a comprehensive review of european epidemiological studies on particulate … · 2017-01-23 · a...
TRANSCRIPT
report no. 4/11
I
A comprehensive review of European epidemiological studies on particulate matter exposure and health Prepared for CONCAWE’s Special Task Force on Health Effects of Air Pollution (H/STF-27) by:
E. Negri S. Gallus Department of Epidemiology, Mario Negri Institute, Milan, Italy
P. Boffetta International Agency for Research on Cancer, Lyon, France
J.K. McLaughlin International Epidemiology Institute, Rockville, MD, USA
C. La Vecchia Department of Epidemiology, Mario Negri Institute, Milan, Italy Institute of Medical Statistics and Biometry, University of Milan, Italy H/STF-27:
R. Taalman (Chairman) M. Banton J. Brown J.-P. Gennart D. King C. Money P. Roberts K. Rose E. Rushton G. Minsavage (Technical Coordinator) Reproduction permitted with due acknowledgement CONCAWE Brussels June 2011
report no. 4/11
II
ABSTRACT
In 2006/2007 CONCAWE’s Health Management Group (through H/STF-27) commissioned a literature survey and review with the primary aim of summarising the current state of science on particulate matter in ambient air and its possible effects on health. The survey and review were undertaken by the Department of Epidemiology, Mario Negri Institute, Milan, Italy1, the International Agency for Research on Cancer, Lyon, France2, the International Epidemiology Institute, Rockville, MD, USA3 and the Institute of Medical Statistics and Biometry, University of Milan, Italy4. The authors of the study are Eva Negri1, Silvano Gallus1, Paolo Boffetta2, Joseph K. McLaughlin3 and Carlo La Vecchia1,4
The body of this report consists of a brief overview of the findings written by H/STF-27, followed by the detailed information in Appendix developed by the mentioned experts.
KEYWORDS
Air pollution, particulate matter, PM10, PM2.5, epidemiology, cardiovascular disease, cancer
INTERNET
This report is available as an Adobe pdf file on the CONCAWE website (www.concawe.org).
NOTE Considerable efforts have been made to assure the accuracy and reliability of the information contained in this publication. However, neither CONCAWE nor any company participating in CONCAWE can accept liability for any loss, damage or injury whatsoever resulting from the use of this information. This report does not necessarily represent the views of any company participating in CONCAWE.
report no. 4/11
III
CONTENTS Page
SUMMARY IV
1. INTRODUCTION 1
2. EUROPEAN STUDIES ON SHORT-TERM PM EXPOSURE 2
3. EUROPEAN STUDIES ON LONG-TERM PM EXPOSURE 3
4. CONCLUSIONS 4 4.1. RECOMMENDATIONS FOR FUTURE RESEARCH ON
HEALTH EFFECTS OF PM IN EUROPE 4
5. GLOSSARY 5
6. REFERENCES 8
APPENDIX 1 DETAILED REPORT 9
EUROPEAN EPIDEMIOLOGICAL STUDIES ON PARTICULATE MATTER EXPOSURE AND HEALTH: A COMPREHENSIVE REVIEW OF THE LITERATURE
report no. 4/11
IV
SUMMARY
There are a limited number of papers on the long term effect of air pollution on morbidity and mortality in Europe, particularly with reference to small particles with aerodynamic diameters less than 2.5 microns (PM2.5). Most information comes from US cohort studies, including the American Cancer Society Cancer Prevention Study II, the Harvard Six Cities Study, the Adventists’ Health Study of Smog, and the Veterans’ Cohort Mortality Study.
Ambient levels of several relevant pollutants are more variable within Europe than in the USA, and are in several areas comparably high. Selected European cohort studies, including the Netherlands Cohort Study on Diet and Cancer and the European Prospective Investigation on Cancer and Nutrition study found some association between indicators of air pollution such as PM10 or NO2 and lung cancer risk, but the results were inconsistent and inadequate to address the health effects of exposure to PM2.5.
In addition to the effect on mortality, there are open issues on the potential impact of air pollution on childhood asthma, allergy and airway disease. In consideration of the difficulties in estimating the prevalence of the conditions in various populations, these issues require additional focus.
In order to provide an indication on possible further analyses of existing European datasets, and on future new studies, a critical review of existing literature (with a focus on European data) was performed.
The project resulted in a detailed report (see Appendix 1) and in a paper published in the European Journal of Cancer Prevention [1].
report no. 4/11
1
1. INTRODUCTION
A comprehensive review of European epidemiological studies published between 1996 and 2006 was conducted.
A search string was defined to select relevant scientific literature. The results of these searches were screened by two independent reviewers and only original studies conducted in Europe with some direct measure of ambient particulate matter (PM) exposure were selected for this review. Studies investigating the effect of long-term exposure on cardiovascular disease (CVD) were included. Cancer studies that provided proxies of PM exposure (e.g. distance of the residence from a major road) were also included. Approximately 150 studies met the inclusion criteria.
report no. 4/11
2
2. EUROPEAN STUDIES ON SHORT-TERM PM EXPOSURE
The outcomes considered were mortality (total and by cause), hospital admissions for cardiovascular and respiratory diseases, and physician visits for respiratory diseases.
Several studies have investigated the association between day-to-day variations in PM concentrations and mortality or emergency hospital admissions. Many of these studies, however, had limitations, either because they were based on small populations or had problems in the statistical approach used, which in several instances tended to artefactually maximise the investigated association. Different and non-comparable measures of PM (PM with diameter <10 µm (PM10), Black Smoke (BS), Total Suspended Particulate (TSP) etc.) were used in various studies.
Most of the European studies found a direct association between mortality and various indicators of PM levels. The multicentric APHEA2 study provided the best estimate of the relation between PM and day-to-day mortality in Europe.
The interpretation of studies on CVD admissions and PM exposure is complicated by the use of different lag times and the combination of emergency and elective hospital admissions (as well as other variations in the definition of outcome) in some of the studies. Despite these limitations, the evidence from European studies supports the hypothesis of an association between air pollution and hospital admissions for CVD. The results on respiratory admissions provided some support for an association between daily variations in air pollution and hospitalization for respiratory disease, although the specificity of the association with PM was not clarified.
Information on the separate effects of coarse, fine and ultrafine PM is limited. Given that more and more areas within Europe have, in recent years, started measuring fine and/or ultrafine PM, it is likely that in the next few years studies will investigate the effects of coarse, fine and ultrafine PM separately. Investigations aimed at separating the effects of PM of different diameters will have to deal with the thorny statistical and interpretational problems of the analysis of highly correlated variables, a problem which has also arisen in the investigation of several correlated pollutants, e.g. PM, NO2, CO and SO2.
Very few results are available from Central and Eastern European countries.
report no. 4/11
3
3. EUROPEAN STUDIES ON LONG-TERM PM EXPOSURE
Few European studies have investigated the long-term effects of PM exposure on health. Ecological studies are suggestive of a link and are generating hypotheses, but, given the potential biases inherent in their design, they cannot provide a solid base for inference.
Record linkage studies include a large number of cases, but the lack of individual variables, particularly smoking data, renders this type of study difficult to interpret.
Only a limited number of analytic studies have been conducted so far. Many of these were based on cohorts designed for other purposes but have exploited the availability of PM concentration data in the area under study. This has led to studies that were often based on small population size, often with exclusions of parts of the study population for which exposure data could not be retrieved or were not available.
In almost all long term studies conducted so far, the exposure to ambient PM has been estimated only on the base of the subjects’ home address. No additional information was available on characteristics of the home (e.g. floor of residence) or of the subject’s individual exposure pattern. Misclassification of exposure is likely to occur and to affect results.
Mortality was associated with various measures of long-term exposure to air pollution in cohort studies from the Netherlands, Germany and France, and in a record linkage study from Norway. The excesses were mainly attributed to cardiovascular and respiratory causes.
The evidence on long-term PM exposure and CVD morbidity is limited and difficult to interpret.
A few studies suggested associations between long-term PM exposure and asthma, lung function and various respiratory symptoms in children, although results were often not significant and not consistent. For adults, no clear evidence emerged for chronic bronchitis or asthma. Studies on air pollution and lung function, on the other hand, reported possible associations.
No clear indication of association emerged from the few studies investigating the relationship between long-term air pollution and lung or other cancers. Overall, the evidence is limited and inconsistent for this association. Limitations in the design of several studies conducted so far hinder any definite conclusion regarding cancer and long-term exposure to air pollution.
report no. 4/11
4
4. CONCLUSIONS
4.1. RECOMMENDATIONS FOR FUTURE RESEARCH ON HEALTH EFFECTS OF PM IN EUROPE
Previous studies on short-term effects of PM in Europe allow a general quantification of the risks. However, information on the possible confounding or modifying effect of covariables, and, in particular, data on course, fine and ultrafine particles are limited. Still, with the exception of Central and Eastern Europe where few data are available, further research on short-term effects of air pollution should not be given high priority.
Only a limited number of analytic studies have been conducted in Europe on long-term effects of air pollution, and these were mainly cohorts designed for different purposes. Their ability to provide valid results was limited by sub-optimal strategies for exposure assessment leading to misclassification, low statistical power, and heterogeneity of approaches. While it is possible that additional exposure data can be retrieved for these and other existing longitudinal studies, it is uncertain whether this approach would lead to major advances in our knowledge of health effects. In any case, priority should be given to the co-ordination and harmonization of exposure data in existing cohorts.
The establishment of new ad hoc studies represents the preferred strategy to provide new insights on long-term effects of air pollution. In areas where estimation of long-term PM exposure could be reconstructed, a case-control design should be considered. The design and implementation of new cohort studies would provide, in the long-term, the most reliable information, but would require great effort in terms of time and resources. Such new studies would rely on uniform methods to measure air pollutants, and be based on an increased number of new measuring stations. Participants should be periodically re-contacted and, in addition to information on possible outcomes, new exposure information based on a uniform methodology should be collected. Such new cohorts should give priority to the inclusion of populations from Central and Eastern Europe, and in particular, areas where air pollution might be higher and adverse health effects may be detected sooner than in less polluted areas of Europe.
report no. 4/11
5
5. GLOSSARY
ACS American Cancer Society
ACS-CPS II American Cancer Society – Cancer Prevention Study II
AHSMOG Adventists Health Study of Smog
AIC Akaike Information Criterion
AMI Acute Myocardial Infarction
APHEA Air Pollution and Heath: a European Approach
BMI Body Mass Index
BS Black Smoke
CI Confidence Interval
CO Carbon Oxide
COPD Chronic Obstructive Pulmonary Disease
CV Cardiovascular
CVD Cardiovascular Disease
ECRHS I European Community Respiratory Health Study I
EPA Environmental Protection Agency (US)
EPIC European Prospective Investigation on Cancer and Nutrition
FEF25-75 Forced Expiratory Flow between 25th and 75th percent of the FVC
FEV1 Forced Expiratory Volume in one second
FVC Forced Vital Capacity
GAM Generalized Additive Model
GLM Generalized Linear Model
HEAPSS Health Effects of Air Pollution among Susceptible Sub-population
HEI Health Effects Institute
HR Hazard Ratio
ICD International Classification of Diseases
IHD Ischemic Heart Disease
report no. 4/11
6
IQR Inter-Quartile Range
Lag x xth day Lag
Lag x-y average x-y days Lag
LCL Lower Confidence Limit
MC Mass Concentration
MEF25-75 Midexpiratory flow between 25 and 75th percent of the FVC
MeSH Medical Subject Headings
NC Number Concentration
NMMAPS National Morbidity, Mortality and Air Pollution Study
NO Nitrogen oxide
NO2 Nitrogen dioxide
NO3 Nitrate
NOx Oxides of nitrogen
NS Not Significant
OLS Ordinary Least Squares
OR Odds Ratio
O3 Ozone
PAARC Pollution Atmosphérique et Affections Respiratoires
PM Particulate Matter
PM10 PM with diameter <10 µm
PMx PM less than x μm in aerodynamic diameter
PNC Particle Number Concentration
RH Relative humidity
RR Relative Risk
SAPALDIA Study on Air Pollution and Lung Diseases in Adults
SES Socioeconomic status
SO2 Sulphur dioxide
report no. 4/11
7
SO4 Sulfate
TSP Total Suspended Particles
UCL Upper Confidence Limit
report no. 4/11
8
6. REFERENCES
1. Gallus, S. et al (2008) European studies on long-term exposure to ambient particulate matter and lung cancer. European Journal of Cancer Prevention 17, 191-194
report no. 4/11
9
APPENDIX 1 DETAILED REPORT
EUROPEAN EPIDEMIOLOGICAL STUDIES ON PARTICULATE MATTER EXPOSURE AND HEALTH: A COMPREHENSIVE REVIEW OF THE LITERATURE
Prepared by: Eva Negri1, Silvano Gallus1, Paolo Boffetta2, Joseph K. McLaughlin3, Carlo La Vecchia1,4
1 Department of Epidemiology, Mario Negri Institute, Milan, Italy 2 International Agency for Research on Cancer, Lyon, France 3 International Epidemiology Institute, Rockville, MD, USA 4 Institute of Medical Statistics and Biometry, University of Milan, Italy
This report was supported by an unrestricted grant of CONCAWE to the International Epidemiology Institute. The authors thank Dimitrios Trichopoulos, University of Athens and Harvard School of Public Health, for comments.
report no. 4/11
10
CONTENTS Page
SUMMARY 13
1. INTRODUCTION 16
2. METHODS 17 2.1. RETRIEVAL OF DATA 17 2.2. PRESENTATION OF FINDINGS 18 2.3. STUDIES CONSIDERED 19 2.4. KEY ABBREVIATIONS AND OTHER TERMS 19
3. SHORT-TERM PARTICULATE MATTER (PM) EXPOSURE AND MORTALITY 21 3.1. THE APHEA STUDIES 1 AND 2 21 3.2. LOCATION OF THE STUDIES 22 3.3. STUDY DESIGNS 22 3.4. INDEX OF EXPOSURE TO PARTICULATE MATTER 22 3.5. STATISTICAL MODELS USED 22 3.6. LAG TIME 24 3.7. CO-POLLUTANTS 26 3.8. CONFOUNDING/MODIFICATION EFFECT OF OTHER
ENVIRONMENTAL VARIABLES 29 3.9. CAUSE-SPECIFIC MORTALITY 29 3.10. MORTALITY DISPLACEMENT AND MEDIUM-TERM
EFFECTS OF PM 30 3.11. ULTRAFINE, FINE AND COARSE PM 32 3.12. SEX DIFFERENCES 35 3.13. AGE DIFFERENCES 36 3.14. SOCIOECONOMIC DIFFERENCES 38 3.15. STUDIES IN PATIENTS WITH SELECTED DISEASES (COPD) 39 3.16. REGIONAL DIFFERENCES 39 3.17. THRESHOLDS AND DOSE-RISK RELATION 40 3.18. COMPARISON OF EUROPEAN WITH LEADING NORTH
AMERICAN STUDIES 42 3.19. CONCLUSIONS 42
4. SHORT-TERM PM EXPOSURE AND CARDIOVASCULAR DISEASE (CVD) HOSPITAL ADMISSIONS 44 4.1. THE APHEA2 STUDY 44 4.2. THE HEAPSS STUDY 46 4.2.1. Analysis on the general population 46 4.2.2. Analysis on survivors of first AMI 50 4.3. NATIONAL MULTICITY STUDIES 51 4.4. STUDY DESIGN, STATISTICAL MODEL AND LAG TIME 52 4.5. INDEX OF EXPOSURE TO PM, FINE, ULTRAFINE AND
COARSE PM AND CO-POLLUTANTS 52 4.6. OUTCOME DEFINITION AND EFFECT FOR DIFFERENT
CAUSES OF HOSPITAL ADMISSION AND AGE GROUPS 53 4.7. CONCLUSIONS 54
5. SHORT-TERM PM EXPOSURE AND RESPIRATORY HOSPITAL ADMISSIONS 55 5.1. THE APHEA1 STUDY 55 5.2. THE APHEA2 STUDY 57 5.3. NATIONAL MULTICITY STUDIES 59
report no. 4/11
11
5.4. FINE AND COARSE PM 59 5.5. CONCLUSIONS 60
6. SHORT-TERM PM EXPOSURE AND OUTPATIENT VISITS 61
7. ANALYTIC STUDIES ON LONG-TERM PM EXPOSURE AND MORTALITY 62 7.1. STUDY DESIGNS, SAMPLE SIZE, TIME FRAME AND
POPULATION DEFINITION 62 7.2. ASSESSMENT OF EXPOSURE TO PM AND CO-
POLLUTANTS 62 7.3. STATISTICAL MODEL AND CONTROL OF CONFOUNDING 63 7.4. MAIN RESULTS 63 7.5. EFFECT MODIFICATION BY AGE, SEX, SOCIOECONOMIC
STATUS OR OTHER VARIABLES 68 7.6. COMPARISON OF EUROPEAN WITH LEADING NORTH
AMERICAN STUDIES 69 7.7. CONCLUSIONS 70
8. ECOLOGIC STUDIES ON LONG-TERM PM EXPOSURE AND MORTALITY 71
9. LONG-TERM PM EXPOSURE AND CARDIOVASCULAR MORBIDITY 74
10. LONG-TERM PM EXPOSURE AND RESPIRATORY MORBIDITY OR SYMPTOMS 76 10.1. LONG-TERM PM EXPOSURE AND ASTHMA IN CHILDREN 76 10.2. LONG-TERM PM EXPOSURE AND LUNG FUNCTION IN
CHILDREN 76 10.3. LONG-TERM PM EXPOSURE AND RESPIRATORY
SYMPTOMS IN CHILDREN 78 10.4. LONG-TERM PM EXPOSURE AND RESPIRATORY
MORBIDITY OR SYMPTOMS IN ADULTS 81 10.5. CONCLUSIONS 82
11. LONG-TERM PM EXPOSURE AND CANCER 83 11.1. ANALYTIC STUDIES ON LUNG CANCER 83 11.2. ECOLOGIC STUDIES ON LUNG AND OTHER CANCERS 86 11.3. STUDIES ON AIR POLLUTION AND CANCER IN
CHILDREN 86 11.4. COMPARISON OF EUROPEAN WITH LEADING NORTH
AMERICAN STUDIES 86 11.5. CONCLUSIONS 87
12. DISCUSSION AND RECOMMENDATIONS FOR FUTURE STUDIES 88 12.1. STUDIES ON SHORT-TERM EXPOSURE 88 12.2. STUDIES ON LONG-TERM EXPOSURE 89 12.3. RECOMMENDATIONS FOR FUTURE RESEARCH ON
HEALTH EFFECTS OF AIR POLLUTION IN EUROPE 90
13. REFERENCES 173
report no. 4/11
12
APPENDIX 1.1 SHORT-TERM PARTICULATE MATTER (PM) AND MORTALITY (DERIVED FROM EPA 1997) 186
APPENDIX 1.2 SHORT-TERM PARTICULATE MATTER (PM) AND RESPIRATORY MORBIDITY 188
APPENDIX 1.3 LONG-TERM PARTICULATE MATTER (PM) AND MORTALITY 189
APPENDIX 1.4 LONG-TERM PARTICULATE MATTER (PM) AND RESPIRATORY MORBIDITY 190
APPENDIX 1.5 LONG-TERM PARTICULATE MATTER (PM) AND LUNG CANCER 192
APPENDIX 1.6 REFERENCES TO APPENDICES 193
report no. 4/11
13
SUMMARY
The US Environmental Protection Agency (EPA) reviewed data from studies on particulate air pollution (particulate matter, PM) and health published between 1996 and 2002 (US EPA, 2004). Herein, we have conducted a comprehensive review of the European epidemiological studies published between 1996 and 2006.
