Röösli, Martin PhD; Künzli, Nino MD, PhD; Schindler, Christian PhD; Theis, Gaston PhD; Oglesby, Lucy PhD; Mathys, Patrick PhD; Camenzind, Markus; Braun-Fahrländer, Charlotte Prof. MD
Although outdoor air has been shown to be contaminated by carcinogens, the extent of cancer attributable to ambient air pollution has been subject to persistent controversy in the last decades. 1 As the general population is exposed to ambient air pollution, the contribution of outdoor air pollution to cancer, especially lung cancer, is of public health concern. Therefore, risk assessment is needed for guiding regulatory decisions.
Cancer risk attributable to air pollution can be assessed by combining the distribution of exposure in the population of interest with a potency factor that describes the increase in risk per unit increase in exposure. 2 In the last years the use of “unit risks” as potency factors has been established in cancer risk assessment studies. 3–9 The unit risk factor describes the cancer risk associated with lifelong exposure to 1 μg/m3 of the substance of interest, assuming a linear dose–response function without a threshold value. 10 Unit risk factors for a variety of substances were evaluated and published by the International Agency for Research on Cancer (IARC), the United States Environmental Protection Agency (USEPA), the Office of Environmental Health Hazard Assessment, California (OEEHA), and the Länderausschuss für Immissionsschutz, Nordrhein-Westfalen, Germany (LAI). 11–14 The quantitative evaluation of the unit risk factors has put strong emphasis on animal experimental data, even though the extrapolation of animal bioassay data to the general population is known to be afflicted with large uncertainties. Risk estimates from varying extrapolation models can differ by five orders of magnitude in the low exposure range for the same data. 15 The magnitude of uncertainty in risk assessment is therefore likely to be greater when using animal data as compared to epidemiologic study results. 16–18
A further problem of unit risk-based risk assessment is that it quantifies cancer attributable to air pollution by the sum of the risks of each single carcinogen. However, it cannot be verified whether all relevant carcinogens are included in a study; moreover, potential interactions between pollutants in the outdoor air mixture are not taken into account. There is substantial evidence suggesting that environmental carcinogens interact synergistically in causing cancer. Synergistic effects are well documented for smoking combined with asbestos, 19–21 radon, 22–24 and alcohol. 25,26 However, different compounds may also interact antagonistically.
In the context of complex mixtures risk assessments it has been stated that knowing the carcinogenic potency of one single compound is more of an academic than a practical importance;27 yet combined effects in complex mixtures of ambient air pollution have hardly been investigated. An assessment of potential health risks, associated with exposure to complex mixtures, requires more than an understanding and quantification of the effects of individual compounds contained in the mixture. 28,29 IARC describes this problem as follows: “Estimating the human cancer risks of exposure to complex mixtures presents formidable methodological problems. However, such exposures are thought to account for a large proportion of cancers, in particular because of widespread exposure to such mixtures within populations.”30
Using a proxy measure of total carcinogenic ambient air pollution instead of summing up single carcinogenic agent may be a way of implicitly including combined effects in the risk quantification. Some studies have recently quantified the health risk of air pollution taking particles with diameter < 10 μm (PM10) as a surrogate for all ambient air pollutants. 31–33 However, none of these studies focused on cancer risk from air pollution.
The excess of lung cancer incidence attributable to a 10 μg/m3 increase in the average PM10 level was estimated from published epidemiologic cohort studies. Based on these estimates, the annual number of lung cancer cases in the Cantons of Basel-Stadt and Basel-Landschaft (Switzerland) attributable to air pollution were calculated, taking into account the PM10 exposure distribution in this population. We then compared the results of this epidemiology-based method with estimates for the same population based on unit risk factors.
Epidemiology-Based Risk Assessment Method
All cohort studies published so far, having quantified the association between lung cancer mortality and particulate matter, were taken into account to derive the concentration–response estimate. These were the American Cancer Society Study, 34 the Six Cities Study, 35 and the Adventists Studies. 36 We also took into account more recently published analyses (Table 1). 37,38
For each study, we calculated the excess number of lung cancer cases (ER10) per 100,000 person years and per 10 μg/m3 increase of the average PM10 concentration from the reported relative risk (RR10) and the average lung cancer incidence in the study population (Iav). It has been shown that for small risks (RR <1.5), this can be done most accurately by using Equation 1 39: Thereby lung cancer incidence was assumed to be equal to the lung cancer mortality. A pooled value for the excess rate of all three studies was obtained using an inverse variance weighting. 40,41 The three study-specific ER10 were tested for homogeneity using Cochran’s Q statistic. 41 As the between-study variance was not significant (P = 0.66), we did not adjust for potential random study effects.
