- Email: [email protected]
A case study of personal exposure to nitrogen dioxide using a new high sensitive diffusive sampler
Science of the Total Environment 366 (2006) 55 – 64 www.elsevier.com/locate/scitotenv
A case study of personal exposure to nitrogen dioxide using a new high sensitive diffusive sampler A. Piechocki-Minguy a,*, H. Plaisance a, C. Schadkowski b, I. Sagnier c, J.Y. Saison c, J.C. Galloo a, R. Guillermo a a
Ecole des Mines de Douai, 941, rue Charles Bourseul, B.P. 838, 59508 DOUAI Cedex, France b APPA, 13 rue Faidherbe 59508 LILLE, France c ATMO NORD PAS DE CALAIS, 5 boulevard de la Liberte´, 59000 LILLE, France Received 13 April 2005; received in revised form 23 July 2005; accepted 4 August 2005 Available online 12 September 2005
Abstract Personal NO2 exposure measurements were achieved during two campaigns in a large northern France city. These campaigns were following an innovating approach based on sequential exposure measurements by diffusive samplers distinguishing four categories of microenvironments (bhomeQ, bother indoor placesQ, btransportQ and boutdoorsQ). The objective of these campaigns was to obtain NO2 personal exposure data in different microenvironments and to examine the determinants of personal exposure to this pollutant. Each campaign comprised two 24-h sampling periods: one during a working day and the second during the weekend. The average total NO2 personal exposure ranged from 17 Ag m 3 for the summer weekend samplings to 38 Ag m 3 for the winter weekday samplings. The highest levels were found in transports and outdoors, the intermediate ones in other indoor places and the lowest in homes. Despite their weak levels, indoor environments contributed for more than 78% to total NO2 personal exposure because of more time spent in these living places. A Multiple Correspondence Analysis (MCA) highlighted the determinants of NO2 personal exposure in the bhomeQ and btransportQ microenvironments. This led to a classification of NO2 personal exposure levels according to different means of transport: from the lowest to the highest exposure levels, train, tramway or underground, bicycle, car or motorcycle. In homes, the rise of NO2 personal exposures is mainly due to the use of gas stoves and gas heating and the absence of automatic airing system. A classification of NO2 personal exposure levels was set up according to the characteristics of homes. An analysis of correlations between the home NO2 personal exposures and outdoor concentrations measured by fixed ambient air monitoring stations showed weak relations suggesting that the data of these stations are poor predictors of NO2 personal exposures in homes. D 2005 Elsevier B.V. All rights reserved. Keywords: Nitrogen dioxide; Sequential measurements; Microenvironments; Personal exposure determinants; Passive sampling; Diffusive sampler
* Corresponding author. Tel.: +33 3 27 71 26 14; fax: +33 3 27 71 29 14. E-mail address: [email protected] (A. Piechocki-Minguy). 0048-9697/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2005.08.009
56
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
1. Introduction Exposure is a key element in the knowledge about health effects of air pollution. For many pollutants, its evaluation remains incomplete and associated with large uncertainties. Exposure to an airborne pollutant may be defined as the contact with the pollutant for a given time period (Sexton and Ryan, 1988; Brunekreef, 1992). Exposure levels can be estimated directly or indirectly. Direct approaches are based on the personal measurements obtained by the use of portable devices. Indirect approaches use various models which reconstruct exposure from existing databases of concentrations measured in living places and from time–activity patterns. Passive sampling is an effective means of measuring personal exposure to nitrogen dioxide (NO2). Nevertheless, a lot of passive samplers like diffusion tubes (Palmes et al., 1976) and badges (Yanagisawa and Nishimura, 1982) require long sampling periods (at least 24 h). As a consequence, the individual measurements achieved with these samplers can only provide an integrated estimation of exposure on the whole sampling time without being able to distinguish the exposure levels in different microenvironments. Recent advancements have led to the development of a new generation of diffusive samplers which present high uptake rates for NO2 (exceeding 50 cm3 min 1). This is the case of Radiello, a radial diffusive sampler developed by Fundazione Salvatore Maugeri and also of an Ecole des Mines de Douai (EMD) sampler recently developed by our own laboratory (Piechocki-Minguy et al., 2003). Using an appropriate method for analysis, it is possible to reduce the sampling time to 1 h. With these diffusive samplers, a monitoring of exposure based on a series of sequential short-time measurements in different microenvironments can be carried out. By this direct approach, the information on personal exposure is more accurate since a distinction is made between the microenvironments in the sampling design. The purpose of this study is to use the EMD sampler to characterize NO2 personal exposure of subjects living in a large French city by a direct approach which separates the four main microenvironment categories (transport, home, other indoor
places and outdoor environments). Personal NO2 levels were evaluated in these four microenvironments. Data were used to assess the contribution of each microenvironment to total exposure and to analyse the environmental determinants of NO2 personal exposure in two living places: home and transport. Home personal NO2 exposures were also compared with the outdoor concentrations measured at fixed ambient air quality monitoring sites.
2. Materials and methods 2.1. Diffusive sampler method A new high uptake rate nitrogen dioxide passive sampler was developed by our laboratory (see Fig. 1). It is composed of a porous cartridge impregnated with the trapping solution and fitted in a cylindrical protective box equipped with caps at its extremities. The large sampling area (cartridge surface) and the two circular openings provide a high uptake rate to this passive sampler. The cartridge is impregnated with a TEA (triethanolamine) aqueous solution to trap NO2 (10% (V/V) TEA and 0.3% (V/V) Brij 35 (polyoxyethylen dodecyl ether)). The extraction is achieved in test tubes by plunging the cartridge into 5 mL of ultra pure water and vortex mixing. The mass of nitrogen dioxide sampled is analysed as the amount of nitrite (NO2) in the extraction solution by ion chromatography. This technique was pre-
Holder
Detachable bar
Caps with circular edges
Porous cartridge Protective box
Clip to maintain the sampler on its holder
Protrusions
Fixed bar
Fig. 1. Cross section of the passive sampler developed in the Ecole des Mines de Douai laboratory.