By means of Medline literature searches, we identified a number of studies, which were then screened by two independent reviewers for inclusion. We selected original studies conducted in Europe with some direct measure of ambient particulate matter (PM) exposure. For studies investigating the effect of long-term exposure on cardiovascular disease (CVD) and cancer, we included also studies providing proxies of PM exposure (e.g. distance of the residence from a major road). Approximately, 150 studies met the inclusion criteria. Studies published before 1996, and included in the EPA report are presented in an appendix, but are not commented on in the text.
1) Studies on short-term PM exposure
The outcomes considered were mortality (total and by cause), hospital admissions for cardiovascular and respiratory diseases, and physician visits for respiratory diseases.
Several studies have investigated the association between day-to-day variations in PM concentrations and mortality or emergency hospital admissions. Many of these studies, however, had limitations, either because they were based on small populations or had problems in the statistical approach used, which in several instances tended to artefactually maximise the investigated association. Different and non-comparable measures of PM (PM with diameter <10 µm (PM10), Black Smoke (BS), Total Suspended Particulate (TSP) etc.) were used in various studies.
Most of the European studies found a direct association between mortality and various indicators of PM levels. The multicentric APHEA2 study provided the best estimate of the relation between PM and day-to-day mortality in Europe. The APHEA2 study was based on a large population derived from several different areas in Europe, and the study data have been thoroughly re-analyzed to evaluate the influence of different statistical models, different time lags (up to 40 days), and the modification effects of some city characteristics, such as temperature or humidity, but also the percentage of people aged >65 years and of smokers. In the APHEA2 study, the estimated increase in total number of deaths associated with an increase of 50 µg/m3 of PM10 (lag 0-1 days) ranged from 2.1% to 3.3%, according to the statistical model used. For an increase in BS of 50 µg/m3 the estimated increase ranged from 1.4% to 2.8%. The associations of PM indicators with mortality approximately halved after adjustment for NO2. Medium-term (40 days) analyses showed that the observed effect is likely not due to mortality displacement, i.e., a mere anticipation of time of deaths of a few days. On the contrary, medium-term effect estimates tended to be two times higher than short-term ones, indicating a possible underestimation of the global effect of PM when only a few days are considered. This problem appeared more pronounced for respiratory than for CVD deaths.
The interpretation of studies on CVD admissions and PM exposure is complicated by the use of different lag times and the combination of emergency and elective hospital admissions (as well as other variations in the definition of outcome) in some of the studies. Despite these limitations, the evidence from European studies supports the hypothesis of an association between air pollution and hospital
report no. 4/11
14
admissions for CVD. The results on respiratory admissions provided some support for an association between daily variations in air pollution and hospitalization for respiratory disease, although the specificity of the association with PM was not clarified.
Although age at death is, with sex, the only individual characteristic readily available from death certification data, the modifying effect of age on the relation between short-term PM exposure and mortality/morbidity has not been fully explored. The data from the APHEA2 study suggest that the effect of air pollution is greater at older ages.
Confounding by individual characteristics is not a major issue, since in time series comparisons are made from one day to another, using the same background populations. In the few studies that have analyzed data using a case-crossover design in addition to a time series approach, the results were comparable.
Limited or inadequate information is available in Europe on modification of the effects of air pollution exerted by individual variables (other than age and sex), e.g. socioeconomic status (SES), lifestyle habits (e.g. smoking), underlying medical conditions (e.g. COPD (chronic obstructive pulmonary disease), coronary disease, diabetes), occupational exposures, etc.
Information on the separate effects of coarse, fine and ultrafine PM is limited. Given that more and more areas within Europe have, in recent years, started measuring fine and/or ultrafine PM, it is likely that in the next several years studies will investigate the effects of coarse, fine and ultrafine PM separately. Investigations aimed at separating the effects of PM of different diameters will have to deal with the thorny statistical and interpretational problems of the analysis of highly correlated variables, a problem which has also arisen in the investigation of several correlated pollutants, e.g. PM, NO2, CO and SO2.
Presently, very few results are available from Central and Eastern European countries.
2) Studies on long-term exposure
Few European studies have investigated the long-term effects of PM exposure on health. Ecologic studies are suggestive and hypothesis generating, but, given the potential biases inherent in their design, they cannot provide a solid base for inference.
Record linkage studies include a large number of cases, but the lack of individual variables, particularly smoking data, renders this type of study difficult to interpret.
Only a limited number of analytic studies have been conducted so far. Many of these were based on cohorts designed for different purposes, which have exploited the availability of PM concentration data in the area under study. This has led to studies that were often based on small population size, often with exclusions of parts of the study population for which exposure data could not be retrieved or were not available.
In short-term time series, comparisons are ideally made between different days within individuals, and adjustment for day of week and holidays is often performed. In contrast, studies of long-term exposure are necessarily based on comparisons between individuals, and on cumulative exposure assessment. Misclassification of exposure is likely to occur and to affect results. In almost all studies conducted so
report no. 4/11
15
far, the exposure to ambient PM has been estimated only on the basis of the home address of the subjects. No additional information was available on characteristics of the home (e.g. type of dwelling) or of the subject’s individual exposure pattern.
Mortality was associated with various measures of long-term exposure to air pollution in cohort studies from the Netherlands, Germany and France, and in a record linkage study from Norway. The excesses were mainly attributed to cardiovascular and respiratory causes.
The evidence on long-term PM exposure and CVD morbidity is scant and difficult to interpret.
A few studies suggested associations between long-term PM exposure and asthma, lung function and various respiratory symptoms in children, although results were often not significant and not consistent. For adults, no clear evidence emerged for chronic bronchitis or asthma. Studies on air pollution and lung function, on the other hand, reported positive results.
No clear indication of association emerged from the few studies investigating the relation between long-term air pollution and lung or other cancers. Overall, the evidence is scant and inconsistent for this association. Limitations in the design of several studies conducted so far hinder any definite conclusion regarding cancer and long-term exposure to air pollution.
3) Recommendations for future research on health effects of air pollution in Europe
Previous studies on short-term effects of air pollution in Europe allow a general quantification of the risks. However, information on the possible confounding or modifying effect of covariates, and mainly data on coarse, fine and ultrafine particles are limited. Still, with the exception of Central and Eastern Europe, where few data are available, further research on short-term effects of air pollution should not be given high priority.
Only a limited number of analytic studies have been conducted in Europe on long-term effects of air pollution, and these were mainly cohorts designed for different purposes. Their ability to provide valid results was limited by sub-optimal strategies for exposure assessment leading to misclassification, low statistical power, and heterogeneity of approaches. While it is possible that additional exposure data can be retrieved for these and other existing longitudinal studies, it is uncertain whether this approach would lead to major advances in knowledge of health effects. In any case, priority should be given to the co-ordination and harmonization of exposure data in existing cohorts.
The establishment of ad hoc studies represents the preferred strategy to provide new insights on long-term effects of air pollution. In areas where estimation of long-term PM exposure could be reconstructed, the case-control design should be considered. The design and implementation of new cohort studies would provide, in the long-term, the most reliable information, but would require great effort in terms of time and resources. Such new studies would rely on uniform methods to measure air pollutants, and be based on an increased number of new measuring stations. Participants should be periodically re-contacted and in addition to information on possible outcomes, new exposure information based on a uniform methodology should be collected. Such new cohorts should give priority to the inclusion of populations from Central and Eastern Europe, where air pollution is higher and adverse health effects may be detected sooner than in less polluted areas of Europe.
report no. 4/11
16
1. INTRODUCTION
Although most epidemiological research on health effects of air pollution has focused on short-term exposure, a few studies have considered long-term exposure. The US Environmental Protection Agency (EPA) reviewed data of studies on particulate air pollution (particulate matter, PM) and health published between 1996 and 2002 (US EPA, 2004).
There is relatively little published on the long-term effect of air pollution on morbidity and mortality in Europe, particularly with reference to small aerodynamic particles with diameters less than 2.5 microns (PM2.5). Most information comes from US cohort studies, including the American Cancer Society - Cancer Prevention Study II (ACS-CPS II), the Harvard Six Cities Study, the Adventists Health Study of Smog (AHSMOG), and the Veterans’ Cohort Mortality Study.
Ambient levels of several relevant pollutants are more variable within Europe than in the USA, and are, in several areas, comparably high. Selected European cohort studies, including the Netherlands Cohort Study on Diet and Cancer and the European Prospective Investigation on Cancer and Nutrition (EPIC) study found some association between indicators of air pollution such as PM10 or NO2 and lung cancer risk, but the results were inconsistent and inadequate to address the health effects of exposure to PM2.5.
In addition to the effect on mortality, there are open issues on the potential impact of air pollution on childhood asthma, allergy and airway disease. In consideration of the difficulties in estimating the prevalence of these conditions in various populations, these issues require additional focus.
In order to provide indication on possible further analyses of existing European datasets, and on future new studies, we performed a critical review of existing literature (with focus on European data). This leads to recommendations on the design of future epidemiological studies on health effects of air pollution in Europe.
We have therefore updated the literature from Europe up to 2006, and conducted a comprehensive review of the European epidemiological studies published between 1996 and 2006.
Most of the original studies considered the association between short-term exposure to ambient pollutants (including PM) and morbidity/mortality. Several components and measures of air pollution were analysed with reference to incidence or mortality from cardiovascular disease (CVD), hospital admissions for respiratory disease, incidence of asthma and other respiratory conditions, and decreased pulmonary function. However, the results were not consistent across studies, and precise quantification of risks remains unattainable.
Only a few studies provided data on long-term exposure to PM and morbidity/mortality in adults.
The main results of these studies are summarised in the present report.
report no. 4/11
17
2. METHODS
2.1. RETRIEVAL OF DATA
We retrieved from PubMed the abstracts of all the journal articles on European epidemiologic studies (96 papers) included in the EPA report. We identified the Medical Subject Headings (MeSH) terms common to all the 96 papers, and defined a search string to select the scientific literature for the period 2002-2006 as follows:
("air pollutants" [MeSH] OR "air pollution" [MeSH]) AND humans [MeSH] AND ("cardiovascular diseases" [MeSH] OR neoplasms [MeSH] OR "respiratory tract diseases" [MeSH] OR hospitalization [MeSH]).
Using PubMed, on November 1, 2006 we retrieved 3,168 papers, which were then included in an electronic database using Endnote (version 9).
On the basis of title, abstract (when available), and keywords (MeSH terms), two independent reviewers (Eva Negri and Silvano Gallus) independently classified all the 3,168 papers, in order to identify studies of high interest using a 1-5 score:
1. Paper not pertinent or of limited interest (eliminate) 2. Probably not pertinent or of limited interest (eliminate if the other reviewer
also scores 1 or 2) 3. Not possible to evaluate on the basis of title/abstract/keywords only 4. Of high interest (important review or methodologic papers or original studies
out of Europe) 5. Original study from Europe
From the first selection, 2,383 papers were eliminated (both reviewers’ score <3).
Among the other 785 papers, approximately 100 papers were scored 5 by both reviewers. To these 785 papers we added 532 papers not yet “indexed for MEDLINE” (“in process”, therefore without MeSH terms), using on November 25, 2006, the following search string:
(air [tiab] OR pollut* [tiab] OR "particulate matter" [tiab] OR ambient [tiab]) AND (particulate* [tiab] OR PM [tiab] OR PM10 [tiab] OR PM2.5 [tiab] OR PM5 [tiab]) NOT(medline[sb]).
On the basis of title, abstract (when available), and keywords (MeSH terms), Eva Negri and Silvano Gallus performed a second independent double classification of the 1,317 papers using a score from 1 to 10:
1. Paper of limited interest (eliminate) 2. Methodologic study of limited interest 3. Original study not conducted in Europe of limited interest 4. Review of limited interest 5. Not possible to evaluate on the basis of title/abstract/keywords, only 6. Original study not conducted in Europe of high interest 7. Methodologic study of high interest 8. Review of high interest 9. Original European study on co-pollutants only 10. Original European study on PM.
report no. 4/11
18
From the second selection we eliminated (both reviewers’ score <5 OR (one reviewer’s score=5 AND the other’s <5) 738 papers. Among the other 579 papers, 144 were considered original European studies by both reviewers. Most of these studies evaluated the association with short-term exposure to ambient pollutants including PM.
Other searches targeting cancer were performed, using a specific search string on PubMed. Furthermore, all the publications from the most important international studies (e.g., APHEA and HEAPSS) were searched. Finally, reference lists of a number of published papers (original articles and reviews) were examined.
For all the 579 papers remaining in the database, the PDF of the original article was retrieved. All papers scored 9 or 10 by at least one reviewer were examined and classified according to outcome.
2.2. PRESENTATION OF FINDINGS
We used the tables on short-term PM exposure of the EPA 2004 report as basis (US EPA, 2004), and updated them with the new studies not included in the EPA 2004 report. From the original EPA tables we have abstracted only studies conducted in Europe. In the tables, the studies already present in the EPA 2004 report are shown on a grey background, while the new studies have a white background. The EPA 2004 report did not include tables summarizing findings from studies on long-term PM exposure and mortality, and CVD and cancer morbidity. We summarized papers published from 1996 also for these outcomes in a similar way.
The EPA 2004 report also included the re-analyses of some studies published in a HEI report (HEI, 2003). This report originated from the fact that problems were found in the statistical software S-Plus routine used in many studies to fit generalized additive models (GAM): the default convergence criteria used in the S-Plus GAM function resulted in biased estimates and under some circumstances the standard errors of the estimates were underestimated (Dominici et al, 2002). The authors were invited to reanalyze their data using more stringent convergence criteria. The corrected analyses were published in the HEI report and compared with the original ones. Fully parametric models (GLM with natural or penalized splines) were also explored as alternative ways of estimation in the re-analyses in the HEI report.
We have modified the order in which the studies appeared in the EPA tables, since we have grouped them in the following categories: 1) Large international multicity studies 2) Nationwide studies 3) Sub-national studies, grouped by country and geographic area (Northern, Central, Southern, Eastern Europe) 4) Studies on selected populations.
In the Appendix we reported the description of European epidemiologic studies published before 1996, and cited in the EPA 1997 report (US EPA, 1997)
report no. 4/11
19
2.3. STUDIES CONSIDERED
For the present revision of the literature we considered the following studies:
I. Studies on short-term PM exposure and mortality, including all the studies providing estimates for PM.
II. Studies on short-term PM exposure and CVD hospital admission including all the studies providing estimates for PM.
III. Studies on short-term PM exposure and respiratory hospital admission including all the studies providing estimates for PM.
IV. Studies on short-term PM exposure and respiratory medical visits including all the studies providing estimates for PM.
V. Studies on long-term PM exposure and mortality, including all the studies providing estimates for PM.
VI. Ecologic studies on long-term PM exposure and mortality, including all the studies providing estimates for PM.
VII. Studies on long-term PM exposure and CVD morbidity, including all the studies providing estimates for PM, or proxies of PM, as those studies having as outcome the distance between a major road and the residence.
VIII. Studies on long-term PM exposure and respiratory morbidity or symptoms, including all the studies providing estimates for PM.
IX. Studies on long-term PM exposure and cancer, including all the studies providing estimates for PM, other co-pollutants or proxies of PM, as those studies having as outcome the distance between a major road and the residence.
We have not considered studies on short-term PM exposure and pulmonary function tests or other symptoms in asthmatics or non-asthmatics.
2.4. KEY ABBREVIATIONS AND OTHER TERMS
AIC Akaike Information Criterion AMI Acute Myocardial Infarction APHEA Air Pollution and Heath: a European Approach BMI Body Mass Index BS Black Smoke CI Confidence Interval CO Carbon Oxide COPD Chronic Obstructive Pulmonary Disease CVD Cardiovascular Disease EPA Environmental Protection Agency (US) FEF25-75 Forced Expiratory Flow between 25th and 75th percent of the FVC FEV1 Forced Expiratory Volume in one second FVC Forced Vital Capacity GAM Generalized Additive Model GLM Generalized Linear Model HR Hazard Ratio ICD International Classification of Diseases IQR Inter-Quartile Range
report no. 4/11
20
IHD Ischemic Heart Disease Lag x xth day Lag Lag x-y average x-y days Lag LCL Lower Confidence Limit MEF25-75 Midexpiratory flow between 25 and 75th percent of the FVC NO Nitrogen oxide NO2 Nitrogen dioxide NOx Oxides of nitrogen NS Not Significant OLS Ordinary Least Squares OR Odds Ratio O3 Ozone RH Relative humidity PM Particulate Matter PMx PM less than x μm in aerodynamic diameter PNC Particle Number Concentration RR Relative Risk SO2 Sulphur dioxide TSP Total Suspended Particles UCL Upper Confidence Limit
report no. 4/11
21
3. SHORT-TERM PARTICULATE MATTER (PM) EXPOSURE AND MORTALITY
Table I (modified and updated from EPA 2004 Table 8A-1) presents the studies conducted in Europe that have considered the association between short term exposure to PM and mortality. The APHEA studies I and II were the only large international multicity studies on short-term effects of PM on mortality in Europe. For this reason, a brief description of the APHEA studies is given below, and particular prominence is given to the results of the APHEA2 study throughout this chapter.
3.1. THE APHEA STUDIES 1 AND 2
The APHEA1 study ended before the EPA 2004 report, and its main results were included there (Katsouyanni et al, 1997; Samoli et al, 2001; Samoli et al, 2003).
The APHEA1 study was conducted in 12 European cities over the period 1975-1992. However, BS indicators were available for 8 cities and PM10 for 6 cities only. In a first publication there was a significant heterogeneity between European areas, with a positive association between BS and PM10, and total mortality in western European cities and no effect in central/eastern European ones (Katsouyanni et al, 1997). Two subsequent re-analyses of the data using different statistical models (GAM models with smoothing terms for seasonal trends and weather) considerably reduced heterogeneity between the two areas (Samoli et al, 2001; Samoli et al, 2003). In those re-analyses, BS and PM10 were positively associated to mortality overall and in the two areas separately, although the size of the effect depended on the statistical model used, and particularly on the method used to adjust for meteorological variables and other confounders.
Several publications concerning the APHEA2 study were already included in the EPA 2004 report (Touloumi et al, 1997; Katsouyanni et al, 2001; Katsouyanni et al, 2003; Zanobetti and Schwartz, 2003), and new publications have appeared thereafter (Aga et al, 2003; Zanobetti et al, 2003; Touloumi et al, 2005; Analitis et al, 2006). Moreover, we have added to Table I the data concerning a publication (Zanobetti et al, 2002) that should have been included in the EPA 2004 report, but was not listed there. In fact, in Table 8A-1 of the EPA 2004 report, the article by Zanobetti and Schwartz 2003, which was one of the re-analyses conducted within the HEI project, was indicated as “reanalysis of the above study”, but that study (Zanobetti et al, 2002) was missing from the table.
The APHEA2 study used data over the period 1990-1997 (most cities had data for at least 5 years) and included 29 European cities for a total population of over 43 million.
Data on BS were available for 14 cities, and PM10 data for 21 cities.
The estimated effects of BS and PM10 on total non accidental mortality, reanalyzed using more stringent convergence criteria, were presented in the HEI report (Katsouyanni et al, 2003). The APHEA2 study found a direct association between mortality and the average level of PM10 on the same day and on the day before. The total mortality excess risk per 50 µg/m3 of PM10 varied between 2.1% (95% CI 1.2-3.0) and 3.3% (95% CI 2.7-3.9), according to the statistical model considered.
report no. 4/11
22
The new analyses of the APHEA2 study not included in the EPA 2004 report have concerned the comparison of short and medium-term effects of exposure to PM10 on mortality (Zanobetti et al, 2003), the effects of BS and PM10 exposure on cardiovascular and respiratory mortality (Analitis et al, 2006), the effect of PM on mortality in the elderly (Aga et al, 2003) and the effect of different modeling techniques of influenza epidemics on the association between PM10 and total and cardiovascular mortality (Touloumi et al, 2005).
