Air pollution has long been suspected of having adverse health effects, particularly in terms of respiratory disease. 1–5 The association of air pollution with mortality is now well established. 6–10 With respiratory morbidity, the situation is less clear. While most studies have found that respiratory symptoms or morbidity are related to levels of air pollution, 11–13 the findings are not consistent with respect to the role of individual pollutants and the magnitude of the effect. In addition to particulates and sulfur dioxide, 14–19 nitrogen oxides, ozone and organic compounds have all been implicated in the development of respiratory symptoms in children. 11,12,20–25 This paper concentrates on ambient nitrogen dioxide (NO2) and sulfur dioxide (SO2).
Studies of air pollution and health are generally of two types: time-series and geographical studies. In time-series analyses, correlations of daily levels of air pollution and health outcome are investigated within one population. Time-series analyses have received relatively more attention in recent decades, 6,23,27–32 and the methodology is now well developed. 26,33 The large advantage of time-series studies is that they eliminate the potential confounding by individual-level factors, such as smoking: it is unlikely that these factors would fluctuate in the studied population over the relatively short periods used in the analyses. On the other hand, these studies measure acute health effects rather than the chronic effects of long-term exposure. It is difficult to distinguish whether the increase in daily counts of health events reflects exacerbations of existing disease, or development of new disease in previously healthy persons. 32
Geographical studies, which compare frequency of health outcomes between geographical areas with different levels of air pollution, 34 have also provided evidence that mortality 7,9,35–37 and respiratory morbidity 15,18 are higher in areas with higher levels of pollution. The advantage of this design is that chronic effects can be investigated. These studies, however, can be severely confounded by individual factors such as smoking or socio-economic status. Most studies have attempted to control for at least some potential confounders at the ecological level, but the utility of such an approach is far less clear than at the individual level. Few studies were able to adjust at the individual level. 7,9
The second major problem of geographical studies is the measurement of exposure. Most studies have compared relatively large areas; one of the assumptions is that all individuals in one area have identical or similar exposures. This assumption may often be violated. Concentrations of outdoor air pollution vary substantially within the study units, and it is likely that persons living in different parts of one study area will be exposed to different levels of ambient air pollution. This issue is particularly pertinent in the case of NO2, ambient concentrations of which depend strongly on proximity to roads. 38–40 Moreover, outdoor concentrations of NO2, even if measured at an individual’s house, predict personal exposure poorly. 38,40 Personal exposures to NO2 reflect more closely indoor concentrations, 40–42 that in turn are higher in houses using gas for cooking and/or heating. 25,41,42 Thus, the indoor environment also needs to be taken into account. Few studies 43,44 have estimated exposure to air pollutants at the individual level.
The Small Area Variations in Air pollution and Health (SAVIAH) Study 45–50 attempted to overcome some of the limitations of geographical studies (problems with confounding and exposure measurement) while preserving their strength (ability to study chronic effects). The primary focus was on methodology, with the objective of evaluating methodologies in the fields of small area epidemiology, geographical information systems (GIS), air pollution modeling, and biostatistics.
The underlying concept is that a study at small area level can be an intermediate step between individual-based studies and geographical studies across large areas. 34,51 As small areas are likely to be more homogeneous with respect to socio-economic profile as well as other factors (type of housing, heating, cooking), the potential for confounding should be less than when large populations are compared. 52 Similarly, when ambient air pollution is measured by a sufficiently dense network of sites (eg, blocks of houses), extrapolation to individual exposures is more credible than at a large area scale (eg, counties).
A novel feature of this project was the use of GIS to integrate the relevant data collected from different sources, and to assign exposure estimates to individuals. 49,53–55 In most developed countries, data at a small area level are becoming increasingly available, and their analysis within a GIS framework could contribute substantially to health-related research. This paper reports analyses of geographical variation in NO2 and SO2 outdoor concentrations and prevalence of wheezing or whistling in last 12 months in the Czech part of this international project.
The SAVIAH study was a multi-center study on air pollution and respiratory health conducted in four centers (Huddersfield, England; Prague, Czech Republic; Poznan, Poland; and Amsterdam, Netherlands) in 1993–1994. The Czech component of the project was based in two districts of Prague. The selected districts have a population of 163,700 (14% of the total Prague population); cover some 48 km 2 (10% of the area of Prague); have hilly terrain (altitude of 172–355 m); contain residential, industrial, recreational, and agricultural areas; and show wide variation in levels of air pollution.