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
ferred to the commonly used spectrophotometry method (Palmes et al., 1976; Atkins and Quirino, 1992) on account of a better sensitivity (Plaisance et al., 2004). The ion chromatography analytical technique was moreover improved to lower the detection limit by using a hydroxide eluent with a gradient program. The analytical method is described in detail elsewhere (Piechocki-Minguy et al., 2003; Plaisance et al., 2004). The sampler performances were tested in an exposure chamber and in field conditions. It can be used on very short sampling durations. Its detection limit is 11 Ag m 3 for a 1-h measurement. The effects of several environmental factors (temperature, wind speed, concentration and exposure duration) on the sampler response were evaluated. A sampling rate model, which took into account the influencing parameters, was established by applying multiple regression (Marquardt, 1963) to the whole of the laboratory test results. This general model gave two sampling rate estimations, one for indoor environments (0.89 cm3 s 1) and the second for outdoor conditions (1.00 cm3 s 1). The field comparisons between the measurements of 4 h exposed samplers and those achieved by chemiluminescent NOx monitors confirmed the sampling rates estimations found through exposure chamber tests. The sampler measurement uncertainty in the field conditions was evaluated at 28% for nitrogen dioxide concentration levels superior to 40 Ag m 3 applying the ISO 13752 standard (ISO 13752, 1998) to the field test results. The whole of validation tests is shown elsewhere (Piechocki-Minguy et al., 2003). 2.2. Sampling design of personal exposure campaigns The EMD sampler was used for two personal exposure campaigns which took place in greater Lille, a large city in northern France. The first campaign was carried out in winter 2001. It included a small number of participants (13) and its aim was chiefly to assess the feasibility of personal exposure measurements with the sampler. The second one was carried out during the following summer and gathered a larger number of participants (31). Before the campaigns, the selected participants were trained in the use of the EMD sampler (for handling the cartridge and the sampler). The volunteers filled in a questionnaire concerning the localization of their home and
57
workplace, some characteristics of their home (type of heating, type of stove, airing) and some information on their way of life (usual transport, smoking status. . .). Each campaign included two 24-h sampling periods: one during the working days (from Thursday 12:00 to Friday 12:00) and the second during the weekends (from Saturday 12:00 to Sunday 12:00). For each sampling period (weekend sampling and working day sampling), the participants received an EMD sampler and four porous cartridges: one cartridge for each of the following microenvironments: home, other indoor places (including workplace, shops, restaurants . . .), transport (car, bus, train . . . .) and outdoors (including on-foot-journeys). The participants carried the sampler all sampling period long, changed the porous cartridge each time they changed microenvironment and noted the corresponding time on a time–activity-diary. Thus, the personal exposure was divided into four parts corresponding to NO2 personal exposure levels in four microenvironment categories. 2.3. Multiple correspondence analysis The Multiple Correspondence Analysis (MCA) was used to examine the relationships between the NO2 exposure levels and selected qualitative variables characterizing the microenvironments (provided by questionnaire data). MCA allows the analysis of an indicator matrix (X) which is composed of I individuals characterized by K modalities for J variables. The general term of this matrix (X) is xi,j = 1 if the individual i is characterized by the k modality and xi,j = 0 otherwise for i = 1, . . . , I and k = 1, . . . , K. MCA involves the projection of the K-modality cloud in a sub-space of fewer dimensions which is composed of N (with N = K J) mutually orthogonal factorial axes (i.e. non-correlated axes) (Lebart et al., 1993; Saporta, 1990). The factorial axes construction and selection are performed so that the whole data set inertia is maximum, regarding these axes. The director vectors and the eigenvalues of the factorial axes are deduced from the diagonalization of the Burt matrix (namely XVX) (Saporta, 1990). The projections of K modalities on the first principal plan (defined as the two factorial axes of maximum inertia) are analysed and an inves-
58
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
Table 1 Mean times spent by the study participants in each microenvironment for the summer and winter NO2 personal exposure campaigns Season
Winter
Type of day
Working day Weekend Summer Working day Weekend
Microenvironments bHomeQ Other indoor Transport Outdoors (%) places (%) (%) (%) 62
32
5
1
88 64
5 29
3 5
4 2
84
4
4
8
tigation on the relations between the NO2 exposure levels and descriptive variables of microenvironment can be achieved.
3. Results and discussion 3.1. Time–activity and questionnaire data Table 1 shows the percentages of time spent by the volunteers in the four microenvironments during the personal exposure campaigns. These results were obtained thanks to the time–activity-diaries. The participants of our study spent more than 85% of their time in indoor places. The times spent in the microenvironments btransportQ and boutdoorsQ are very weak (respectively less than 5% and 10% of total time). These results are in agreement with other studies (Schwab et al., 1990; Lebret et al., 1995).
The proportion of participants living in a flat is much higher (on average 45%) than the regional average given by 1999 national census (25%). This difference can be explained by the fact that the participants of our study mainly live in town (Table 2). 3.2. NO2 exposure levels in the four microenvironments Tables 3 and 4 summarize the NO2 personal exposures recorded in the four microenvironments for winter and summer campaigns respectively. Boxand-whisker plots of these results are presented in Fig. 2 (winter campaign) and Fig. 3 (summer campaign). A hierarchy in the NO2 exposure levels according to microenvironment type clearly appears. The highest levels are found in transport, and outdoors, the intermediate values in other indoor places and the lowest ones in homes. Note that the WHO guidelines for health protection (200 Ag m 3 on 1 h) was exceeded for one person in the microenvironment btransportQ and one person in bother indoor placesQ. An overall increase of exposure levels is observed in all the microenvironments in winter. This can be explained by two factors: on the one hand, by the use of gas appliances for heating in indoor places, and on the other hand, by the meteorological conditions in winter which are in favour of an outdoor concentration level increase. The exposure levels are found to be higher in bother indoor placesQ than in homes. Indeed, a differ-
240 200
Working day Weekend
160 120 80 40 0 Home
Other indoor places
Transport
Outdoors
Total exposure
Fig. 2. Box-and-whisker plots of the winter personal exposure campaign NO2 measurement results (Ag m3).
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
300 250
59
Working day Workin Weekend
200 150 100 50 0 Home
Other indoor places
Transport
Outdoors
Total exposure
Fig. 3. Box-and-whisker plots of the summer personal exposure campaign NO2 measurement results (Ag m3).
ence exists between the spatial distribution of these two categories of indoor environments: the participant homes are partly located outside the town in low polluted areas, whereas bother indoor placesQ include various building types, mainly located in urban and suburban areas next to major NO2 sources. The weekend and working day levels comparison provides some information on the traffic emission impact on the different microenvironment concentrations. It is known that in cities a traffic emission decrease of about 20% to 50% occurs between working days and weekends (Pont and Fontan, 2001). The btransportQ and boutdoorsQ levels follow this traffic Table 2 Participant home and living mode characteristics gathered by the questionnaire filled in by the participants of the winter and summer campaigns Volunteer home characteristics
Dwelling unit type
Heating type
Cooking type
Smoking status
Individual house Flat Campus room Gas Electric Fuel Gas Electric Gas and electric No smoker Non-smoker subjected to environmental tobacco smoke
Sampling periods Winter (n = 13)
Summer (n = 31)
55% 45% 0% 39% 46% 15% 38% 62% 0% 27% 8%
45% 48% 6% 65% 35% 0% 22% 68% 10% 23% 6%
emission pattern for the two campaigns. On the contrary, an inverse trend can be observed for indoor environments: the weekend exposure levels are higher than or equal to the weekday ones. Consequently, the indoor levels seem to be weakly affected by traffic emissions. For each participant, total NO2 personal exposure was calculated according to the following equation: X Ci;j ti;j Ei ¼
j
X
ti;j
ð1Þ
j
where E i is the total exposure of the participant i (Ag m 3), C i, j is the NO2 concentration at which the participant i is exposed in the microenvironment j (Ag m 3) and t i, j is the time (s) spent in the microenvironment j by the participant i. The total exposure values are shown in Tables 3 and 4 in the last two columns. They are close to home levels on weekends and between the two indoor environment levels for working days. This is mainly related to the long time spent by people in these microenvironments (from 20 to 23 h per day). 38% of the winter campaign participants had a total personal exposure for working days which exceeded the annual limit value of 40 Ag m 3 set for 2010. This was also the case of 30% of this campaign participants during the weekend. Only 3% of the summer campaign volunteers exceeded this limit value for the working days and no exceeding was observed for the summer campaign during the weekend.