Long-term PM10 exposure of the 450,900 persons 42 living in the Basel area (Cantons of Basel-Stadt and Basel-Landschaft) was assessed by using a dispersion model, 43 which had been developed for a tri-national study. 31 The model was based on emission data, taking into account primary and secondary particulates and had been validated against measured PM10 concentrations. It calculates mean PM10 concentration per kilometer squared. It is described in detail elsewhere. 44 The population weighted average of PM10 in the study area was 25.0 μg/m3.
Calculation of the Attributable Number of Cases.
To quantify the number of lung cancer cases attributable to air pollution in the Basel area, we used a PM10 reference concentration of 7.5 μg/m3, thus, health effects below this level were not quantified. The same assumption was used in the tri-national study. 31 This assumption was applied because, first, PM10 concentrations below this reference level were not measured in the epidemiologic studies from which the exposure-response association was derived. Second, these studies have very limited power to assess the shape of the concentration–response function in the lower bound of the observed exposure. Third, this level may also include the natural background PM10 concentration. Furthermore, this assumption allowed us to compare the result of this cancer risk assessment with the results of the health risk assessment of the tri-national study.
The number of lung cancer cases attributable to air pollution (NER) derived by the epidemiology-based method was calculated with the following formula: where P>30 is the population size of persons older than 30 years (analogous to the age distribution in the cohorts from which the ER10 was derived), ER10 is the excess rate per 10 μg/m3PM10 increase and (Cav −Cref) is the average exposure of the population minus the reference concentration. Two third of the population in Basel was older than 30 years. 42
Risk assessment Method Based on Unit Risk Factors
Summary estimates of unit risks were obtained by calculating the geometrical mean of the published unit risks from IARC, USEPA, OEEHA, and LAI (Table 2). To estimate minimum and maximum cancer risks, additional calculations were performed using the lowest and highest published unit risk factors in Table 2 for each substance. For diesel exhaust, unit risk estimate based on animal studies (mainly rat studies) differed systematically from unit risk estimate from human studies. 45 Thus, the unit risk factor finally used to quantify the cancer risk of diesel exhaust was determined as geometrical mean value of an unit risk factor derived from rat studies and one derived from the occupational diesel exhaust studies. 45
By means of the PM10 dispersion model, our study population was classified into categories of low, average and high exposure. Persons living in areas with an average PM10 concentration of less than 20 μg/m3 were assumed to be exposed to a generally low amount of pollution (50,000 of 451,000 inhabitants). Persons living in areas with a PM10 level between 20 and 25 μg/m3 were defined as exposed to an average amount of pollution (160,000 inhabitants), and persons living in areas with a PM10 concentration above this range were defined as generally highly exposed (240,000 inhabitants). Between 1997 and 1999 a series of carcinogenic substances were measured repeatedly at 11 different sites in the Basel area, representing the three exposure categories. Concentrations of nickel, lead, arsenic and cadmium were chemically determined from PM10 filters using energy-dispersive x-ray fluorescence spectrometry (Ni, Pb), inductively coupled plasma atomic emission spectroscopy (Cd), and atom absorption spectroscopy (As), respectively. Various polycyclic aromatic compounds from PM10 filters (cyclopenta[c,d]pyrene, benzo[b]naphto[2,1-day]thiophene, benzo[a]anthroacene, benzo[k,b]fluoranthene, benzo[a]pyrene, benzo[e]pyrene, perylene, benzo[ghi]perylene, dibenz[a,h]anthracene, indeno[1,2,3-cd]pyrene, anthranthene and coronene) as well as benzene, 1,4-dichlorobenzene, trichloromethane, trichloroethene, and tetrachloroethene (sampled by passive samplers) were analyzed with gas chromatography and mass spectroscopy (GC/MS). 1,3-butadiene was sampled with a passive canister and analyzed by gas chromatography and mass spectroscopy. Diesel exhaust exposure was obtained by multiplying the elemental carbon concentration, quantified from PM10 filters using a thermographic method, with a factor of 1.8, reflecting the proportion of EC in the diesel exhaust. 46
The details of the sampling procedure is described elsewhere. 47,48 Based on these data, the average concentrations in the three exposure categories were estimated and the population weighted average of each carcinogen was calculated (Table 3).
