ARTICLE IN PRESS

Environmental Research 109 (2009) 651–656

Contents lists available at ScienceDirect

Environmental Research

0013-93

doi:10.1

� Corr

E-m

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

Validity of a traffic air pollutant dispersion model to assess exposureto fine particles

Aude Kostrzewa �, Patrice Reungoat, Chantal Raherison

Laboratoire Sante Travail Environnement (LSTE), EA 3672, ISPED-Universite Victor Segalen Bordeaux2, case 11, 146 rue Leo Saignat, 33076 Bordeaux cedex, France

a r t i c l e i n f o

Article history:

Received 3 July 2008

Received in revised form

19 May 2009

Accepted 28 May 2009

Keywords:

Traffic air pollution

Fine particles

Dispersion model

ExTra index

Validation

51/$ - see front matter & 2009 Elsevier Inc. A

016/j.envres.2009.05.007

esponding author. Fax: +33 557 571698.

ail address: Aude.Kostrzewa@isped.u-bordeau

a b s t r a c t

Introduction: Fine particles (PM2.5) are an important component of air pollution. Epidemiological

studies have shown health effects due to ambient air particles, particularly allergies in children. Since

the main difficulty is to determine exposure to such pollution, traffic air pollutant (TAP) dispersions

models have been developed to improve the estimation of individual exposure levels. One such model,

the ExTra index, has been validated for nitrogen oxide concentrations but not for other pollutants. The

purpose of this study was to assess the validity of the ExTra index to assess PM2.5 exposure.

Methods: We compared PM2.5 concentrations calculated by the ExTra index to reference measures

(passive samplers situated under the covered part of the playground), in 15 schools in Bordeaux, in

2000. First, we collected the input data required by the ExTra index: background and local pollution

depending on traffic, meteorology and topography. Second, the ExTra index was calculated for each

school. Statistical analysis consisted of a graphic description; then, we calculated an intraclass

correlation coefficient.

Results: Concentrations calculated with the ExTra index and the reference method were similar. The

ExTra index underestimated exposure by 2.2mg m�3 on average compared to the reference method. The

intraclass correlation coefficient was 0.85 and its 95% confidence interval was [0.62; 0.95].

Conclusions: The results suggest that the ExTra index provides an assessment of PM2.5 exposure similar

to that of the reference method. Although caution is required in interpreting these results owing to the

small number of sites, the ExTra index could be a useful epidemiological tool for reconstructing

individual exposure, an important challenge in epidemiology.

& 2009 Elsevier Inc. All rights reserved.

1. Introduction

Air pollution consists of numerous substances. In recent years,air pollution due to industrial or gas equipment has decreased whilefine and ultrafine particle pollution, principally due to traffic, hasincreased (Ebelt et al., 2001). Experimental studies have shown thatparticles may induce oxidative stress and inflammatory reactions,which could explain the activation and exacerbation of respiratoryeffects in sensitive populations (Seaton et al., 1995; Donaldson andMacNee, 2001; Sydbom et al., 2001).

Human exposure analysis can be measured by direct orindirect methods (Lioy, 1995). The direct methods consistof personal monitoring or biological markers. They are time-consuming and expensive for epidemiological studies, and cannotestimate air pollution due to transport-related pollution. Indirectmeasures consist of collective monitoring, questionnaires, diariesor models. The use of background stations for measuring pollutant

ll rights reserved.

x2.fr (A. Kostrzewa).

concentrations is a form of collective monitoring which cannotestimate individual exposure or air pollution due to transport-related pollution. As for questionnaires and diaries, they lackprecision and cannot estimate air pollution due to transport-related pollution. Models have therefore been developed toimprove the estimation of exposure levels (Hertel and Berkowicz,1989). They have rarely been used in epidemiology to study theimpact of traffic air pollution (TAP) on children’s respiratorymorbidity (Pershagen et al., 1995; Pikhart et al., 2000; Braueret al., 2002; Gehring et al., 2002; Nicolai et al., 2003) and haveonly rarely been used to estimate ambient concentrations ofparticulate air pollution (Brauer et al., 2002; Gehring et al., 2002;Nicolai et al., 2003).

