Air pollution exposure monitoring and estimation Part IV.‡ Urban exposure in children† Jocelyne Clench-Aas,* Alena Bartonova, Knut E. Grønskei and Sam-Erik Walker Norwegian Institute for Air Research, P.O. Box 100, N-2027 Kjeller, Norway. E-mail: jocelyne.clench-aas@nilu.no; Fax: +47 63 89 80 50; Tel:+47 63 89 80 00 Received 7th April 1999, Accepted 25th June 1999. In the winter of 1994, 2300 school-age children in Oslo participated in a panel study of the role of traYc pollution on the exacerbation of diseases of the respiratory system and other symptoms of reduced health and well being in children.The children filled out a diary daily with information for five time points over six weeks. In order to quantify exposure–eVect relationships for the symptoms, individual exposure to NO2 and particulate matter (PM2.5) was estimated, using the DINEX method a combination of information from the diary as to the children’s whereabouts during the five time points each day, coupled with continuous dispersion modelling.An individual exposure estimate for each time point for each child was defined. Individual exposure estimated using dispersion modelling can be used to examine patterns of exposure such as isolating geographic areas with higher concentrations or describing concentrations of pollution by time of day. The diary allowed the time-use of the children to be described.reduced general health. The study was performed during the 1. Introduction winter of 1994 at 12 schools. The chosen study design enhanced In all air pollution related studies, it is important to define the possibility of measuring dose-response relationships. and measure exposure.It is of vital interest in properly defining exposure-eVect relationships that exposure is estimated as correctly as possible. The principal source of air pollution in Oslo is vehicular traYc. It is therefore natural to focus on the 2. Materials principal pollutants from traYc, such as NOx, NO2 and 2.1 Study design particulate matter (PM).Air pollution exposure is the average concentration of air pollution that each individual is in reality To study the role of exposure to air pollution on the incidence exposed to over a period of time (hour, day, etc.).1 and prevalence of asthma and other airway diseases in children, Children are among the most sensitive sub-population both a study involving a series of investigations was initiated in Oslo.because of physiological factors that influence uptake and Ambient pollution exposure to particulate matter (PM2.5) because of their lifestyle which results in them being highly and nitrogen dioxide was estimated using dispersion models active outdoors and thus more exposed to pollutants than coupled to measurements, for each child’s home address, for are adults.schools, homes and other neighbouring places visited by the Panel studies, which investigate individuals repeatedly, may child during the 6 week period. be used to study the relationship of short-term variations in An important feature of this study is the opportunity to exposure and disease aggravation.In these studies, each indi- combine the information collected in the panel study with that vidual is his/her own control. Health and well-being while collected in an initial cross-sectional study. The cross-sectional exposed to pollution is compared to health and well-being questionnaire based investigation (the NOx–Ox study) on when not exposed. Studies done in children have usually children attending grades 1–6 in the primary schools in Oslo shown weak but significant associations between exposure to was chosen to study the eVects of air pollution on respiratory O3, SO2 , SO4 2-, PM10, PM2.5 and NO2 and symptom health.Parents of children attending two groups of schools reporting, medication use or decreased lung function.2–22 Not were invited to participate during the fall of 1993.The first all studies, however, find acute eVects of air pollution.23–28 group was in the more pollution-exposed centre of the city. As part of NILU’s (Norwegian Institute for Air Research’s) The second group lay in the peripheral higher areas of the research programme (NOx–Ox), financed by the Norwegian city, known to be less exposed to pollution.All schools in the Research Council (NFR), NILU has developed methods and centre of the city and those schools considered least exposed performed research on exposure to NOx and NO2 pollution were invited. Seventeen schools were willing to participate. in