The study population comprised all children 7 to 10 years of age who lived and attended school in the study area. Children were recruited through all primary schools in the study area (excluding special schools).
We collected three types of data (1) individual data, (2i) data on small geographical areas, and (2) data on air pollution.
(1) Data at Individual Level
In the last 3 months of 1993, parents of children were asked to complete a questionnaire covering socio-economic circumstances (crowding, car ownership, education of parents), type of housing (dampness at home, type of cooking, heating, and parental smoking [as the only indicators of indoor air pollution]), family history of atopy, and health of the child. Prevalence of wheezing or whistling in the last 12 months was assessed by the following question: “Has your child in the last 12 months had wheezing or whistling in the chest?”
(2) Aggregated Data on Geographical Areas
Aggregated data from the 1991 Census and other urban planning sources were available for geographical areas (not individuals or households). The smallest area was enumeration district (ED), covering some 200 households on average. There are approximately 700 EDs in the study area. Because of small numbers of surveyed children in EDs, we aggregated adjacent EDs a priori into 75 larger “study areas” with similar numbers of children (average of 50). Data from the census included type of cooking and heating, proportion of population by different levels of education, crowding (persons per room), presence of water and gas supply at home, presence of toilets, bathroom at home, and car ownership. Urban planning sources provided data on traffic flow in roads and streets, land cover type (residential or industrial), and geographical data (eg, altitude).
(3) Air Pollution Concentrations
We measured NO2 concentrations in three 2-week campaigns (October 1993, February 1994, May 1994) by Palmes tubes 56 at 80 sites and SO2 by Willems badges 57 at 50 sites. We chose the locations to represent local topography, distribution of air pollution sources and location of schools, rural and urban background, and street levels. We used four fixed continuous monitors within the study area for validation of the passive sampling. For NO2, means from all three surveys were used in analyses reported here. For SO2, we used only data from the second survey (February 1994), as this was the most extensive survey and more sampling points allowed more precise modeling of the spatial distribution of SO2 (the correlation between data used and mean of all surveys was 0.86). We conducted analyses similar to those presented here for the other two SO2 surveys and their average, and these analyses gave similar results.
Integration into Geographical Information System
We integrated all data described above into a geographic information system (GIS) 49,53,54 in Arc/Info software, in a digitized map at a scale of 1:10,000. The addresses of responders, locations of schools, and air pollution monitoring sites were digitized, linked to the GIS, and integrated into the map. Data on annual average number of cars per hour for all except local streets (provided by the Institute of Traffic Engineering and the Institute of City Informatics 58), landcover scores describing the type of built-up land (density of houses), and altitude (from Municipal Authority, Prague) were added into the GIS.
Within the GIS, we used data from the air pollution surveys to construct an air pollution map of the study area, and to estimate outdoor concentrations in each “study area” (mean) and at children’s homes and schools. NO2 and SO2 were estimated using different modeling strategies to take account of the different dispersal characteristics and main emission sources (traffic for NO2, and industrial sources and home heating for SO2).
Regression Model of NO2 Pollution
Full details of the regression mapping methodology have been published elsewhere. 49 Briefly, for each monitoring site, we calculated average concentrations of NO2 from the three surveys. We used the logarithm of the mean values at the 80 monitored sites as a dependent variable in regression analysis, with traffic volume, altitude, and landcover scores as independent variables. This model allowed estimation of mean annual air pollution concentrations at any point on the map from known variables, and hence construction of the air pollution map. We computed the predicted value at each point in the GIS using the values of independent variables and the corresponding regression coefficients. Some 84% of the variation of NO2 levels measured at the 80 locations could be explained by the independent variables. 49
Estimation of SO2
SO2 tends to show a more smoothly varying pattern than NO2. We therefore performed mapping of SO2 using kriging, a spatial interpolation technique widely used in environmental studies in recent years. 59,60 Kriging was also performed in Arc/Info, and we used a spherical model to fit the semivariogram, and a window size to include the 20 nearest sampling points. Maps of the area and of the spatial distribution of air pollution can be obtained from the authors.