The ExTra index is a TAP dispersion model (Sacre et al., 1995)produced by the French Scientific Centre for Building Physics(CSTB) and the French National Institute for Transport and SafetyResearch (INRETS). The ExTra index estimates ambient concentra-tions of transport-related pollutants in front of worksites andplaces of urban dwelling in combination with time-activitydiaries. It has been validated for nitrogen oxide (NOx) concentra-tions (Reungoat et al., 2003) but not for other pollutants.

ARTICLE IN PRESS

A. Kostrzewa et al. / Environmental Research 109 (2009) 651–656652

The purpose of this study was to assess the validity of theExTra index in order to assess PM2.5 exposure.

2. Methods

2.1. Study design

Validation was based on comparing PM2.5 concentrations obtained from

calculations using the ExTra index versus those obtained by reference measure-

ments performed in the study International Study of Asthma and Allergies in

Childhood, part II (ISAAC II) in Bordeaux in 2000.

2.2. Context of study

2.2.1. Reference measurements: ISAAC II study

The ISAAC study (Weiland et al., 2004) is an international multicenter cross-

sectional study investigating the distribution and risk factors of allergic diseases.

In phase II (ISAAC II), standardized health data were collected among children

in about 50 countries by a parental questionnaire and a medical examination.

In France, the ISAAC study, named the Six Cities Study, has been realized on

109 schools distributed on six cities (Penard-Morand et al., 2005; Raherison et al.,

2007). Measurement of PM2.5 concentrations obtained the week of the clinical

examination in Bordeaux served as a surrogate for the children’s usual exposure.

The concentrations of PM2.5 were measured from Monday to Friday in classrooms,

schoolyards and air pollution monitoring stations, usually the week during which

the children underwent examination. These measurements were reproducible

(data not shown). PM2.5 were collected on a glass fibre filter coated with Teflon

using a continuous-flow pump (4 l/min) and a KTL cyclone (BGI Inc) allowing

particulates o2.5mm to be captured. Filters were weighed on electronic scales

with a precision of 1mg in a room at constant temperature (20 1C) and relative

humidity (50%) by the Laboratoire d’Hygiene de la Ville de Paris (Paris, France). Our

air assessments were consistent with measurements provided by the national

monitoring stations for air quality control (data not shown).

2.2.2. Tool to validate: the ExTra index

The ExTra index assesses individual exposure to TAP. It can be calculated for

most primary pollutants (carbon monoxide, hydrocarbons, polycyclic aromatic

hydrocarbons and particles plus nitrogen oxide) and it reflects ambient

concentrations of TAP, which constitutes a complex mixture of pollutants. It can

classify a subject’s exposure according to sites (house(s), and worksite(s) for the

adults and house(s), school(s) and daycare centre(s) for the children). For each site,

the ExTra index has two components: regional pollution and local pollution.

Regional pollution is the background component and is always taken into account

by the ExTra index. Local pollution principally depends on traffic and concerns the

site’s configuration. When the site is on the street, the ExTra index takes into

account both background and local pollution. When the site is not on the street,

only background pollution is considered. Two pollutant dispersion models are

PM25 = -3.4874 +0.8793PM10

0

10

20

30

40

50

60

70

80

90

0

PM

25

periode

PM1010 20 30 40 50 6

Fig. 1. Regression model for winter pe

used to calculate the ExTra index. The Danish model Operation Street Pollution

Model (OSPM) is used for road canyons bordered on one or both sides by a row of

buildings or detached houses (Hertel and Berkowicz, 1989), whereas the California

Line Source Dispersion Model (CALINE 3 model) is used for open spaces near roads

and highways (Benson, 1992). Both models account for the three components of

local pollution: traffic conditions, meteorology and topography. The ExTra index

takes into account both emissions (background pollution and traffic) and local

dispersion conditions (meteorology and topography).

2.3. Sites and period of study

In the ISAAC II study, the sites were obtained from a random selection of all

primary schools in Bordeaux. PM2.5 reference concentrations were measured from

10th January to 24th March 2000, during five days for each school. PM2.5

concentrations were calculated with the ExTra index for the same schools and

periods.