60
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
Table 3 Winter personal exposure campaign NO2 measurement results (Ag m 3) (arithmetic means, median, minima and maxima) Microenvironement
Home
Type of day
Working day
Mean Median Minimum Maximum
22 19 11 38
Other indoor places
Transport
Weekend
Working day
Weekend
Working day
Weekend
Working day
Weekend
Working day
Weekend
27 20 10 60
56 55 19 86
92 62 31 229
114 104 71 159
56 60 31 77
115 109 59 191
45 42 11 103
38 35 24 53
29 23 14 61
3.3. Contribution of the different microenvironments to global personal exposure Fig. 4 shows the average contributions of the four investigated microenvironments to total personal exposure for the four sampling days. As can be seen, whatever the sampling day, the indoor places (bhomeQ average contribution added to bother indoor placesQ average contribution) account for more than 78% to total personal exposure. Despite the weak indoor levels, their contribution to personal exposure is the major component of total personal exposure because of the time spent in these indoor environments. The contribution of the other indoor places to global personal exposure for the working days is about 40%. This is close to the contribution of homes. On the contrary the contribution of the other indoor places does not exceed 10% for the weekends. These results highlight the need of accurate data on indoor environments concentrations to assess NO2 personal exposure. 3.4. Determinants of the exposure in the microenvironments btransport Q and bhome Q The determinants of NO2 personal exposure in the microenvironments btransportQ and bhomeQ were examined using a MCA multivariate analysis.
Outdoors
Total exposure
For the transport microenvironment data MCA, the NO2 exposure values were recorded according to the six following modalities: 20–32 Ag m 3 (C1), 33–50 Ag m 3 (C2), 51–68 Ag m 3 (C3), 69–96 Ag m 3 (C4), 97–125 Ag m 3 (C5) and above 125 Ag m 3 (C6). These classes were chosen in order to obtain a compromise between equal numbered and equal interval classes. In the transport analysis, four descriptive variables (transport modes, season, type of day and smoking status) were taken into account and divided into ten modalities as follows: train, tramway or underground, bicycle, car or motorcycle, winter, summer, working day, weekend, smoker and non-smoker. Fig. 5 shows the projection of the modalities on the first factorial plane ( F 1 / F 2) that accounts for 31% of the explained variance. An overview of the results shows that the NO2 personal exposure modalities are related to the transport cluster. Thus, each transport mode can be associated with one or two classes of NO2 exposure, as follows: the journeys by train with the C1 (20–32 Ag m 3) and C2 (33–50 Ag m 3) classes; those by tramway or underground with the C2 (33–50 Ag m 3) and C3 (51–68 Ag m 3) classes, those by bicycle with the C4 (69–96 Ag m 3) class and those by car or motorcycle with the C5 (97–125 Ag m 3) class. In this analysis, the other descriptive variables
Table 4 Summer personal exposure campaign NO2 measurement results (Ag m 3) (arithmetic means, median, minima and maxima) Microenvironment
Home
Type of day
Working day
Mean Median Minimum Maximum
14 11 5 24
Other indoor places
Transport
Outdoors
Total exposure
Weekend
Working day
Weekend
Working day
Weekend
Working day
Weekend
Working day
Weekend
15 13 5 44
29 29 13 79
35 38 10 72
79 64 20 256
49 40 22 99
37 36 9 60
30 26 6 78
22 23 12 32
17 16 8 50
Contributions of the four micro environments to NO2personal exposure (%)
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
Winter campaign
61
Summer campaign
100 90 80 70 60 50 40 30 20 10 0
100 90 80 70 60 50 40 30 20 10 0 Week day
Weekend
Outdoors
Week day
Transport
Weekend
Other indoor places
Home
Fig. 4. Average contributions of the four microenvironments to total NO2 personal exposure for the four sampling days.
(season, type of day and smoking status) do not appear as factors affecting the NO2 personal exposure in transport.
In the same manner, the MCA was applied to homes data by dividing the NO2 exposure values into five modalities: 5–10 Ag m 3 (C1), 11–16 Ag smoker
bicycle
1.33
2nd component - inertia : 15.0%
C4 0.57
winter C3
C5
carmotorcycle
working days tramway-underground -0.20
summer
non smoker
week-end
C2 -0.97
train
C1
-1.73 -1.31
-0.76
-0.20
0.36
0.91
1st component - inertia: 16.0%
Concentration scale (µ g m-3) 20
32
C1
50
C2
68
C3
96
C4
125
C5
259
C6*
*group corresponding to atypic items which were not taken into account for the analysis
Train
Tramway-suburban
Bicycle
Car-motorcycle
Fig. 5. Modalities projection on the first factorial plane ( F 1 / F 2) of the MCA performed on the btransportQ microenvironment data.
62
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
m 3 (C2), 17–23 Ag m3 (C3), 24–37 Ag m 3 (C4) and 38–60 Ag m 3 (C5). Five qualitative variables were considered to describe home microenvironment : smoking status, airing, heating mode, stove type, season and type of day. For this analysis, the variables were divided into fourteen modalities: smoker, nonsmoker, subjected to environmental tobacco smoke, not subjected to environmental tobacco smoke, presence of an automatic airing system, no automatic airing system, electric heating, gas heating, electric stove, gas stove, winter, summer, working day and weekend. The modality projections on the first factorial plane ( F 1 / F 2) are shown in Fig. 6. These two MCA produced factors, F 1 and F 2 , account for 34% to the explained variance. NO2 home personal exposure depends mainly on three qualitative variables: heating mode, stove type, and supply of fresh air. The season also has an effect on home NO2 personal exposures, which is probably related to the impact of
heating, on in winter and off in summer. The C1 (5–10 Ag m 3) and C2 (11–16 Ag m 3) are the exposure classes associated with homes equipped with electric heating, an automatic airing system and an electric stove in summer. The C3 (17–23 Ag m 3) and C4 (24– 37 Ag m 3) classes correspond to persons living in homes equipped with a gas stove, gas heating and with no automatic airing system in winter. The C5 (38–60 Ag m 3) class represents the NO2 personal exposures of smokers in their homes. 3.5. Relationships between the home NO2 personal exposures and the stationary monitoring data The correlations between home NO2 personal exposure levels and outdoor NO2 concentrations recorded at various fixed ambient air monitoring sites were examined in order to evaluate the opportunity of predicting NO2 personal exposures by using C1
0.99 summer
2nd component - inertia: 15.2 %
non smoker 0.20
gas heating
electric cooking
no automatic airing system
C2 working days
C4 C3 -0.58
gas cooking
electric heating
non smoker not daily subjected to tobacco smoke week-end days
automatic airing system
non smoker daily subjected to tobacco smoke
smoker winter
-1.36
C5 -2.14 -0.79
-0.32
0.15
1st
0.63
1.10
component - inertia: 18.8 %
Concentration scale (µg m-3) 5
10
C1 Electric heating
16
C2 Electric cooking Summer
Automatic airing system
23
C3
37
C4 Gas heating
60
C5 Gas cooking winter
No automatic airing system
Fig. 6. Modalities projection on the first factorial plane ( F 1 / F 2) of MCA applied to the bhomeQ microenvironment data.
