Click on the links below to access all the ArticlePlus for this article.
Please note that ArticlePlus files may launch a viewer application outside of your web browser.
Research during the past decades provided strong evidence for the effects of particulate air pollution on mortality and morbidity. Adverse effects on health are not limited to high concentrations, but have been observed at surprisingly low and relatively common concentrations. The most recent evidence on the adverse health effects of particulate matter (PM) and nitrogen dioxide (NO2) was summarized by a World Health Organization (WHO) working group,1 the U.S. Environmental Protection Agency,2 and a WHO meta-analysis.3
Health effects are related to short-term exposure as well as to long-term exposure.1,4–7 However, research during the past decades has primarily focused on short-term effects, often in time-series studies. Two important limitations of time-series studies are that they consider only deaths within a few days8 up to a few months9 after exposure and that they do not account for long-term cumulative exposure.8 Therefore, the need for more epidemiologic studies evaluating potential long-term health effects of exposure to ambient air pollution has been expressed.10
Our knowledge with regard to long-term effects of air pollution on mortality in adults is limited to 4 American,11–14 one Canadian,15 and 2 European16,17 cohort studies. Dockery et al13 and Pope et al12 have found that long-term exposure to fine particulate matter (PM2.5) is associated with an increased mortality, in particular mortality due to cardiopulmonary causes. Filleul et al17 reported associations of cardiopulmonary mortality with long-term exposure to total suspended particulate matter and NO2 for a subset of the study areas. Hoek et al16 and Finkelstein et al15,18 focused on the impact of motor traffic as a major source of outdoor air pollutants such as NO2 19 and suspended PM.20 They reported that living near major roads was associated with increased all-cause and cardiopulmonary mortality.
To further evaluate these associations, the North Rhine-Westphalia State Environment Agency (LUA NRW) initiated a cohort study based on a series of cross-sectional studies performed in the framework of the North Rhine-Westphalian environmental health monitoring program. In a cohort of approximately 4800 women, living in several areas in North Rhine-Westphalia, Germany, we assessed whether long-term exposure to air pollution originating from motorized traffic and industrial sources is associated with total and cause-specific mortality.
Study Population and Study Area
This study was designed as a follow up of a series of cross-sectional studies on the effect of air pollution on women's health performed in the 1980s and 1990s in North Rhine-Westphalia,21 including the Ruhr area, one of Europe's largest industrial areas. For each cross-sectional study, approximately 450 women in their mid-50s with German nationality from each of a number of industrial areas and 2 nonindustrial reference areas (Borken, Dülmen) were randomly selected and invited to participate. The overall response rate for these cross-sectional studies was 70%. Between January 2002 and May 2003, approximately 4800 women from 10 areas in 7 towns, whose addresses were still available, were successfully followed up for vital status and migration. Figure 1 shows timing and participant numbers for the cross-sectional studies along with the number of participants who were successfully followed up.
Vital Status and Causes of Death
Data on vital status and address history were obtained for 4752 of 4874 women (97%) by contacting the registration offices. For women who died in North Rhine-Westphalia (394 of the 399 deaths), data on cause of death were provided by the North Rhine-Westphalia state agency for data processing and statistics. For 4 women who died outside North Rhine-Westphalia, but within Germany, copies of the death certificates were obtained by contacting the municipalities. No cause of death was available for one woman who died outside Germany. Causes of death were coded according to the International Classification of Diseases, Ninth Revision (ICD-9). Total mortality and cause-specific mortality, including cardiovascular (ICD-9 codes 400–440), pulmonary (ICD-9 codes 460–519), cardiopulmonary (ICD-9 codes 400–440 or 460–519), lung cancer (ICD-9 code 162), and all other causes of mortality were the primary outcomes defined according to Hoek et al.16
Estimation of Exposure
Air pollution exposure was defined as the women's exposure to NO2 and to particles less than 10 μm in diameter (PM10) calculated from total suspended particulates (TSP) at their baseline address. We calculated 1-year average concentrations for the year of the baseline examination (which was from 1985 to 1994, see Fig. 1) and 5-year average concentrations for the year of the baseline examination and the preceding 4 years using continuous air pollution measurements (half-hourly measurements for NO2, 24-hour measurements for TSP) performed by the LUA NRW at centrally located air monitoring stations in the study areas. Nitrogen dioxide concentrations were measured by means of chemiluminescence; TSP levels were measured by beta absorption. PM10 was calculated as 0.71 × TSP. The conversion factor was calculated as the average PM10/TSP ratio from data from 7 monitoring sites in the study area, where PM10 and TSP were measured in parallel between 1998 and 2004. PM10/TSP ratios varied between monitoring sites with site-specific ratios ranging from 0.62 to 0.84.