2.4. Data collection

2.4.1. Measurement of PM2.5 concentrations with reference method

In the ISAAC II study, the passive samplers were situated under the covered

part of the playground. All measures were standardized.

2.4.2. Calculation of PM2.5 concentrations by the ExTra index

The input data required to calculate the ExTra index were background

pollution data for all the schools and local pollution data for schools situated on

the street.

�

= 2

0

riod

Three stages were necessary to estimate the PM2.5 background pollution at

each school

o Assessment of maximal background level pollution in Bordeaux centre

The data required were the daily average PM2.5 concentrations measured

by a fixed-site area monitoring station in Bordeaux centre between January

and March 2000. These data were not available for the period of the study.

After consulting experts of the ‘‘Laboratoire d’Hygiene de la Ville de Paris’’

and Aquitaine air quality network (AIRAQ) (expert of local station which

measure pollutants), we have estimated PM2.5 concentrations with a

regression model: the dependant variable was PM2.5 concentration and the

independent variable was PM10 concentration. Data were provided by the

AIRAQ. Two models were built: one for summer period (1st April to

30th September) and one for winter period (1st October to 31st march)

with data measured between July 27th 2001 and July 17th 2004. For

winter period, the model was: PM2.5 ¼ 0.88�PM10�3.49. Determination

coefficient (R2) was 0.93 (Fig. 1). Normality and homoscedasticity were

confirmed. These results were similar with the results found by Park

(Park and Ha, 2008).

o Estimation of decreasing radius for background concentrations

We have defined the centre of the city of Bordeaux, where background

pollution is maximal and the decreasing radius for background concentrations.

7

(1s

N510 Rsq 0.9380AdjRsq0.9379RMSE 2.7677

0 80 90 100 110

t October to 31st March).

ARTICLE IN PRESS

TabPer

Sch

cod

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

MD

A. Kostrzewa et al. / Environmental Research 109 (2009) 651–656 653

The radius depends on the size of the town studied (Eerhens et al., 1993).

Decreasing radius was approximated by AIRAQ, by analogy with a similar

city (Zmirou et al., 2002).

Collection of each school’s coordinates compared to Bordeaux centre.

Each school’s coordinates in relation to Bordeaux centre (Bordeaux City

Hall) were collected to apply the decreasing radius in order to define the

geographic positioning.

le 1iods

ool

e

: m

�

30

35

40

Three stages were necessary to estimate the PM2.5 local pollution

o Traffic

Emissions depend on the type and intensity of traffic. Type of traffic was

defined automatically by the ExTra index according to a study performed

by the INRETS (Lacour and Joumard, 2002). The average daily traffic

density in the street nearest to the school was counted by the municipal

highway services in Bordeaux in 2000. In the absence of numerical data,

the software used a density of 100 vehicles per hour (v/h) (light traffic),

750 v/h (medium traffic), or 1500 v/h (heavy traffic).

o Meteorology

The weather conditions required for the calculations consisted of a wind

frequency table containing 20 speeds and 18 directions during the study

period (between 10th January and 24th March 2000) obtained from the

meteorological agency (Meteo France) in Bordeaux.

o Topography

Topographical data were required for home streets and for streets (i)

having a greater traffic density than the home street and (ii) that crossed

the home street up to a distance of 50 m on either side of the residence.

First, topographical data were determined from a map of Bordeaux. These

data consisted of the direction of the street defined in relation to north and

the left and right sides of the street, the longitudinal axis of the street

always being considered as pointing east. Topographical data were also

determined by visiting each school, where the following parameters were

assessed: (i) the width of the roadway and the sidewalks, (ii) the height of

buildings on the right and left of the street and (iii) the height of the

covered part of the playground. Heights were measured using a clinometer.

10

15

20

25

PM2.