We assigned one ecological and one individual exposure to outdoor NO2 and SO2 to each child using the GIS. We calculated ecological exposures for each geographical area as the means of estimated NO2 and SO2 levels at all points in the study area. We constructed mean individual exposures to outdoor NO2 and SO2 as the average of home and school exposures for each pollutant. We used average pollution concentrations in a buffer area (80 m in diameter) around each child’s house of residence to estimate individuals’ home exposures to outdoor NO2 and SO2, and average concentrations in buffer areas around the child’s school were used to estimate individuals’ school exposure to outdoor NO2 and SO2.
Individual Level Analyses
The effect of potential risk factors on the prevalence of wheezing or whistling was estimated by logistic regression. In addition to estimates of NO2 and SO2 exposures, the following variables known from the literature as important were included: sex, age, noise/fumes from traffic at home, damp at home, parental history of atopy, birth order of child, maternal education, time spent in the countryside, gas/stove cooking and gas/stove heating at home, and maternal smoking. 21,61,62
We modeled the effect of explanatory variables on prevalence of respiratory symptoms at the level of geographical units by weighted least squares regression method. The proportion of households with water supply, with gas supply, and mean number of individuals per room were selected as indicators of social conditions of the geographic unit. Mean NO2 and SO2 concentrations (predicted in the regression analysis and kriging) in 75 “study areas” were used as exposures.
The dataset is an example of a hierarchical structure in which units at one level are nested within units at a higher level. There are two alternative structures. First, children in the study are grouped within “study area.” Alternatively, children in the study are grouped within the schools. These two structures are not nested within each other (schools are not nested within “study areas” and “study areas” are not administered within schools). The hierarchical structure of the data can be represented appropriately using multilevel modeling techniques. 63 This approach takes into account the fact that children from the same school (or the same study district) may be more alike on average than children from different schools or “study areas.” We fitted both models (based on areas or schools) using the MLn procedure. 64 As they gave almost identical results, we report only the model with “study areas” in level two (ecological level). We also compared the use of individually vs ecologically measured adjustment variables.
A total of 4,176 questionnaires were distributed in 24 primary schools, and 3,680 questionnaires were returned (response rate 88.1%). Seventy-seven addresses were incorrect; traffic volume data and landcover scores were not available for areas relating to houses of 61 children and two schools attended by 208 children; SO2 data estimated by kriging were not available for 325 children; and information on wheezing was missing for 75 children. Data were thus available for 3,045 children (83% of returned) who were included in the present analyses; their descriptive characteristics are given in Table 1. The prevalence of wheezing or whistling in the chest in the last 12 months was 11.7% (10.6% in girls and 12.8% in boys). Median (25th and 75th percentiles) concentration measured by passive samplers was 35.8 (27.9 and 45.3) μg/m 3 for NO2 and 73.9 (63.5 and 95.5) μg/m 3 for SO2. Modeled individual home and school exposures to NO2 were positively correlated (R = 0.36).
In the individual level analysis, wheezing prevalence was higher in boys, those with atopic history reported by parents and dampness at home, and lower among children with higher educated mothers and with increasing birth order (Table 1). Maternal smoking, if anything, was associated with slightly lower prevalence of wheeze. We found no strong effect of these covariates in the geographical analyses, although indicators of higher socioeconomic status (water and gas supply) were associated with lower risk of wheeze (Table 2).
Table 3 shows the associations between wheezing and pollutants at the different levels of analysis and with the various adjustments for covariates. In univariate analyses, exposure to both pollutants was positively related to wheezing; odds ratios (95% confidence interval) per 10 μg/m 3 increase in concentrations were 1.16 (0.98–1.38) for NO2 and 1.08 (0.98–1.19) for SO2. With area exposures, odds ratio for SO2 remained unchanged and that for NO2 was reduced to 1.02 (0.87–1.20). With individual socioeconomic variables included in the model, the estimated effects of NO2 and SO2 remained virtually the same. With both pollutants included simultaneously in one model, and with individual level exposure variables, the estimated effects of NO2 were reduced while the effects of SO2 were reduced or eliminated (Table 3, top panel). With both pollutants included simultaneously in one model, and with area level exposure variables, the estimated effects of NO2 became inverse while the odds ratio for SO2 became more positive (Table 3, middle panel). In the full ecological analysis, and with both pollutants in the model, the estimated odds ratios for both NO2 and SO2 were close to one (Table 3, bottom panel).