A map showing the locations of the participants’ baseline addresses and the monitoring stations is available with the online version of this article. Air pollution levels measured in Borken (which was close and most comparable to Dülmen) were used for Dülmen, where no monitoring station was available. Air pollution measurements in Borken started in 1990 to 1991 for NO2 and TSP. Therefore, air pollution concentrations before these dates were imputed by adding an “average” difference in air pollution levels between 1990 to 1991 and the respective year to levels measured in 1990 to 1991. The “average” difference was estimated by a linear regression model with air pollution concentration as the dependent variable, year of measurement as the independent variable, and an autoregressive (AR) correlation between repeated measurements performed at the same measurement site using air pollution concentrations measured from 1981 to 2000 at 15 monitoring stations in the study area (n = 248 observations). The estimated average difference was 1.02 μg/m3 per year for NO2 and 1.36 μg/m3 per year for PM10. PM10 has been directly measured since 1998 at several monitoring stations in the study area.
We were able to geo-code 4615 (97%) of the participants’ baseline addresses. We then calculated proximity of homes to major roads, defined as roads with at least 10,000 cars/d, for 4230 (92% of the geo-coded addresses) through a geographic information system (GIS) using traffic count data for the year 2000 provided by the LUA NRW.
As part of the cross-sectional studies, all participants completed a self-administered questionnaire on education, current symptoms or diseases, medications, smoking, occupational exposures, and use of gas for cooking or water heating. In addition, 81% of the women underwent a physical examination that included measurements of height and weight, which were used to calculate body mass index. Confounder variables were selected a priori as the confounders considered in the American and the Dutch cohort studies.13,16,22 In addition, we checked for confounding by asthma and hypertension at baseline as additional risk factors for (cardiopulmonary) mortality. Because 92% of the women were between 53 and 55 years of age (with an overall range for age of 50–59 years) at baseline, no adjustment for age was done. Socioeconomic status was defined as the maximum educational level of the study participant or her partner classified into 3 categories: low (compulsory school only), medium (lower and upper secondary education), and high (postsecondary education, including university). Passive smoking was defined as environmental tobacco smoke exposure at home or workplace for nonsmoking women. Current smokers were classified into 3 categories according to the number of cigarettes smoked. Smokers with missing information on the number of cigarettes smoked were assigned to the highest category. Completely missing smoking information was defined as an extra category. Body mass index was classified as underweight/normal (<25 kg/m2), overweight (25–30 kg/m2), and obese (≥30 kg/m2) according to the WHO classification.23
Correlations are presented as Spearman rank-order correlation coefficients. Associations between mortality and exposure were analyzed using Cox's proportional hazards models with adjustment for potential confounders. For participants who died, time in the study was calculated as the difference between the date of the original cross-sectional study and date of death; for those who were alive at the end of follow up, time in the study was the difference between the start and the end of follow up; for those who moved during the follow up and were lost to follow up after moving (n = 5), time in the study was the difference between the start and the last date with known vital status and place of residence. For the latter 2 groups of participants, time in the study was regarded as censored. In the analysis of cause-specific mortality, time in the study for participants who died of causes other than those analyzed was also treated as censored. Results are presented as crude and adjusted relative risks (RRs) with 95% confidence intervals (CIs). Statistical analyses were done with the statistical analysis package SAS for Windows release 8.02 (SAS Institute, Cary, NC). Geographic information system calculations were done with Arcinfo (ESRI Inc., Redlands, CA).