5 E

xTra

(µg

.m-3

)

2.5. Statistical analysis

To assess the criterion validity of the ExTra index to assess PM2.5 exposure, we

first described the data graphically(Bland and Altman, 1986), then calculated the

agreement between the two series of PM2.5 concentrations with the intraclass

correlation coefficient (ICC) and its 95% confidence interval (95% CI). Since there

are numerous versions of the ICC, we chose the ICC that was the most appropriate

for our situation, i.e. the ICC which is used when each target is rated by the same

judges (or methods in our situation) who are assumed to be the only judges of

interest (Shrout and Fleiss, 1979).

0

5

0

PM2.5 reference (µg.m-3)

5 10 15 20 25 30 35 40

Fig. 2. Measured and calculated PM2.5 concentrations with the line of equality.

3. Results

In Bordeaux, 16 schools were obtained from a random selectionon the 55 schools of Bordeaux according to ISAAC protocol. 15/16sets of data were available for the reference method and for the

of the study and measured and calculated PM2.5 concentrations at each school

Period of the study PM2.5 concentration measured by the reference

method (mg m�3)

10/01/2000–14/01/2000 30.1

10/01/2000–14/01/2000 15.1

17/01/2000–21/01/2000 21.7

17/01/2000–21/01/2000 22.5

24/01/2000–28/01/2000 11.3

31/01/2000–04/02/2000 12.9

31/01/2000–04/02/2000 20.1

21/02/2000–25/02/2000 21.1

21/02/2000–25/02/2000 35.3

28/02/2000–03/03/2000 MD

28/02/2000–03/03/2000 37.9

06/03/2000–10/03/2000 31.3

06/03/2000–10/03/2000 16.0

13/03/2000–17/03/2000 38.2

20/03/2000–24/03/2000 18.1

20/03/2000–24/03/2000 13.9

issing data.

ExTra index (one of the 16 schools was excluded due to missingdata). Schools were distributed homogeneously throughout thecity of Bordeaux.

During the study period, the average PM2.5 concentrationmeasured with the reference method was 23.1mg m�3 (standarddeviation: 9.2). Minimal and maximal concentrations were,respectively, 11.3 and 38.2mg m�3. Table 1 shows each school’smeasures.

The OSPM model was used for 4/15 schools (codes 5, 6, 10and 13) which were on the street and the CALINE 3 model wasused for one school (code 16). The ExTra index was calculated foreach school (Table 1). The average PM2.5 concentration calculatedby the ExTra index was 20.9mg m�3 (standard deviation: 7.9).Minimal and maximal concentrations were 10.5 and 36.5mg m�3,respectively.

The PM2.5 concentrations calculated by the ExTra index weresimilar to those measured with the reference method (Fig. 2).According to Bland and Altman’s method, the mean difference (d)

, Bordeaux 2000.

PM2.5 concentration calculated by the ExTra

index (mg m�3)

PM2.5 local concentration

(mg m�3)

20.1

15.5

19.0

20.1

10.5 0.1

14.0 0.1

23.0

23.5

36.5

19.0 0.4

36.0

18.5

12.5 0.2

30.5

18.5

15.1 0.9

ARTICLE IN PRESS

A. Kostrzewa et al. / Environmental Research 109 (2009) 651–656654

between the calculated PM2.5 concentrations and the measuredPM2.5 concentrations was �2.2mg m�3 (95% CI: [�4.8; 0.4]). Thestandard deviation (s) was 4.7 and the differences were normallydistributed. Therefore, 95% of differences were situated between–11.3 and 7.0mg m�3 (Fig. 3). The intraclass correlation coefficientwas 0.85 (95% CI: [0.62; 0.95]).

4. Discussion

Series of values obtained with both methods were similar.PM2.5 concentrations calculated by the ExTra index and PM2.5

concentrations measured with the reference method had a greatagreement. The ICC was 0.85 (95% CI: [0.62; 0.95]). The differenceswere large for three schools. These schools were not on the streetand the Index did not use OSPM or CALINE. No clear explanationwas found for the differences observed: they were not on thesame district and were not near an important pollution source(e.g. car park or petrol station).

We chose PM2.5 concentrations to validate the ExTra index fortwo reasons: first, it was unreasonable to apply the results of theNOx validation study to particulate air pollution (Reungoat et al.,2003); second, the health effects of particulate air pollution arebiologically plausible (Seaton et al., 1995; Donaldson and MacNee,2001; Sydbom et al., 2001).