The main objective of the SAVIAH study was to develop and test new methods for mapping and analyzing relationships between air pollution and health. For this purpose, large amounts of data from different sources were integrated into GIS and used in various ways. We found positive associations of NO2 and SO2 with wheezing of individuals when NO2 and SO2 were estimated at individual level and included into the model separately. After mutual adjustment, the estimated effects of both pollutants changed, and they varied also according to the method of classification of exposure to NO2 and SO2. It seems reasonable to assume that exposure defined as the arithmetic mean of exposure at home and school is a better proxy of personal exposure than an area average.
There does not appear to be a simple explanation for our results with respect to pollutants. Of course, at least to some extent, they may reflect random variation. We suggest that a combination of issues related to exposure estimation, study power, interaction between pollutants, and confounding by other factors should also be considered.
Measurement of exposure has been a major problem in studies of health effects of air pollution. Few studies have been able to conduct personal monitoring of air pollution in sufficient numbers of subjects and, even then, only for relatively short periods. 65,66 As an alternative, concentrations of outdoor pollution could be measured at the house and school of each participating child, possibly combined with time-activity analysis. This program would also be difficult to sustain over long periods of time.
In the GIS modeling of exposure, we went further than other studies we are aware of. Although the models of NO2 and SO2 are based on several assumptions, the results are consistent with both expert expectations and an independent model produced by the Institute of City Informatics, Prague. 58 The pollution model also predicted well the mean annual NO2 levels at 10 independent reference sites that were not used in the pollution mapping. 49 Terrain and climatic conditions in Prague predict high levels of pollution in areas closer to the center (in the valley) and at locations with high volumes of car traffic. Again, this finding corresponds with our predictions. The spatial distribution of NO2 was similar in all four surveys, and it is unlikely that the map seriously misclassified areas with respect to NO2 concentrations. In this regard, the rationale for small area studies of air pollution seems well founded.
A more serious limitation, however, is the extrapolation from outdoor pollution (even at an individual’s home or school) to personal exposure. In northern countries, people spend most of their time indoors, most of it in their homes. U.S. study reported that children (students) spent 66% of their time at home, 20% at school, and only a small part of their time (4% in winter and 15% in summer) outdoors. 41 Not surprisingly, outdoor concentrations of NO2 explained only 9–12% of variation in concentrations measured by personal monitors. 41 Another smaller study found that outdoor air quality did not predict personal exposure. 42 This finding suggests that estimates of individual exposure based on outdoor concentrations alone, even if measured or modeled at the small area level, would severely misclassify total personal exposure. Such misclassification is likely to be random, which would bias any effect of NO2 toward unity.
As most geographical studies of air pollution have used ecological measures of both exposure and confounding factors, the impact of the measurement of covariates on adjusted odds ratios is potentially important. Overall, analyses using data at the individual level do not show large changes in the estimated odds ratios as further variables are added in the model (Table 3). In this respect, the assumption that covariates at small area level would approximate individual data seems justified.
We did not find evidence for an interaction between NO2 and SO2. Nevertheless, respiratory symptoms may be related to other pollutants that were not measured in our study, such as particulates, carbon monoxide, benzene, or other volatile organic compounds. If pollutants affect respiratory health synergistically, in combination, and if this mechanism involves unmeasured pollutants, such an effect would have been missed in this study.
Although the above would lead to false negative results, the possibility of a real absence of an effect of NO2 and SO2 on wheeze should be considered. Not all studies have shown clear associations between symptoms and these pollutants in ambient air, 68 particularly as even the effect of indoor NO2 is not clearly demonstrated. Studies in the United States have shown that houses using gas for cooking had mean levels of NO2 substantially (about twice) higher than houses using electricity. 40–42 Our indicators of indoor concentrations of NO2 (smoking and cooking and heating on gas) were not related to wheeze, in contrast to reports showing that use of gas at home is positively related to respiratory illness. 25,40,41,69 Another source of indoor NO2 (as well as particulates) is parental smoking. 70,71 In our study, 34% of mothers and 37% of fathers (or other adults in the child’s home) smoked; however, smoking was not related to prevalence of wheezing: ORs (95% CI) were 0.87(0.68–1.13) for maternal and 1.14 (0.88–1.48) for paternal smoking. Our results are not consistent with studies showing that parental smoking was a risk factor for respiratory illness, 21,72,73 although some studies did not find this association. 74,75 The absence of an association between maternal smoking and children’s wheezing could be due to preferential cessation of smoking among parents of wheezy children. On the other hand, the associations of wheeze with most individual characteristics, such as sex, maternal education, birth order, or family history of atopy, were consistent with other research, 61,67 which supports the validity of the reporting of wheeze in our study.