Description of the Study Population and Exposure
Table 1 shows baseline characteristics of the study participants. At baseline, more than 80% of the women reported that they lived 5 years or longer at their current address. Survivors had a higher socioeconomic status than the deceased as well as a lower prevalence of smoking or exsmoking, bronchial asthma, hypertension, and obesity. Furthermore, their homes were further away from major roads. No difference was found between survivors and deceased with regard to occupational exposures. Median and maximum length of follow up were 12.9 and 18.2 years, respectively. Eighty percent of the women did not move throughout the follow-up period, and another 14% moved within their baseline place of residence. Eight percent of the cohort died during the follow up (Table 2); approximately one third died of cardiovascular causes. Pulmonary causes and lung cancer were rare. More than half of the deceased died of causes other than cardiovascular or lung cancer, approximately half of them from cancer other than lung cancer. There was a considerable range in the women's exposure to NO2 and PM10 (Table 3). NO2 and PM10 were strongly correlated (Spearman correlation coefficient r = 0.5 and 0.8 for 1- and 5-year averages). One-year averages were highly correlated with 5-year averages (r = 0.8 for NO2 and PM10).
Minimum and maximum NO2 and PM10 levels were the same for women with a low, medium, and high level of education, but median levels were lower for women with a high level of education compared with women with a medium or low level of education (38 vs 41 μg/m3 for NO2 and 43 vs 44 μg/m3 for PM10 for both groups). Women with a low level of education lived slightly less often in the proximity of roads compared with women with a medium or high level of education (8.1% vs 8.4% and 9.2%, respectively).
Associations Between Mortality and Exposure to Air Pollution
So far, too few women have died of pulmonary causes or lung cancer to obtain stable effect estimates. Therefore, the main outcomes (besides all-cause mortality) were cardiopulmonary mortality and noncardiopulmonary nonlung cancer mortality. Living within a 50-meter radius of a major road was associated with an increased risk of cardiopulmonary mortality (Table 4). The associations between distance of homes to roads and mortality decreased only slightly when adjusted for socioeconomic status and smoking. Elevated levels of NO2 (1- and 5-year averages) were associated with an increased risk of all-cause mortality and in particular with cardiopulmonary mortality. Furthermore, PM10 was associated with an increased all-cause mortality (5-year average) and mortality due to cardiopulmonary causes (1- and 5-year averages). Effects were somewhat stronger for 5-year averages compared with 1-year averages. After adjustment for confounders, and in particular for smoking, the relative risks became smaller but remained elevated. None of the exposures was associated with noncardiopulmonary nonlung cancer mortality. Additional adjustment for occupational exposures, bronchial asthma, chronic bronchitis, and hypertension did not alter the associations between mortality and exposure (data not shown). We also checked for confounding by body mass index in the subgroup of participants of whom this information was available; the effect estimates remained unchanged. Mutual adjustment of air pollution effects for proximity to roads and vice versa did not change the air pollution effects and did not change the effects of distance to roads (data not shown).
In 1987, 1993, and 1994, cross-sectional studies were performed in other places in addition to Borken (Dormagen, Düsseldorf, Hürth, Köln, Leverkusen, and Wuppertal). However, names and baseline addresses of the participants from these places were not available for the present follow-up study. Therefore, for these years, only women from Borken could be followed up for vital status and address history. We assessed the impact of the 1987, 1993, and 1994 Borken data on the overall results by excluding it from the analysis. The relative risks for all-cause mortality and cardiopulmonary mortality did not change substantially (adjusted RR [95% CI] = 1.35 [0.96–1.89] and 1.75 [1.05–2.91], respectively, for living near major roads; 1.22 [1.04–1.42] and 1.57 [1.19–2.07] for NO2 1-year averages; and 1.27 [1.04–1.55] and 1.78 [1.22–2.60] for NO2 5-year averages). In contrast, relative risks of cardiopulmonary mortality associated with PM10 exposure decreased for 1-year averages (1.23 [0.96–1.59]), but not for 5-year averages (1.49 [1.11–1.99]).