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
the air quality network data. For that purpose, each home NO2 personal exposure value was paired with the NO2 concentration recorded at the monitoring station considered as brepresentativeQ of the home geographical area. Indeed, a previous study (Perdrix et al., 2001) showed that a large part of greater Lille could be covered by four areas in which NO2 levels were homogeneous and comprised in a F30% interval (representativeness criterion) around the concentration recorded by a selected station. Fig. 7a and b
Home NO2 personal exposures (µg.m-3)
a
y = 0.13 x + 11.39 R2 = 0.09
P = 0.0101
70 60
shows home NO2 personal exposures compared with outdoor concentrations measured by the respective selected fixed stations. Fig. 7a takes into account all the two campaign participants and Fig. 7b concerns only the summer participants who live in electric stove and electric heating equipped homes. The correlation is not statistically significant when participants are not selected. The correlation increases when no dominant source is present but it remains weak. Nevertheless, this relation can constitute a basic component to set a NO2 personal exposure estimation model which must take into account the household explicative variables. Other studies (Alm et al., 1998; Quackenboss et al., 1982) have shown that ambient air quality monitoring network data are generally poor predictors of NO2 personal exposure.
50 40
4. Conclusion
30 20 10 0 0
20
40
60
80
100
120
NO2 concentration measured at the representative fixed station (µg.m-3)
b Home NO2 personal exposures (µg.m-3)
63
y = 0.86 x – 18.38 R2 = 0.61
P = 0.0001
70 60 50 40 30 20 10 0 0
10
20
30
40
50
NO2 concentration measured at the representative fixed station (µg.m-3)
Fig. 7. a: Relationships between the home NO2 personal exposures and the outdoor concentrations recorded at fixed ambient air monitoring stations for all the participants. b: Relationships between the home NO2 personal exposures and the outdoor concentrations recorded at fixed ambient air monitoring stations for the summer participants with no major indoor NO2 sources.
This study was designed to investigate the NO2 personal exposure of 45 volunteers living in Lille, a large city in northern France. An innovating approach based on sequential measurements of personal exposure in four microenvironments (home, transport, other indoor places and outdoors) could be carried out using diffusive sampler having a high uptake rate. The highest levels were found in transport and outdoors, the intermediate ones in other indoor places and the lowest in homes. Nevertheless, the indoor pollution remains the main contributor to total NO2 personal exposure (superior to 78%) because the participants spent the main part of their time in indoor environments (from 20 to 23 h per day). The average total NO2 personal exposure ranges from 17 Ag m 3 for the summer weekends to 38 Ag m 3 for the winter working days. As for daily journeys, NO2 personal exposure is mainly determined by the transport mode. The highest levels, which exceeded 100 Ag m 3 were recorded for journeys by car or motorcycle. High home NO2 personal exposures are for the most part explained by gas stove, gas heating and the absence of automatic airing system. Outdoor levels appear as minor factors to explain home NO2 personal exposure. Nevertheless, btransportQ and boutdoorsQ are the microenvironments to consider in health risk assessment because they include various situations of high NO2 exposure levels occurring on short-time periods.
64
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
Some further studies should be made in order to improve the knowledge on the contribution of the main determinants to home NO2 personal exposure. Acknowledgment We sincerely would like to thank our volunteers for their kind participation to this study. References Alm S, Mukala K, Pasanen P, Tiitanen P, Ruuskanen J, Tuomisto J, et al. Personal NO2 exposures of preschool children in Helsinki. J Expo Anal Environ Epidemiol 1998;8:79 – 100. Atkins DHF, Quirino I. A survey of nitrogen dioxide in Paris. European Community Commission Report EUR 13369 EN; 1992. Brunekreef B. The measurement of exposure to air pollution: concepts, methods and stakes. In nume´ro spe´cial de Pollution Atmosphe´rique: 40e`me anniversaire de l’APPA: la pollution atmosphe´rique et l’homme : e´valuer les expositions; 1992;17–29. ISO 13752. Assessment of uncertainty of a measurement method under field conditions using a second method as a reference; 1998. 19 pp. Lebart L, Morineau A, Warwick KM. Multivariate descriptive statistical analysis: correspondence analysis and related techniques. New York7 Wiley Editors; 1993. 231 pp. Lebret E, Fischer P, DeHollander G. Time–activity patterns at population level: regional population and subpopulation variations and their assessment. Air pollution epidemiology report n86: time–activity patterns in exposure assessment. Luxemburg7 Office for Official Publications of the European Community; 1995. p. 9 – 15.
Marquardt DW. An algorithm for least-squares estimation of nonlinear parameters. J Soc Ind Appl Math 1963;11:431 – 41. Palmes ED, Gunnison AF, Dimattio J, Tomczyk C. Personal sampler for nitrogen dioxide. Am Ind Hyg Assoc J 1976;37:570 – 7. Perdrix E, Fourche´ B, Plaisance H. Repre´sentativite´ spatiale des stations urbaines. Rapport du Laboratoire Central de la Qualite´ de l’Air, convention n843 / 2000; 2001. 54 pp. Piechocki-Minguy A, Plaisance H, Garcia-Fouque´ S, Galloo JC, Guillermo R. Validation tests of a new high passive sampler for nitrogen dioxide measurements. Environ Technol 2003;24: 1527 – 35. Plaisance H, Piechocki-Minguy A, Garcia-Fouque´ S, Galloo JC. Influence of meteorological factors on the NO2 measurements by passive diffusion tube. Atmos Environ 2004;38:573 – 80. Pont V, Fontan P. Comparison between weekend and weekday ozone concentration in large cities in France. Atmos Environ 2001;35:1527 – 35. Quackenboss JJ, Kanarek MS, Spengler MS, Letz JD. Personal monitoring for nitrogen dioxide exposure: methodological considerations for a community study. Environ Int 1982;8:249 – 58. Saporta G. Probabilite´, analyse des donne´es et statistique. Paris7 Technip Editors; 1990. 493 pp. Schwab M, Colome SD, Spengler JD, Ryan PB, Billick IH. Activity patterns applied to pollutant exposure assessment: data from personal monitoring study in Los Angeles. Toxicol Ind Health 1990;6:517 – 32. Sexton K, Ryan PB. Assessment of human exposure to air pollution: methods, measurements and models. In: National Academy Press, editor. Air pollution, the automobile, and public health. Washington: Health Effects Institute, National Academy Press; 1988, p. 207–38. Yanagisawa Y, Nishimura HA. Badge-type personal sampler for measurement of personal exposure to NO2 in ambient air. Environ Int 1982;8:1 – 13.