Oslo.29 The programme provided NILU with insight into The panel study, performed in Oslo during the following the dynamics of NOx and oxidant sources, distribution, and winter, forms the database for the present paper.It was aimed chemical processes. at studying the impact of air pollution on the ongoing health As part of this programme, a panel study of school-age of children, focusing on health problems related to the upper children in Oslo was initiated to study the eVects of exposure and lower airways.Symptoms were reported for 5 time periods to air pollution on symptoms of both respiratory disease and each school day, for 6 weeks during the winter by a panel of 2234 first to sixth graders attending 12 of the above mentioned schools. The children filled out the diary at school. The diary †Presented at AIRMON ’99, Geilo, Norway, February 10–14, 1999. ‡For Part III, see ref. 31. was developed so that it was easy for children to fill in the J. Environ. Monit., 1999, 1, 333–336 333required information. In addition, the children received concentrations of NO2 and sometimes particulate matter in ice-skating rinks. instruction in the form of an amusing story. The panel design allows short-term reversible changes in health status to be correlated with corresponding changes in 3.Results air pollution exposure. The aim is to study the temporal covariation of the health responses against a measure of 3.1 Air pollution exposure exposure. In this investigation type, each individual is the unit It is important when examining the eVects of separate com- of research. This reduces problems of confounding factors and pounds that the compounds are not too correlated.Since in allows each individual to be his/her own control. The collection Oslo, traYc is the principal source of pollution, the correlation of repeated data allows health status be compared with current between NO2 and PM2.5 had to be examined. The principal or preceding air exposure. source of NO2 is traYc and for PM2.5, traYc and home- The panel design also allows the eVects of individual air heating.One would expect with only one source to have a pollution components to be distinguished. This is dependent perfect 1 to 1 correlation between NO2 and PM2.5 . However, on choosing and measuring outcome variables sensitive to the sources diVer, and diVerent types of vehicles also diVer in small changes in air pollutant concentrations and using high the relative percentage of emissions of NO2 and PM2.5.enough time resolution. The ability to distinguish between Therefore the two compounds do not correlate completely. pollutants, or identify eVects of their interactions is dependent Fig. 1 shows the relationship between the estimates of NO2 on the pollutants not being strongly correlated with each and PM2.5 either as estimated at the home of the participant other.If they all vary uniformly, it is diYcult if not impossible or as the mean exposure of the child during 6 weeks in the to distinguish between them. winter as estimated using the diary. A study of this nature required approval from a series of As can be seen in Fig. 2 and Fig. 3 the overall geographical responsible institutions: the Director of Oslo Schools, the schools’ principals, parents’ associations, the teachers involved and the parents of each child.Information was disseminated in meetings with the involved parties. Approval for this study was obtained from the Data Inspectorate, and the Ethical Committee of the Norwegian Medical Association. 2.2 Choice of subjects Parents of all children in the participating schools from 1st through 6th grade (primarily from 7 to 12 years of age) received the selection questionnaire, accompanied by a letter explaining the panel study.The teachers delivered and collected the questionnaires. One written reminder was given. Those children with parental consent filled the diary at the school. The study sample was evenly divided by sex.There were a few children outside the typical range of 7 to 12 years. The population was considered randomly distributed since all children at chosen schools were invited to participate. 