Mortality was associated with the presence of bronchial asthma and hypertension at baseline. It is probably related as well to other cardiopulmonary diseases about which no information is available. Selective participation of diseased and nondiseased women in the original cross-sectional studies might have biased the association between exposure to air pollutants and mortality in the present study. Because no information on selective participation is available, we approached this problem by excluding the first 10 years of the follow up, that is, by analyzing the association between air pollution and mortality in a cohort of 3897 women in their mid-60s. The underlying hypothesis is that disease at baseline creates associations between mortality and air pollution at shorter scales. The associations of proximity to roads and exposure to NO2 and PM10 with all-cause and cardiopulmonary mortality remained except for the association of proximity to roads with all-cause mortality. Relative risks increased slightly for 1-year averages of NO2 and all-cause mortality (1.23) and cardiopulmonary mortality (1.70); for 1-year averages of PM10 and all-cause (1.35) and cardiopulmonary mortality (1.72); and for living near major roads and cardiopulmonary mortality (1.80). This indicates that there is little impact of the participant's health status at baseline on the association between air pollution and mortality in the present study.
Effect Modification by Socioeconomic Status and Smoking
We assessed potential modification of traffic and air pollution effects on mortality by socioeconomic status and smoking status. Exsmokers were excluded from this analysis due to the small numbers. We found modestly stronger NO2 and PM10 effects on all-cause mortality for women with a low socioeconomic status, for current smokers, and for nonsmokers with exposure to environmental tobacco smoke (data not shown).
Our study provides evidence that living within a 50-meter buffer of a major road and exposure to elevated concentrations of NO2 and of PM10 are associated with an increased risk of death from cardiopulmonary causes.
Hoek et al16 and Finkelstein et al15,18 found living near a major road to be associated with an increased all-cause and cardiopulmonary mortality. Despite differences in GIS data and definition of “major roads” between the Dutch and Canadian studies and the present study, our study confirms the Dutch and Canadian results; relative risks for the association of all-cause mortality and distance of home to a major road ≤50 meters were very similar in the 3 countries (RR = 1.29 in Germany vs 1.34 and 1.18 in The Netherlands and Canada). For mortality due to cardiopulmonary causes, the effect in the present study was smaller than in the Dutch study and stronger than in the Canadian study (RR = 1.63 vs 1.95 and 1.38). Our results with regard to particulate air pollution are in agreement with the results from the American cohort studies12,13 and the Dutch cohort study.16 In these studies, like in ours, elevated ambient air pollution concentrations were associated with an increased all-cause mortality and cardiopulmonary mortality. When we adjusted air pollution effects for distance to major roads, we found air pollution effects and effects of distance to major roads to be stable. Although NO2 has been suggested to be an indicator for traffic-related pollution,1 the independence of the effect of proximity to roads from the effects of the other air pollutants in the present study seems plausible in that a considerable part of the air pollution in the Ruhr area in North Rhine-Westphalia during the 1980s and early 1990s originated from industrial sources (mainly from power plants, steel industry, and coal mining). Therefore, NO2 might originate from both motorized traffic and industry. Using only proximity to roads to assess exposure to NO2, other factors that influence exposure would be ignored.24
Furthermore, living in the vicinity of roads might represent a combination of factors such as noise and air pollution. It has been suggested that chronic exposure to noise might be a risk factor for cardiovascular diseases,25 and we cannot rule out the possibility that part of the observed effect is attributable to noise. Moreover, proximity to roads might be a proxy for low socioeconomic status, which was found to be associated with an increased mortality in the present study. However, proximity to roads was not related to socioeconomic status in the present study.
The comparison of relative risks in our study and the American, Dutch, and French cohort studies is complicated by the fact that different air pollutants were studied and different methods for exposure assessment were used. However, the relative risk associated with a 16-μg/m3 increase in NO2 exposure in the present study agrees with the corresponding relative risk reported for the Dutch16 and French17 cohorts (after exclusion of 6 monitors influenced by local traffic in the French study) for all-cause mortality (RR = 1.20 and 1.23, respectively), whereas it is notably higher for cardiopulmonary mortality (RR = 1.42 and 1.47, respectively). The American studies12,13 found stronger effects for PM2.5 compared with total suspended particulates and PM10. No data on PM2.5 concentrations were available for our study. To get a rough estimate, we used the conversion factors proposed for Austria26 and the United States27 ranging from 0.3 to 0.5 for PM2.5/TSP. This resulted in relative risks ranging from 1.18 to 1.32 and 1.82 to 2.71 per 10-μg/m3 PM2.5 exposure (1-year average) for all-cause and cardiopulmonary deaths, respectively; these RRs are somewhat higher than the values of 1.06 and 1.09 per 10-μg/m3 PM2.5 reported by Pope et al12 for the cohort of the American Cancer Society. Wichmann28 calculated from the relative risks presented by Pope et al12 that a reduction of PM2.5 concentrations in Germany by 3 μg/m3 would result in a decrease in all-cause mortality by 1.8%, which corresponds to a gain in life expectancy of approximately 2 months. This is comparable to the gain in life expectancy anticipated with the decline in emissions in Germany from 2000 to 2020 by Amann et al.29 The results of the present study indicate that the public health impact of air pollution may be even larger.