Calculation of the Attributable Number of Cases.
The unit risk-based method assumes a linear cumulative association between exposure to carcinogenic substances and the health outcome, 10 similar to the epidemiology-based risk assessment. With the unit risk-based method, the annual number of cases attributable to air pollution (NUR) was obtained by multiplying the population size in the exposure category j (Pj) with the unit risk factor of each pollutant of interest i (URi) and the average exposure level of substance i in the exposure category j (Ci,j) (Equation 3). The total risk attributable to air pollution was then calculated by summing up the risks of all (n) carcinogenic pollutants. In contrast with the epidemiology-based method, reference concentrations were assumed to be zero. This is the usual procedure for an unit risk-based risk assessment, 10 because natural sources of these pollutants are negligible. Division by 70 yields the annual unit risk factor. If one is comparing unit risk based estimates (NUR) with excess risk based estimates (NER), one has to be aware of the fact that the former refers to all types of cancer whereas the latter refers only to lung cancer cases.
Figure 1 shows the study specific lung cancer excess rates per 10 μg/m3 increase in the average PM10 level (ER10) as well as the weighted average. Despite a considerably higher relative risk for lung cancer in the Adventists study compared with the Six-Cities and the American Cancer Society study, the absolute number of lung cancer cases (ER10) was similar because of the lower lung cancer incidence of the Adventists. The pooled value across the three epidemiologic studies was 11.5 cases per 100,000 person years and per 10 μg/m3 increase in PM10 (95% confidence interval: 5.9–17.0;Fig. 1).
Using the pooled excess rate and multiplying it with the PM10 exposure of the Basel population (Equation 2) resulted in 13.3 annual lung cancer cases per 100,000 inhabitants of the Basel area. The 95% confidence interval was 6.9 to 19.8 cases (Table 4). A minimum estimate, based on the American Cancer Society study, yielded 11.7 annual lung cancer cases per 100,000 persons and the maximum estimate, based on the Adventists study, resulted in 19.5 cases.
Unit Risk-Based Estimate
With the unit risk-based method, the summary risk of all carcinogens yielded 1.06 annual cancer cases per 100,000 persons attributable to air pollution. Confidence intervals cannot be estimated using unit risk factors. Minimum and maximum estimates were 0.45 and 2.8 cases, respectively. The highest cancer risk was found for diesel exhaust, followed by 1,3-butadiene, chromium (VI), polycyclic aromatic hydrocarbons (PAH), and benzene (Table 4). These five compounds accounted for 98% of the total cancer risk attributable to air pollution according to the unit risk-based approach. The minimum and maximum estimates reflect the range of unit risk factors published by different agencies. A particularly wide range was estimated for the cancer risk from diesel exhaust according to the used unit risk factor. A unit risk factor based on rat studies was considerably lower than based on human occupational studies.
By means of the epidemiology-based cancer risk assessment, we obtained a point estimate of 13.3 annual lung cancer cases attributable to air pollution per 100,000 persons living in the Basel area. The unit risk-based method yielded a point estimate of 1.1 for all types of cancer, a value an order of magnitude lower than that of the epidemiology-based method. Thus, the differences between the two approaches is substantial and one may ask for the reason of this discrepancy.
At present, the observed lung cancer incidence in the Basel area is 50 annual cases per 100,000 persons (Torhorst J., Cancer registry of Basel, Institute for Pathology, 1999). Compared with this value the epidemiology based estimate is rather high where as the unit risk based estimate is quite low. However, none of the estimates can be ruled out with certainty. An estimate of the total annual mortality cases (all causes) attributable to air pollution in the Basel area, based on the method used in the tri-national study, 31 yielded 59 cases of death per 100,000 persons, 49 and this is not inconsistent with either estimate.
Several assessments of cancer risk from air pollution, based on unit risk factors, have been published so far. They are in line with our unit risk-based result: a study in Switzerland 7 yielded 0.85 (0.21–4.1) annual cancer cases per 100,000 persons. Diesel exhaust was estimated to cause 60% of the total risk attributable to air pollution. An unit risk-based assessment in Germany 3 found a cancer risk of 1.14 excess cases per 100,000 person years in urban environments and 0.21/100,000 in rural areas, mostly caused by diesel exhaust. Furthermore, in Sweden particulate organic material was estimated to cause 0.7 cancer cases per 100,000 person years, and exposure to 1,3-butadiene was associated with 0.31 cases per 100,000 person years. 4 A recent unit risk-based assessment in California 9 yielded an average of 0.43 cancer cases per 100,000 person years attributable to air pollution, with particulate organic material and 1,3-butadiene providing the largest contribution.