2.3 Air pollution exposure estimating The major diYculty in estimating ambient pollution exposure is that pollution concentrations locally can vary substantially. In order to handle this, a measurement programme was supplemented with results of dispersion modelling to estimate more precisely the pollution concentrations at the children’s homes, school and other places they visited (the DINEX method). The procedure used is described elsewhere.1 2.3.1 Individual pollution estimate. Modelling of air quality in the city was done on an hourly basis for the time periods and geographic places of interest.Hourly exposure to NO2 and PM2.5 was estimated for each child, for each day based on the location and other information provided in the diary, together with the ambient concentrations estimated using dispersion models and meteorological data.1,30 In the panel study, the covariance of the two compounds (NO2 and PM2.5) was 0.57. Each child was asked information for each time period concerning: being at home; being at school; being in walking or bicycling distance from home; being other places in Oslo; being outside Oslo; and being in downtown Oslo. 2.3.2 Socio-demographic parameters. In addition for each time children reported whether or not they were near heavy traYc, were in a room where people smoked, were indoors, Fig. 1 Relationship between (a) average NO2 and PM2.5 estimated at slept with the window open or had been in a skating hall.The each home, (b) average NO2 and PM2.5 exposure (STINEX) of each child and (c) (DINEX) individual time interval estimates. latter question was introduced because of the known high 334 J. Environ. Monit., 1999, 1, 333–336distributions of the estimates of NO2 and PM2.5 are not identical.This is a result of the diVerent concentration profiles of emissions from diesel and gasoline driven vehicles, and the diVerent patterns of driving of buses and trucks as opposed to cars. An additional emission source for particles is home heating. Even though the figures are based on overall estimates calculated during the winter prior to the study, the concentrations would not change substantially over 1 year.Fig. 4 shows the mean exposure of the students attending selected schools. Schools M to P lie in less polluted areas surrounding the city. The spread even shows a slight overlap between the exposed and unexposed areas of the city. Fig. 2 Estimated exposure to the 98th percentile of hourly NO2 over Box plots of exposure by time of day show a trend with the city of Olso.lowest values at night (see Fig. 5). 3.2 Time-use Diary studies oVer the possibility of describing the time-use patterns of the participants. The children in this study spent an average of: (1) 57% of the time intervals at home (at their home address, not inclusive of second parent, or other place that they spend the night); (2) 20% of the time intervals at school; (3) 10% of their time in walking or bicycling distance from home; (4) 6% of the time other places in Oslo; (5) 1% of the time downtown Oslo; and (6) 2% of the time outside of Oslo.They are 62% of the time indoors (this question was phrased in a reverse direction from the others and was most prone to misinterpretation). They are exposed to passive smoking 6% of the time; they are near heavy traYc 3% of the time; and they are less than 1% of the time at an ice-skating hall.Fig. 3 Estimated exposure to the 98th percentile of hourly PM2.5 over the city of Oslo. 4. Discussion Isolating an eVect of air pollution exposure on health requires refined methods. The eVects are often, if present, relatively Fig. 4 Box plot of exposure to NO2 and PM2.5 of the participants by the school of attendance.School M through P lie in the outskirts of Fig. 5 Box plot of NO2 and PM2.5 for the diVerent time intervals of the city in less polluted areas. The line represents the median, the box the 25th and 75th percentile and the standard deviation. the study. J. Environ. Monit., 1999, 1, 333–336 335Gnehm,M. Rutishauser and H.U.Wanner, Am. Rev. Respir. Dis., small although important. Improved measures of exposure are 1992, 145, 42. necessary to find these small eVects, and to create exposure- 10 L. M. Neas, D. W. Dockery, H. Burge, P. Koutrakis and F. E. eVect relationships. Speizer, Am. J. Epidemiol., 1996, 143, 797. In the panel study of children a diary method was chosen 11 W. S. Linn, D.A. Shamoo, H. R Anderson, R.-C. Peng, E. L. to estimate exposure. This method allows exposure to be Avol, J. D. Hackney and H. Gong, Jr., J. Expo. Anal. Environ. Epidemiol., 1996, 6, 449. estimated for a greater number of children, over a longer time 12 M. E. Gordian, H. Ozkaynak, J. Xue, S. S. Morris, and J. D. period and without changing their normal lifestyle as compared Spengler, Environ.Health Perspect, 1996, 104(3), 290–297. to personal monitoring equipment. 