A case study of personal exposure to nitrogen dioxide using a new high sensitive diffusive sampler A. Piechocki-Minguy a,*, H. Plaisance a, C. Schadkowski b, I. Sagnier c, J.Y. Saison c, J.C. Galloo a, R. Guillermo a a
Ecole des Mines de Douai, 941, rue Charles Bourseul, B.P. 838, 59508 DOUAI Cedex, France b APPA, 13 rue Faidherbe 59508 LILLE, France c ATMO NORD PAS DE CALAIS, 5 boulevard de la Liberte´, 59000 LILLE, France Received 13 April 2005; received in revised form 23 July 2005; accepted 4 August 2005 Available online 12 September 2005
Abstract Personal NO2 exposure measurements were achieved during two campaigns in a large northern France city. These campaigns were following an innovating approach based on sequential exposure measurements by diffusive samplers distinguishing four categories of microenvironments (bhomeQ, bother indoor placesQ, btransportQ and boutdoorsQ). The objective of these campaigns was to obtain NO2 personal exposure data in different microenvironments and to examine the determinants of personal exposure to this pollutant. Each campaign comprised two 24-h sampling periods: one during a working day and the second during the weekend. The average total NO2 personal exposure ranged from 17 Ag m 3 for the summer weekend samplings to 38 Ag m 3 for the winter weekday samplings. The highest levels were found in transports and outdoors, the intermediate ones in other indoor places and the lowest in homes. Despite their weak levels, indoor environments contributed for more than 78% to total NO2 personal exposure because of more time spent in these living places. A Multiple Correspondence Analysis (MCA) highlighted the determinants of NO2 personal exposure in the bhomeQ and btransportQ microenvironments. This led to a classification of NO2 personal exposure levels according to different means of transport: from the lowest to the highest exposure levels, train, tramway or underground, bicycle, car or motorcycle. In homes, the rise of NO2 personal exposures is mainly due to the use of gas stoves and gas heating and the absence of automatic airing system. A classification of NO2 personal exposure levels was set up according to the characteristics of homes. An analysis of correlations between the home NO2 personal exposures and outdoor concentrations measured by fixed ambient air monitoring stations showed weak relations suggesting that the data of these stations are poor predictors of NO2 personal exposures in homes. D 2005 Elsevier B.V. All rights reserved. Keywords: Nitrogen dioxide; Sequential measurements; Microenvironments; Personal exposure determinants; Passive sampling; Diffusive sampler
* Corresponding author. Tel.: +33 3 27 71 26 14; fax: +33 3 27 71 29 14. E-mail address: [email protected] (A. Piechocki-Minguy). 0048-9697/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2005.08.009
56
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
1. Introduction Exposure is a key element in the knowledge about health effects of air pollution. For many pollutants, its evaluation remains incomplete and associated with large uncertainties. Exposure to an airborne pollutant may be defined as the contact with the pollutant for a given time period (Sexton and Ryan, 1988; Brunekreef, 1992). Exposure levels can be estimated directly or indirectly. Direct approaches are based on the personal measurements obtained by the use of portable devices. Indirect approaches use various models which reconstruct exposure from existing databases of concentrations measured in living places and from time–activity patterns. Passive sampling is an effective means of measuring personal exposure to nitrogen dioxide (NO2). Nevertheless, a lot of passive samplers like diffusion tubes (Palmes et al., 1976) and badges (Yanagisawa and Nishimura, 1982) require long sampling periods (at least 24 h). As a consequence, the individual measurements achieved with these samplers can only provide an integrated estimation of exposure on the whole sampling time without being able to distinguish the exposure levels in different microenvironments. Recent advancements have led to the development of a new generation of diffusive samplers which present high uptake rates for NO2 (exceeding 50 cm3 min 1). This is the case of Radiello, a radial diffusive sampler developed by Fundazione Salvatore Maugeri and also of an Ecole des Mines de Douai (EMD) sampler recently developed by our own laboratory (Piechocki-Minguy et al., 2003). Using an appropriate method for analysis, it is possible to reduce the sampling time to 1 h. With these diffusive samplers, a monitoring of exposure based on a series of sequential short-time measurements in different microenvironments can be carried out. By this direct approach, the information on personal exposure is more accurate since a distinction is made between the microenvironments in the sampling design. The purpose of this study is to use the EMD sampler to characterize NO2 personal exposure of subjects living in a large French city by a direct approach which separates the four main microenvironment categories (transport, home, other indoor
places and outdoor environments). Personal NO2 levels were evaluated in these four microenvironments. Data were used to assess the contribution of each microenvironment to total exposure and to analyse the environmental determinants of NO2 personal exposure in two living places: home and transport. Home personal NO2 exposures were also compared with the outdoor concentrations measured at fixed ambient air quality monitoring sites.
2. Materials and methods 2.1. Diffusive sampler method A new high uptake rate nitrogen dioxide passive sampler was developed by our laboratory (see Fig. 1). It is composed of a porous cartridge impregnated with the trapping solution and fitted in a cylindrical protective box equipped with caps at its extremities. The large sampling area (cartridge surface) and the two circular openings provide a high uptake rate to this passive sampler. The cartridge is impregnated with a TEA (triethanolamine) aqueous solution to trap NO2 (10% (V/V) TEA and 0.3% (V/V) Brij 35 (polyoxyethylen dodecyl ether)). The extraction is achieved in test tubes by plunging the cartridge into 5 mL of ultra pure water and vortex mixing. The mass of nitrogen dioxide sampled is analysed as the amount of nitrite (NO2) in the extraction solution by ion chromatography. This technique was pre-
Holder
Detachable bar
Caps with circular edges
Porous cartridge Protective box
Clip to maintain the sampler on its holder
Protrusions
Fixed bar
Fig. 1. Cross section of the passive sampler developed in the Ecole des Mines de Douai laboratory.