3.2. LOCATION OF THE STUDIES
Besides the APHEA studies, which were based on several cities located throughout Europe, there were one nationwide study, conducted in the Netherlands (Hoek et al, 2000; Hoek et al, 2001; Hoek, 2003, Fisher, 2003 #726) and a number of studies conducted in one or more cities/areas, from the UK, Ireland, Finland, the Netherlands, Germany, the Czech Republic, France, Italy and Spain.
3.3. STUDY DESIGNS
Almost all studies were time series correlating daily mortality with particulate levels on one or more days. Only a study in the general population and a study in patients with COPD adopted a crossover design.
3.4. INDEX OF EXPOSURE TO PARTICULATE MATTER
Several different PM measures were used, including Total Suspended Particulate (TSP), Black Smoke (BS), PM10, PM7, PM13, PM15, PM2.5, PM2.5-10, Number Concentration (NC) and Mass Concentration (MC) of ultrafine (0.01-0.1 µm) and fine (0.1-2.5 µm) PM. A few studies also estimated associations with SO4
2- and NO3-.
In the APHEA1 study the indicators of PM exposure used were BS, available for 8 cities, and PM10, estimated for 6 cities and computed as follows: 0.55*TSP (3 cities), PM13 (2 cities) and PM7 (1 city).
In the APHEA2 study data on BS were available for 14 cities, and PM10 was estimated for 21 cities as follows: measured directly as PM10 (11 cities), by regression to BS (2 cities) or TSP (2 cities), or as % of TSP (6 cities).
3.5. STATISTICAL MODELS USED
Different statistical models for time series analyses were used in various studies. The GAM has been the favored method in recent years. The HEI report (HEI, 2003) provides a comprehensive assessment of the impact of different statistic approaches, and in particular several ways of adjusting for time-related factors, such as long term trends in pollution levels and mortality.
The APHEA2 study has been analyzed using different statistical approaches, and was re-analyzed in the HEI report (Katsouyanni et al, 2003). Shortly, overall estimates were obtained using a two stage model. In the first stage an estimate of the association between either PM10 or BS was obtained for each city using the same statistical model. In the second stage city-specific effect estimates were regressed on city-specific covariates to obtain an overall estimate and to explore confounding, using both fixed and random effects models. The single city models
report no. 4/11
23
were adjusted for seasonality, long-term trends, temperature, humidity, influenza, day of the week, holidays and unusual events if necessary. The lag was set a priori to the average of days 0 and 1.
The results of the effects on daily total number of deaths of PM10 are reproduced below (Table 1 from Katsouyanni et al, 2003). As compared to the original estimates, based on the default convergence criteria, the use of more strict convergence criteria did not materially modify the estimated betas (-4%). The use of a parametric approach by means of natural splines led to increased standard errors and a 30-40% reduction in the estimated betas, and to a 10-15% reduction when penalized splines were used. Although the size of the effect was reduced using alternative statistical methods, this reanalysis confirmed the findings of the original investigation (Katsouyanni et al, 2001), i.e. that there is a positive association between daily number of deaths and PM10 levels.
Significant heterogeneity between cities was found for all models considered.
report no. 4/11
24
3.6. LAG TIME
The definition of the lag time, i.e. the time interval between exposure (as determined by ambient levels of particulate) and outcome, remains still a controversial issue.
On one hand, there is the need to investigate which is the “correct” time to be used in the model. Moreover, the interval between exposure and effect may differ according to cause of death considered, reflecting different mechanisms of action. Rossi and colleagues (Rossi et al, 1999) in their analysis of the effects of ambient TSP levels on mortality in Milan, Italy, focused on the difference in lag times on cause-specific mortality. They reported a significant association on the concurrent day for deaths from respiratory infection and from heart-failure, while an association with myocardial infarction and COPD were found for the means of 3 to 4 days prior to death. In the APHEA2 study, the extension of the lag time from 2 to 6 days led to an increase of 20-30% of the estimated effect for CVD deaths, while the increase was more marked (75-90%) for respiratory deaths (Analitis et al, 2006).
On the other hand, many investigators tried a number of different single day and cumulative day lag times, and chose to present the results with the strongest association. The authors generally did not apply statistical methods (e.g. to alter statistical significance levels according to the number of tests performed) to allow for the multiple testing. This is, from a statistical point of view, questionable.
The APHEA1 study was also affected by this problem, since the “best” individual day models (days 0, 1, 2, 3) and the “best” average of 2-4 consecutive days was selected for each centre (Katsouyanni et al, 1997). The lag time used was also allowed to vary across centres.
Conversely, in the APHEA2 study the lag time was set a priori to the average of days 0 and 1 (except for analyses on medium term and harvesting effects). Thus the results of the APHEA2 study are unaffected by the problem of multiple comparison due to the use of several different lag times.
A few studies presented a table showing how the estimated beta varies according to the chosen lag time.
report no. 4/11
25
For example, Figure 2 from Bremner (Bremner et al, 1999) above shows the estimated effects on total mortality of individual day lags 0, 1, 2 and 3 of cumulative lags mean 0-1, 0-2, and 0-3 for PM10 and BS (and gaseous pollutants). In this case, the estimated effects are all positive albeit non significant, although the point estimates varied. In other cases, however, the direction of the effect and the statistical significance vary according to the chosen lag (Ponka et al, 1998).
For the reasons explained above, we prioritize studies in which the lag time has been set a priori over those in which several lag times have been considered.
Most of the studies reported the association between daily deaths and pollution concentration on the same day or 1-2 days before. However, a few European studies investigated the medium-term effects of air pollution levels considering lag times up to 40 days (Zanobetti et al, 2002). The findings of these studies are reported in section 10 of the present chapter.
report no. 4/11
26
3.7. CO-POLLUTANTS
Many studies measured other pollutants besides PM. The ones reported in most studies were SO2, NO2, O3 and CO. However, relatively few studies investigated the confounding effect of other pollutants on the relation between PM and mortality.
The APHEA1 study collected data on levels of SO2, but they did not introduce SO2 in the model because it was too strongly correlated to BS and PM10.
The APHEA2 study measured SO2, O3 and NO2, besides BS and PM10, and presented two-pollutant models in Table 2 of the paper by Katsouyanni et al, 2001 (see below).
For both PM10 and BS, adjustment for O3 resulted in an increase in the estimated coefficients, particularly for BS (+9% for PM10 and +40% for BS), adjustment for SO2 resulted in a slight decrease of the association (-13% for PM10 and -18% for BS), while the association with mortality halved after adjustment for NO2 (-49% for PM10 and -49% for BS). In all two-pollutant models, however, the associations between PM10 and BS with total mortality remained statistically significant.
Katsouyanni and colleagues (Katsouyanni et al, 2001) also fitted multivariate second stage regression models with the estimated effect parameters of PM and NO2 for each city as dependent variables and average long-term NO2 concentrations as a potential effect modifier. That analysis showed that, after adjustment for confounding of the daily fluctuations of NO2 on the PM effect parameters, NO2 continued to act as an effect modifier: the estimated increase in total daily mortality for an increase of 10 µg/m3 of PM10 was 0.19% in a city with low average NO2 concentration, and 0.80% in a city with high average NO2 concentration. Corresponding estimates for BS were 0.26% and 0.73%. Also the ratio of PM10 to NO2 was an effect modifier.
These estimates were affected by the problems due to the use of incorrect convergence criteria pointed out by Dominici and colleagues (Dominici et al, 2002). In the reanalysis presented in the HEI report (Katsouyanni et al, 2003), corresponding results were not shown. The authors however stated that exploration of the pattern of effect modification using parametric methods (natural and penalized splines) preserved exactly the same pattern of effect modification as the original analysis.
Hoek and colleagues (Hoek et al, 2000; Hoek, 2003), in their nationwide study in the Netherlands also extensively investigated the confounding effect of several
report no. 4/11
27
co-pollutants (O3, SO2, NO2, CO) on the relation between total mortality and PM10 and BS levels. They also measured fine particle component sulfate (SO4
2-) and nitrate (NO3
-).
The Table 2 of the paper by Hoek (Hoek, 2003) below reproduces part of the results from the reanalysis in HEI report. We will comment on the results for the “GAM strict” model. For natural splines results were similar.
In the one pollutant models (lag 0-6 days) all pollutants considered were significantly and positively associated with mortality, with the exception of SO4
2- for which the positive association was not significant. The point estimates were 1.023 for an increase of 100 µg/m3 PM10, 1.026 for 50 µg/m3 of BS, 1.022 for 25 µg/m3 of SO4
2-, and 1.030 for 50 µg/m3 of NO3
-.
Table 5 from Hoek (Hoek, 2003) (see below) presents results for two pollutant models.
The RR for all 4 indicators of particulate increased after adjustment for O3 or CO. Conversely, after adjusting for SO2 or NO2, the RR for PM10 was below 1, and that of BS was more than halved (1.012) and no longer statistically significant. Adjustment for SO2 or NO2 did not substantially change the estimated effects of SO4
2- (1.019 and 1.022, respectively) and NO3- (1.027 and 1.029), although the
standard error of the parameters increased with the introduction of the other pollutant.
report no. 4/11
28
It is also important to notice that the effect of SO2 and NO2 tended to remain positively and significantly associated with total mortality after adjustment for PM indicators.
report no. 4/11
29
In conclusion, these data suggest that PM10 is not the only pollutant of concern, and the PM10-effect should be carefully adjusted for co-pollutant effects.
3.8. CONFOUNDING/MODIFICATION EFFECT OF OTHER ENVIRONMENTAL VARIABLES
In most studies the authors adopted various methods to adjust for seasonality and long-term trends in air pollutants and daily mortality, while investigating the day to day association. Most studies also adjusted for meteorological variables like temperature at various lags and humidity. Other confounders often considered were day of the week, holydays and influenza epidemics.
One study conducted a sensitivity analysis examining the effect of different methods of modeling influenza epidemics on the estimated association between BS and total mortality in Barcelona, Spain (Tobias and Campbell, 1999). It found that the coefficient for BS decreased and became not statistically significant, when influenza was modeled by the daily number of cases, instead that by dummy variables identifying epidemic periods. Conversely, a reanalysis of the APHEA2 data controlling for influenzas by ten different methods found that the significant and positive association between PM10 and daily number of deaths remained and even increased after controlling for influenza, regardless of the method used for control (Touloumi et al, 2005).
In the APHEA2 study, the effect of PM was larger in warmer and drier cities, although temperature was a more important effect modifier than humidity (Katsouyanni et al, 2001). The HEI reanalysis (Katsouyanni et al, 2003) confirmed the same pattern of effect modification. When the authors investigated cause-specific mortality, average annual temperature and humidity were effect modifiers for CV death, but not for respiratory causes (Analitis et al, 2006)
In the Netherlands study, the effects of PM were stronger in summer than in winter (Hoek et al, 2000; Hoek, 2003). A stronger association in warmer periods was also reported in a few other studies (Anderson et al, 1996; Michelozzi et al, 1998; Anderson et al, 2001).
3.9. CAUSE-SPECIFIC MORTALITY
In the majority of studies, the endpoint considered was total non accidental mortality (ICD9 codes 1-799). A number of studies also considered CVD (ICD9 codes 390-459) and respiratory non cancer mortality (ICD9 codes 460-519) separately. In general, CVD deaths account for roughly half of total non accidental deaths, and often results for total and CVD mortality are similar. Conversely, respiratory mortality accounts for a much smaller proportion of total mortality. Thus, the results for respiratory mortality are much more affected by random variation, particularly in smaller studies.
Analitis and colleagues (Analitis et al, 2006) analyzed the effects of BS and PM10 separately for CVD and respiratory mortality in the APHEA2 study.
In Table 1 of their paper (below), they summarized the main results.
report no. 4/11
30
The results for the two causes were, in proportional terms, remarkably similar. For example, under the random effect model, the estimated % increase in daily number of deaths for an increase of 10 µg/m3 of PM10 was 0.76% (0.47, 1.05) for CVD causes and 0.71% (0.22, 1.20) for respiratory causes. Corresponding values for BS were 0.62% and 0.84%. Estimates under the fixed effect model were lower, but still statistically significant for both causes of death. Their analysis suggested that the major impact of PM on CVD mortality occurs in the first two days (lags 0, 1), while the effects on respiratory mortality persist for several days.
Several other smaller studies considered CVD and respiratory mortality separately, but results were not always easy to interpret, given the smaller numbers of deaths, and hence the greater random variation in cause-specific mortality analyses. In the study based on 9 French cities (5 of which had data on BS) the excess RR (lag 0-1) associated with an increase of 50 µg/m3 was 2.9% (1.3, 4.4) for total, 3.1% (0.5, 5.5) for CVD and 2.7% (-2.6, 8.3) for respiratory mortality (Le Tertre et al, 2002b).
In the Netherlands study, Hoek and colleagues (Hoek et al, 2001) analyzed associations with specific CVD causes of death. They found that deaths due to heart failure (ICD9 428), arrhythmia (ICD9 427), cerebrovascular causes (ICD9 430-436) and thrombocytic causes (ICD9 415.1, 433-4, 444, 452-3) were more strongly associated with PM (as well as with gaseous pollutants) than other CVD causes of death, including ischaemic heart diseases. This was confirmed in the HEI reanalysis (Hoek, 2003).
Cause-specific mortality was also considered in a study from Milan, Italy (Rossi et al, 1999). The authors found that respiratory infections and heart failure deaths were best predicted by TSP on the concurrent day, while, for deaths from myocardial infarction and COPD, lag 3-4 was the best predictor. The estimated RRs, however, were similar for heart failure (7% for 100 µg/m3 TSP) and myocardial infarction (10%).
3.10. MORTALITY DISPLACEMENT AND MEDIUM-TERM EFFECTS OF PM
Although several time series found a direct relation between levels of PM and mortality, it has been questioned whether these excess deaths occur in those who would have died in a few days anyway (harvesting effect). If this is true, one would
report no. 4/11
31
expect an increase in mortality in the few days following a day with high levels of PM, followed by a decrease in the medium term. Zanobetti and colleagues (Zanobetti et al, 2002) analyzed the issue using the APHEA2 data. Due to time and resource constraints, it was decided a priori to limit the analysis of mortality displacement to the ten largest cities from different countries. Using the usual two stage methodology of the APHEA studies, different lag modes were fitted to the data. First a model for lag 0-1 was fitted, in order to estimate short-term effects. Then, different models for lag 0-40, measuring the cumulative PM10 exposure of the onset day and the previous 40 days, were fitted: an unconstrained distributed lag model, where one parameter for each day was used, and polynomial distributed lags of the third and fourth degree.
Estimated PM10 effect per 10 µg/m3 are given below, for the original analysis (Zanobetti et al, 2002) and for the HEI reanalysis (Zanobetti and Schwartz, 2003) using more strict convergence criteria and penalized splines.
Beta (SE)
Lag, model Original GAM Strict GAM Penalized Splines
0-1 0.70 (0.14) 0.67 (0.14) 0.57 (0.15)
0-40, unconstrained 1.61 (0.39) Not given Not given
0-40, cubic 1.57 (0.67) Not given Not given
0,40 4th degree 1.61 (0.30) 1.45 (0.30) 1.08 (0.40)
Since the estimated effect of PM on mortality more than doubles when lag 0-40 days is used, instead of lag 0-1, the authors concluded that i) the effects observed in daily time series are not due primarily to short-term mortality displacement, and ii) the short-term risk assessment based on lag 0-1 associations likely underestimates the PM effect on mortality. The HEI reanalysis did confirm the same results, although all estimates were reduced for penalized splines. Similar results were reported in studies conducted in Milan, Italy (Zanobetti et al, 2000) and Dublin, Ireland (Goodman et al, 2004).
Zanobetti and colleagues (Zanobetti et al, 2000) also examined mortality displacement separately for cardiovascular and respiratory disease. In Table 2 of their paper (below), they found that the effect size more than doubled for CVD deaths, and increased 5-fold for respiratory deaths, when they used lag 0-40 instead than lag 0-1.
report no. 4/11
32
They also found a different pattern of mortality risk over time for CVD and respiratory deaths, with the elevation in risk of death after exposure declining more slowly over time for respiratory rather than CVD mortality.
3.11. ULTRAFINE, FINE AND COARSE PM
Only a few studies specifically considered fine or coarse PM separately.
Anderson and colleagues (Anderson et al, 2001) considered the association between PM10, BS, PM2.5, PM2.5-10 and SO4
2- on daily mortality in a study conducted between 1994-1996 in the west Midland conurbation of the UK. To compare estimates, they used as measurement unit for each pollutant the difference between the 10th and the 90th percentiles. The relative risk estimates at lag 0-1 were not significant for any of the PM indicators considered in the all season analysis (see Table 3 of their article below).
The authors noted, however, that all measures of PM except PM2.5-10 showed significant positive effects in the warm season (figure 1 below).
Although not significant, the estimates for coarse PM were not substantially different from those of other PM indicators, particularly for the warm season. Moreover, the correlation between PM10 and PM2.5 was 0.92, thus the effects of the two measures were difficult to distinguish.
report no. 4/11
33
A study from Erfurt, Germany, conducted between 1995 and 1997 assessed the association and role of particles of different size with respect to mortality in an urban setting (Wichmann et al, 2000). One specific aim was to compare the role of ultrafine (aerodynamic diameter <0.1 µm) and fine (0.1-2.5 µm) PM on mortality.
Number concentration (NC) was measured for each particle size range, while mass concentration (MC) was calculated for each particle size range from the NC assuming sphericity of particles and constant density.
In Table 18 of their report (see below) the authors show the estimated effect for the best fitting single day model and for polynomial distributed lag models over 5 or 6 days.
report no. 4/11
34
The authors concluded that fine and ultrafine particles had direct effects on mortality which were comparable in strength, and the effect was stronger if the contributions of all size classes were added up, mainly for respiratory diseases, followed by CVD diseases. When they considered two-pollutant models, they concluded that “the
report no. 4/11
35
observed correlations of gaseous pollutants and mortality were explained by colinearity of the gases to particles rather than by an independent causal effect”.
The authors also stated that there were indications from their data that the effects of fine particles occur more immediately (lag 0-1) and the effects of ultrafine particles more delayed (lag 4-5). The HEI Review Committee agreed with the authors’ conclusion that fine and ultrafine particles were associated with mortality, but did not agree with the authors’ suggestion of a different lag effect (immediate vs delayed) for the two particle size. The Committee also did not agree with the authors’ interpretation that the association of SO2 with mortality was an artifact due to confounding by PM.
The Committee also pointed out that the study had strengths like the use of a more precise PM metric, but also had concerns regarding the statistical analysis and interpretation, mainly linked to the problems of multiple comparisons, the sensitivity of the overall estimates to confounders like influenza, and to the definition of lag.
The HEI reanalysis using more strict convergence criteria for GAM and fully parametric methods (GLM) confirmed the major interpretations and conclusions of the original study (Stölzel et al, 2003).
An extension of the study to the period 1995-2001 currently published online ahead of print (Stolzel et al, 2006), found associations between ultrafine particles NC and total (RR=1.029, 95% CI 1.003, 1.055 for an increase of 9748 cm-3) and cardio-respiratory (RR=1.031, 95% CI 1.003, 1.060) mortality. No association between fine particle MC and mortality was found.
Particle number concentration (PNC) - as a proxy for ultrafine particles - was also considered in a case-crossover study conducted between 1998 and 2000 in Rome on out-of-hospital coronary deaths (Forastiere et al, 2005). The study found a direct association between out-of-hospital fatalities and PNC, PM10 and CO. However, PNC were measured for the period April 2001-Jun 2002 and estimated retrospectively for the study period by means of a regression model with PNC as the dependent variable and other pollutants (PM10, CO, NO2 and O3) and meteorological readings as explanatory variables. Moreover, given that the correlation between measured and estimated PNC was 0.90, and that between CO and PNC was 0.89, estimated PNCs were not a better proxy of “true” PNC than CO. Nevertheless, the same study was re-analyzed as a time series and yielded similar results (Stafoggia et al, 2005).