However, some approaches to assess cancer risk from air pollution resulted in higher estimates and are rather in line with our epidemiology-based risk assessment. Recently, a Swedish case-control study concluded that the proportion of lung cancer cases attributable to traffic-related air pollution in Stockholm can be as high as 10%. 50 Ecologic studies evaluating smoking-adjusted lung cancer risks between urban and rural sites generally found a difference of 50%. 5,51 This would correspond to 15 excess lung cancer cases per 100,000 person years in urban areas. 5 However, it is not clear whether this number can be fully attributed to air pollution, as control of confounding is problematic in such studies.
Obviously these higher cancer risk estimates are based on human data. Thus, it is striking that our epidemiology-based approach, using data from human cohorts, provides also a substantially higher point estimate than the unit risk estimate mainly based on toxicological data. Moreover, the same picture was observed in the unit risk factors of diesel exhaust. Unit risk factors based on human occupational studies 52–55 are considerably higher than unit risk factors based on rat studies. 45 In fact, our unit risk-based cancer estimate would be very similar to the epidemiology-based estimate, if only human studies had been considered to derive the unit risk factor of diesel exhaust. An upper bound estimate of the cancer risk from diesel exhaust using a unit risk factor based on human occupational studies yielded 13 annual cases per 100,00 person year.
Likewise, an estimate based on human data revealed a higher risk than one based on toxicological data in a risk assessment study of cancer mortality from smoking. 56 Thus, it is conceivable that the difference between impact assessment based on human and animal data may reflect conceptual differences between toxicology and epidemiology.
One reason that could explain an underestimation of the toxicology-based unit risk assessment is the possibility that an important carcinogen was not included in the unit risk-based assessment. However, this seems unlikely as none of the previous studies has identified such a strong additional carcinogenic air pollutant.
A further explanation could be that a large part of the carcinogenic potency from air pollution is not caused by single pollutants but by interaction processes of various components in the ambient air pollution mixture. Therefore, the total effect of all components influencing different organ systems in humans might be larger than the sum of the individual effects of each component. This would imply that it is not possible to quantify the carcinogenicity of air pollution by simply adding up the carcinogenic potencies of each pollutant as done in the unit risk method. Thus, only studies on human populations under real exposure conditions could adequately assess the cancer risk of the general population attributable to ambient air pollution. In such studies (eg, Adventists, the Six-Cities, and the American Cancer Society Study as well as the occupational diesel exhaust studies) PM10 or diesel exhaust may then be interpreted as a proxy of the total cancerogenity of ambient air pollution.
However, other arguments may suggest an overestimation of the epidemiology based assessment. The key assumption of the epidemiology-based method (ie, that PM10 can be used as a proxy for the total carcinogenic effect of ambient air pollution) cannot be adequately proven at present. Generally, site to site correlation between various pollutants is relatively high. However, it might be problematic to extrapolate results of long term effect studies from the US to European populations, as there may be systematic differences in the composition of the ambient air pollution mixture between the United States and Europe. However, according to recent time series studies of air pollution and health there is no evidence of specific discrepancies between European and US studies, although there is some heterogeneity, both within the United States and Europe. 57,58 The first European cohort study has, so far, not enough power to assess the association of air pollution long-term exposure with lung cancer incidence. 59 However, the study confirms the US findings for total mortality and air pollution.
The Swedish case-control study 50 qualitatively confirms the US findings for lung cancer. However, the substantially different approach limits the use of this study to quantitatively validate our US-based results.
An overestimation of the epidemiology-based assessment could also be the result of residual confounding in the cohort studies. Although the important confounding factors were taken into account, it cannot be excluded that the reported relative risks are overestimating the true exposure–response association as a result of residual confounding. In particular, as the majority of lung cancer cases is caused by smoking, it would appear that even a relatively small error in controlling for the effect of smoking could lead to the apparent excess of lung cancer seen in the epidemiological analyses. However, one has also to acknowledge that many error patterns in air pollution epidemiology may lead to underestimation of effects, thus, our epidemiology based estimate can not be taken as the upper bound of the unknown value.