This study has demonstrated the feasibility of collecting and integrating data from different sources at small area scale using GIS techniques, which are potentially useful for the study of health outcomes. Several conclusions can be drawn. First, mapping of the spatial distribution of NO2 and SO2 in the study area appeared to give credible results; they were consistent with both measurements made at independent sampling points and independent models. Second, adjustment for ecologically derived (small area) covariates did not lead to markedly different estimated odds ratios, compared with covariates measured at individual data. Third, interpolation to personal exposures may be problematic, and future studies may require the addition of indoor or personal measurements and time activity analysis to increase their validity and statistical power.
The SAVIAH study was a multi-center study funded by the European Union Third Framework Environment Programme. It was led by Paul Elliott, and co-principal investigators were David Briggs, Bohumir Kriz, Pawel Gorynski (National Institute of Hygiene, Warsaw, Poland), and Erik Lebret (National Institute of Public Health and Environmental Protection, Bilthoven, Netherlands). Other members of the project team were: Jana Danova, Martin Celko, Vladimir Prikazsky, Hynek Pikhart, Martin Bobak, Jan Pretel, Karel Pryl (Czech Republic), Bogdan Wojtyniak (Poland), Marco Martuzzi, Chris Grundy, Susan Collins, Emma Livesely, Kirsty Smallbone (United Kingdom), Henrik Harssema, Gerda Dornbos, Paul Fischer, Hans van Reeuwijk, Andre van der Veen, and Leon Gras (Netherlands). All members of this team made important contributions to the project. We thank the Czech National Institute of Public Health, the Czech Hydrometeorological Institute, Prague Institute of Urban Development, local authorities, and health authorities in both Prague districts for their support and assistance.
1. Fairbairn AS, Reid DD. Air pollution
and other local factors in respiratory disease. Br J Prev Soc Med 1958; 12:94–103.
2. Reid DD. Air pollution
as a cause of chronic bronchitis. Symposium No.6, Sect. I, Medical and epidemiological aspects of air pollution
. Proc R Soc Med 1964; 57:965–968.
3. Lambert PM, Reid DD. Smoking, air pollution
and bronchitis in Britain. Lancet 1970; 1:853–857.
4. Colley JRT, Douglas JWB, Reid DD. Respiratory disease in young adults: influence of early childhood lower respiratory tract illness, social class, air pollution
and smoking. BMJ 1973; 518:195–198.
5. Holland WW, Bennett AE, Cameron IR, Florey CV, Leeder SR, Schilling RS, Swan AV, Walter RE. Health effects of particulate air pollution
: reappraising the evidence. Am J Epidemiol 1979; 110:527–659.
6. Schwartz J. Air pollution
and daily mortality in Birmingham, Alabama. Am J Epidemiol 1993; 137:1136–1147.
7. Dockery DW, Pope III CA, Xu X, Spengler JD, Fay ME, Ferris BG, Speizer FE. An association between air pollution
and mortality in six U.S. cities. N Engl J Med 1993; 329:1753–1759.
8. Dockery DW, Pope III CA. Acute respiratory effects of particulate air pollution
. Ann Rev Public Health 1994; 15:107–132.
9. Pope III CA, Thun MJ, Namboodiri MM, Dockery DW, Evans JS, Speizer FE, Heath CW. Particulate air pollution
as a predictor of mortality in a prospective study of US adults. Am J Respir Crit Care Med 1995; 151:669–674.
10. Committee on the Medical Effects of Air Pollutants. Non-biological particles and health. Department of Health. London, 1995.
11. Melia RJW, du Florey VC, Chinn S, Morris RW, Goldstein BD, John HH, Clark D. Investigations into the relations between respiratory illness in children
, gas cooking and nitrogen dioxide
in the UK. Tokai J Exp Clin Med 1985; 10:375–378.
12. Rutishauser M, Ackermann U, Braun C, Gnehm HP, Wanner HU. Significant association between outdoor NO2
and respiratory symptoms
in preschool children
. Lung 1990; 168(suppl):347–352.
13. Wjst M, Reitmeir P. Road traffic and adverse effects on respiratory health in children
. BMJ 1993; 307:596–600.
