondazione IRCCS Istituto
Nazionale dei Tumori, Milan,
Italy (S Grioni BSc, V Krogh
MD);
Department of
Epidemiology
and Public Health, Swiss
Tropical and Public Health
Institute, University
of Basel,
Basel, Switzerland
(M-Y Tsai PhD); Department of
Environmental and
Occupational Health Sciences,
University of
Washington,
Seattle, W
A, USA (M-Y Tsai);
Human Genetics Foundation,
Turin, Italy (F Ricceri PhD); Unit
of Cancer Epidemiology, AO
Citta’ della Salute e della
Scienza–University of Turin and
Center for Cancer Prevention,
Turin, Italy (C Sacerdote PhD,
C Galassi MD, E Migliore MSc);
Environmental Health
Reference Centre–Regional
Agency for Environmental
Prevention of Emilia-Romagna,
Modena, Italy (A Ranzi PhD);
Health Division of Gipuzkoa,
Research Institute of
BioDonostia, Donostia-San
Sebastian, Spain
(I Tamayo MSc); CIBERESP,
Consortium for Biomedical
Research in Epidemiology and
Public Health, Madrid, Spain
(P Amiano MSc,
M Dorronsoro MD); and Hellenic
Health Foundation, Athens,
Greece (Prof A Trichopoulou)
Correspondence to:
Dr Ole Raaschou-Nielsen, Danish
Cancer Society Research Center,
2100 Copenhagen, Denmark
ole@cancer.dk
See
Online for appendix
robustness of the results by repeating the meta-analysis
after exclusion of the two largest cohorts. The proportional
hazards assumption of the Cox model was not violated
(appendix, p 19).
We used a common STATA script for all analyses,
except for spline models, which were fi tted with
R software. The versions of software used to analyse
individual cohorts are listed in the appendix (pp 2–18).
Role of the funding source
The sponsors had no role in the study design, data
collection, data analysis, data interpretation, or writing of
the report. Authors with access to the raw data included
JS and AO (EPIC-Umeå), BO (HUBRO), JP (SNAC-K,
SALT, Sixty, and SDPP), ZJA (DCH), RB (EPIC-MORGEN
and EPIC-PROSPECT), WWX (EPIC-Oxford and
EPIC-Varese), GW (VHM&PP), FR (EPIC-Turin), CG
and EM (SIDRIA-Turin), GC (SIDRIA-Rome), IT
(EPIC-San Sebastian), and KK (EPIC-Athens). The
corresponding author had full access to all analysis
results from each cohort and fi
nal responsibility for the
decision to submit for publication.
Results
The 17 cohorts in nine European countries that
contributed to this study contained 312 944 cohort
members and contributed 4 013 131 person-years at risk
and 2095 incident lung cancer cases that developed
during follow-up (average follow-up was 12·8 years).
More details of each cohort, including characteristics of
the participants, available variables, and their distribution
are provided in the appendix (pp 2–18). Most of the
cohort studies recruited participants in the 1990s
(appendix, pp 2–18). The number of participants and the
number of those who developed cancer varied
substantially between cohorts, with the Danish (DCH)
and Austrian (VHM&PP) cohorts contributing more
than half the lung cancer cases (table 1). The cohort areas
represented a wide range of air pollution concentrations,
with three to 12 times higher mean air pollution levels in
some southern European areas than in some northern
European areas (table 1). The variation in exposure
within study areas was substantial (fi
gure 2; appendix
pp 26–28). The mean age at enrolment in each cohort
ranged from 43 to 73 years (table 1).
The meta-analysis showed an association with risk for
lung cancer that was statistically signifi
cant for PM
10
concentration (hazard ratio [HR] 1·22 [95% CI
1·03–1·45] per 10 μg/m³) in confounder model 3. For
PM
2·5
concentration, the HR was 1·18 (0·96–1·46)
per 5 μg/m³, and for traffi c load at major roads within
100 m the HR was 1·09 (0·99–1·21) per 4000 vehicle-km
per day in confounder model 3 (table 2). The results
from model 1, with adjustment only for age, sex, and
calendar time, showed stronger associations; the eff ect
of adjustment was due mainly to the smoking variables.
Results of models 2 and 3 showed no association
between risk for lung cancer and NO
2
, NOx, or traffi c
intensity at the nearest street (table 2). Restriction to the
14 cohorts for whom estimates of exposure to particulate
matter were available gave similar results for NO
2
(HR 1·01, 95% CI 0·94–1·09) and NOx (HR 1·03,
0·97–1·10). Figure 3 shows the HRs for each cohort
from the meta-analyses for PM
10
and PM
2·5
. Although
the HRs varied substantially across cohorts, the 95% CIs
for each cohort always included the overall meta-
analysis estimate, and we did not identify any signifi cant
heterogeneity between cohorts. The meta-analysis HRs
Increase Number
of
cohorts
HR (95% CI) Measures of heterogeneity
between cohorts
(model 3)
Model 1* Model 2† Model 3‡
I
2
p value
PM
10
10 µg/m³ 14 1·32 (1·12–1·55) 1·21 (1·03–1·43) 1·22 (1·03–1·45) 0·0% 0·83
PM
2·5
5 µg/m³ 14 1·34 (1·09–1·65) 1·17 (0·95–1·45) 1·18 (0·96–1·46) 0·0% 0·92
PM
coarse
5 µg/m³ 14 1·19 (0·99–1·42) 1·08 (0·89–1·31) 1·09 (0·88–1·33) 33·8% 0·11
PM
2·5absorbance
10
–5
/m 14 1·25 (1·05–1·50) 1·09 (0·87–1·37) 1·12 (0·88–1·42) 19·0% 0·25
NO
2
10 µg/m³ 17 1·07 (1·00–1·14) 0·99 (0·93–1·06) 0·99 (0·93–1·06) 0·0% 0·70
NOx 20 µg/m³ 17 1·08 (1·02–1·14) 1·01 (0·95–1·06) 1·01 (0·95–1·07) 0·0% 0·62
Traffi c density on nearest road 5000 vehicles
per day
15 1·02 (0·98–1·06) 1·00 (0·97–1·04) 1·00 (0·97–1·04) 0·0% 0·90
Traffi c load on major roads
within 100 m
4000 vehicle-km
per day
16 1·10 (1·00–1·21) 1·07 (0·97–1·18) 1·09 (0·99–1·21) 0·0% 0·92
We included only participants without missing data in any of the variables included in model 3, so the datasets were identical for analyses with all three models.
See appendix (p 25) for numbers of participants and lung cancer cases contributing to each meta-analysis result. HR=hazard ratio. PM
10
=particulate matter with
diameter <10 µm. PM
2·5
=particulate matter with diameter <2·5 µm.
PM
coarse
=particulate matter with diameter 2·5–10 µm. PM
2·5absorbance
=soot. NO
2
=nitrogen dioxide.
NOx=nitrogen oxides (the sum of nitric oxide and nitrogen dioxide). *Model 1: age (timescale in Cox model), sex, calendar time. †Model 2: model 1 + smoking
status, smoking intensity, square of smoking intensity, smoking duration, time since quitting smoking, environmental tobacco smoke, occupation, fruit intake,
marital status, education level, and employment status. ‡Model 3: model 2 + area-level socioeconomic status.
Table 2: Meta-analyses of associations between air pollutants and traffi c indicators and the risk for lung cancer