13 R. T. Burnett, R. E. Dales, M. E. Raizenne, D. Krewski, P. W. A more refined method of exposure estimating that takes Summers, G. R. Robergs, M. Raad-Young, T. Dann, and advantage of geographical diVerences in source profile of J. Brook, Environ. Res., 1994, 65(2), 172. emissions, allows developing dose-response functions for 13 A.Pinter, P. Rudnai, E. Sarkany, M. Goczan and A. Paldy, Cent Eur. J. Public Health, 1996, 4 Suppl, 17. eVects occurring at low concentrations. The individual pol- 14 R. Buchdahl, A. Parker, T. Stebbings, and A. Babiker, BMJ, 1996, lution exposure estimate also allows pollution exposure by for 312(7032), 661. example school or geographic district to be assessed. 15 I. Romieu, F. Meneses, S. Ruiz, J. J. Sienra, M. C. White, and Socio-demographic parameters showed wide variation in R. A. Etzel, Am. J. Respir. Crit. Care Med., 1996, 154(2 Pt 1), 300. this study. The time use information collected using a diary 16 W. S. Linn, D. A. Shamoo, H. R. Anderson, R. C. Peng, E. L. with children is fairly unique. The 60% time interval spent at Avol, J.D. Hackney, and H. Gong, Jr., J. Expo. Anal. Environ. Epidemiol., 1996, 6(4), 449. home, however, may be too low, since it did not properly 17 W. Roemer, G. Hoek and B. Brunekreef, Am. Rev. Respir. Dis., account for time spent with a second parent in divorced homes 1993, 147(1), 118. (only one home was registered, time spent at the other parent 18 A. Peters, I.F. Goldstein, U. Beyer, K. Franke, J. Heinrich, D. W. went into a more general category). However, the other Dockery, J. D. Spengler, and H. E.Wichmann, Am. J. Epidemiol., categories were more realistic, 20% of time intervals at school, 1996, 144(6), 570. and only 3% were near heavy traYc. 19 J. F. Scarlett, K. J. Abbott, J. L. Peacock, D. P. Strachan and H. R. Anderson, Thorax, 1996, 51(11), 1109.The ambient pollution exposure method is a valuable 20 A. Peters, D. W. Dockery, H. J. Heinrich, and H. E. Wichmann addition to methodology for panel studies aimed at examining Environ. Health Perspect, 1997, 105(4), 430. the health eVects of ambient pollution. 21 M. H. Gielen, S. C. van der Zee, J. H. van Wijnen, C. J. van Steen and B. Brunekreef, Am. J. Respir.Crit. Care Med., 1997, 155(6), 2105. 5. Acknowledgements 22 J. R. Goldsmith, M. D. D. Friger, and M. Abramson, Arch. The study was financed by the Norwegian Research Council Environ. Health, 1996, 51(5), 359. 23 G. Bjerknes-Haugen, J. Clench-Aas, S. O. Samuelsen, A. and the Norwegian Institute for Air Research. Bartonova and L. S. Bakketeig, Cent. Eur. J. Public Health, 1995, 3(1), 13. 6. References 24 S. Vedal, M. B. Schenker, A. Munoz, J. M. Samet, S. Batterman and F. E. Speizer, Am. J. Publ. Health, 1987, 77, 694. 1 J. Clench-Aas, A. Bartonova, T. Bøhler, K. E. Grønskei and 25 H. Luttmann, U. Gromping, L. Kreienbrock, C. Treiber-Klotzer S. Larssen, 1999, J.Environ. Monit., 1999, 1, 313. and H. E. Wichmann, Zentralbl. Hyg. Umweltmed., 1995, 198(2), 2 O. Herbarth, Environ.Health Perspect. 1995, 103(9), 852. 172. 3 L. M. Neas, D. W. Dockery, P. Koutrakis, D. J. Tollerud and 26 G. Hoek and B. Brunekreef, Environ. Res., 1994, 64, 136. F. E. Speizer, Am. J. Epidemiol., 1995, 141, 111. 27 G. Hoek and B. Brunekreef, Am. J. Respir. Crit. Care Med., 1995 4 C. A. Pope, III, D. W. Dockery, J. D. Spengler and M. E. 151, 27. Raizenne, Am. Rev. Respir. Dis., 1991, 144, 668. 28 W. Roemer. et al., Eur. Respir. Rev., 1998, 8(52), 125. 5 C. A. Pope, III and D. W. Dockery, Am. Rev. Respir. Dis., 1992, 29 S. Larssen, K. E. Grønskei, F. Gram, L.O.Hagen, and S. E. 145, 1123. Walker, in: Air Pollution Modeling and Its Application X. ed., S. E. 6 J. Schwartz, D. W. Dockery, L. M. Neas, D. Wypij, J. H. Ware, Gryning and M. M. Millan, Plenum Press, New York, 1994, J. D. Spengler, P. Koutrakis, F. E. Speizer and B. G. Ferris, Jr, pp. 91. Am. J. Respir. Crit. Care. Med., 1994, 150, 1234. 30 S. E. Walker, L. H. Slørdal, C. Guerreiro, F. Gram and K. E. 7 M. J. Studnicka, T. Frischer, R. Meinert, A. Studnicka-Benke, Grønskei, 1999, J. Environ. Monit., 1999, 1, 321. K. Hajek, J. D. Spengler and M. G. Neumann, Am. J. Respir. 31 J. Clench-Aas, C. Guerreiro and A. Bartonova, J. Environ. Monit., Crit. Care. Med., 1995, 151, 423. 1999, 1, 327. 8 B. D. Ostro, M. J. Lipsett, J. K. Mann, H. Braxton-Owens and M. C. White, Inhal. Toxicol., 1995, 7, 711. 9 C. Braun-Fahrla�nder, U. Ackermann-Liebrich, J. Schwartz, H. P. Paper 9/02779C 336 J. Environ. Monit., 1999, 1, 333–3