Since pollutant measurements and particularly particle mea-surements are technically complex and difficult to handle, weused existing measurements. However, the ones we used were notinitially collected to validate the ExTra index. Despite the randomselection and homogeneous distribution of the schools, this studyhas some limitations. First, all the schools were located in the cityof Bordeaux, so there was no geographic or climatic diversity.Second, the study concerned only 15 schools so the statisticalpower may be low. Third, the study sites were all schools. Schoolsare situated in particular places (only 5/15 schools were on thestreet) and are often quite far from streets with heavy traffic. Ifthese limitations are taken into account, the ExTra index seems tobe valid for assessing exposure to PM2.5 in particular conditions.

In the ISAAC II study, pollutant measures were standardizedwith an adapted method that may be considered as a referencemethod in 2000 (Council of the European Union, 1999). Moreover,a comparison between building heights measured by a clinometerand estimated from the number of floors led to very similarresults (Reungoat et al., 2003). However, we chose to use objectivedata obtained exclusively by visiting each site. Two dispersionmodels can be used to calculate the ExTra index (OSPM andCALINE 3), which is important to distinguish particular config-urations (open spaces, squares, crossroadsy). A limitation to ourstudy concerns the periods of study and the approximations. The

-15

-10

-5

0

5

10

15

0

Cal

cula

ted

PM2.

5- m

easu

red

PM2.

5

5 10 15 20 2

Fig. 3. Differences between PM2.5 concentrations calculated with the ExTra in

period of validation was four days, which is short: concentrationscalculated by modelling are more sensitive to data imprecisionover short periods (Raaschou-Nielsen et al., 2000). Aboveall, PM2.5 background pollution was estimated with the PM10

concentration in a regression model. Since no quantified datawere available for streets with light traffic, we had to useapproximations given by the ExTra index.

Other studies of model validity have already been performed:two studies concerned one site(Larssen et al., 1993; Walker et al.,1999), two others concerned several sites all located in one town(Nyberg et al., 2000; Carr et al., 2002) and other studies concernedseveral sites and several towns (Briggs et al., 2000; Raaschou-Nielsen et al., 2000; Brauer et al., 2003; Reungoat et al., 2003).All these studies lasted longer than ours (Larssen et al., 1993;Walker et al., 1999; Briggs et al., 2000; Raaschou-Nielsen et al.,2000; Carr et al., 2002; Brauer et al., 2003; Reungoat et al., 2003).The ExTra Index has been validated for NOx (ICC between 0.75and 0.91, depending on the town (Reungoat et al., 2003). Onestudy published the validity of a regression model to assessthe exposure to PM2.5 (Brauer et al., 2003) in the Netherlands,Germany and Sweden; the percentage of the explained variationwas 0.50–0.73. A recent study has compared a stochastic and adispersion modelling for NO2 and PM2..5. For PM2..5 the correlationbetween stochastic and dispersion modelling was 0.79 and bothmodels yield similar results regarding exposure estimation totraffic-related air pollutants (Cyrys et al., 2005)

Validity is only one element of the performance of tools. It isalso important to take reproducibility, applicability and interestinto account when making comparisons with other indices usedpreviously (Salmi, 2002). Reproducibility has not yet been studiedbecause of the time constraints that this would involve. Applic-ability depends on simplicity of use, cost and acceptability.Regarding simplicity of use, the ExTra index might appear tohave shortcomings but the data required to assess it are readilyavailable from different sources. Collecting topographical parameterscan be simplified by using two approaches: self-administeredquestionnaires to estimate heights and cadastral maps to estimatewidths. The costs of the ExTra index depend on the time taken tocollect data per se and the associated financial costs. Most data arefree in France but this is not the case for cadastral maps andmeteorological data.