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "Air pollution and lung cancer incidence in 17 european cohorts: prospective analyses from the european study of cohorts for air pollution effects (escape)" ?

Raaschou-Nielsen, Zorana J Andersen, Rob Beelen, Evangelia Samoli, Massimo Stafoggia, Gudrun Weinmayr, Barbara Hoff mann, Paul Fischer, Mark J Nieuwenhuijsen, Bert Brunekreef, Wei W Xun, Klea Katsouyanni, Konstantina Dimakopoulou, Johan Sommar, Bertil Forsberg, Lars Modig, Anna Oudin, Bente Oftedal, Per E Schwarze, Per Naf 

Q2. What are the strengths of this study?

The strengths of their study include the use of 17 cohort studies in several locations in Europe with very diff erent air pollution exposure levels and also the use of standardised protocols for exposure assessment and data analysis. 

Q3. What did the authors use to estimate the exposure of air pollutants?

The authors used data on air pollution for 2008–11 in the development of their land-use regression models but applied them to addresses of participants at baseline (mainly 10–15 years earlier). 

Q4. How can the overall risk estimate be generalised to all European populations?

How widely the overall risk estimates from this meta-analysis can be generalised to all European populations is uncertain, but the absence of signifi cant heterogeneity among the HRs obtained for the single cohorts suggests that the overall estimate can be generalised. 

Q5. What was the effect of urbanisation on the HR for PM10?

Adjustment for extent of urbanisation, which could be done in seven cohorts, led to a small change in the HR for PM10, which was, however, due almost entirely to selection of contributing cohorts and not to adjustment for urbanisation per se (appendix p 24). 

Q6. What was the effect of the adjustment on the pulmonary cancer risk?

The results from model 1, with adjustment only for age, sex, and calendar time, showed stronger associations; the eff ect of adjustment was due mainly to the smoking variables. 

Q7. What was the effect of the relocation of the study?

In their study, exposure was assessed at the enrolment address; relocation during follow-up might have led to misclassifi cation of the exposure relevant to later development of lung cancer.