Because it is not feasible to measure personal exposure for large study populations, measurement data from existing air pollution monitoring networks are frequently used in epidemiologic studies. In 2 of the U.S. studies,12,13 exposure to air pollution was assigned on an area level using monitoring station data. Thus, variability of air pollution concentrations within each study area was not taken into account. In the third U.S. cohort,14 air pollution levels were interpolated from fixed network monitoring stations to zip code centroids. This accounts for small-scale variations of air pollution levels only if the monitoring network is dense enough. Hoek et al16 used GIS data to assess exposure to traffic-related air pollution on the personal level. In their study, distance to roads (as determined by GIS) was the strongest predictor of mortality. For the present study, we used a mixed approach defining exposure by means of monitoring station data (on the area level) and by means of GIS (on the individual level). The fact that mortality is associated with both air pollution concentrations measured at a central monitoring station and with proximity of homes to roads indicates that air pollution not only plays an important role on an area level, but also on a smaller scale.
To our knowledge, this is the first German cohort study to assess the long-term effects of air pollution on mortality. Only women in their mid-50s with German nationality were included to reduce the influence of potential confounding factors such as occupational exposures, smoking, biologic–physiological differences, and other lifestyle factors that are potentially related to nationality. Moreover, we combined exposure assessment on the area level (by means of monitoring site data) with individual assessment of exposure to traffic-related air pollutants (by means of GISs) to take into account within-study area variability of exposure. Furthermore, more than 80% of the women did not move during the follow-up period and more than 80% of the women reported at baseline that they lived at their current address for more than 5 years. Thus, other than changes due to temporal trends in air pollution levels, baseline exposure represents exposure throughout the follow up and to some extent exposure during even longer time periods for the majority of the women (although certainly not lifetime exposure).
Mortality rate was low compared with other studies due to the relatively young age of women at recruitment. Therefore, deaths from pulmonary causes and lung cancer could not be analyzed separately. An extension of the follow-up period will be needed to have the power to assess certain subgroups. The pollution concentrations in our analysis represent exposures experienced by the study participants only at the beginning of the follow up. The air pollution effects on mortality, however, might reflect the cumulative effect of long-term exposure.
The study population consists of women living in industrial and nonindustrial cities in the Ruhr area. Cities were chosen to represent a range of air pollution concentrations. One potential limitation of this design might be that women from industrial places differ from women living in nonindustrial places not only in terms of air pollution concentrations (which is the case for NO2 and proximity to roads in the present study, although not for PM10), but also in terms of socioeconomic and lifestyle factors associated with mortality. There was no strong association between exposure to air pollutants and socioeconomic status in the present study, which is in agreement with another study conducted in Germany and The Netherlands.30 We controlled for confounding by socioeconomic factors by adjustment for education and a number of occupational exposures, but we cannot rule out residual confounding by other unmeasured lifestyle factors. In conclusion, living close to major roads and chronic exposure to NO2 and PM10 may be associated with an increased mortality due to cardiopulmonary causes.
The authors thank Elke Link (IUF at the Heinrich-Heine-University of Düsseldorf) for her assistance with GIS data; Thomas Kuhlbusch (IUTA, Duisburg) for calculation of the conversion factor for PM10 to TSP; and Peter Bruckmann, Ulrich Pfeffer, and Alfred Doppelfeld (LUA-NRW, Essen) for providing air pollution monitoring data.
1. Health Aspects of Air Pollution With Particulate Matter, Ozone and Nitrogen Dioxide
. Report on a WHO working group. 5042688. Copenhagen: WHO Regional Office for Europe; 2003.
2. Air Quality Criteria for Particulate Matter
. Research Triangle Park, NC: US Environmental Protection Agency; 2004.