Most recent epidemiological studies have used more crudeindices of exposure to TAP. Exposure assessment has includedtraffic density or distance from a busy street (Wjst et al., 1993;Brunekreef et al., 1997; Van Vliet et al., 1997). These variousapproaches have many advantages: they are simpler, less time-consuming and less expensive than the performance of calcula-tions with pollution dispersion models. However, they have rarelybeen validated and when they have, the performance of modelled

Measured PM2.5

d − 1.96s = -11.3

d + 1.96s = 7.0

d = -2.2

5 30 35 40 45

dex against PM2.5 concentrations measured with the reference method.

ARTICLE IN PRESS

A. Kostrzewa et al. / Environmental Research 109 (2009) 651–656 655

exposure was higher (Raaschou-Nielsen et al., 2000; Reungoatet al., 2003; Heinrich et al., 2005).

Another dispersion model, STREET, is largely used in France. Itwould be interesting in further work, to compare our results withthis model (Penard-Morand et al., 2006).

Despite the limitations of this study, the ExTra index seems tobe valid for assessing exposure to fine particles. Another studywould be useful to extend these results. In the future, it would beinteresting to use the ExTra index to assess lifelong average TAPconcentrations in children in the ISAAC II study. Topographicaldata would have to be collected using a questionnaire for eachhome, school or daycare center address attended by children, butthe method would be able to estimate individual exposure toTAP and to study robustly the impact of TAP on asthma andallergies on children. Therefore, use of the ExTra index shouldreduce misclassification in TAP exposure and make it possible tocontrol the confounding effect of TAP exposure in studiesanalysing other environmental risks (e.g. electromagnetic fieldsand waste incinerators). Models like the ExTra index go a longway to resolving an important issue in epidemiology: exposureassessment.

Acknowledgments

This study was funded by Fondation Recherche Medicale.The authors would like to thank the following for the data they

provide and their helpful comments: Mr. Le Moullec (Laboratoired’hygiene de la ville de Paris), Mr. Flori (Scientific Centre forBuilding Physics), Mme Campagne (Aquitaine air quality net-work), Mr. Bunales (Aquitaine air quality network), Mr. Jaffre,la Communaute Urbaine de Bordeaux, Meteo France and all theheadmasters of the schools involved.

We are particularly indebted to pupils and parents withoutwhom the study would not have been possible. We wish to thankthe Education Nationale (Dr Delmas (rectorat of Bordeaux),the schools doctors and the school principals. The French SixCities study was supported by the National Institute for Healthand Medical Research (INSERM) (Programme Determinantsde la Sante), the Ministry of Health (DGS), the EnvironmentalProgramme PRIMEQUAL-PREDIT of the Ministry of Environment,the Agency for Environment and Energy Management (ADEME),the French Agency for Environmental and Occupational HealthSafety (AFSSET), the mutual insurance company of the stateeducation system (the Mutuelle Generale de l’Education Nationale(MGEN)) and the French At Home Respiratory Support association(the Association Nationale pour le Traitement A Domicile del’Insuffisance Respiratoire chronique (ANTADIR).

References

Benson, P.E., 1992. A review of the development and application of the CALINE3and CALINE4 models. Atmos. Environ. 26B (3), 379–390.

Bland, J.M., Altman, D.G., 1986. Statistical methods for assessing agreementbetween two methods of clinical measurement. Lancet 1 (8476), 307–310.

Brauer, M., Hoek, G., Meliefste, K., Fischer, P., Gehring, U., et al., 2003. Estimatinglong-term average particulate air pollution concentrations: application oftraffic indicators and geographic information systems. Epidemiology 14 (2),228–239.

Brauer, M., Hoek, G., van Vliet, P., Meliefste, K., Fischer, P.H., Wijga, A., et al., 2002.Air pollution from traffic and the development of respiratory infections andasthmatic and allergic symptoms in children. Am. J. Respir. Crit. Care Med. 166(8), 1092–1098.

Briggs, D.J., de Hoogh, C., Gulliver, J., Wills, J., Elliott, P., Kingham, S., et al., 2000. Aregression-based method for mapping traffic-related air pollution: applicationand testing in four contrasting urban environments. Sci. Total Environ. 253(1–3), 151–167.