A study from the coal basin of the Czech Republic (Peters et al, 2000) found a significant association between PM10 and mortality only at lag 1 (RR=1.098, 95% CI 1.07, 1.197 for 100 µg/m3 increase). For each given lag, estimates for a 100 µg/m3 increase of PM2.5 were lower than for a corresponding increase in PM10 (RR=1.059, 95% CI 0.980, 1.144 at lag 1).
3.12. SEX DIFFERENCES
Only a few studies considered sex differences in the association between PM and mortality. Although one study from Northern Bohemia, Czech Republic (Kotesovec et al, 2000), found different results in men and women (significant associations only for men below age 65 years, and for women only above age 65 years) these differences may be due to random variation in subgroup analysis. Another study
report no. 4/11
36
from Rome found similar associations between PM10 and fatal out-of-hospital coronary events for the two sexes (Forastiere et al, 2005).
3.13. AGE DIFFERENCES
In the APHEA2 study, the proportion of elderly in various cities was an effect modifier of the relation between PM10 and mortality, and the mortality excess risk for a 10 µg/m3 increase in PM10 was 0.54% (0.38, 0.69) in a city at the 25th percentile of the distribution of proportion of elderly, and 0.76% (0.63, 0.90) in a city at the 75th percentile (Katsouyanni et al, 2001). Corresponding estimates for BS were 0.48% (0.34, 0.62) and 0.70% (0.51, 0.90).
When the analyses were carried out separately for CVD and respiratory mortality, the modifying effect was stronger for respiratory mortality (Analitis et al, 2006). For BS, the proportion of elderly explained less than 10% of the heterogeneity between cities for CVD mortality, while for respiratory mortality the effect at the 25th percentile was 0.44% (-0.01, 0.88) and at the 75th percentile was 1.57% (0.99, 2.16).
When analysis of the APHEA2 data was limited to the population aged 65 years or more (Aga et al, 2003) larger effects of PM10 and BS were found than in the all age analysis (see Table 1 from their article below).
The increases in estimates when analysis was restricted to the elderly population were around 10% for PM10 and around 20% for BS.
In the analysis restricted to the elderly, the same pattern of effect modification as in the overall analysis was reported, with larger PM effects in cities with higher NO2 levels, higher temperature, lower humidity, higher age-standardized annual mortality rate, higher proportion of individuals ≥65 years and from Northwestern and Southern Europe.
An issue that still remains unclear is the quantification of the effect of PM on mortality in the population younger than 65 years. In the cities included in the
report no. 4/11
37
APHEA2 study the proportion of deaths occurring in the elderly ranged between 67% and 88% (Aga et al, 2003). Since the majority of deaths occurred in the elderly, mortality at all ages and in the elderly are highly correlated. Restricting mortality at younger ages may result in substantially different PM effect estimates, which however are clearly affected by larger random variation.
The APHEA2 analysis of the medium-term effects of PM (Zanobetti et al, 2003) found no association between PM10 and total mortality at age 15-64 years, but significant associations at ages 65-74 years and ≥75 years.
The association between PM10 and BS levels on the one hand and daily mortality on the other by age group was considered also in the Netherlands study (Fischer et al, 2003). Separate estimates were provided for the age groups <45, 45-64, 65-74 and ≥75 years for total mortality and for deaths due to CVD, COPD and pneumonia (see below figure 1 of Fisher et al., 2003). The RRs were calculated for a difference between the 1st and 99th percentiles (i.e., 80 µg/m3 for PM10 and 40 µg/m3 for BS).
From Fig 1 of Fischer et al., 2003. The successive age groups are <45, 45-64, 65-74 and ≥75 years.
When they analyzed specific causes of deaths, the pattern for CVD deaths was similar to overall mortality. Large random variation affected the estimates for COPD and pneumonia. However, for pneumonia, estimates appeared higher below age 65 years for PM10, where the estimates in subsequent age categories were 1.43, 1.71*, 1.24 and 1.12* (the asterisk indicated that the CI does not include 1). No clear pattern across age groups for pneumonia mortality emerged for BS (RRs=0.74, 1.26, 0.95 and 1.12*, respectively).
The few smaller studies investigating the effect in different age groups were affected by substantial random variation, and yielded conflicting results: some found indications that the effect of PM was larger or limited to elderly people (Michelozzi et al, 1998; Goodman et al, 2004; Forastiere et al, 2005) some to younger people (Ponka et al, 1998; Bremner et al, 1999; Wichmann et al, 2000) and some did not find noticeable differences between age groups (Hoek et al, 1997; Prescott et al, 1998).
report no. 4/11
38
3.14. SOCIOECONOMIC DIFFERENCES
One case-crossover study investigated the effect of PM10 on daily mortality by socioeconomic status (SES) among residents in Rome aged over 35 years between 1998 and 2001 (Forastiere et al, 2005). For each death from natural causes, the day on which the death occurred was the “case” day, and the same days of the week within the same month were chosen as control days. Hospital discharge data for the 2 years before deaths were used to estimate history of selected diseases requiring hospitalization, including diabetes, hypertension, heart failure and COPD. As a surrogate for individual SES, two area-based indicators were used, i.e. the median per capita income and the average SES index of the census block of residence (Rome is divided in 5,736 census blocks, with an average of 480 inhabitants). The SES index was based on several characteristics of the census block, like educational level, occupational category, unemployment rate, family size, crowding and proportion of rented/owned dwellings. The two SES indexes were than divided in four groups by the 20th, 50th and 80th percentiles.
In the population of Rome the percentage of people living in areas with high levels of PM10, CO, NOx and benzene increased with increasing SES indicator, particularly for SES index: 9% of those with low SES, and 12%, 21% and 40% in those with mid-low, mid-high and high SES, respectively. This was due to the fact that people with high SES tended to live nearer to the more polluted city centre.
There was a monotonous inverse relation between both SES indicators and hospitalization rates for many chronic conditions - including malignant neoplasms, several cerebrovascular diseases, COPD and diabetes - two years to 29 days before deaths.
Figure 2 of the article presents the estimated % increase in daily mortality by SES:
report no. 4/11
39
The effects of ambient PM10 on mortality were similar for the first three SES categories, but there was no association in the high SES category.
The authors concluded that the effect of PM was stronger in lower social classes. Given the observed pattern of exposure and of history of hospitalization for several severe conditions, they argued that the different effect of PM in SES groups was likely due to the differential burden of chronic health conditions conferring a greater susceptibility to lower SES groups.
However, as the authors pointed out, the monotonous increase in comorbidity across SES strata is not mirrored by a corresponding increase in PM effect, which appears, if anything, dichotomic.
3.15. STUDIES IN PATIENTS WITH SELECTED DISEASES (COPD)
Three articles (Garcia-Aymerich et al, 2000; Sunyer et al, 2000; Sunyer and Basagana, 2001) investigated the relation between ambient PM and mortality in a cohort of patients with COPD from Barcelona, Spain.
All patients aged 35 years or more attending the emergency room services in Barcelona for COPD between 1985 and 1989 were recruited (9,987 people).
In a time series analysis on the 3,245 patients who died in the period 1985-1889 (Garcia-Aymerich et al, 2000) the estimated percent increase in total mortality associated with an increase of 20 µg/m3 in the patients in the cohort was 2.3% (-0.8, 15) and was, if anything, lower that the one estimated in the general population (5.6%, 95% CI 2.7, 8.7). Given the higher mortality rates in patients with COPD, in absolute terms, the effect may be higher than in the general population.
A case-crossover study (Sunyer et al, 2000) was conducted in patients from the cohort who died in the period 1990-1995 (2,305 patients). PM levels on the day of death were compared to those one week before and one week after. The estimated percent increase in mortality for an increase in BS of 20 µg/m3 was 11.2% (1.7, 21.5) for all causes of deaths, 7.7% (1.3, 26.4) for CV mortality, and 18.2% (2.5, 36.5) for respiratory deaths.
When the same design was used to estimate PM10 effects (Sunyer and Basagana, 2001), the percent increase in total mortality for a 27 µg/m3 (IQR) increase was 11.4% (0, 24). Although adjustment for other pollutants (NO2, O3, CO) resulted in larger SE of the parameters, the point estimates were not substantially modified and ranged from 10% to 13%.
However, it is important to notice that subjects with COPD are less likely to be exposed to outdoor air levels of pollutants because they spend more time indoors.
3.16. REGIONAL DIFFERENCES
Regional differences were investigated within the APHEA Project in eight western and five central-eastern European countries, using different statistical models (Samoli et al, 2001). Using sinusoidal terms for seasonal control and polynomial terms for meterologic variables, the effects of air pollution on daily mortality were lower in central-eastern European cities (Katsouyanni et al, 1996; Katsouyanni et al, 1997). Data were re-analysed using generalized additive models (GAM), in order to
report no. 4/11
40
smooth differences in risk estimates across various countries. Thus, when excluding days with pollutant levels above 150 µg/m3, an increase in BS by 50 ug/m3 was associated with a 3.1% (2.6, 3.6) increase in mortality overall, a 3.1% (2.3, 3.9) increase in western countries, and a 2.9% (1.8, 4.1) increase in central-eastern countries (Samoli et al, 2001).
The APHEA2 study, including 29 European cities, has provided the best framework to investigate regional variation of PM effect in Europe. The median levels of PM10 concentration among the 21 cities providing data ranged between 14 to 166 ug/m3. For BS median concentrations ranged between 9 and 64 ug/m3 among the 14 cities providing data. The estimated increase in mortality for an increase of 10 µg/m3 ranged from -0.6% to 1.5%, and there was significant heterogeneity between the single city estimates. Several city-specific factors acted as effect modifiers of the relation between PM10 and mortality, the effects being greater in cities with high average NO2 levels, high mean temperature and low standardized mortality. Most of the heterogeneity between cities was explained by the inclusion of these three factors in the model (Katsouyanni et al, 2001).
3.17. THRESHOLDS AND DOSE-RISK RELATION
Combining estimates for cities included in the APHEA2 study and of a few other studies, Anderson and colleagues investigated the relation between mean PM10 concentration level and estimated short-term effect on total mortality (Ross Anderson et al, 2004) [Ross Anderson et al., 2004]. In Figure 1 of the report (shown below), the estimated PM10 effect on all-causes mortality is plotted against annual average levels of PM10 in individual European areas. No apparent relationship was evident, indicating that the relative risk does not depend on the average levels of PM10, at least for a certain range of concentration.
report no. 4/11
41
As Brunekreef and Holgate pointed out, current knowledge, in particular from European studies, on short-term exposure to air pollution and mortality does not allow to understand whether thresholds concentrations exist below which air pollution has no effect on population health. In fact, most of the European studies analysing the association between air pollution levels and risk for health analyse time-series data. In these studies, total excess risks are usually estimated according to an increment of various measures of PM. This way of presentation of findings does not take into account the absolute pollution levels. In fact, in case of non-linearity, an increase from 100 to 110 µg/m3 does not have the same population impact as an increase from 10 to 20 µg/m3. This also demonstrates that time-series methodology is not suitable for detection of thresholds. (Brunekreef and Holgate, 2002).
In most studies, the statistical analysis took care of allowing a substantial variation in the possible shape of the confounders. For the PM effects however, generally linear, or in some cases log-linear, relationship with the dependent variable was assumed in the analysis. In the APHEA2 study, days with over 150 µg/m3 were excluded because the investigators deemed that a linear relation may not hold at higher doses. In their analysis of the National Morbidity, Mortality and Air Pollution Study (NMMAPS), Dominici and colleagues found that the national average exposure-response curve was linear, although some modest departures from the linear model were observed in some regions (Dominici et al, 2003).
report no. 4/11
42
3.18. COMPARISON OF EUROPEAN WITH LEADING NORTH AMERICAN STUDIES
The two largest multicity studies from Europe (APHEA2) and the United States (National Morbidity Mortality Air Pollution Study, NMMAPS) have been reanalyzed using similar methodology within the HEI project (HEI, 2003).
The table below illustrates the estimated increase in total mortality from nonexternal causes in the two studies under different statistical models.
Estimated percent mortality change for an increase in daily PM10 of 10 µg/m3 in Europe and the United States
Model NMMAPS (lag 1) APHEA2 (lag 0-1)
GAM estimated with default S-Plus parameters
0.4 (0.3-0.6) 0.6 (0.4-0.8)
GAM with more stringent criteria for convergence
0.3 (0.2-0.8) 0.6 (0.4-0.8)
GLM with natural cubic splines 0.2 (0.1-0.3) 0.4 (0.2-0.6)
As can be seen, in both studies the estimated excess mortality changes according to the model used. In Europe, estimates tend however to be higher than in the Unites States. Some differences in the estimation process may explain, at least in part the differences obtained, including the different lag times used and the confounders included in the model). Differences in the quality of the air pollution measurements or in the pattern of exposure of the underlying population may also contribute to the discrepancies in results. Katsouyanni and colleagues (Katsouyanni et al, 2003), however, suggested that a possible explanation may be the difference in relative sources and contribution of PM10 in Europe and in the US. In particular diesel cars are rare in the US, while in many European cities they approach 50% of cars.
Since the APHEA2 study found that other city-specific characteristics acted as effect modifiers on the PM-mortality relation, it is also possible that differences in these factors between Europe and the US explain the different strength of the observed relation.
3.19. CONCLUSIONS
In spite of the many studies conducted in Europe, several uncertainties on the relation between ambient PM levels remain, even because of some still unresolved methodological issues.
Most of the European studies (Table I) found a direct association between mortality and various indicators of PM levels.
In the largest, multicity study (APHEA2) the estimated increase in total number of deaths associated with an increase of 50 µg/m3 of PM10 (lag 0-1) ranged between 2.1% and 3.3% according to the statistical model used (Katsouyanni et al, 2003).
report no. 4/11
43
In the Netherlands study (Hoek, 2003), the corresponding estimates ranged between 0.9% and 1.2% according to statistical model and lag time (1 or 0-6) considered. For both studies, the 95% confidence intervals did not include unity.
For an increase in BS of 50 µg/m3 the APHEA2 study’s estimates ranged between 1.4% and 2.8% and in the Netherlands study’s between 1.9% and 2.6%.
Medium-term (40 days) analyses have shown that the observed effect is likely not due to mortality displacement, i.e., a mere anticipation of time of deaths of a few days. On the contrary, medium-term effect estimates tended to be two times higher than short-term one, indicating a possible underestimation of the global effect of PM when only few days are considered. This problem may be more pronounced for respiratory deaths than for CVD ones.
Given the marked correlations between PM and gaseous pollutants, the reciprocal confounding effect is an important issue in the light of a causal interpretation of the observed association. Introduction of highly correlated variables in statistical regression models creates problems of instability and can seriously affect the estimates.
This may explain, at least in part, the inconsistencies observed between the major studies. In the APHEA2 study the effect of PM was only slightly reduced after adjustment for SO2, but almost halved, remaining statistically significant, after adjustment for NO2. In the Netherlands study, no association with PM10 was shown after adjustment for SO2 or NO2.
Besides the issue of confounding, other pollutants can also act as effect modifier on the relation between PM and mortality, as suggested by the APHEA2 study. This relation can be further modified by other meteorological variables, like temperature and humidity.
Individual variables - like age, general health status and SES - can also affect this relation.
Finally, the European studies aiming at separating the effects of fine (and ultrafine) and coarse PM are still too few and small-scaled to allow clear conclusions.
report no. 4/11
44
4. SHORT-TERM PM EXPOSURE AND CARDIOVASCULAR DISEASE (CVD) HOSPITAL ADMISSIONS
Table II (modified and updated from EPA 2004 Table 8B-1) presents European studies investigating the association between short-term ambient PM exposure and CVD hospital admissions. These include two international multicity studies, the APHEA2 (Le Tertre et al, 2002a) and the HEAPSS (von Klot et al, 2005; Lanki et al, 2006), two national multicity studies from France (Eilstein et al, 2004) and Italy (Biggeri et al, 2005) and a number of smaller studies from various countries.
4.1. THE APHEA2 STUDY
The APHEA2 study has been already discussed in the chapter on short term effects of PM on mortality. The analysis on CVD hospital admissions (Le Tertre et al, 2002a) was based on 8 cities/areas, i.e. Barcelona in Spain, Birmingham and London in the UK, Milan and Rome in Italy, Paris in France, Stockholm in Sweden and the whole of the Netherlands, for a total population close to 38 million.
The study period started between 1989 and 1994 and ended between 1994 and 1997, depending on the area. The number of days contributing to the analysis varied between city and PM exposure indicator, ranging between 930 and 2,904. Ambient PM exposure indicators used were BS, available for 5 cities, and PM10, measured directly in 5 cities, as PM13 in Paris and as 0.75*TSP in Milan and Rome. Days with levels of PM exposure indicators above 150 µg/m3 were excluded, because of the possible lack of linearity at higher doses.
Autoregressive Poisson models, adjusting for long-term trends, season, influenza and meteorology were used to estimate PM effects for each city. Pooled estimates were then obtained combining the city-specific effect estimates by means of fixed and random effect models. The lag was set a priori at 0-1.
Emergency admissions were used in all cities except Milan, Paris and Rome, where only general admission data were available. In order to exclude elective admissions, some diagnostic codes were excluded, based on the distinction between emergency and elective admissions in London for the period 1992-94. Causes with more than 70% of elective admissions were excluded, plus the code 414 (“other forms of ischemic heart disease”), for which 65% of admissions were elective. The outcomes considered were all cardiac causes (ICD9 390-429), ischemic heart disease (ICD9 410-413) and stroke (ICD9 430-438).
report no. 4/11
45
Table 5 of the article (above) gives the pooled results for a 10 µg/m3 increase in PM10 or BS.
The percent increase of cardiac admissions for a 10 µg/m3 increase was 0.5% (0.3, 0.7) for PM10 and 0.8% (0.5, 1.2) for BS under the fixed effect model, and the estimates were higher when the analysis was limited to individuals above age 65 years. For IHD the estimates were 0.3% (-0.1, 0.6) below 65 years of age and 0.6% (0.3, 0.8) above age 65 years for PM10; and 0.1% (-0.4, 0.5) below 65 years and 1.1% (0.7, 1.5) above age 65 years for BS. No effect of PM10 or BS was evident for stroke admissions. Estimates from random effect models were either similar or moderately higher.
The authors also fitted two pollutant-models (see Tables 6 and 7 of the article below):
report no. 4/11
46
Although in the titles of Tables 6 and 7 the term “total daily number of deaths” appears, the results likely refer to hospital admissions, as written in the text (mortality data were not presented in this paper). The effects of PM10 on cardiac, cardiac over 65 years, and IHD over 65 years admissions decreased after adjustment for CO, and disappeared after adjustment for NO2 or BS. Adjustment for NO2 about halved the effects of BS, which became non significant. In contrast, the point estimates of the BS effects on cardiac emergency admissions became bigger after adjustment for CO or PM10.
According to the authors, the stronger confounding of PM10 effect by NO2 – a secondary pollutant mainly influenced by diesel and non-catalyzed gasoline fuelled engines – than by CO – a primary traffic pollutant representing mainly gasoline fuelled engines – suggests that diesel exhaust was primarily responsible. The authors also suggested that this conclusion was supported by the fact that BS, which is dominated by diesel particles, was little affected by confounding from CO or PM10. This may be debatable, as BS may also reflect the contribution of other non traffic-related sources (coal, industrial, etc.).