14. Melia RJW, du Florey VC, Chinn S. Respiratory illness in British schoolchildren and atmospheric smoke and sulphur dioxide 1973–7. II. Longitudinal findings. J Epidemiol Community Health 1981; 35:168–173.
15. Melia RJW, du Florey VC, Swan AV. Respiratory illness in British schoolchildren and atmospheric smoke and sulphur dioxide 1973–7. I. Cross-sectional findings. J Epidemiol Community Health 1981; 35:161–167.
16. Lawther PJ, Waller RE, Henderson M. Air pollution
and exacerbations of bronchitis. Thorax. 1970; 25:525–539.
17. Lunn JE, Knowelden J, Handyside AJ. Patterns of respiratory illness in Sheffield infant school children
. Br J Prev Soc Med 1967; 21:7–16.
18. Dockery DW, Speizer FE, Stram DO, Ware JH, Spengler JD, Ferris BG. Effects of inhalable particles on respiratory health of children
. Am Rev Respir Dis 1989; 139:587–594.
19. Ware JH, Ferris BG, Dockery DW, Spengler JD, Stram DO, Speizer FE. Effects of ambient sulphur dioxide and suspended particles on respiratory health of preadolescent children
. Am Rev Respir Dis 1986; 133:834–842.
20. Jaakkola JJK, Paunio M, Virtanen M, Heinonen OP. Low-level air pollution
and upper respiratory infections in children
. Am J Public Health 1991; 81:1060–1063.
21. Ware JH, Dockery DW, Spiro A, Speizer FE, Ferris BG. Passive smoking, gas cooking and respiratory health of children
living in six cities. Am Rev Respir Dis 1984; 129:366–374.
22. Ware JH, Spengler JD, Neas LM, Samet JM, Wagner GR, Coultas D, Ozkaynak H, Schwab M. Respiratory and irritant health effects of ambient volatile organic compounds. Am J Epidemiol 1993; 137:1287–1301.
23. Schwartz J. PM-10, ozone, and hospital admissions for the elderly in Minneapolis-St Paul, Minnesota. Arch Environ Health 1994; 49:366–374.
24. Hoek G, Fischer P, Brunekreef B, Lebret E, Hofschreuder P, Mennen MG. Acute effects of ambient ozone on pulmonary function of children
on the Netherlands. Am Rev Respir Dis 1993; 147:111–117.
25. Neas LM, Dockery DW, Ware JH, Spengler JD, Speizer FE, Ferris BJ Jr. Association of indoor nitrogen dioxide
with respiratory symptoms
and pulmonary functions in children
. Am J Epidemiol. 1991; 134:204–219.
26. Schwartz J, Spix C, Touloumi G, Bacharova L, Barumamdzadeh T, le Tertre A, Piekarski T, Ponce de Leon A, Ponka A, Rossi G, Saez M, Schouten JP. Methodological issues in studies of air pollution
and daily counts of deaths or hospital admissions. J Epidemiol Community Health 1996; 50(suppl 1):S3–S11.
27. Katsouyanni K, Touloumi G, Spix C, Schwartz J, Balducci F, Medina S, Rossi G, Wojtyniak B, Sunyer J, Bacharova L, Schouten JP, Ponka A, Anderson HR. Short-term effects of ambient sulphur dioxide and particulate matter on mortality in 12 European cities: results from time series data from the APHEA project. Air pollution
and Health: a European Approach. BMJ 1997; 314:1658–1663.
28. Schwartz J, Dockery DW. Particulate air pollution
and daily mortality in Steubenville, Ohio. Am J Epidemiol 1992; 135:12–19.
29. Schwartz J, Dockery DW. Increased mortality in Philadelphia associated with daily air pollution
concentrations. Am Rev Respir Dis 1992; 145:600–604.
30. Schwartz J. Air pollution
and hospital admissions for the elderly in Birmingham, Alabama. Am J Epidemiol 1994; 139:589–598.
31. Pope III CA, Schwartz J, Ransom MR. Daily mortality and PM10 pollution in Utah Valley. Arch Environ Health 1992; 47:211–217.
32. Samet JM, Zeger SL, Berhane K. The association of Mortality and Particulate Air Pollution
. In: Health Effects Institute. Particulate air pollution
and daily mortality. Replication and validation of selected studies. The Phase I Report of the Particle Epidemiology Evaluation Project. Cambridge, MA: Health Effects Institute, 1995.