Brunekreef, B., Janssen, N.A., de Hartog, J., Harssema, H., Knape, M., van Vliet, P.,1997. Air pollution from truck traffic and lung function in children living nearmotorways. Epidemiology 8 (3), 298–303.

Carr, D., von Ehrenstein, O., Weiland, S., Wagner, C., Wellie, O., Nicolai, T., et al.,2002. Modeling annual benzene, toluene, NO2, and soot concentrations on thebasis of road traffic characteristics. Environ. Res. 90 (2), 111–118.

Council of the European Union. Council Directive 99/30/EC of 22 April 1999,relating to limit values for sulphur dioxide, nitrogen dioxide and oxides ofnitrogen, particulate matter and lead in ambient air: Official Journal L 163, 29/06/1999, p. 41.

Cyrys, J., Hochadel, M., Gehring, U., Hoek, G., Diegmann, V., Brunekreef, B., Heinrich,J., 2005. GIS-based estimation of exposure to particulate matter and NO2 in anurban area: stochastic versus dispersion modeling. Environ. Health Perspect.113 (8), 987–992.

Donaldson, K., MacNee, W., 2001. Potential mechanisms of adverse pulmonary andcardiovascular effects of particulate air pollution (PM10). Int. J. Hyg. Environ.Health 203 (5–6), 411–415.

Ebelt, S., Brauer, M., Cyrys, J., Tuch, T., Kreyling, W.G., Wichmann, H.E., et al., 2001.Air quality in postunification Erfurt, East Germany: associating changes inpollutant concentrations with changes in emissions. Environ. Health Perspect.109 (4), 325–333.

Eerhens, H.C., Sliggers, C.J., Van den Hout, K.D., 1993. The car model:the Dutch method to determine city street air quality. Atmos. Environ. 27B,389–399.

Gehring, U., Cyrys, J., Sedlmeir, G., Brunekreef, B., Bellander, T., Fischer, P., et al.,2002. Traffic-related air pollution and respiratory health during the first 2 yrsof life. Eur. Respir. J. 19 (4), 690–698.

Heinrich, J., Gehring, U., Cyrys, J., Brauer, M., Hoek, G., Fischer, P., et al., 2005.Exposure to traffic related air pollutants: self reported traffic intensity versusGIS modelled exposure. Occup. Environ. Med. 62 (8), 517–523.

Hertel, O., Berkowicz, R., 1989. Modelling pollution from traffic in a street canyon.Evaluation of data and model development. DMU LUFT-A129. DK-4000Roskilde, National Environmental Research Institute.

Lacour, S., Joumard, R., 2002. Parc, usage et emissions des vehicules routiers enFrance de 1970 a 2020. BR-LTE0208. INRETS, [in French]

Larssen, S., Tonnesen, D., Clench-Aas, J., Aarnes, M.J., Arnesen, K., 1993. A model forcar exhaust exposure calculations to investigate health effects of air pollution.Sci. Total Environ. 134 (1–3), 51–60.

Lioy, P.J., 1995. Measurement methods for human exposure analysis. Environ.Health Perspect. 103 (Suppl. 3), 35–43.

Nicolai, T., Carr, D., Weiland, S.K., Duhme, H., von Ehrenstein, O., Wagner, C., et al.,2003. Urban traffic and pollutant exposure related to respiratory outcomes andatopy in a large sample of children. Eur. Respir. J. 21 (6), 956–963.

Nyberg, F., Gustavsson, P., Jarup, L., Bellander, T., Berglind, N., Jakobsson, R., et al.,2000. Urban air pollution and lung cancer in Stockholm. Epidemiology 11 (5),487–495.

Park, D.U., Ha, K.C., 2008. Characteristics of PM10, PM2.5, CO2 and CO monitored ininteriors and platforms of subway train in Seoul, Korea. Environ. Int. 34 (5),629–634.

Penard-Morand, C., Charpin, D., Raherison, C., Kopferschmitt, C., Caillaud, D.,Lavaud, F., et al., 2005. Long-term exposure to background air pollution relatedto respiratory and allergic health in school children. Clin. Exp. Allergy 35 (10),1279–1287.