4.2. THE HEAPSS STUDY
The Health Effects of Air Pollution among Susceptible Sub-population (HEAPSS) originally investigated the effect of traffic related air pollution on cardiac hospital readmissions in a cohort of subjects from five European cities who were admitted for first AMI (von Klot et al, 2005). In a subsequent analysis the group also investigated the association between air pollutants and hospitalization for first AMI, i.e., the event responsible for the inclusion into the HEAPSS cohort (Lanki et al, 2006). Thus, this second study is based on the general population, and not on a selected cohort.
4.2.1. Analysis on the general population
The HEAPSS study included five European cities, Augsburg in Germany, Barcelona in Spain, Helsinki in Finland, Rome in Italy and Stockholm in Sweden. For each city, the study period was within the range 1992 and 2000, and included at least three years.
report no. 4/11
47
PM was measured as PM10 and PNC. In Augsburg, PM10 was estimated as 0.83*TSP, and in Barcelona data were missing for two years and were estimated by linear prediction from TSP and BS. Monitoring of PNC started in 1999 in Augsburg, and in April 2001 in the other cities. Thus, during the study period no direct measurement of PNC was available. City-specific models for estimating PNC from other pollutants and meteorological variables were developed for the period when PNC was measured, and applied then retrospectively to predict daily PNC concentrations during the study period (Paatero et al, 2005). The correlation of measured and predicted half of PNC, using split-halves technique ranged between 0.72 and 0.89 in the five cities, and that of day-to-day differences in measured and predicted PNC between 0.61 and 0.88 (Paatero et al, 2005). Levels of NO2, CO and O3 were also recorded and analyzed.
The data on hospitalizations for first AMI were collected using AMI registers in Augsburg and Barcelona, and hospital discharge registers in the other three cities. The ager range was over 35 in all cities. Augsburg included patients up to 74 years, and Barcelona up to 79 years. In order to increase specificity of AMI diagnoses, hospitalizations shorter than 3 days were excluded. Elective hospitalizations were also excluded in Helsinki and Stockholm. To select first AMI only, the recorded patient history was used in AMI registers, and a record linkage with the hospital discharge data of the previous three years was used in the other three cities. The presence of codes indicating a previous infarction (ICD9 412) as secondary diagnosis was also a reason for exclusion.
City-specific models were fitted in the first stage. Poisson regression, adjusting for non-linear effects by means of penalized splines in GAM was used for the analysis. Time trend, apparent temperature, barometric pressure, day of the week, holiday, and population reduction in summer were considered as confounders. Lags 0, 1, 2 and 3 were considered and presented in the article. The estimates were then combined by means of either fixed or random effect models in case of heterogeneity between city-specific estimates.
Table 4 of the paper (below) shows the relative risk of hospitalization for a first AMI for a 10 µg/m3 increase in PM10 and for a 10,000 particles/cm3 in PNC.
report no. 4/11
48
PM10 was not associated with hospitalization for first AMI in all five cities, nor when the analysis was limited to the three cities using hospital discharge data to define the outcome. PNC was not associated in the overall analysis, but there was a 1.3% (0.0, 2.6) increase in hospitalization for first AMI at lag 0, albeit not at other lags, when the analysis was restricted to hospital discharge register data.
report no. 4/11
49
The authors excluded Augsburg and Barcelona since the upper limit on age in the inclusion criteria reduced the power in these two cities, and this could be the reason for the lack of association. However, including more cities should have, if anything, increased the power of the study. Consequently, even if nominally based on five cities, the main conclusions are based on analyses on three cities only. Figure 1 of the paper on the left gives the city specific associations. Although the authors claimed that there was no heterogeneity between the three cities (Helsinki, Rome, and Stockholm) at lag 0, there appeared to be an heterogeneity between Rome and Stockholm, since the point estimate of each of these cities was not included in the 95% CI of the estimate of the other. In fact, Rome and Stockholm are the only two
report no. 4/11
50
cities where some significant associations were observed. For Rome, a positive one at lag 0, and a strong inverse one at lag 3. For Stockholm, an inverse one at lag 1 and a positive one at lag 3.
The analysis was then conducted separately for fatal and non fatal AMI and for patients aged below and above 75 years, only for the three cities using hospital discharge data, and the results were presented in Table 5 of the article (see below).
No appreciable association emerged in this subgroup analysis for PM10. Significant associations emerged for fatal AMI below age 75, and for non fatal AMI above age 75 years. Effects were more pronounced during the warm season. Two pollutant models were not performed.
Given the pattern of association with PNC (and CO), the authors concluded that their study supports the hypothesis that exposure to traffic related air pollution increases the risk of AMI, and that the most consistent associations were observed among fatal cases aged <75 years and in the warm season.
4.2.2. Analysis on survivors of first AMI
An article by von Klot and colleagues presented the results of the HEAPSS study for hospital readmission for cardiac causes after a first AMI (von Klot et al, 2005). The cohort of subjects on which this analysis was based included all subjects who had had a first non fatal AMI during the study period, i.e., all the subjects who had had an outcome in the analysis presented above (Lanki et al, 2006) and did not die within 28 days from the event. The enrollment period coincided with the one of the previous analysis, and the follow-up period was extended up to 2000-2001 for all centres. The cohort consisted of 22,006 subjects. Endpoints were defined as hospital readmission within the study area for 1) AMI, 2) angina pectoris and 3) cardiac causes, i.e., a combined endpoint including the previous two, heart failure and dysrhythmia. The start of the follow-up was the 29th day after the index event,
report no. 4/11
51
and the end of follow-up was either the endpoint of interest, death, migration out of study area, loss to follow-up or end of study period.
The levels of PM10 and PNC were defined as in Lanki et al (Lanki et al, 2006).
GAMs with penalized splines to control for confounders were used in city-specific models. The outcome was daily number of readmissions and the natural logarithm of the number of persons at risk each day was included as an offset. Pooled estimates were obtained by fixed or random effect models. Lag 0 was used for presenting results.
Table 3 of the article (below) presents the main results of their study.
PNC and PM10 were significantly associated with readmissions for cardiac causes. Point estimates for AMI only were higher than for cardiac causes, but not significant, because of smaller numbers of events resulting in wider CIs. Estimates for angina pectoris were above unity, but lower than for cardiac causes, and not significant.
In two-pollutant models, effects of PNC, CO and NO2 were not affected by PM10 or O3, while the association with PM10 was only slightly attenuated by traffic related pollutants.
4.3. NATIONAL MULTICITY STUDIES
Two multicity studies from France (Eilstein et al, 2004) and Italy (Biggeri et al, 2005) investigated the association between PM and hospital admissions for CVD.
The French study comprised nine cities, but the analysis on BS and CVD admissions was based on 5 cities, and that on PM10 and CVD admissions on 4. No significant associations were found for BS and PM10 at lags 0-1 and 0-5.
The Italian study was based on 8 cities. For two cities, PM10 was measured directly, while for the other six PM10 was estimated as a fraction of TSP. The pooled estimate of the percent change in number of admissions for cardiac causes (ICD9 390-429) for a 10 µg/m3 increase in PM10 (lag 0-1) was 0.77% (0.40, 1.15) under the fixed
report no. 4/11
52
effect model and 0.82% (0.32, 1.32) under the random effect model. Results were similar to the random effect model when several Bayesian models were fitted.
4.4. STUDY DESIGN, STATISTICAL MODEL AND LAG TIME
As for mortality, almost all the studies were time series.
It is likely that some of the early studies (including the APHEA2 study) were affected by the statistical problems connected with the use of S-plus routines. However, in the case of mortality studies, the HEI reanalysis had shown that results changed only slightly by the use of more strict convergence criteria. Greater changes were observed by the use of different statistical models for the control of continuous confounders (HEI, 2003).
Some studies were also affected by multiple comparison problems due to the use of different lag times. The APHEA2 study defined lag time a priori while the HEAPSS study showed results for different lag times, and combined data across cities using the same lag time for each city. No evaluation of medium-term effects of PM on CVD hospital admissions was available.
4.5. INDEX OF EXPOSURE TO PM, FINE, ULTRAFINE AND COARSE PM AND CO-POLLUTANTS
The PM indexes used in the APHEA2 study were BS and PM10, and PM10 and PNC in the HEAPSS study. In both studies, PM10 was not measured directly in all cities, but was estimated either from PM13 or TSP.
Within the APHEA2 study, the RR estimates for both PM10 and BS are given for an increase of 10 µg/m3. The IQR however, was generally higher for PM10 (range 8.7-23.6 µg/m3) than for BS (range 7.3-17.5 µg/m3). However, when BS and PM10 were included together in a model, the effect of PM10 disappeared, while the effect of BS increased for cardiac causes overall, and was only slightly decreased for IHD.
In the HEAPSS study, no clear effect of PM10 was evident, except for certain subgroups. A problem in evaluating the results for PNC is the fact that PNC was estimated from regression on TSP or PM10 and other co-pollutants, rather than estimated directly. It is not clear how this may have affected the results. Overall, when all five cities were included, no clear effect emerged, while PNC was directly and significantly associated with AMI admissions when the analysis was restricted to the 3 cities that used hospital admission registers to ascertain the outcome, excluding the two cities where AMI registers were used. However, as previously pointed out, the results for PNC are not entirely convincing, given the exclusions a posteriori.
The conclusion of the HEAPSS study that AMI admissions are mostly related to traffic related air pollution are supported by a case-crossover study on non fatal AMI from Augsburg, Germany (Peters et al, 2004), that did not measure PM exposure, but found that transient exposure to traffic appeared to increase the risk of AMI.
In another article, the data from Rome included in the HEAPSS study were analyzed with a case-crossover design, confirming that in Rome PNC was associated with AMI risk at lag 0 (D'Ippoliti et al, 2003).
report no. 4/11
53
One study from the West Midlands conurbation of the UK (Anderson et al, 2001) examined the association between fine (PM2.5) and coarse (PM2.5-10) PM on CVD emergency admissions separately. For the cardiovascular causes considered - i.e. all cardiovascular (ICD9 390-459), cardiac (ICD9 390-429), IHD (ICD9 410-414) at age 65+ years and stroke (ICD9 430-438) at age 65+ years – no consistent associations emerged with either BS, PM10, PM2.5 or PM2.5-10 (see Table 4 from the article, below).
The only significant finding was a strong negative association between PM2.5-10 (and SO2) and daily admissions for stroke at age 65+ years.
A small study from Trondheim, Norway, found no association between PM2.5 and cardiovascular admissions (Mannsaker et al, 2004).
As concerns adjustment for other co-pollutants, the APHEA2 presented result for two–pollutant models, while the HEAPSS briefly discussed them in the discussion.
4.6. OUTCOME DEFINITION AND EFFECT FOR DIFFERENT CAUSES OF HOSPITAL ADMISSION AND AGE GROUPS
The main problem in the definition of outcomes was the difficulty to distinguish, in some studies, between emergency and elective admissions. In the APHEA2 study, emergency admissions were not available for three cities (Milan, Paris and Rome). The authors tried to exclude admission diagnosis where a high proportion (over 65-70%) were elective, based on data from London. It is difficult to estimate how this
report no. 4/11
54
may have affected the estimates. However, from the graphs presenting city-specific results, no difference was evident between these three cities and the others.
Within the range of CVD morbidity, different outcomes were considered in the various studies. The APHEA2 found associations with cardiac (ICD9 390-429) deaths overall, stronger over 65 years of age, and for IHD (ICD 410-413) over age 65 years. No clear association emerged for IHD below age 65 years, and for stroke over 65 years.
The outcome considered in the HEAPSS study was AMI (ICD 410). This outcome is overlapping with the IHD outcome considered in the APHEA2 study. To estimate comparability, in Italy in 1999 there were 89,079 hospital admissions for AMI, and 147,927 for the codes 411-413 [source Website of the Italian Ministry of Health, http://www.ministerosalute.it/programmazione/sdo/ric_informazioni/]. In contrast to the APHEA2 study, the HEAPSS study found the strongest associations for fatal AMI below age 65 years. This result should be interpreted with caution because apparent associations from subgroup analyses may arise by chance alone, given the multiple comparisons performed and the greater random variation due to the reduced sample size. Moreover, in the HEAPSS study the association with PNC in the older age group appeared, if anything, stronger for non-fatal events. These subgroup results are based on three cities only, since the other two cities were excluded because they had data on younger subjects only (even though those were the ones among whom the strongest relation was found in the other three cities).
The two multicity studies were based on partially overlapping populations, since three cities (Barcelona, Rome and Stockholm) were included in both projects.
As for mortality studies, there was some evidence from several studies that the effect of air pollution may be stronger in warmer or drier seasons/areas.
Finally, the HEAPSS study also analyzed the effects of PM on a susceptible population, i.e. survivors from a first AMI, and found stronger associations in this group than in the general population.
4.7. CONCLUSIONS
In the multicentric APHEA2 study, a significant increase in cardiac admission was reported for PM exposure, which might have resulted, in part, from confounding by NO2, a marker of traffic pollution. Similar results were reached in the general-population analysis of hospitalization for myocardial infarction in the HEAPSS multicentric study. Results in survivors of a first myocardial infarction were stronger than those in the general population. A multicentric study from Italy found an association with PM exposure, while a similar French study reported no association.
The interpretation of studies on cardiac admissions and PM exposure is complicated by the use of different lag times and the combination of emergency and elective hospital admissions (as well as other aspects of the definition of outcome) in some of the studies. Despite these limitations, the evidence from European studies supports the hypothesis of an association between air pollution and hospital admissions for cardiovascular diseases.
report no. 4/11
55
5. SHORT-TERM PM EXPOSURE AND RESPIRATORY HOSPITAL ADMISSIONS
Table III (modified and updated from EPA 2004 Table 8B-2) presents European studies investigating the association between short-term ambient PM exposure and respiratory hospital admissions. These include two international multicity studies, the APHEA1 (Sunyer et al, 1997) and APHEA2 (Atkinson et al, 2001), two national multicity studies from France (Eilstein et al, 2004) and Italy (Biggeri et al, 2005), and a number of small studies from various countries. The two national multicity studies are the same already briefly discussed in the chapter on CV hospital admissions, since both outcomes were presented in the same article.
5.1. THE APHEA1 STUDY
An outline of the APHEA1 project has been already given in the chapter on mortality. One article investigated emergency admissions for asthma in Barcelona, Helsinki, London and Paris, i.e. the four cities participating to the APHEA project that collected data on asthma (Sunyer et al, 1997). Barcelona collected data on asthma emergency room visits in the age group 15-64 years, while London and Helsinki on daily asthma emergency admissions in children (0-14 years) and adults (15-64 years), and Paris on total asthma admissions in both age groups. Data for BS were not available for Helsinki. Thus, the analysis in children was limited to two cities (London and Paris), to which Barcelona was added for adults. The same two stage procedure (in the first stage an estimate of the association between either PM10 or BS was obtained for each city using the same statistical model; in the second stage city-specific effect estimates were regressed on city-specific covariates to obtain an overall estimate and to explore confounding, using both fixed and random effects models) was used as for the mortality analysis. Similarly, several one day and cumulative lags were fitted, and the best fitting one day and cumulative model was chosen for each centre. The study period was from 1986-87 to 1989-91, depending on city.
Table 2 of the paper (below) gives city specific and combined results for the age group 15-64 years.
report no. 4/11
56
The pooled percent daily increase for a 50 µg/m3 increase in BS was 2.1% (-1.5, 5.9) for single day lags, and 3.0% (-1.9, 8.1) for cumulative lags. Single day lags used were 3 for Barcelona, and 0 for London and Paris. Cumulative lags combined were 0-3 for Barcelona and Paris, and 0-1 for London.
Table 3 of the paper (below) presents results for children.
The pooled estimates under age 15 years were 3.0% (-2.1, 8.4) for single day lags, and 4.6% (-2.2, 12.0) for cumulative lags. Single day lags used were 0 for London and 2 for Paris, and cumulative ones 0-3 for London and 0-2 for Paris.
The reason given for the use of different lags in each city is that local factors like wind direction, size of the city or location of the monitoring sites vary between cities. It is however not clear why lags should differ between children and adults: in London the chosen cumulative lag was 0-1 for adults and 0-3 for children and in Paris the single day lag was 0 for adults and 2 for children. For adults, the pooled estimate using lag 0 in all cities was 2.0% (-1.7, 5.7), and 0.7% (-2.8, 4.5) for lag 1.
When two-pollutant models were considered (Table 4 of the paper, below), the effect of BS disappeared after adjustment for NO2 or SO2, while the effects of these two pollutants were still present after adjustment for BS.
report no. 4/11
57
5.2. THE APHEA2 STUDY
In the APHEA2 study, one article was dedicated to the association between PM10 and respiratory admissions (Atkinson et al, 2001). The methods, including the eight cities/areas included, the study period, PM10 estimation and the statistical model used were the same as for the CVD admission analysis (Le Tertre et al, 2002a).
Four different outcomes were considered, i.e., daily admissions for asthma (ICD9 493) at ages 0-14 and 15-64 years, COPD and asthma (ICD9 490-496) at age 65+ years, and all respiratory disease admissions (ICD9 460-519) at age 65+ years. The authors specify that, wherever possible, emergency admissions resulting in overnight hospital stay were specified, in order to exclude elective admissions and those resulting only in an emergency room visit. However, unlike for CVD admissions, no investigation of the proportion of elective admissions for specific admission diagnoses, and consequent exclusion of diagnoses with high prevalence of elective admissions, was performed.
In the following table (Table 3 of the article) the percent change in daily admissions associated with a 10 µg/m3 increase in PM10 or BS for each city is given, together with the pooled estimates. With few exceptions, the city-specific estimates were positive for all four endpoints, although only for a few of them the confidence interval did not include zero increase. For 5 of the 8 pooled results there was evidence of heterogeneity between cities.
For asthma in children the summary estimates were 1.2% (0.2, 2.3) for PM10 and 1.3% (0.3, 2.4) for BS, while for asthma in adults they were 1.1% (0.3, 1.8) for PM10 and 0.7% (-0.3, 1.8). Thus, for PM10 there was no evidence that children were
report no. 4/11
58
relatively more susceptible than adults to PM-induced asthma exacerbation. In the age group 65+ years, the pooled estimates for COPD plus asthma were 1.0% (0.4, 1.5) for PM10 and 0.2% (-1.3, 1.6) for BS. Corresponding estimates for all respiratory diseases were 0.9 % (0.6, 1.3) and 0.1% (-0.7, 0.9), respectively.
report no. 4/11
59
Two-pollutant models were also fitted for PM10 (Table 4 of the paper, below).
For asthma in children and adults, adjustment for NO2 had the strongest effect on the PM10 estimates, which were substantially reduced. Ozone was the major confounder for COPD plus asthma in elderly people, while no substantial change in the point estimate for all respiratory diseases in the elderly was observed (although the inclusion in the model of correlated variables resulted in inflated standard errors of the estimates).
Finally the effect modification of several city-specific variables was considered. For asthma in children, humidity (inversely), smoking prevalence (inversely) and percent of population over 65 years (directly) reduced the heterogeneity χ2 by over 40%. For the two endpoints considered for the elderly, higher mean ozone levels resulted in stronger PM effects.
5.3. NATIONAL MULTICITY STUDIES
In the same articles in which results for CVD admissions were presented, the French (Eilstein et al, 2004) and the Italian (Biggeri et al, 2005) studies investigated also the association between PM and hospital admissions for respiratory diseases.
The French study found a positive significant association between BS day to day variations and all respiratory admissions in the age group 0-14 years, and no effect at age 65+ years. The Italian study found a direct association between PM10 and respiratory admissions at all ages, with however large heterogeneity between cities.