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
ferred to the commonly used spectrophotometry method (Palmes et al., 1976; Atkins and Quirino, 1992) on account of a better sensitivity (Plaisance et al., 2004). The ion chromatography analytical technique was moreover improved to lower the detection limit by using a hydroxide eluent with a gradient program. The analytical method is described in detail elsewhere (Piechocki-Minguy et al., 2003; Plaisance et al., 2004). The sampler performances were tested in an exposure chamber and in field conditions. It can be used on very short sampling durations. Its detection limit is 11 Ag m 3 for a 1-h measurement. The effects of several environmental factors (temperature, wind speed, concentration and exposure duration) on the sampler response were evaluated. A sampling rate model, which took into account the influencing parameters, was established by applying multiple regression (Marquardt, 1963) to the whole of the laboratory test results. This general model gave two sampling rate estimations, one for indoor environments (0.89 cm3 s 1) and the second for outdoor conditions (1.00 cm3 s 1). The field comparisons between the measurements of 4 h exposed samplers and those achieved by chemiluminescent NOx monitors confirmed the sampling rates estimations found through exposure chamber tests. The sampler measurement uncertainty in the field conditions was evaluated at 28% for nitrogen dioxide concentration levels superior to 40 Ag m 3 applying the ISO 13752 standard (ISO 13752, 1998) to the field test results. The whole of validation tests is shown elsewhere (Piechocki-Minguy et al., 2003). 2.2. Sampling design of personal exposure campaigns The EMD sampler was used for two personal exposure campaigns which took place in greater Lille, a large city in northern France. The first campaign was carried out in winter 2001. It included a small number of participants (13) and its aim was chiefly to assess the feasibility of personal exposure measurements with the sampler. The second one was carried out during the following summer and gathered a larger number of participants (31). Before the campaigns, the selected participants were trained in the use of the EMD sampler (for handling the cartridge and the sampler). The volunteers filled in a questionnaire concerning the localization of their home and
57
workplace, some characteristics of their home (type of heating, type of stove, airing) and some information on their way of life (usual transport, smoking status. . .). Each campaign included two 24-h sampling periods: one during the working days (from Thursday 12:00 to Friday 12:00) and the second during the weekends (from Saturday 12:00 to Sunday 12:00). For each sampling period (weekend sampling and working day sampling), the participants received an EMD sampler and four porous cartridges: one cartridge for each of the following microenvironments: home, other indoor places (including workplace, shops, restaurants . . .), transport (car, bus, train . . . .) and outdoors (including on-foot-journeys). The participants carried the sampler all sampling period long, changed the porous cartridge each time they changed microenvironment and noted the corresponding time on a time–activity-diary. Thus, the personal exposure was divided into four parts corresponding to NO2 personal exposure levels in four microenvironment categories. 2.3. Multiple correspondence analysis The Multiple Correspondence Analysis (MCA) was used to examine the relationships between the NO2 exposure levels and selected qualitative variables characterizing the microenvironments (provided by questionnaire data). MCA allows the analysis of an indicator matrix (X) which is composed of I individuals characterized by K modalities for J variables. The general term of this matrix (X) is xi,j = 1 if the individual i is characterized by the k modality and xi,j = 0 otherwise for i = 1, . . . , I and k = 1, . . . , K. MCA involves the projection of the K-modality cloud in a sub-space of fewer dimensions which is composed of N (with N = K J) mutually orthogonal factorial axes (i.e. non-correlated axes) (Lebart et al., 1993; Saporta, 1990). The factorial axes construction and selection are performed so that the whole data set inertia is maximum, regarding these axes. The director vectors and the eigenvalues of the factorial axes are deduced from the diagonalization of the Burt matrix (namely XVX) (Saporta, 1990). The projections of K modalities on the first principal plan (defined as the two factorial axes of maximum inertia) are analysed and an inves-
58
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
Table 1 Mean times spent by the study participants in each microenvironment for the summer and winter NO2 personal exposure campaigns Season
Winter
Type of day
Working day Weekend Summer Working day Weekend
Microenvironments bHomeQ Other indoor Transport Outdoors (%) places (%) (%) (%) 62
32
5
1
88 64
5 29
3 5
4 2
84
4
4
8
tigation on the relations between the NO2 exposure levels and descriptive variables of microenvironment can be achieved.
3. Results and discussion 3.1. Time–activity and questionnaire data Table 1 shows the percentages of time spent by the volunteers in the four microenvironments during the personal exposure campaigns. These results were obtained thanks to the time–activity-diaries. The participants of our study spent more than 85% of their time in indoor places. The times spent in the microenvironments btransportQ and boutdoorsQ are very weak (respectively less than 5% and 10% of total time). These results are in agreement with other studies (Schwab et al., 1990; Lebret et al., 1995).
The proportion of participants living in a flat is much higher (on average 45%) than the regional average given by 1999 national census (25%). This difference can be explained by the fact that the participants of our study mainly live in town (Table 2). 3.2. NO2 exposure levels in the four microenvironments Tables 3 and 4 summarize the NO2 personal exposures recorded in the four microenvironments for winter and summer campaigns respectively. Boxand-whisker plots of these results are presented in Fig. 2 (winter campaign) and Fig. 3 (summer campaign). A hierarchy in the NO2 exposure levels according to microenvironment type clearly appears. The highest levels are found in transport, and outdoors, the intermediate values in other indoor places and the lowest ones in homes. Note that the WHO guidelines for health protection (200 Ag m 3 on 1 h) was exceeded for one person in the microenvironment btransportQ and one person in bother indoor placesQ. An overall increase of exposure levels is observed in all the microenvironments in winter. This can be explained by two factors: on the one hand, by the use of gas appliances for heating in indoor places, and on the other hand, by the meteorological conditions in winter which are in favour of an outdoor concentration level increase. The exposure levels are found to be higher in bother indoor placesQ than in homes. Indeed, a differ-
240 200
Working day Weekend
160 120 80 40 0 Home
Other indoor places
Transport
Outdoors
Total exposure
Fig. 2. Box-and-whisker plots of the winter personal exposure campaign NO2 measurement results (Ag m3).
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
300 250
59
Working day Workin Weekend
200 150 100 50 0 Home
Other indoor places
Transport
Outdoors
Total exposure
Fig. 3. Box-and-whisker plots of the summer personal exposure campaign NO2 measurement results (Ag m3).
ence exists between the spatial distribution of these two categories of indoor environments: the participant homes are partly located outside the town in low polluted areas, whereas bother indoor placesQ include various building types, mainly located in urban and suburban areas next to major NO2 sources. The weekend and working day levels comparison provides some information on the traffic emission impact on the different microenvironment concentrations. It is known that in cities a traffic emission decrease of about 20% to 50% occurs between working days and weekends (Pont and Fontan, 2001). The btransportQ and boutdoorsQ levels follow this traffic Table 2 Participant home and living mode characteristics gathered by the questionnaire filled in by the participants of the winter and summer campaigns Volunteer home characteristics
Dwelling unit type
Heating type
Cooking type
Smoking status
Individual house Flat Campus room Gas Electric Fuel Gas Electric Gas and electric No smoker Non-smoker subjected to environmental tobacco smoke
Sampling periods Winter (n = 13)
Summer (n = 31)
55% 45% 0% 39% 46% 15% 38% 62% 0% 27% 8%
45% 48% 6% 65% 35% 0% 22% 68% 10% 23% 6%
emission pattern for the two campaigns. On the contrary, an inverse trend can be observed for indoor environments: the weekend exposure levels are higher than or equal to the weekday ones. Consequently, the indoor levels seem to be weakly affected by traffic emissions. For each participant, total NO2 personal exposure was calculated according to the following equation: X Ci;j ti;j Ei ¼
j
X
ti;j
ð1Þ
j
where E i is the total exposure of the participant i (Ag m 3), C i, j is the NO2 concentration at which the participant i is exposed in the microenvironment j (Ag m 3) and t i, j is the time (s) spent in the microenvironment j by the participant i. The total exposure values are shown in Tables 3 and 4 in the last two columns. They are close to home levels on weekends and between the two indoor environment levels for working days. This is mainly related to the long time spent by people in these microenvironments (from 20 to 23 h per day). 38% of the winter campaign participants had a total personal exposure for working days which exceeded the annual limit value of 40 Ag m 3 set for 2010. This was also the case of 30% of this campaign participants during the weekend. Only 3% of the summer campaign volunteers exceeded this limit value for the working days and no exceeding was observed for the summer campaign during the weekend.