33. Schwartz J. Air pollution
and daily mortality: a review and meta analysis. Environ Res 1994; 64:36–52.
34. English D. Geographical epidemiology and ecological studies. In: Elliott P, Cuzick J, English D, eds. Geographical & Environmental Epidemiology. Methods for Small-Area Studies. New York: Oxford University Press, 1996: 3–13.
35. Lave LB, Seskin EP. Air pollution
and human health. Baltimore: John Hopkins University Press, 1977.
36. Evans JS, Tosteson T, Kinney PL. Cross-sectional mortality and air pollution
risk assessment. Environ Int 1984; 10:53–83.
37. Bobak M, Leon DA. Air pollution
and infant mortality in the Czech Republic, 1: 986–88. Lancet 1992; 340:1010–1014.
38. Bocken P, Michorius J, van Reeuwijk H, Schellevis L. A passive sample for measuring NO2
in ambient air: networks, performance and intercomparison of methods in Prague city. Report V-130. Department of Air Pollution
. Wageningen: Agricultural University Wageningen, 1992.
39. Eerens HC, Sliggers CJ, van den Hout KD. The CAR model: the Dutch method to determine city street air quality. Atmos Environ 1993; 27B:389–399.
40. Quackenboss JJ, Krzyzanowski M, Lebowitz MD. Exposure assessment approaches to evaluate respiratory health effects of particulate matter and nitrogen dioxide
. J Expo Anal Environ Epidemiol 1992; 1:83–107.
41. Quackenboss JJ, Kanarek MS, Spengler JD, Letz R. Personal monitoring for nitrogen dioxide
exposure: methodological considerations for a community study. Environ Int 1982; 8:249–258.
42. Quackenboss JJ, Spengler JD, Kanarek MS, Letz R, Duffy CP. Personal exposure to nitrogen dioxide
: relationship to indoor/outdoor quality and activity patterns. Environ Sci Technol 1986; 20:775–782.
43. Infante-Rivard C. Childhood asthma and indoor environmental risk factors. Am J Epidemiol 1993; 137:834–844.
44. Berglund M, Braback L, Bylin G, Jonson JO, Vahter M. Personal NO2
exposure monitoring shows high exposure among ice-skating schoolchildren. Arch Environ Health 1994; 49:17–24.
45. Elliott P, Briggs D, Lebret E, Gorynski P, Kriz B. Small area variations in air quality and health (The SAVIAH Study): design and methods. Epidemiology 1995; 6(suppl):S32.
46. Bobak M, Pikhart H, Kriz B, Danova J, Celko M, Prikazsky V, Pryl K, Pretel J. Description of the Czech part of the SAVIAH Study (Abstract). Epidemiology 1995; 6(suppl):S61.
47. Fischer PA, Kriz B, Martuzzi M, Wojtyniak B, Lebret E, van Reeuwijk H, Pikhart H, Briggs D, Gorynski P, Elliott P. Risk factors indoors and prevalences of childhood respiratory health in four countries in Western and central Europe. Indoor Air 1998; 8:244–254.
48. Pikhart H. Association between child respiratory health and air-pollution in Prague (1993–94) (MSc Thesis). London School of Hygiene and Tropical Medicine, 1995.
49. Briggs DJ, Collins S, Elliott P, Fischer P, Kingham S, Lebret E, Pryl K, van Reeuwijk H, Smallbone K, van der Veen A. Mapping urban air pollution
using GIS: a regression-based approach. Int J Geogr Inf Sci 1997; 11:699–718.
50. Pikhart H, Prikazsky V, Bobak M, Kriz B, Celko M, Danova J, Pryl K, Pretel J. Association between ambient nitrogen dioxide
and respiratory symptoms
in Prague, Czech Republic. Preliminary results from the Czech part of the SAVIAH study. Cent Eur J Public Health 1997; 5:82–85.
51. Cuzick J, Elliott P. Small-area studies: purpose and methods. In: Elliott P, Cuzick J, English D, eds. Geographical & Environmental Epidemiology. Methods for Small-Area Studies. New York: Oxford University Press, 1996; 14–21.
52. Diggle P, Elliott P. Disease risk near point sources: statistical issues for analyses using individual or spatially aggregated data. J Epidemiol Community Health 1995; 49(suppl 2):S20–27.