5.4. FINE AND COARSE PM
The study from the West Midlands conurbation of the UK (Anderson et al, 2001) examined the association between fine (PM2.5) and coarse PM (PM2.5-10) with respiratory emergency admissions. The endpoints considered were daily emergency hospital admissions for respiratory causes (ICD9 460-519) at all ages and in the age groups 0-14, 15-64 and 65+ years, asthma in the age groups 0-14 and 15-64 years and COPD at ages 65+ years (Table 5 of the paper, below).
report no. 4/11
60
For all respiratory causes, significant positive associations were found in the age group 0-14 years for PM10 and BS, and borderline positive associations for PM2.5 and PM2.5-10, although the strongest positive association was found with SO2.
For asthma in children, there were significant direct associations with PM10 and BS, while the association with PM2.5 and PM2.5-10 were not significant (Table 6 of the paper, above). The only other significant association was a negative one between asthma at 15-65 years and coarse PM.
5.5. CONCLUSIONS
In the multicentric European APHEA1 study, an association was reported between NO2 and SO2 exposure and respiratory hospital admissions. In the APHEA2 study several pollutants and respiratory diseases were considered, with positive associations detected for most pollutant-outcome combinations, albeit mostly not significant. Additional support for an association between air pollution, principally gaseous pollutants, and respiratory admissions is provided by a multicentric Italian study.
report no. 4/11
61
6. SHORT-TERM PM EXPOSURE AND OUTPATIENT VISITS
A few studies examined the relation between PM levels and outpatient visits in emergency department or at home, or dispensing of respiratory drugs. Most studies used as outcomes visits for respiratory diseases overall or asthma, although also allergic rhinitis and cardiovascular causes were considered in some studies. These studies are described in Table IV (modified and updated from EPA 2004 Table 8-B3).
report no. 4/11
62
7. ANALYTIC STUDIES ON LONG-TERM PM EXPOSURE AND MORTALITY
In 2003, at the time of the EPA report, only one cohort study from the Netherlands (Hoek et al, 2002) was available. In more recent years, other four studies from Sweden (Rosenlund et al, 2006), Norway (Naess et al, 2007), Germany (Gehring et al, 2006), and France (Filleul et al, 2005) have been published.
The description and results of each of these five studies are shown in Table V. In the following sections the methods and results of the four longitudinal studies are compared.
The Swedish study (Rosenlund et al, 2006) is a population based case-control study on first AMI. Since separate analyses have also been performed for fatal AMI, the study has been included also in this table. Its results, however, will be discussed in chapter VII, together with other studies investigating the association between long-term exposure to PM and CVD morbidity.
7.1. STUDY DESIGNS, SAMPLE SIZE, TIME FRAME AND POPULATION DEFINITION
The Norwegian study (Naess et al, 2007) is a record linkage study based on all inhabitants of Oslo (n=143,842) aged 51-90 years on January 1, 1992 and followed up until 1998.
The study from the Netherlands (Hoek et al, 2002) is a case-cohort study based on a random sample of 5,000 subjects selected from the over 120,000 participants to the Netherlands Cohort Study on Diet and Cancer aged 55-69 years in 1986, recruited from 204 of 714 municipalities with computerized population registries and covered by a cancer registry, who completed a self-administered questionnaire in 1986. Ninety percent of the sample (4,492 subjects) was successfully followed up until 1994.
The German study (Gehring et al, 2006), is based on about 4,800 women aged 50-59 years who participated to several cross-sectional surveys conducted in various towns of North Rhine Westphalia between 1985 and 1994 and followed up until 2002-2003. The overall response rate to the surveys was 70%, and data on follow-up and address were obtained for 97% of the sample.
The study from France (Filleul et al, 2005) is based on the PAARC (Pollution Atmosphérique et Affections Respiratoires) cohort study and includes 14,284 adults residing in 7 French cities, aged 25-59 years at enrollment in 1974 and followed up until 2001 for vital status. Cause of death was available until 1998 for 96% of the sample.
7.2. ASSESSMENT OF EXPOSURE TO PM AND CO-POLLUTANTS
For all the studies, exposure assessment was based solely on the home address, without considering other possible sources of exposure (other locations or occupation). The French study, however, excluded from the sample heads of households who were manual workers, to avoid confounding from other heavy
report no. 4/11
63
exposures. Occupation was controlled for in some studies (see section on statistical model and control of confounding).
Average ambient concentration at the home address over a period of several years was considered as indicator of exposure in all the studies. The follow-up studies evaluated exposure for 1-5 years around the baseline.
Several methods were used for estimating ambient PM concentrations based either on data from air monitor stations placed ad hoc (Filleul et al, 2005) or on air dispersion modeling systems (Hoek et al, 2002; Gehring et al, 2006; Naess et al, 2007), even corrected for distance to a major road (Hoek et al, 2002; Gehring et al, 2006).
Several PM indexes were used, including PM10, PM2.5, BS and TSP. Although other gaseous pollutants were measured in several studies, no mutual adjustments were performed.
7.3. STATISTICAL MODEL AND CONTROL OF CONFOUNDING
Cox proportional hazard models were used in the four longitudinal studies. In general, no other environmental variables were included. Sex and age were controlled for in all the studies. The record linkage study from Norway (Naess et al, 2007) recovered information on education and occupation (manual/non manual) from census data for 77% of the sample. Smoking and education/socioeconomic status were controlled for in most of the other studies, together with other potential confounders (physical activity, diabetes, body mass, diet, occupational exposure).
7.4. MAIN RESULTS
The hazard ratio (HR), as an estimate of RR, for deaths from all causes from the Norwegian record linkage study according to approximate quartiles of long-term PM2.5 exposure are presented in Table 2 of the article (Naess et al, 2007) (below), separately by sex and age at enrollment (51-70 and 71-90 years).
Compared to those in the lowest exposure categories, the HRs were 0.96 (0.89, 1.04), 1.12 (1.03, 1.22) and 1.48 (1.36, 1.60) in subsequent quartiles of exposure for men aged 51-70 years. Corresponding estimates in older men were 0.99 (0.93, 1.06), 1.10 (1.03, 1.17) and 1.19 (1.12, 1.27). Estimates in women were very similar to those of men of the corresponding age group. Adjustment for education and occupational class did not materially modify the results.
report no. 4/11
64
Cause- and sex-specific HRs for PM10 and PM2.5 (and NO2) are presented in Table 3 of article for the age group 51-70 years and in Table 4 for the age group 71-90 years (Naess et al, 2007) (below).
report no. 4/11
65
These results are difficult to interpret for two reasons. First, it is not clear what are the categories compared in the models. In the headings of the table the authors say “quartile increase”. It is not clear whether this means that these are risks for an increase of exposure from one quartile (i.e. the exposure variable was entered as an ordered variable coded 1, 2, 3 or 4 according to the exposure quartile) or whether the forth and first quartiles were compared. In the first and more likely case, Table 2 points to a limitation of this analysis, since there was no increase in risk between the first two quartiles, and the biggest increase was observed between the third and fourth quartiles, particularly for younger subjects. In the second case, half of the
report no. 4/11
66
population (i.e. those in the second and third quartile) would not contribute to the HR estimates.
Moreover, division of the population in four subgroups according to age and sex renders the results difficult to interpret, in particular in the absence of clear justifications and formal tests of heterogeneity between subgroups.
Table 4 of the corresponding paper, below, shows the main results from the Netherlands study (Hoek et al, 2002) according to two different models. In the first model, average regional background concentration of BS and distance of the residence from a major road were entered separately, while in the second model these two were combined by adding a constant to BS exposure estimates for those living near a major road.
Subjects living near a major road had an increased risk of all cause mortality, particularly after adjustment for several confounders and in individuals who had lived 10 years or longer at their 1986 address (RR=1.53, 95% CI 1.01, 2.33). Most of the excess risk observed in total mortality appeared to be attributable to cardiopulmonary causes of death. Additional analyses of the Netherlands study have been reported in an abstract form (Beelen et al, 2007). Using a case-cohort approach, adjusted RRs for an increase of BS by 10 µg/m3 were 1.03 (0.91, 1.17) for all causes except accidents and violence, 1.04 (0.90, 1.20) for cardiopulmonary, 1.16 (0.91, 1.48) for respiratory, 0.99 (0.80, 1.23) for lung cancer, and 1.03 (0.91, 1.16) for non-cardiopulmonary, non-lung cancer mortality. Corresponding RRs in the full cohort analyses, which included confounder information only on age, gender, active smoking and area-level socio-economic status, were 1.05 (1.00, 1.10) for all natural causes, 1.06 (0.98, 1.15) for cardiopulmonary, 1.22 (0.99, 1.49) for respiratory, 1.03 (0.88, 1.20) for lung cancer, and 1.00 (0.96, 1.11) for non-cardiopulmonary, non-lung cancer mortality (Beelen et al, 2007).
In the German study too, the effects of PM and traffic appeared stronger for - or limited to - cardiopulmonary mortality (Gehring et al, 2006). In that study, a similar approach to the Dutch study was used, i.e., background exposure and distance of the residence from a road with >10,000 cars/day were estimated independently. Background exposure was computed as average concentrations for the year of recruitment and for the four years preceding recruitment.
report no. 4/11
67
The adjusted RRs for a 7 µg/m3 increase in PM10 in the baseline year were 1.08 (0.94, 1.25) for all cause mortality, 1.34 (1.06, 1.71) for cardiopulmonary mortality and 0.92 (0.80, 1.10) for non cardiopulmonary non lung cancer mortality (see Table 4 of the corresponding paper, above). When exposure was measured for five years the RR for cardiopulmonary mortality increased to 1.59 (1.23, 2.04). Women living near a major road had a 70% (2, 181) increased risk of dying for cardiopulmonary causes. In the text the authors state that mutual adjustment of air pollution effects and proximity to roads did not change these estimates.
The French study (Filleul et al, 2005) estimated average exposure for the period 1974-76 (see Table 4 of the paper, below). When all 24 areas were included in the analysis, long-term exposure to BS or TSP did not affect the risk of mortality from all non-accidental causes, cardiopulmonary causes or lung cancer, all the RR ranging between 0.97 and 1.01.
report no. 4/11
68
However, the authors repeated the analysis excluding subjects from six areas for which the ratio NO/NO2 (measured in ppb) was more than 3. The reason for this exclusion was that NO is the principal component of the NOx emissions by cars and the high NO/NO2 ratio indicated that the monitoring station was close to a very busy roadside. Thus, the exposure measure was heavily influenced by local traffic and non representative of the mean exposure of the population of the entire area. When the analysis was restricted to 18 areas, the RR for a 10 µg/m3 increase in TSP was 1.05 (1.02, 1.08) for total non-accidental mortality, 1.00 (0.92, 1.10) for lung cancer and 1.06 (1.01, 1.12) for all cardiopulmonary causes. Corresponding estimates for BS were 1.07 (1.03, 1.10), 1.03 (0.92, 1.15) and 1.05 (0.98, 1.12).
7.5. EFFECT MODIFICATION BY AGE, SEX, SOCIOECONOMIC STATUS OR OTHER VARIABLES
As we have already commented, the Norwegian study (Naess et al, 2007) shows all results separately for four subgroups defined by sex and age without however a formal evaluation of differences between subgroups. For total mortality, results appeared similar betweens sexes, and the associations with PM2.5 were stronger in the younger age group, although, mainly based on models for NO2, the authors argue for the opposite: The HRs for all cause mortality in the highest PM2.5 exposure quartile were 1.44 (1.32, 1.58) in younger men and 1.18 (1.10, 1.26) in older men, 1.41 (1.27, 1.57) in younger women and 1.11 (1.05, 1.17) in older women.
The German study (Gehring et al, 2006) was based on women who were of similar age at recruitment, and thus could not investigate modification effects by age and sex. In the text the authors write that they “found modestly stronger NO2 and PM10 effects on all-cause mortality for women with a low SES, for current smokers, and for nonsmokers exposed to environmental tobacco smoke”.
In the French study (Filleul et al, 2005), subgroup analyses by sex, smoking status, occupational exposure to dust, gases and fumes, education and age were shown in a graph (see Figure 2 of the paper, below). No clear differences were observed with age, sex, or smoking. The highest point estimates were observed in more educated or occupationally exposed subjects.
report no. 4/11
69
7.6. COMPARISON OF EUROPEAN WITH LEADING NORTH AMERICAN STUDIES
European studies suggest that mortality from cardiopulmonary diseases is associated with long-term exposure to air pollution, but provide limited evidence that various measures of PM are related to total mortality. This is in apparent contrast with the leading North American studies, consistently providing direct associations between fine particles and total mortality. In the Harvard Six Cities study, 8,096 white participants from various cities of the USA were followed since the mid-1970 to 1998. The adjusted RRs for an increase of PM2.5 of 10 µg/m3 were 1.16 (1.07, 1.26) for total mortality, 1.28 (1.13, 1.44) for cardiovascular diseases, 1.08 (0.79, 1.49) for respiratory, 1.27 (0.96, 1.69) for lung cancer and 1.02 (0.90, 1.17) for other causes (Laden et al, 2006). In the ACS-CPS-II study, that linked air pollution data with the individual data of approximately 500,000 adults from the USA, followed from 1982 to the end of 1998 (Pope et al, 2002) (Pope et al, 2004), the adjusted RRs for an increase of PM2.5 by 10 µg/m3 were 1.06 (1.02, 1.11) for all-cause mortality, 1.12 (1.08, 1.15) for cardiovascular diseases plus diabetes, 0.92 (0.86, 0.98) for respiratory, 1.14 (1.04, 1.23) for lung cancer and 1.01 (0.95, 1.08) for all other causes. In that study, however, an interaction with education was evident. In fact, for subjects in the highest category of education (>high school), no association was found between increasing levels of PM2.5 and total, cardiopulmonary and lung
report no. 4/11
70
cancer mortality. This is in contrast with the findings from the French study (Filleul et al, 2005), where the highest point estimates were observed in more educated subjects. Apart for chance and/or bias, at least part of the discrepancies could be explained by the different PM measure investigated. In fact, most European studies provided data on PM10 only, and not on finer PM.
7.7. CONCLUSIONS
In a Norwegian cohort study, a significant association was reported between PM exposure and mortality, although the interpretation of the results is not straightforward. Mortality was associated with various measures of air pollution in cohort studies from the Netherlands, Germany and France, and the excess risk in these studies was mainly attributed to cardiovascular and respiratory diseases.
report no. 4/11
71
8. ECOLOGIC STUDIES ON LONG-TERM PM EXPOSURE AND MORTALITY
One study compared mortality rates before and after the ban of coal sales in Dublin (Clancy et al, 2002). Given the deterioration of the air quality in the 1980s after a switch from oil to other fuels, mainly bituminous coal for domestic space and water heating, the Irish Government banned the marketing, sale and distribution of bituminous coals within the city of Dublin on September 1990. The Figure 1 of the paper, below, shows the consequent declines in BS and SO2.
Average BS concentrations declined by 35.6 µg/m3 (70%) after the ban.
The authors compared age-adjusted death rates (directly standardized on the 1991 Irish census population) for the 72 months before (Sep 1984 to Aug 1990) and after (Sep 1990 to Aug 1996) the ban.
report no. 4/11
72
Table 2 of the paper, below, presents mortality rates for the two periods, overall and by season, for total non-accidental causes, and for cardiovascular, respiratory and other causes, separately.
Significant decreases were observed for all non-accidental and cardiovascular mortality in all seasons, and the decreases were more marked in winter. Significant decreases were observed for respiratory mortality overall, in winter and spring, while no significant change was observed overall for other causes. The authors concluded that control of PM could substantially diminish daily death. However, it is worth noticing that, during the same period, mortality declined in several European countries. Thus, a causal link between the decline in mortality and the ban of coal sales is far from established. On the other hand, the authors interpreted the fact that the biggest declines were observed in winter, when the use of coal was highest, as suggestive of their hypothesis.
Two studies compared air pollution levels in 1,030 census enumeration districts in Sheffield, UK, to mortality and emergency hospital admission rates from coronary heart disease (Maheswaran et al, 2005b) and stroke (Maheswaran et al, 2005a). For each district a 5-year average PM10 concentration for the period 1994-1999 was computed. The districts were then divided according to quintiles of PM10 concentrations. The mean PM10 concentration in the highest quintile was 23.3 µg/m3, and 16.0 µg/m3 in the lowest one. Mortality and emergency hospitalization rates
report no. 4/11
73
were then computed for the districts’ quintiles. For coronary heart disease, the rate ratios (adjusted for sex, age, deprivation and smoking prevalence) for the highest quintile compared to the lowest one were 1.08 (0.96, 1.20) for mortality and 1.01 (0.90, 1.14) for hospital admissions. Corresponding values for stroke were 1.33 (1.14, 1.56) and 1.13 (0.99, 1.29).
report no. 4/11
74
9. LONG-TERM PM EXPOSURE AND CARDIOVASCULAR MORBIDITY
Table VII summarizes results of studies on the effect of long-term exposure to PM and cardiovascular morbidity.
A population-based case-control study on 1,393 cases of first AMI and 1,870 population controls resident in Stockholm county, aged 45-70 years was conducted between 1992 and 1994 (Rosenlund et al, 2006). Response rates to the mailed questionnaire varied between 70-85% depending on sex and case-control status. For each subject, exposure to PM10 and PM2.5 was reconstructed from 1960 to a year prior to enrollment (1992-1994), i.e., for over 20 years, using data on traffic around the home address only. Only data for long-term exposure to PM10 were used in the analysis, given the high correlation between PM10 and PM2.5 (r=0.998).
Logistic regression models adjusted for the matching variables, i.e. age, sex and hospital catchment area, and smoking, physical inactivity, diabetes and SES were used to compute odds ratios (OR). Hypertension, body mass, job strain, diet, passive smoking, alcohol and coffee intake and occupational exposure to motor exhaust and other combustion products were also evaluated, but did not appear to confound the relation with PM.
Table 2 of the article (below) presents the main results of the study.
Overall, there was no association between PM10 and AMI (OR=1.00, 95% CI 0.79, 1.27). The point estimate was above unity for fatal cases (OR=1.39, 95% CI 0.94, 2.07) and below unity for non fatal cases (OR=0.92, 95% CI 0.71, 1.19) although none of these estimates was statistically significant. A borderline significant association was found when fatal cases were further restricted to those who died out of hospital (OR=1.84, 95% CI 1.00, 3.40). These were 84 out of 1,357 cases (6%). Although the authors interpret this finding as supportive of a possible association between air pollution exposure and mortality, random variation in small subgroups cannot be excluded.
Moreover, it is not clear how covariates were measured for fatal cases in general, and specifically in this group that died out of hospital. If proxy interviews were used for deceased cases, and direct interviews for controls, the effect of differential collection of covariates has to be considered.
report no. 4/11
75
The two ecologic studies from Sheffield, UK, discussed in the previous chapter found rate ratios of 1.01 (0.90, 1.14) for emergency hospital admissions for coronary heart disease and of 1.13 (0.99, 1.29) for emergency hospital admissions for stroke for subjects living in census enumeration districts with the highest levels of PM10 concentrations, as compared to the lowest ones.
Two studies investigated the association between long-term exposure to PM and prevalence of coronary or ischemic heart disease (Solomon et al, 2003; Hoffmann et al, 2006). Being based on prevalence data, these studies should be interpreted with caution.
A cross sectional postal survey on 1,166 women aged 45 years or more was conducted in 11 wards in the UK for which 30 years long concentration data on BS were available (Solomon et al, 2003). After adjustment for potential confounders, women living in areas with high BS concentrations did not report higher prevalence of ischaemic heart disease, as compared to women living in low BS areas (OR=1.00, 95% CI 0.7, 1.4).