60
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
Table 3 Winter personal exposure campaign NO2 measurement results (Ag m 3) (arithmetic means, median, minima and maxima) Microenvironement
Home
Type of day
Working day
Mean Median Minimum Maximum
22 19 11 38
Other indoor places
Transport
Weekend
Working day
Weekend
Working day
Weekend
Working day
Weekend
Working day
Weekend
27 20 10 60
56 55 19 86
92 62 31 229
114 104 71 159
56 60 31 77
115 109 59 191
45 42 11 103
38 35 24 53
29 23 14 61
3.3. Contribution of the different microenvironments to global personal exposure Fig. 4 shows the average contributions of the four investigated microenvironments to total personal exposure for the four sampling days. As can be seen, whatever the sampling day, the indoor places (bhomeQ average contribution added to bother indoor placesQ average contribution) account for more than 78% to total personal exposure. Despite the weak indoor levels, their contribution to personal exposure is the major component of total personal exposure because of the time spent in these indoor environments. The contribution of the other indoor places to global personal exposure for the working days is about 40%. This is close to the contribution of homes. On the contrary the contribution of the other indoor places does not exceed 10% for the weekends. These results highlight the need of accurate data on indoor environments concentrations to assess NO2 personal exposure. 3.4. Determinants of the exposure in the microenvironments btransport Q and bhome Q The determinants of NO2 personal exposure in the microenvironments btransportQ and bhomeQ were examined using a MCA multivariate analysis.
Outdoors
Total exposure
For the transport microenvironment data MCA, the NO2 exposure values were recorded according to the six following modalities: 20–32 Ag m 3 (C1), 33–50 Ag m 3 (C2), 51–68 Ag m 3 (C3), 69–96 Ag m 3 (C4), 97–125 Ag m 3 (C5) and above 125 Ag m 3 (C6). These classes were chosen in order to obtain a compromise between equal numbered and equal interval classes. In the transport analysis, four descriptive variables (transport modes, season, type of day and smoking status) were taken into account and divided into ten modalities as follows: train, tramway or underground, bicycle, car or motorcycle, winter, summer, working day, weekend, smoker and non-smoker. Fig. 5 shows the projection of the modalities on the first factorial plane ( F 1 / F 2) that accounts for 31% of the explained variance. An overview of the results shows that the NO2 personal exposure modalities are related to the transport cluster. Thus, each transport mode can be associated with one or two classes of NO2 exposure, as follows: the journeys by train with the C1 (20–32 Ag m 3) and C2 (33–50 Ag m 3) classes; those by tramway or underground with the C2 (33–50 Ag m 3) and C3 (51–68 Ag m 3) classes, those by bicycle with the C4 (69–96 Ag m 3) class and those by car or motorcycle with the C5 (97–125 Ag m 3) class. In this analysis, the other descriptive variables
Table 4 Summer personal exposure campaign NO2 measurement results (Ag m 3) (arithmetic means, median, minima and maxima) Microenvironment
Home
Type of day
Working day
Mean Median Minimum Maximum
14 11 5 24
Other indoor places
Transport
Outdoors
Total exposure
Weekend
Working day
Weekend
Working day
Weekend
Working day
Weekend
Working day
Weekend
15 13 5 44
29 29 13 79
35 38 10 72
79 64 20 256
49 40 22 99
37 36 9 60
30 26 6 78
22 23 12 32
17 16 8 50
Contributions of the four micro environments to NO2personal exposure (%)
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
Winter campaign
61
Summer campaign
100 90 80 70 60 50 40 30 20 10 0
100 90 80 70 60 50 40 30 20 10 0 Week day
Weekend
Outdoors
Week day
Transport
Weekend
Other indoor places
Home
Fig. 4. Average contributions of the four microenvironments to total NO2 personal exposure for the four sampling days.
(season, type of day and smoking status) do not appear as factors affecting the NO2 personal exposure in transport.
In the same manner, the MCA was applied to homes data by dividing the NO2 exposure values into five modalities: 5–10 Ag m 3 (C1), 11–16 Ag smoker
bicycle
1.33
2nd component - inertia : 15.0%
C4 0.57
winter C3
C5
carmotorcycle
working days tramway-underground -0.20
summer
non smoker
week-end
C2 -0.97
train
C1
-1.73 -1.31
-0.76
-0.20
0.36
0.91
1st component - inertia: 16.0%
Concentration scale (µ g m-3) 20
32
C1
50
C2
68
C3
96
C4
125
C5
259
C6*
*group corresponding to atypic items which were not taken into account for the analysis
Train
Tramway-suburban
Bicycle
Car-motorcycle
Fig. 5. Modalities projection on the first factorial plane ( F 1 / F 2) of the MCA performed on the btransportQ microenvironment data.
62
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
m 3 (C2), 17–23 Ag m3 (C3), 24–37 Ag m 3 (C4) and 38–60 Ag m 3 (C5). Five qualitative variables were considered to describe home microenvironment : smoking status, airing, heating mode, stove type, season and type of day. For this analysis, the variables were divided into fourteen modalities: smoker, nonsmoker, subjected to environmental tobacco smoke, not subjected to environmental tobacco smoke, presence of an automatic airing system, no automatic airing system, electric heating, gas heating, electric stove, gas stove, winter, summer, working day and weekend. The modality projections on the first factorial plane ( F 1 / F 2) are shown in Fig. 6. These two MCA produced factors, F 1 and F 2 , account for 34% to the explained variance. NO2 home personal exposure depends mainly on three qualitative variables: heating mode, stove type, and supply of fresh air. The season also has an effect on home NO2 personal exposures, which is probably related to the impact of
heating, on in winter and off in summer. The C1 (5–10 Ag m 3) and C2 (11–16 Ag m 3) are the exposure classes associated with homes equipped with electric heating, an automatic airing system and an electric stove in summer. The C3 (17–23 Ag m 3) and C4 (24– 37 Ag m 3) classes correspond to persons living in homes equipped with a gas stove, gas heating and with no automatic airing system in winter. The C5 (38–60 Ag m 3) class represents the NO2 personal exposures of smokers in their homes. 3.5. Relationships between the home NO2 personal exposures and the stationary monitoring data The correlations between home NO2 personal exposure levels and outdoor NO2 concentrations recorded at various fixed ambient air monitoring sites were examined in order to evaluate the opportunity of predicting NO2 personal exposures by using C1
0.99 summer
2nd component - inertia: 15.2 %
non smoker 0.20
gas heating
electric cooking
no automatic airing system
C2 working days
C4 C3 -0.58
gas cooking
electric heating
non smoker not daily subjected to tobacco smoke week-end days
automatic airing system
non smoker daily subjected to tobacco smoke
smoker winter
-1.36
C5 -2.14 -0.79
-0.32
0.15
1st
0.63
1.10
component - inertia: 18.8 %
Concentration scale (µg m-3) 5
10
C1 Electric heating
16
C2 Electric cooking Summer
Automatic airing system
23
C3
37
C4 Gas heating
60
C5 Gas cooking winter
No automatic airing system
Fig. 6. Modalities projection on the first factorial plane ( F 1 / F 2) of MCA applied to the bhomeQ microenvironment data.