53. Burrough PA. Principles of Geographical Information Systems for Land Resources Assessment. Oxford Science Publications, 1991.
54. Environmental Systems Research Institute. Understanding GIS. The ARC/Info Method. ESRI, USA, 1991.
55. Briggs D, Elliott P. The use of geographical information systems in studies on environment and health. World Health Stat Q 1995; 48:85–94.
56. Palmes ED, Gunnison AF, DiMattio J, Tomczyk C. Personal sampler for nitrogen dioxide
. Am Ind Hyg Assoc J 1976; 37:570–577.
57. Hartog KD. Laboratory procedures for measuring SO2
Willems-badges and NO2
Palmes-tubes. Department of Air Quality, Wageningen Agricultural University. Report No: IV-202, 1995.
58. Praha - zivotni prostredi 1995.(Prague – Environment 1995). Institute of City Informatics, Prague, Czech Republic, 1995.
59. Oliver MA, Webster R. Kriging: a method of interpolation for geographical information systems. Int J Geogr Inf Syst 1990; 4:313–332.
60. Briggs DJ. Mapping environmental exposure. In: Elliott P, Cuzick J, English D, eds. Geographical & Environmental Epidemiology. Methods for Small-Area Studies. New York: Oxford University Press, 1996; 158–176.
61. Clifford RD, Radford M, Howell JB, Holgate ST. Prevalence of respiratory symptoms
among 7 and 11 year old schoolchildren and associations with asthma. Arch Dis Child 1989; 64:1118–1125.
62. Volkmer RE, Ruffin RE, Wigg NR, Davies N. The prevalence of respiratory symptoms
in South Australian preschool children
. II. Factors associated with indoor air quality. J Paediatr Child Health 1995; 31:116–120.
63. Goldstein H. Multilevel statistical models. 2nd ed. Kendalls Library of Statistics 3, 1995.
64. Woodhouse G, Rasbash J, Goldstein H, Yang M, Howart J, Plewis I. A guide to MLn for new users. Institute of Education, University of London, 1995.
65. Hoek G, Meijer R, Scholten A, Noij D, Lebret E. The relationship between indoor nitrogen dioxide
concentration levels and personal exposure: a pilot study. Int Arch Occup Environ Health 1984; 55:73–78.
66. Linaker CH, Chauhan AJ, Inskip H, Frew AJ, Sillence A, Coggon D, Holgate ST. Distribution and determinants of personal exposure to nitrogen dioxide
in school children
. Occup Environ Med 1996; 53:200–203.
67. Bobak M, Koupilova I, Williams HC, Leon DA, Danova J, Kriz B. Prevalence of asthma, atopic eczema and hay fever in 5 Czech cities with different level of air pollution
. [in Czech]. Prakticky Lekar 1995; 75:480–485.
68. Pershagen G, Rylander E, Norberg S, Eriksson M, Nordvall SL. Air pollution
involving nitrogen dioxide
exposure and wheezing bronchitis in children
. Int J Epidemiol 1995; 24:1147–1153.
69. Jarvis D, Chinn S, Luczynska C, Burney P. Association of respiratory symptoms
and lung function in young adults with use of domestic gas appliances. Lancet 1996; 347:426–431.
70. Neas LM, Dockery DW, Ware JH, Spengler JD, Ferris BG, Speizer FE. Concentration of indoor particulate matter as a determinants of respiratory health in children
. Am J Epidemiol 1994; 139:1088–1099.
71. Neas LM, Dockery DW, Speizer FE. Concentration of indoor particulate matter as a determinant of respiratory health in children
(Reply to letter). Am J Epidemiol 1995; 141:582.
72. Goren AI, Hellmann S. Respiratory conditions among schoolchildren and their relationship to environmental tobacco smoke and other combustion products. Arch Environ Health. 1995; 50:112–118.
73. Dales RE, Burnett R, Zwanenburg H. Adverse health effects among adults exposed to home dampness and molds. Am Rev Respir Dis 1991; 143:505–509.
74. Fergusson DM, Horwood LJ. Parental smoking and respiratory illness during early childhood: a six-year longitudinal study. Pediatr Pulmonol 1985; 1:99–106.
75. Timonen KL, Pekkanen J, Korppi M, Vahteristo M, Salonen RO. Prevalence and characteristics of children
with chronic respiratory symptoms
in eastern Finland. Eur Respir J 1995; 8:1155–1160.