Another study used baseline data from a newly started cohort study on 4,814 subjects aged 45-75 years from three large adjacent cities in the Rhein-Rhur region, Germany, to investigate the relation between background PM2.5 exposure, residence close to a major road, and prevalence of coronary heart disease (Hoffmann et al, 2006). After adjustment for several potential confounders, the OR for those with high traffic exposure was 1.75 (1.16, 2.62), and for background PM2.5 exposure 0.55 (0.14, 2.11), although the measurement unit for the latter OR is not given.
report no. 4/11
76
10. LONG-TERM PM EXPOSURE AND RESPIRATORY MORBIDITY OR SYMPTOMS
Many European studies have investigated the relation between PM and respiratory symptoms (Table VIII, modified and updated from EPA 2004 Table 8B8). The majority of these studies were on children, but some also in adults. Several different endpoints have been considered.
10.1. LONG-TERM PM EXPOSURE AND ASTHMA IN CHILDREN
A study from the Netherlands (Brauer et al, 2002) recruited about 4,000 children (during the seconds trimester of pregnancy) and evaluated the development of respiratory infections and asthmatic and allergic symptoms using questionnaires completed by the parents when the children were 2 years old. Outdoor concentrations of traffic-related PM2.5 and soot at the home address of the children were estimated by means of a validated model. At two years of age, 176 children (5%) had developed physician-diagnosed asthma. After adjustment for several potential confounders, the OR for an increase equal to the IQR (3.2 µg/m3 for PM2.5 and 2x10-5/m for soot, corresponding to 2.9 µg/m3 of elemental carbon) were 1.12 (0.84, 1.50) for PM2.5 and 1.12 (0.88, 1.43) for soot.
A cross-sectional survey conducted in France (Penard-Morand et al, 2005) on a random sample of 6,672 schoolchildren aged 9-11 years found an adjusted OR of asthma during the past year for an increment of PM10 by 10 µg/m3 of 1.23 (0.77, 1.95), that was virtually unchanged after further adjustment for SO2.
Among the 3,193 French children aged 5-9 years and included in the PAARC survey (Baldi et al, 1999), 195 (6%) were found to be asthmatic. The adjusted OR for an increase of 50 µg/m3 in annual mean concentrations was 0.99 (0.81, 1.20) for TSP and 1.08 (0.85, 1.38) for BS.
10.2. LONG-TERM PM EXPOSURE AND LUNG FUNCTION IN CHILDREN
Spirometry was performed twice a year between 1994 and 1997 in 975 Austrian schoolchildren of grades 2-3 at baseline (Horak et al, 2002). After adjusting for potential confounders (sex, atopy, passive smoking, initial height, height difference, site, initial lung function) an increase of summer PM10 by 10 µg/m3 was associated with a decrease in FEV1 growth of 84 mL yr-1 and 329 mL s-1 yr-1 for MEF25–75. No association was found for FVC or in winter (see Table 4 of the paper below).
report no. 4/11
77
A study from the UK (Kulkarni et al, 2006) obtained airway macrophages and assessed the carbon content from 64 healthy children. PM10 concentrations near the home address, FEV1 and FVC were also measured. They found that an increase of 1.0 μg/m3 in modeled primary PM10 was associated with a 0.10 μm2 (0.01, 0.18) increase in the carbon content of airway macrophages. Increased primary PM10 was inversely associated with the percentage of the predicted FEV1 (P = 0.04) and the FEF25–75 (P = 0.05). After adjusting for the carbon content of PM10, the carbon content of airway macrophages remained inversely associated with the percentage of the predicted values of FEV1, FVC, and FEF25–75 (P = 0.02, P = 0.04, and P = 0.02, respectively), whereas PM10 was no longer significantly associated with lung function (Table 2 of the paper, below).
report no. 4/11
78
Finally, a study from eastern Germany (Frye et al, 2003) conducted three cross-sectional surveys of lung function in schoolchildren aged 11-14 between the early and late 1990s, and found that lung function of children has increased, while TSP and other pollutant levels had decreased during the study period.
10.3. LONG-TERM PM EXPOSURE AND RESPIRATORY SYMPTOMS IN CHILDREN
Two cohort studies investigated the issue.
The cohort of about 4,000 children from the Netherlands, that has been already discussed for asthma (Brauer et al, 2002), evaluated the development of respiratory infections and asthmatic and allergic symptom using questionnaires completed by the parents when the children were 2 years old.
report no. 4/11
79
Outdoor concentrations of traffic-related PM2.5 and soot at the home address of the children were estimated by means of a validated model. Adjusted OR for the various symptoms and diseases were computed for an increase in exposure equal to the IQR i.e. 3.2 µg/m3 for PM2.5 and 2x10-5/m for soot, corresponding to 2.9 µg/m3 of elemental carbon. The results are presented in Table 4 of the article (above). Statistically significant increases in risk were found between PM2.5 and ear, nose and throat infections (OR=1.20 95% CI 1.01, 1.42) and doctor-diagnosed flu/serious cold (OR=1.00, 1.27). For soot, the only significant association was with ear, nose and throat infections (OR=1.15 95% CI 1.00, 1.33).
A cohort of 4,400 children aged 1-5 years was recruited in Leichestershire, UK in 1998 (Pierse et al, 2006). Parents filled in a respiratory symptom questionnaire at baseline and again in 2001. Exposure to primary PM10 was calculated from the home address using a dispersion model based on estimated traffic emissions from traffic flow data. Adjusted OR for an increase in primary PM10 of 1 µg/m3 were computed.
report no. 4/11
80
The prevalence of symptoms was first evaluated cross-sectionally in the two surveys (Table 2 of the paper, above). The authors then evaluated the OR of developing symptoms in the study period (Table 3 of the paper, below). The ORs were 1.62 (1.31, 2.00) for cough without a cold, 1.19 (0.96, 1.47) for night time cold, and 1.42 (1.02, 1.97) for wheeze.
A number of cross-sectional studies investigated the association between exposure to PM and prevalence of a number of respiratory symptoms. These studies are often difficult to interpret, given the unclear time relation between exposure and outcome and the high number of outcomes considered, and consequent problems related to multiple testing. Their results are summarized in Table VIII.
report no. 4/11
81
10.4. LONG-TERM PM EXPOSURE AND RESPIRATORY MORBIDITY OR SYMPTOMS IN ADULTS
A random sample of about 7,000 persons residing in 21 centres in 10 countries, who participated to the European Community Respiratory Health Study I (ECRHS I) in 1991-93, were re-interviewed in 2000-2002 to investigate the onset of symptoms for chronic bronchitis (response rate 65%). Table 4 (below) of the article presents the main findings.
No association was found between centre level of PM2.5 and chronic phlegm prevalence at follow up. The ORs were 0.97 (0.70, 1.35) for males and 0.99 (0.85, 1.17) for females, but the measurement unit was not given. A constant exposure to traffic was associated with prevalence of chronic phlegm in women (OR=1.86, 95% CI 1.24, 2.77), but not in men (OR=0.88, 95% CI 0.56, 1.38).
The PAARC study (Baldi et al, 1999) collected data also on 20,310 French adults aged 25-59 years, 1,261 (6%) of whom were found to be asthmatic. No association
report no. 4/11
82
was found between long-term exposure to TSP (OR for an increase of 50 µg/m3=1.01, 95% CI 0.92, 1.11) or BS (OR=1.07, 95% CI 0.96, 1.20).
The SAPALDIA study (Zemp et al, 1999) is a cross-sectional investigation in a random sample of 9,651 Swiss adults aged 18-60 years. Asthma prevalence was inversely associated with TSP and PM10 in never smokers, while some respiratory symptoms showed a positive association with PM10.
A cross-sectional postal survey on 1,166 women aged 45 years or more was conducted in 11 wards in the UK for which 30 years long concentration data on BS were available (Solomon et al, 2003). After adjustment for potential confounders, women living in areas with high BS concentrations reported lower prevalence of asthma, as compared to women living in low BS areas (OR=0.7, 95% CI 0.5, 1.0). Productive cough was not associated with BS (OR=1.0, 95% CI 0.7, 1.5).
Three studies investigated the relation between air pollution and lung function in adults. In the SAPALDIA study (Ackermann-Liebrich et al, 1997) significant inverse effects were found between PM10 and FVC and FEV1 in the whole population and separately between never, ex and current smokers. The Health Survey for England (Wheeler and Ben-Shlomo, 2005), including 39,251 participants aged 16-79 years, found that low social class and poor air quality were independently associated with decreased lung function (FEV1), but not asthma prevalence. For men, but not for women, the effects of air pollution on lung function appeared more marked in lower social classes. In a series of cross-sectional surveys on 4,757 women living in the Rhine-Ruhr Basin in Germany (Schikowski et al, 2005), a 7 µg/m3 increase in PM10 was associated with a 5.1% (2.5, 7.7) decrease in FEV1, a 3.7% (1.8, 5.5) decrease in FVC and an OR of 1.33 (1,03, 1.72) for COPD.
Finally, among the first 3,904 participants to the EPIC study residing in Athens, those living in the most polluted areas had a two-fold increase in prevalence of chronic bronchitis, emphysema or COPD (Karakatsani et al, 2003).
10.5. CONCLUSIONS
Three studies on PM exposure and asthma in children from the Netherlands and France reported weak, non-significant associations. Two studies on PM exposure and lung function in children from Austria and UK reported associations with some of the indicators of lung function, which were not confirmed by a further study from eastern Germany. Two cohort studies from the Netherlands and UK reported significant associations with various symptoms.
Studies in adults did not provide consistent evidence of an association between PM exposure and chronic bronchitis or asthma. Studies on air pollution and lung function, on the other hand, reported positive results.
report no. 4/11
83
11. LONG-TERM PM EXPOSURE AND CANCER
Table IX summarizes data on long-term PM exposure and cancer. Most studies considered lung cancer. Data for other neoplasms are sparse. A few studies considered exposure to traffic and childhood cancer, although the underlying etiologic hypothesis is in general connected to benzene, and not PM, exposure. They are listed in Table IX, but not discussed here.
11.1. ANALYTIC STUDIES ON LUNG CANCER
Three (Hoek et al, 2002; Filleul et al, 2005; Naess et al, 2007) of the four longitudinal studies that were discussed in the chapter on long-term exposure to PM and mortality (chapter V) also presented data on lung cancer mortality.
In the cohort study from the Netherlands (Hoek et al, 2002) 60 deaths from lung cancer were recorded. The adjusted relative risk for an increase of 10 µg/m3 in BS was 1.06 (0.43, 2.63), and 1.25 (0.42, 3.72) for a 30 µg/m3 increase in NO2. The number of cases was too small to compute a risk for the indicator variable of living near a major road. Preliminary results of new analyses of the Netherlands study showed that the adjusted RRs for an increase of BS by 10 µg/m3 was 0.99 (0.80, 1.23) in the case-cohort analysis, and 1.07 (0.96, 1.19) in the full cohort analysis, based on 1,935 lung cancer deaths (Beelen et al, 2007).
In the Norwegian record linkage study (Naess et al, 2007), risks were computed separately by sex and age group at baseline (51-70 and 71-90 years), for lung cancer too. For men, no significant excess was found, and the estimated RR for a “quartile increase” in PM10 was 1.07 (0.97, 1.18) in both age groups. Conversely, significantly increased risks were found for women in both age groups: 1.27 (1.13, 1.43) for younger and 1.17 (1.03, 1.33) for older women. Estimates for PM2.5 were virtually identical to those for PM10.
In the French PAARC study cohort (Filleul et al, 2005) 178 deaths from lung cancer were recorded, and the risk of lung cancer associated to an increase in exposure of 10 µg/m3 of TSP was 0.97 (0.94, 1.01) when all 24 areas were considered, and 1.00 (0.92, 1.10) when the analysis was restricted to subjects living in the 18 areas with the ratio NO/NO2 <3. Corresponding values for BS were 0.97 (0.93, 1.01) and 1.03 (0.92, 1.15). The only significant results for lung cancer was for NO2, when the analysis was restricted to 18 areas (RR=1.48, 95% CI 1.05, 2.06).
A nested case-control study within EPIC (European Prospective Investigation on Cancer and Nutrition) cohort also investigated the association between long-term exposure to PM10 and the risk of lung cancer (Vineis et al, 2006). The over 500,000 healthy volunteers aged 35-74 of the EPIC cohort were recruited in 10 European countries between 1993 and 1998. Identification of cancer cases was mostly based on cancer registries during a median follow-up time of 7 years. A nested case-control study was conducted within the cohort. For each lung cancer case within selected areas, who was a non smoker or an ex smoker since more than 10 years, 3 controls were selected, matched on sex, age, smoking status, country of recruitment and follow-up time. Assessment of exposure to traffic-related air pollution was made for the period 1990-1999 using available data from monitoring stations from the various areas under study. Distance of the home from a major road was also computed. Given the different availability of exposure data from the various areas, analyses were based on different number of cases. Of the 271 cases
report no. 4/11
84
and 737 controls included in the study, data on PM10 exposure were available for 113 cases and 312 controls.
Table IV of the article (below) shows the results for pollutant concentrations. After adjustment for the matching variables, the OR was 0.91 (0.70, 1.18) for an increase in PM10 of 10 µg/m3 and 0.98 (0.66, 1.45) for exposures over 27 µg/m3, compared to less than 27 µg/m3. The marked change observed when cotinine was added to the model was interpreted by the authors as an effect of random variation due to small numbers.
The OR was 1.38 (0.87, 2.19) for those living near a heavy traffic road.
A case-control study on lung cancer was conducted among men who died in Trieste, Italy, between 1979 and 1981 and between 1985 and 1986 (Barbone et al, 1995). Cases were 755 men who died of lung cancer, and controls 755 men who had died for other causes, non COPD nor various cancers of the upper aerodigestive tract, urinary tract, pancreas, liver or gastrointestinal system. Total daily deposition of particulate measured at the closest monitor station from the last subject’s residence was used as measure of the subject’s exposure. Compared to <0.18 g/m2/day of deposition of particulate the OR was 1.1 (0.8, 1.5) for 0.18-0.30 and 1.4 (1.1, 1.8) for >0.30 g/m2/day.
Table VIII/1 shows results of the various analytical studies.
report no. 4/11
85
Table VIII/1 Analytic studies investigating the relation between long-term PM exposure and lung cancer.
Study, design Sex, age at recruitment
No. of cases
Risk indicator, measurement unit
Relative risk (95% CI)
Barbone et al., 1995, case-control, Italy
M 755 Particulate deposition 0.18-0.30 vs <0.18 g/m2/day
1.1 (0.8-1.5)
Particulate deposition >0.30 vs <0.18 g/m2/day
1.4 (1.1-1.8)
Hoek et al., 2002, NLCS cohort, the Netherlands
MF, 55-69 60 BS, 10 µg/m3 1.06 (0.43-2.63)
Beelen et al., 2007 MF, 55-69 1,935 BS, 10 µg/m3 1.07 (0.96-1.19) Filleul et al., 2005, PAARC cohort, France
MF, 25-59 178 TSP, 10 µg/m3 (24 areas) 0.97 (0.94-1.01)
TSP, 10 µg/m3 (18 areas) 1.00 (0.92-1.10) BS, 10 µg/m3 (24 areas) 0.97 (0.93-1.01) BS, 10 µg/m3 (18 areas) 1.03 (0.92-1.15) Vineis et al., 2006, case-control nested in the EPIC cohort, 7 European countries
MF, 35-74 Non- or ex-
smokers 113 PM10, 10 µg/m3 0.91 (0.70-1.18)
PM10 >27 vs <27 µg/m3
0.98 (0.66-1.45)
Naess et al., 2007, cohort, Norway
M, 51-70 449 PM10, quartile increase 1.07 (0.97-1.18)
PM2.5, quartile increase 1.07 (0.97-1.18) F, 51-70 295 PM10, quartile increase 1.27 (1.13-1.43) PM2.5, quartile increase 1.27 (1.13-1.43) M, 71-90 424 PM10, quartile increase 1.08 (0.98-1.20) PM2.5, quartile increase 1.07 (0.97-1.18) F, 71-90 285 PM10, quartile increase 1.17 (1.03-1.33) PM2.5, quartile increase 1.16 (1.02-1.32)
Three other studies on the association of air pollution and lung cancer did not directly measure the effect of PM. A case-control study from Poland (Jedrychowski et al, 1990) found ORs of 1.00 (0.75, 1.33) and 1.46 (1.06, 1.99) for subjects with intermediate and high exposure to air pollution, compared to those with low exposure, based on an index combining TSP and SO2 levels. A case-control study from Sweden (Nyberg et al, 2000) did not find significant associations with NO2 (OR=1.05, 95% CI 0.93, 1.18) or SO2 (OR=1.00, 95% CI 0.96, 1.05). A cohort study from Norway (Nafstad et al, 2003) found a RR of 1.08 (1.02, 1.15) for a 10 µg/m3 increase in exposure to NO2. No association was found for SO2 (OR=1.01, 95% CI 0.94, 1.08).
The evidence concerning the association between long-term PM exposure and lung cancer risk is still scanty. Some studies were based on small numbers of cases. The Norwegian linkage study (Naess et al, 2007) included a large number of cases, but
report no. 4/11
86
did not have information on possible confounding factors, particularly smoking. Moreover, an association was found only for women, but not for men, in the absence of a plausible explanation.
In general, the potential influence of residual confounding by smoking must always be taken into account when dealing with weak associations for lung cancer.
11.2. ECOLOGIC STUDIES ON LUNG AND OTHER CANCERS
A study (Nawrot et al, 2007) found a correlation of 0.57 (p=0.028) between age and smoking standardized lung cancer mortality rates in men and annual mean PM2.5 concentrations in 15 European countries. No significant association was found in women (r=0.37, p=0.18).
No clear evidence for an association between residence along main roads and the incidence of cancer in adults was found in Amsterdam (Visser et al, 2004). Overall, cancer incidence in the area around Amsterdam Schipol Airport was similar to national incidence (Visser et al, 2005). A study from northern Italy (Parodi et al, 2005) investigated lung cancer incidence near a coke oven plant, finding no clear excess risk.
11.3. STUDIES ON AIR POLLUTION AND CANCER IN CHILDREN
A few studies evaluated the association between exposure to traffic and childhood cancer, yielding conflicting results (Raaschou-Nielsen et al, 2001; Crosignani et al, 2004; Knox, 2005; Knox, 2006). The main etiologic hypotheses, however, are related to exposure to benzene and polycyclic aromatic hydrocarbons, rather than to PM.
11.4. COMPARISON OF EUROPEAN WITH LEADING NORTH AMERICAN STUDIES
The findings of European studies on PM and lung cancer are not consistent, nor was a clear pattern of risk observed. More important, no clear association emerged in the major European cohort studies including measures of PM exposure. This is at apparent variance with findings from major US studies, which reported direct associations between fine particles and lung cancer mortality. In fact, the adjusted RRs for lung cancer mortality for an increase of PM2.5 by 10 µg/m3 were 1.27 (0.96, 1.69) in the Harvard Six Cities study (Laden et al, 2006), and 1.14 (1.04, 1.23) in the ACS study (Pope et al, 2002). In the ACS study, the association with PM2.5 was limited to men, in agreement with the AHSMOG study, a cohort based on 6,338 nonsmoking white Californian adults and 36 incident lung cancer cases, which found that the RR for an IQR increase of mean concentrations of PM10 was 5.21 (1.94, 13.99) in men. The corresponding estimate for women was not significant and ranged between 1.0 and 1.2 (Beeson et al, 1998). This is in apparent contrast with the findings from the Norwegian record linkage study (Naess et al, 2007), where no significant excess risk was found for men, while there was a positive association in women.