The observed heterogeneity of effects across socioeconomic subgroups is a further source of uncertainty in the application of the US estimates to another population. 37 The underlying reason of this heterogeneity is not yet understood, thus, the use of these estimates in our population is a further source of errors, which may go in either direction. It also highlights the limitations of toxicology-based assessments, which are unable to address differences in risk gradients across subgroups, which are based on complex definitions such as socioeconomic factors.
The estimates of the attributable cases in each study were similar, despite large differences in the relative risk. As shown, this may be due to differences in the baseline diseases frequency in different populations, supporting our approach to pool the estimates of the excess risk from each study rather then pooling the relative risks. 39
Furthermore, one may argue that cohort studies overestimate the exposure-response function, if exposure assignment is based on current rather than historic pollution levels. For pollutants with steadily decreasing concentrations this may play a role. Further studies on exposure trends, latency periods and effects are needed, particularly in Europe.
Every approach to assess cancer risk in a given population has its inherent limitations. The presented epidemiology-based method may be an integrative way to deal with combined effects of an air pollutant mixture, which is often considered a key issue in risk assessment. However, it remains still unclear why risk assessment studies based on human data provide systematically higher cancer risks than approaches based on animal data. In light of the precautionary principle, given the uncertainty, epidemiology based impact assessment may provide important estimates.
This study was funded by the foundation MGU (Man - Society - Environment) of the University of Basel and by the Swiss Federal Agency of Environment, Forest and Landscape, Bern.
Martin Röösli has received support for grants and research from Novartis Foundation, the University of Basel, and the Swiss Federal Agency of Environment, Forest, and Landscape.
1. Cohen AJ. Outdoor air pollution and lung cancer. Environ Health Perspect. 2000; 108: 743–750.
2. Hertz-Picciotto I. Environmental risk assessment. In: Talbott EO, Craun GF, eds. Introduction to Environmental Epidemiology. Boca Raton, FL: Lewis Publishers, CRC Press, Inc.; 1995.
3. Kappos AD, Schmitt OA. Das Konzept des Länderausschusses fur Immissionschutz zur Regulierung kanzerogener Luftschadstoffe in der Aussenluft. Gesundheitswesen 1993; 55: 40–45.
4. Törnqvist M, Ehrenberg L. On cancer risk-estimation of urban air-pollution. Environ Health Perspect. 1994; 102: 173–182.
5. Hemminki K, Pershagen G. Cancer risk of air pollution: epidemiological evidence. Environ. Health Perspect. 1994; 102: 187–192.
6. French C, Peters W, Maxwell B, et al. Assessment of health risks due to hazardous air pollutant emissions from electric utilities. Drug Chem Toxicol. 1997; 204: 375–386.
7. Yetergil D. Externe Kosten von Krebserkrankungen durch kanzerogene Luftschadstoffe. Reihe 15. Düsseldorf: VD; 1998.
8. Woodruff TJ, Axelrad DA, Caldwell J, Morello-Frosch R, Rosenbaum A. Public health implications of 1990 air toxics concentrations across the United States. Environ Health Perspect. 1998; 1065: 245–251.
9. Morello-Frosch RA, Woodruff TJ, Axelrad DA, Caldwell JC. Air toxics and health risks in California: the public health implications of outdoor concentrations. Risk Analysis. 2000; 20: 273–291.
10. WHO. Air Quality Guidelines for Europe. WHO Regional Publications, European Series No. 23. Copenhagen: World Health Organization, Regional Office for Europe; 1987.
11. OEEHA. Technical support document for describing available cancer potency factors. California: Office of Environmental Health Hazard Assessment, Sacramento, CA; 1999.
12. USEPA. List of Substances on IRIS (Integrated Risk Information System)p 1999.
13. WHO. Air Quality Guidelines for Europe. 2nd ed. WHO Regional Publications. Copenhagen: World Health Organization, Regional Office for Europep 2001.
14. LAI. Krebsrisiko durch Luftverunreinigungen: Entwicklung von Beurteilungsmassstäben für kanzerogene Luftverunreinigungen im Auftrage der Umweltministerkonferenz. Düsseldorf: Länderausschuss für Immissionsschutz. Ministerium für Umwelt, Raumordnung und Landwirtschaft des Landes Nordrhein-Westfalen; 1992.
15. Maynard RL, Cameron KM, Fielder R, McDonald A, Wadge A. Setting air quality standards for carcinogens: an alternative to mathematical quantitative risk assessment—Discussion paper. Regul Toxicol Pharmacol. 1997; 26: S60–S70.