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
the air quality network data. For that purpose, each home NO2 personal exposure value was paired with the NO2 concentration recorded at the monitoring station considered as brepresentativeQ of the home geographical area. Indeed, a previous study (Perdrix et al., 2001) showed that a large part of greater Lille could be covered by four areas in which NO2 levels were homogeneous and comprised in a F30% interval (representativeness criterion) around the concentration recorded by a selected station. Fig. 7a and b
Home NO2 personal exposures (µg.m-3)
a
y = 0.13 x + 11.39 R2 = 0.09
P = 0.0101
70 60
shows home NO2 personal exposures compared with outdoor concentrations measured by the respective selected fixed stations. Fig. 7a takes into account all the two campaign participants and Fig. 7b concerns only the summer participants who live in electric stove and electric heating equipped homes. The correlation is not statistically significant when participants are not selected. The correlation increases when no dominant source is present but it remains weak. Nevertheless, this relation can constitute a basic component to set a NO2 personal exposure estimation model which must take into account the household explicative variables. Other studies (Alm et al., 1998; Quackenboss et al., 1982) have shown that ambient air quality monitoring network data are generally poor predictors of NO2 personal exposure.
50 40
4. Conclusion
30 20 10 0 0
20
40
60
80
100
120
NO2 concentration measured at the representative fixed station (µg.m-3)
b Home NO2 personal exposures (µg.m-3)
63
y = 0.86 x – 18.38 R2 = 0.61
P = 0.0001
70 60 50 40 30 20 10 0 0
10
20
30
40
50
NO2 concentration measured at the representative fixed station (µg.m-3)
Fig. 7. a: Relationships between the home NO2 personal exposures and the outdoor concentrations recorded at fixed ambient air monitoring stations for all the participants. b: Relationships between the home NO2 personal exposures and the outdoor concentrations recorded at fixed ambient air monitoring stations for the summer participants with no major indoor NO2 sources.
This study was designed to investigate the NO2 personal exposure of 45 volunteers living in Lille, a large city in northern France. An innovating approach based on sequential measurements of personal exposure in four microenvironments (home, transport, other indoor places and outdoors) could be carried out using diffusive sampler having a high uptake rate. The highest levels were found in transport and outdoors, the intermediate ones in other indoor places and the lowest in homes. Nevertheless, the indoor pollution remains the main contributor to total NO2 personal exposure (superior to 78%) because the participants spent the main part of their time in indoor environments (from 20 to 23 h per day). The average total NO2 personal exposure ranges from 17 Ag m 3 for the summer weekends to 38 Ag m 3 for the winter working days. As for daily journeys, NO2 personal exposure is mainly determined by the transport mode. The highest levels, which exceeded 100 Ag m 3 were recorded for journeys by car or motorcycle. High home NO2 personal exposures are for the most part explained by gas stove, gas heating and the absence of automatic airing system. Outdoor levels appear as minor factors to explain home NO2 personal exposure. Nevertheless, btransportQ and boutdoorsQ are the microenvironments to consider in health risk assessment because they include various situations of high NO2 exposure levels occurring on short-time periods.
64
A. Piechocki-Minguy et al. / Science of the Total Environment 366 (2006) 55–64
Some further studies should be made in order to improve the knowledge on the contribution of the main determinants to home NO2 personal exposure. Acknowledgment We sincerely would like to thank our volunteers for their kind participation to this study. References Alm S, Mukala K, Pasanen P, Tiitanen P, Ruuskanen J, Tuomisto J, et al. Personal NO2 exposures of preschool children in Helsinki. J Expo Anal Environ Epidemiol 1998;8:79 – 100. Atkins DHF, Quirino I. A survey of nitrogen dioxide in Paris. European Community Commission Report EUR 13369 EN; 1992. Brunekreef B. The measurement of exposure to air pollution: concepts, methods and stakes. In nume´ro spe´cial de Pollution Atmosphe´rique: 40e`me anniversaire de l’APPA: la pollution atmosphe´rique et l’homme : e´valuer les expositions; 1992;17–29. ISO 13752. Assessment of uncertainty of a measurement method under field conditions using a second method as a reference; 1998. 19 pp. Lebart L, Morineau A, Warwick KM. Multivariate descriptive statistical analysis: correspondence analysis and related techniques. New York7 Wiley Editors; 1993. 231 pp. Lebret E, Fischer P, DeHollander G. Time–activity patterns at population level: regional population and subpopulation variations and their assessment. Air pollution epidemiology report n86: time–activity patterns in exposure assessment. Luxemburg7 Office for Official Publications of the European Community; 1995. p. 9 – 15.
Marquardt DW. An algorithm for least-squares estimation of nonlinear parameters. J Soc Ind Appl Math 1963;11:431 – 41. Palmes ED, Gunnison AF, Dimattio J, Tomczyk C. Personal sampler for nitrogen dioxide. Am Ind Hyg Assoc J 1976;37:570 – 7. Perdrix E, Fourche´ B, Plaisance H. Repre´sentativite´ spatiale des stations urbaines. Rapport du Laboratoire Central de la Qualite´ de l’Air, convention n843 / 2000; 2001. 54 pp. Piechocki-Minguy A, Plaisance H, Garcia-Fouque´ S, Galloo JC, Guillermo R. Validation tests of a new high passive sampler for nitrogen dioxide measurements. Environ Technol 2003;24: 1527 – 35. Plaisance H, Piechocki-Minguy A, Garcia-Fouque´ S, Galloo JC. Influence of meteorological factors on the NO2 measurements by passive diffusion tube. Atmos Environ 2004;38:573 – 80. Pont V, Fontan P. Comparison between weekend and weekday ozone concentration in large cities in France. Atmos Environ 2001;35:1527 – 35. Quackenboss JJ, Kanarek MS, Spengler MS, Letz JD. Personal monitoring for nitrogen dioxide exposure: methodological considerations for a community study. Environ Int 1982;8:249 – 58. Saporta G. Probabilite´, analyse des donne´es et statistique. Paris7 Technip Editors; 1990. 493 pp. Schwab M, Colome SD, Spengler JD, Ryan PB, Billick IH. Activity patterns applied to pollutant exposure assessment: data from personal monitoring study in Los Angeles. Toxicol Ind Health 1990;6:517 – 32. Sexton K, Ryan PB. Assessment of human exposure to air pollution: methods, measurements and models. In: National Academy Press, editor. Air pollution, the automobile, and public health. Washington: Health Effects Institute, National Academy Press; 1988, p. 207–38. Yanagisawa Y, Nishimura HA. Badge-type personal sampler for measurement of personal exposure to NO2 in ambient air. Environ Int 1982;8:1 – 13.