Penard-Morand, C., Armengaud, A., et al., 2006. Assessment of school children’sexposure to traffic-related air pollution in the French six cities study using adispersion model. Atmos. Environ. 40 (13), 2274–2287.

Pershagen, G., Rylander, E., Norberg, S., Eriksson, M., Nordvall, S.L., 1995. Airpollution involving nitrogen dioxide exposure and wheezing bronchitis inchildren. Int. J. Epidemiol. 24 (6), 1147–1153.

Pikhart, H., Bobak, M., Kriz, B., Danova, J., Celko, M.A., Prikazsky, V., et al., 2000.Outdoor air concentrations of nitrogen dioxide and sulfur dioxide andprevalence of wheezing in school children. Epidemiology 11 (2), 153–160.

Raaschou-Nielsen, O., Hertel, O., Vignati, E., Berkowicz, R., Jensen, S.S., Larsen, V.B.,et al., 2000. An air pollution model for use in epidemiological studies:evaluation with measured levels of nitrogen dioxide and benzene. J. Expo. Anal.Environ. Epidemiol. 10 (1), 4–14.

Raherison, C., Penard-Morand, C., Moreau, D., Caillaud, D., Charpin, D., Kopfersmitt,C., et al., 2007. In utero and childhood exposure to parental tobacco smoke, andallergies in school children. Respir. Med. 101 (1), 107–117.

Reungoat, P., Chiron, M., Gauvin, S., Le Moullec, Y., Momas, I., 2003. Assessment ofexposure to traffic pollution using the ExTra index: study of validation.Environ. Res. 93 (1), 67–78.

Sacre, C., Chiron, M., Flori, J.P., 1995. Elaboration d’un indice d’exposition a lapollution atmospherique d’origine automobile a l’usage des etudes epidemio-logiques. EN-CLI95.9C: CSTB, INRETS, [in French].

Salmi, L.R., 2002. Les criteres specifiques a l’evaluation diagnostique. In: Lecturecritique et communication medicale scientifique Comment lire, presenter,rediger et publier une etude clinique ou epidemiologique (Elsevier), Paris,111–123 [in French].

Seaton, A., MacNee, W., Donaldson, K., Godden, D., 1995. Particulate air pollutionand acute health effects. Lancet 345 (8943), 176–178.

Shrout, P.E., Fleiss, J.L., 1979. Intraclass correlations: uses in assessing raterreliability. Psychol. Bull. 86 (2), 420–428.

Sydbom, A., Blomberg, A., Parnia, S., Stenfors, N., Sandstrom, T., Dahlen, S.E.,2001. Health effects of diesel exhaust emissions. Eur. Respir. J. 17 (4),733–746.

Van Vliet, P., Knape, M., de Hartog, J., Janssen, N., Harssema, H., Brunekreef, B., 1997.Motor vehicle exhaust and chronic respiratory symptoms in children livingnear freeways. Environ. Res. 74 (2), 122–132.

ARTICLE IN PRESS

A. Kostrzewa et al. / Environmental Research 109 (2009) 651–656656

Walker, S.E., Slordal, L.H., Guerreiro, C., Gram, F., Gronskei, K.E., 1999. Air pollutionexposure monitoring and estimation. Part II. Model evaluation and populationexposure. J. Environ. Monit. 1 (4), 321–326.

Weiland, S.K., Bjorksten, B., Brunekreef, B., Cookson, W.O., von Mutius, E., Strachan,D.P., 2004. Phase II of the international study of asthma and allergies inchildhood (ISAAC II): rationale and methods. Eur. Respir. J. 24 (3), 406–412.

Wjst, M., Reitmeir, P., Dold, S., Wulff, A., Nicolai, T., von Loeffelholz-Colberg, E.F.,et al., 1993. Road traffic and adverse effects on respiratory health in children.Br. Med. J. 307 (6904), 596–600.

Zmirou, D., Gauvin, S., Pin, I., Momas, I., Just, J., Sahraoui, F., et al., 2002. Fiveepidemiological studies on transport and asthma: objectives, design anddescriptive results. J. Expo. Anal. Environ. Epidemiol. 12 (3), 186–196.