3. Anderson HR, et al. Meta-analysis of Time-Series Studies and Panel Studies of Particulate Matter (PM) and Ozone (O3)
. Report of a WHO task group. Geneva: World Health Organization; 2004.
4. Air Quality Criteria for Particulate Matter
(Fourth External Review Draft). Research Triangle Park, NC: US Environmental Protection Agency; 2003.
5. Brunekreef B, Holgate ST. Air pollution and health. Lancet
6. Schwartz J. Air pollution and children's health. Pediatrics
7. Katsouyanni K. Ambient air pollution and health. Br Med Bull
8. Kunzli N. The public health relevance of air pollution abatement. Eur Respir J
9. Schwartz J. Harvesting and long term exposure effects in the relation between air pollution and mortality. Am J Epidemiol
10. Kunzli N, Tager IB. Long-term health effects of particulate and other ambient air pollution: research can progress faster if we want it to. Environ Health Perspect
11. Lipfert FW, Perry HM Jr, Miller JP, et al. The Washington University-EPRI Veterans’ Cohort Mortality Study: preliminary results. Inhal Toxicol
. 2000;12(suppl 4):41–73.
12. Pope CAI, Burnett RT, Thun MJ, et al. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA
13. Dockery DW, Pope CA III, Xu X, et al. An association between air pollution and mortality in six US cities. N Engl J Med
14. Abbey DE, Nishino N, McDonnell WF, et al. Long-term inhalable particles and other air pollutants related to mortality in nonsmokers. Am J Respir Crit Care Med
15. Finkelstein MM, Jerrett M, Sears MR. Traffic air pollution and mortality rate advancement periods. Am J Epidemiol
16. Hoek G, Brunekreef B, Goldbohm S, et al. Association between mortality and indicators of traffic-related air pollution in The Netherlands: a cohort study. Lancet
17. Filleul L, Rondeau V, Vandentorren S, et al. Twenty five year mortality and air pollution: results from the French PAARC survey. Occup Environ Med
18. Finkelstein MM, Jerrett M, Sears MR. Environmental inequality and circulatory disease mortality gradients. J Epidemiol Community Health
19. Briggs D, Collins S, Elliott P, et al. Mapping urban air pollution using GIS: a regression-based approach. Int J Geographical Information Science
20. Brauer M, Hoek G, van Vliet P, et al. Estimating long-term average particulate air pollution concentrations: application of traffic indicators and geographic information systems. Epidemiology
21. Dolgner R, Krämer U. Wirkungskatasteruntersuchungen. In: Wichmann H-E, Schlipköter H-W, Fülgraff G, eds. Handbuch der Umweltmedizin
. Landsberg: Ecomed-Verlag; 1993.
22. Pope CA III, Thun MJ, Namboodiri MM, et al. Particulate air pollution as a predictor of mortality in a prospective study of US adults. Am J Respir Crit Care Med
23. World Health Organization. Obesity: Preventing and Managing the Global Epidemic—Report of a WHO Consultation on Obesity, 3–5 June 1997, Geneva
. WHO Technical Report Series Report No. 894. The Stationary Office Books; 2001.
24. Ross Z, English PB, Scalf R, et al. Nitrogen dioxide prediction in Southern California using land use regression modeling: potential for environmental health analyses. J Expo Anal Environ Epidemiol
25. Ising H, Babisch W, Kruppa B. Noise-induced endocrine effects and cardiovascular risk. Noise Health
26. Gomiscek B, Hauck H, Stopper S, et al. Spatial and temporal variations of PM1, PM2. 5, PM10 and particle number concentration during the AUPHEP-project. Atmos Environ
27. Lall R, Kendall M, Ito Km et al. Estimation of historical annual PM2. 5 exposures for health effects assessment. Atmos Environ
28. Wichmann H-E. Positive gesundheitliche Auswirkungen des Einsatzes von Partikelfiltern bei Dieselfahrzeugen—Risikoabschätzungen für die Mortalität in Deutschland. Umweltmed Forsch Prax
30. Heinrich J, Gehring U, Cyrys J, et al. Exposure to traffic related air pollutants: self reported traffic intensity versus GIS modelled exposure. Occup Environ Med