16. Hertz-Picciotto I. Epidemiology and quantitative risk assessment: a bridge from science to policy. Am J Public Health. 1995; 854: 484–493.
17. Smith AH. Epidemiologic input to environmental risk assessment. Arch Environ Health. 1988; 43: 124–127.
18. Smith AH. Epidemiology: bias everywhere? and not one drop of science? In: Graham JD, ed. The Role of Epidemiology in Regulatory Risk Assessment. Amsterdam: Elsevier. 1995; 39–42.
19. Vainio H, Boffetta P. Mechanisms of the combined effect of asbestos and smoking in the etiology of lung cancer. Scand J Work Environ Health. 1994; 20: 235–242.
20. Erren TC, Jacobsen M, Piekarski C. Synergy between asbestos and smoking on lung cancer risks. Epidemiology. 1999; 10: 405–411.
21. Albin M, Magnani C, Krstev S, Rapiti E, Shefer I. Asbestos and cancer: an overview of current trends in Europe. Environ Health Perspect. 1999; 107( Suppl 2): 289–98.
22. Hornung RW, Deddens J, Roscoe R. Modifiers of exposure-response estimates for lung cancer among miners exposed to radon progeny. Environ Health Perspect. 1995; 103: 49–53.
23. Hornung RW, Deddens JA, Roscoe RJ. Modifiers of lung cancer risk in uranium miners from the Colorado Plateau. Health Physics. 1998; 741: 12–21.
24. Thomas D, Pogoda J, Langholz B, Mack W. Temporal Modifiers of the Radon-Smoking Interaction. Health Physics. 1994; 663: 257–262.
25. Rothman K, Keller A. The effect of joint exposure to alcohol and tobacco on risk of cancer of the mouth and pharynx. J Chronic Dis. 1972; 25: 711–716.
26. Choi SY, Kahyo H. Effect of cigarette-smoking and alcohol-consumption in the etiology of cancer of the oral cavity, pharynx and larynx. Int J Epidemiol 1991; 204: 878–885.
27. Möller L, Schuetzle D, Autrup H. Future-research needs associated with the assessment of potential human health risks from exposure to toxic ambient air-pollutants. Environ Health Perspect. 1994; 102: 193–210.
28. Berenbaum MC. Consequences of synergy between environmental carcinogens. Environ Res. 1984; 38: 310–318.
29. Wahrendorf J. Use of quantitative epidemiologic data in regulatory approaches to air-pollution. Environ Health Perspect. 1994; 102: 183–185.
30. Vainio H, Sorsa M, McMichael AJ. Complex Mixtures and Cancer Risk. Lyon: International Agency for Research on Cancer; 1990.
31. Künzli N, Kaiser R, Medina S, et al. Public health impact of outdoor and traffic-related air pollution: a tri-national European assessment. Lancet. 2000; 356: 795–801.
32. Aunan K, Patzay G, Aaheim HA, Seip HM. Health and environmental benefits from air pollution reductions in Hungary. Sci Total Environ. 1998; 212: 245–268.
33. Zmirou D, Deloraine A, Balducci F, Boudet C, Dechenaux J. Health effects costs of particulate air pollution. J Occup Environ Med. 1999; 4110: 847–856.
34. Pope CA, 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. 1995; 1513: 669–674.
35. Dockery DW, Pope CA, Xu XP, et al. An association between air-pollution and mortality in 6 united-states cities. N Engl J Med. 1993; 32924: 1753–1759.
36. 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. 1999; 1592: 373–382.
37. Pope CA III, Burnett RT, Thun MJ, et al. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA. 2002; 2879: 1132–41.
38. Krewski D, Burnett RT, Goldberg MS, et al. Reanalysis of the Harvard Six Cities Study and American Cancer Society Study of Particulate Air Pollution and Mortality: Special Report. Cambridge, MA: Health Effect Institute. http://www.healtheffects.org/pubs-special.htm
39. Röösli M, Künzli N, Schindler C, Braun-Fahrländer C. Which Effect Measure Should Be Used for Impact Assessment in a New Population Context? J Human Ecol. Risk Assess. 2003; 9( 3): 709–719.
40. Petitti DB. Meta-Analysis, Decision Analysis, and Cost-Effectiveness Analysis. New York: Oxford University Press; 1994.
41. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986; 7: 177–188.
42. Bundesamt für Statistik. Annual Statistics Report for Switzerland (Statistisches Jahrbuch). Zürich: Verlag Neue Zürcher Zeitung (NZZ), 1997.
43. Heldstab J, Haan Pd, Künzle T. PM10-Modell für Kantone. Studie im Auftrag der Luftreinhaltefachstellen der Kantone ZH, LU und BL/BS. Zürich: INFRAS; 1999.
44. Filliger P, Puybonnieux-Texier V, Schneider J. PM10 Population Exposure. Technical Report on Air Pollution. Health Costs due to Road Traffic-related Air Pollution. An impact assessment project of Austria, France and Switzerland. Bern, Paris, Wien: Prepared for the Third WHO Ministerial Conference of Environment & Health, London. Bern: Federal Department of Environment, Transport, Energy and Communications, 1999; Internet: http://www.who.dk/london99/transport04.htm
45. OEEHA. Part B: Health risk assessment for diesel exhaust. California: Office of Environmental Health Hazard Assessment; 1998.
46. Pierson WR, Gorse RAJ, Szkarlat AC, et al. Mutagenicity and chemical characteristics of carbonaceous particulate matter from vehicles on the road. Environ Sci Technol. 1983; 17: 31–44.
47. Röösli M, Theis G, Künzli N, et al. Temporal and spatial variation of the chemical composition of PM10 at urban and rural sites in the Basel area, Switzerland. Atmospheric Environment 2001; 3521: 3701–3713.
48. Röösli M, Camenzind M, Theis G, Künzli N, Monn C, Braun-Fahrlähnder C. Gesundheitsrisiken durch Luftverunreinigungen in der Stadt Basel—Analyse der Immissionsmessungen. 1. Bericht der Studie BRISKA. Liestal: Lufthygieneamt beider Basel; 1999. http://www.unibas.ch/ispmbs/pdf/briska1.pdf
49. Röösli M, Camenzind M, Theis G, Künzli N, Monn C, Braun-Fahrlähnder C. Gesundheitsrisiko durch Luftschadstoffe in der Stadt Basel - Abschätzung der kanzerogenen und nicht-kanzerogenen Gesundheitsrisiken. 2. Bericht der Studie BRISKA. Basel: Institut für Sozial- und Präventivmedizin, 2000. http://www.unibas.ch/ispmbs/pdf/briska2.pdf
50. Nyberg F, Gustavsson P, Järup L, et al. Urban air pollution and lung cancer in Stockholm. Epidemiology. 2000; 115: 487–495.
51. Katsouyanni K, Pershagen G. Ambient air pollution exposure and cancer. Cancer Causes Control. 1997; 8: 284–291.
52. Steenland K, Deddens J, Stayner L. Diesel exhaust and lung cancer in the trucking industry: exposure-response analyses and risk assessment. Am J Indust Med. 1998; 343: 220–228.
53. Stayner L, Dankovic D, Smith R, Steenland K. Predicted lung cancer risk among miners exposed to diesel exhaust particles. Am J Indust Med. 1998; 343: 207–219.
54. Garshick E, Schenker MB, Munoz A, et al. A case-control study of lung-cancer and diesel exhaust exposure in railroad workers. Am Rev Respir Dis. 1987; 1356: 1242–1248.
55. Garshick E, Schenker MB, Munoz A, et al. A Retrospective Cohort Study of Lung-Cancer and Diesel Exhaust Exposure in Railroad Workers. Am Rev Respir Dis. 1988; 1374: 820–825.
56. Hertz-Picciotto I, Schulz M, Korte J, Ball L. Integration of epidemiology with toxicology for assessing low-dose cancer risks. Am J Epidemiol. 2000; 151: S26.
57. Katsouyanni K, Touloumi G, Samoli E, et al. Confounding and effect modification in the short-term effects of ambient particles on total mortality: results from 29 European cities within the APHEA2 project. Epidemiology. 2001; 125: 521–531.
58. Samet JM, Dominici F, Curriero FC, Coursac I, Zeger SL. Fine particulate air pollution and mortality in 20 US Cities, 1987–1994. N Engl J Med. 2000; 34324: 1742–1749.
59. Hoek G, Brunekreef B, Goldbohm S, Fischer P, van den Brandt PA. The association between mortality and indicators of traffic-related air pollution in a Dutch cohort study. Lancet. 2002; 360: 1203–1209.