However, most European studies provided data on PM10 only, and not on finer particulates. Other limitations of exposure measurement may account for at least part of these apparent discrepancies.
report no. 4/11
87
11.5. CONCLUSIONS
Five analytical epidemiological studies from Europe provided results on various measures of particulate matter (PM) and lung cancer. A case-control study from Trieste, Italy found that, compared to <0.18 g/m2/day of deposition of particulates, the RR was 1.1 (0.8, 1.5) for 0.18-0.30 and 1.4 (1.1, 1.8) for >30 g/m2/day. In a cohort study from the Netherlands the RR was 1.06 (0.43, 2.63) for an increase of 10 µg/m3 in black smoke. In a French cohort study the RR of lung cancer associated with an increase in exposure of 10 µg/m3 of total suspended particulate (TSP) was 0.97 (0.94, 1.01). In a nested case-control study within a multicentric European cohort the RR was 0.91 (0.70, 1.18) for an increase in PM10 of 10 µg/m3 and 0.98 (0.66, 1.45) for exposures over 27 µg/m3, compared to less than 27 µg/m3. In a Norwegian record linkage study no significant excess risk was found for men, while there was an association in women. A few other studies considered indirect indicators of air pollution, and some ecological studies linked information on air pollution and cancer incidence or mortality. Although some of the latter studies reported positive associations, various limitations in their design hinder a causal inference.
report no. 4/11
88
12. DISCUSSION AND RECOMMENDATIONS FOR FUTURE STUDIES
12.1. STUDIES ON SHORT-TERM EXPOSURE
Given the availability in several areas of data on daily PM concentrations, and of mortality, and in some instances, hospital admissions, several studies have investigated the association between day-to-day variations in PM concentrations and mortality or emergency hospital admissions. Many of these studies, however, have limitations, either because they were based on small populations or had questionable statistical analysis, which in several instances tended to maximize the investigated association. The latter would affect results of a systematic reanalysis of published studies, and the computing of a pooled measure of association. The use of different measures of PM (PM10, BS, TSP etc.) would further complicate the combination of results from various studies. A pooled re-analysis of various studies, based on the original data, and not on published material, would be more informative and valid, although it would require a great effort.
The APHEA2 study is the largest study on the relation between short-term PM exposure and mortality, and provides the best estimate of the relation between PM10 and day-to-day mortality in Europe. The data of the APHEA2 study were based on a large population from several different areas in Europe, and have been thoroughly re-analyzed to evaluate the influence of different statistical models, the medium-term effects of PM (up to 40 days), and the modification effects of some city characteristics, such as temperature or humidity, but also the percentage of people aged >65 years or of smokers.
Although age at death is, with sex, the only individual characteristic readily available from death certification data, the modifying effect of age on the relation between short-term PM exposure and mortality/morbidity has not been fully clarified. The data from the APHEA2 study suggest that the effect of air pollution is larger at older ages. Deaths in younger age groups are relatively rare, and even the largest study, i.e., the APHEA2, had problems with small numbers in some age categories. Only the combination of many datasets, either within Europe, or including studies from other areas of the world, would provide sufficient power to investigate the effects of air pollution in specific age groups, in particular when specific causes of death are investigated.
The issue of confounding by individual characteristics should not be a major issue, since, in time series data, comparisons are made from one day to another, using the same background populations. In the few studies that have analyzed data using a case-crossover design in addition to a time series approach, the results were comparable.
Scant or inadequate information is available in Europe on modification of the effects of air pollution exerted by individual variables (other than age and sex), e.g. socioeconomic status (SES), lifestyle habits (e.g. smoking), underlying medical conditions (e.g. COPD, coronary disease, diabetes), occupational exposures etc. In Rome a case-crossover design investigated the modification effect of SES using two area-based SES indicators, derived from census data (Forastiere et al, 2005). Although yielding interesting results, that analysis was based on small numbers and on a single population. That type of analysis was based on routinely collected data, and could, in principle, be extended to other datasets.
report no. 4/11
89
Information on the separate effects of coarse, fine and ultrafine PM is still limited. Given that more and more areas within Europe have, in recent years, started the measurement of fine and/or ultrafine PM, it is likely that in the next years several studies will investigate the effects of coarse, fine and ultrafine PM separately. This should be the case, for example, of the HEAPSS study. In the cities included in this study, PNC measurements have started in 2002. In a recent publication, PNC values for a previous period have been estimated retrospectively. New data from the HEAPSS study, from more recent periods and based on actual measurements of PNC, should be available in the next future. Investigations aimed at separating the effects of PM of different diameters will have to deal with the problem of the analysis of highly correlated variables, a problem which has also arisen in the investigation of several correlated pollutants, e.g. PM, NO2, CO and SO2.
As of now, very few results are available from central and eastern European countries.
12.2. STUDIES ON LONG-TERM EXPOSURE
Very few studies have investigated the long-term effects of PM exposure on health. Ecologic studies may be suggestive and hypothesis generating, but, given the potential biases inherent in their design, they cannot provide a solid base for inference.
Record linkage studies, such as the one recently published from Norway (Naess et al, 2007) may include a high number of cases, but the lack of individual variables, first of all smoking habit, renders this type of study difficult to interpret.
Only a limited number of analytic studies have been conducted so far. Many of these were based on cohorts designed for different purposes, which have exploited the availability of PM concentration data in the area under study. This has led to studies that were often based on small numbers, often with exclusions of parts of the study population for which exposure data could not be retrieved.
It is possible that for other longitudinal studies data on exposure could be retrieved and analyzed, but limited availability of reliable data on relevant exposures limits the usefulness of this approach. Thus, only the design of ad hoc studies could provide new insights on the issue.
A somewhat less attractive approach is the pooling of existing cohorts, given the heterogeneity of information on variables used to reconstruct air pollution exposure indicators and on potential confounders. Although technically feasible and relatively inexpensive, it is unclear whether this approach would provide more informative data on long-term effects of air pollution than traditional meta-analyses of already published data. Therefore, the establishment of new multicentric cohorts in Europe remains an appealing approach to investigate long-term effect of air pollution in this region.
In short-term time series, comparisons are ideally made between different days within individuals, and adjustment for day of week and holidays is often performed. In contrast, studies on long-term exposure are necessarily based on comparisons between individuals, and on cumulative exposure assessment. Misclassification of exposure is likely to occur and to affect results. To comment about the different models used for attributing exposure at a certain address is beyond the scope of this report. In almost all the studies conducted so far, the exposure to ambient PM
report no. 4/11
90
has been estimated only on the base of the subject’s home address. No additional information was available on characteristics of the home (e.g. floor of residence) or of the subject’s exposure pattern. Collecting and combining information on hours spent at home, at work, outdoors etc. during the day could improve individual estimates of long-term exposure. The collection of data on potential confounders specifically for the analysis of the association between PM exposure and mortality/morbidity would also provide important additional information.
12.3. RECOMMENDATIONS FOR FUTURE RESEARCH ON HEALTH EFFECTS OF AIR POLLUTION IN EUROPE
Previous studies on short-term effects of air pollution in Europe allow a general quantification of the risks. However, information on the possible confounding or modifying effect of covariates, and mainly data on course, fine and ultrafine particles is limited. Still, with the exception of Central and Eastern Europe, where few data are available, further research on short-term effects of air pollution should not be given high priority.
Only a limited number of analytic studies have been conducted in Europe on long-term effects of air pollution, and these were mainly cohorts designed for different purposes. Their ability to provide valid results was limited by sub-optimal strategies for exposure assessment leading to misclassification, low statistical power, and heterogeneity of approaches. While it is possible that additional exposure data can be retrieved for these and other existing longitudinal studies, it is uncertain whether this approach would lead to major advances. In any case, priority should be given to the co-ordination and harmonization of exposure data in existing cohorts.
The establishment of new studies represents the preferred strategy to provide new insights on long-term effects of air pollution. In areas where estimation of long-term PM exposure could be reconstructed, the case-control design should be considered. The design and implementation of new cohort studies would provide, in the long-term, the most reliable information, but would require great effort in terms of time and resources. Such new studies would rely on uniform methods to measure air pollutants, and be based on an increased number of new measuring stations. Participants should be periodically re-contacted and in addition to information on possible outcomes, new information on exposure based on uniform methodology should be collected. Such new cohorts should give priority to the inclusion of populations from Central and Eastern Europe, where air pollution is higher and adverse health effects may be detected sooner than in less polluted areas of Europe.
report no. 4/11
91
TAB
LE I
- SH
OR
T-TE
RM
PA
RTI
CU
LATE
MA
TTER
EXP
OSU
RE
AN
D M
OR
TALI
TY (D
eriv
ed fr
om E
PA T
AB
LE 8
A-1
). R
efer
ence
, Loc
atio
n,
Year
s, P
M In
dex,
Mea
n or
Med
ian
(µg/
m3 ).
Stud
y D
escr
iptio
n. M
ode
ling
met
hods
: lag
s,
smoo
thin
g, c
o-po
lluta
nts
and
cova
riate
s.
Res
ults
and
Com
men
ts.
PM In
dex,
lag,
per
cent
exc
ess
risk%
(95%
LC
L, U
CL)
APH
EA 1
Kat
souy
anni
et
al.
(199
7)
(Kat
souy
anni
et a
l, 19
97)
12
Eur
opea
n ci
ties
(AP
HE
A),
19
75-1
992
(stu
dy
year
s di
ffere
nt
from
city
to c
ity).
M
edia
n B
S le
vels
ran
ged
fr
om 1
3 in
Lon
don
to 7
3 in
Ath
ens
and
Kra
cow
.
Tot
al
daily
d
eath
s re
gres
sed
on
BS
or
S
O2
usin
g P
oiss
on
GLM
m
odel
s,
adju
stin
g fo
r se
ason
al
cycl
es,
day-
of-w
eek,
in
fluen
za
epid
emic
, ho
liday
s, t
empe
ratu
re a
nd h
um
idity
. F
inal
ana
lysi
s do
ne w
ith a
utor
egre
ssiv
e P
oiss
on
mod
els
to
allo
w
for
ove
rdis
pers
ion
an
d au
toco
rrel
atio
n. P
ollu
tion
effe
cts
exa
min
ed
at 0
th
roug
h 3
day
lags
an
d m
ulti-
day
ave
rage
s th
ereo
f. W
hen
city
-spe
cific
co
effic
ient
s w
ere
te
sted
to
be
ho
mog
eneo
us,
over
all
estim
ates
w
ere
ob
tain
ed
fixed
ef
fect
s m
odel
s.
Whe
n
sign
ifica
nt h
eter
ogen
eity
wa
s pr
esen
t, so
urc
e of
he
tero
gene
ity s
ough
t by
exa
min
ing
a pr
edef
ined
lis
t of
ci
ty-s
pec
ific
varia
ble
s,
incl
udin
g an
nual
an
d se
ason
al m
eans
of
pollu
tion
and
wea
ther
va
riabl
es,
num
ber
of
mon
itorin
g si
tes,
co
rrel
atio
n be
twee
n m
easu
rem
ents
fr
om
diffe
rent
si
tes,
ag
e-st
anda
rdiz
ed
mor
talit
y,
prop
ortio
n of
el
derly
pe
ople
, sm
okin
g pr
eval
ence
, a
nd g
eogr
aphi
c di
ffere
nce
(Nor
th-
Sou
th,
Eas
t-W
est)
. A
ran
dom
effe
cts
mod
el w
as
fit w
hen
hete
roge
neity
cou
ld n
ot b
e ex
plai
ned.
Sub
stan
tial
varia
tion
in
pollu
tion
leve
ls
(win
ter
mea
n S
O2
rang
ed f
rom
30
to 3
30
µg/
m3 ),
clim
ate,
and
sea
sona
l pat
tern
s w
ere
ob
serv
ed
acro
ss
citie
s.
Sig
nific
ant
hete
roge
neity
was
fou
nd f
or t
he e
ffect
s of
B
S
and
SO
2,
but
only
th
e se
para
tion
betw
een
we
ster
n an
d ce
ntra
l-eas
tern
E
urop
ean
citie
s re
sulte
d in
m
ore
hom
ogen
eous
su
bgro
ups.
S
igni
fica
nt
hete
roge
neity
for
SO
2 re
mai
ned
in w
est
ern
citie
s. C
umul
ativ
e ef
fect
s of
pro
long
ed (
two
to
fo
ur
days
) ex
posu
re
to
air
pollu
tant
s re
sulte
d in
est
imat
es c
ompa
rabl
e w
ith t
he
one
day
effe
cts.
The
effe
cts
of b
oth
SO
2 an
d B
S w
ere
str
onge
r du
ring
the
sum
mer
an
d w
ere
inde
pend
ent.
Tot
al m
orta
lity
exce
ss d
eath
s in
si
ngle
day
for
wes
tern
Eur
opea
n
citie
s:
BS
(pe
r 25
µg
/m3 ):
1.4
% (
1.0,
1.8
) P
M10
(pe
r 50
µg/
m3 ):
2%
(1
, 3
)
In
cent
ral/e
aste
rn
Eur
opea
n
citie
s,
BS
(p
er
25
µg/
m3 ):
0.
3%
(0.0
5, 0
.5).
Sam
oli
et
al.
(200
1)
(Sam
oli e
t al,
2001
) A
PH
EA
1
citie
s (s
ee
Kat
souy
anni
(1
997)
(K
atso
uyan
ni e
t al
, 19
97)
At
lea
st
five
year
s be
twee
n 19
80-1
992.
The
P
M l
evel
s ar
e th
e sa
me
as
tho
se i
n K
atso
uyan
ni
et
al.
(199
7)
(Kat
souy
anni
et
al
, 19
97).
In o
rder
to
furt
her
inve
stig
ate
the
sour
ce o
f th
e re
gion
al
hete
roge
neity
of
P
M
effe
cts,
an
d to
ex
amin
e th
e se
nsiti
vity
of
the
RR
s, t
he A
PH
EA
da
ta w
ere
re-a
naly
zed.
Unl
ike
prev
ious
mod
el in
w
hic
h si
nuso
idal
ter
ms
for
seas
onal
con
trol
and
po
lyno
mia
l te
rms
for
wea
ther
, th
e in
vest
igat
ors
this
tim
e us
ed
a G
AM
mo
del
with
sm
ooth
ing
te
rms
for
seas
onal
tre
nd a
nd w
eath
er,
whi
ch i
s m
ore
com
mon
ly u
sed
appr
oac
h in
rec
ent
year
s.
The
est
imat
ed R
Rs
for
cent
ral-e
aste
rn c
ities
w
ere
lar
ger
than
tho
se o
bta
ined
fro
m t
he
prev
ious
m
odel
. A
lso,
re
stric
ting
the
anal
ysis
to
day
s w
ith c
once
ntra
tion
< 1
50
µg/
m3
furt
her
redu
ced
the
diffe
renc
es
betw
een
the
w
este
rn
and
cent
ral-e
aste
rn
Eur
opea
n ci
ties.
The
aut
hors
con
clud
ed t
hat
part
of
the
hete
roge
neity
in t
he e
stim
ated
air
pollu
tion
effe
cts
betw
een
w
este
rn
and
ce
ntra
l ea
ster
n ci
ties
in
prev
ious
pu
blic
atio
ns
wa
s ca
used
by
the
sta
tistic
al
appr
oach
and
the
data
ran
ge.
Tot
al
mor
talit
y pe
r 50
µ
g/m
3 of
B
S; G
AM
app
roac
h:
all c
ities
: 2.5
% (
2.1,
2.9
) w
est
ern
citie
s: 3
.1%
(2.
3, 3
.8)
cent
ral-e
aste
rn c
ities
: 2.
3%
(1.
7,
2.9)
old
met
hod:
al
l citi
es: 1
.3%
(0.
9, 1
.7)
we
ster
n ci
ties:
2.9
% (
2.1,
3.7
) ce
ntra
l-eas
tern
citi
es:
0.6
% (
0.1,
1.
1).
report no. 4/11
92
Sam
oli
et
al.
(200
3)
(Sam
oli e
t al,
2003
) R
e-an
alys
is
of
abov
e
stud
y.
Re-
anal
ysis
of
ab
ove
stud
y us
ing
strin
gent
co
nver
genc
e cr
iteria
as
wel
l as
natu
ral s
plin
es.
BS
ris
k es
timat
es u
sing
GA
M w
ere
redu
ced
by
~
10%
w
hen
strin
gent
co
nver
genc
e cr
iteria
w
ere
appl
ied.
U
se
of
GLM
/nat
ural
sp
lines
re
sulte
d in
fu
rthe
r an
d gr
eate
r re
duct
ions
.
Res
ults
co
rres
pond
ing
to
abov
e
usin
g th
e
GA
M
with
st
ringe
nt
conv
erge
nce
crite
ria:
all c
ities
: 2.3
% (
1.9,
2.7
)
we
ster
n ci
ties:
2.7
% (
2.0,
3.4
) ce
ntra
l-eas
tern
citi
es:
2.1
% (
1.5,
2.
7).
Cor
resp
ondi
ng
GLM
/nat
ural
sp
lines
res
ults
wer
e:
all c
ities
: 1.2
% (
0.7,
1.7
) w
est
ern
citie
s: 1
.6%
(0.
8, 2
.4)
cent
ral-e
aste
rn c
ities
: 1.
0%
(0.
3,
1.7)
.
Z
miro
u et
al
. (1
998)
(Z
mir
ou e
t al,
1998
) 10
E
urop
ean
citie
s (A
PH
EA
).
1977
-199
2 (s
tudy
ye
ars
diffe
rent
fr
om c
ity to
city
).
Med
ian
BS
leve
ls r
ange
d
from
13
in L
ondo
n to
73
in K
raco
w.
Car
diov
ascu
lar,
re
spira
tory
, an
d di
gest
ive
mor
talit
y se
ries
in 1
0 E
urop
ean
citie
s an
aly
zed
to e
xam
ine
caus
e-sp
ecifi
city
of
air
pol
lutio
n. T
he
mor
talit
y se
ries
wer
e an
alyz
ed f
or a
ssoc
iatio
ns
with
P
M
(BS
, e
xcep
t T
SP
in
M
ilan
an
d B
ratis
lava
; P
M13
in
Lyon
), N
O2,
O3,
an
d S
O2.
P
oiss
on
GL
M
mod
els,
la
g/av
erag
ing
of
pollu
tion,
and
com
puta
tion
of c
ombi
ned
effe
cts
acro
ss t
he c
ities
don
e in
the
sam
e w
ay a
s by
K
atso
uyan
ni
et
al.
(199
7)
(Kat
souy
anni
et
al
, 19
97),
abo
ve.
The
car
diov
ascu
lar
and
resp
irato
ry m
orta
lity
serie
s w
ere
ass
ocia
ted
with
BS
and
SO
2 in
w
est
ern
Eur
opea
n ci
ties,
but
not
in
the
five
cent
ral
Eur
opea
n ci
ties.
NO
2 di
d no
t sh
ow
cons
iste
nt
mor
talit
y as
soci
atio
ns.
RR
s fo
r re
spira
tory
cau
ses
wer
e at
leas
t eq
ual t
o, o
r gr
eate
r th
an
thos
e fo
r ca
rdio
vasc
ula
r ca
uses
. N
o
pollu
tant
ex
hibi
ted
an
y as
soci
atio
n w
ith d
iges
tive
mor
talit
y.
Poo
led
card
iova
scul
ar
mor
talit
y pe
rcen
t e
xce
ss
deat
hs
per
25
µ
g/m
3 inc
reas
e in
BS
for
wes
tern
E
urop
ean
citie
s: 1
.0%
(0.
3,
1.7)
; fo
r re
spira
tory
m
orta
lity,
it
was
2.
0% (
0.8,
3.2
) in
lag
0 (
the
lags
ap
pare
ntly
var
ied
acro
ss c
ities
).
APH
EA 2
Kat
souy
anni
et
al.
(200
1)
(Kat
souy
anni
et a
l, 20
01)
29
Eur
opea
n ci
ties,
19
90-1
997
(var
iabl
e fr
om c
ity to
city
).
Med
ian
PM
10
rang
ed
from
14
(S
tock
holm
) to
66
(P
ragu
e).
Med
ian
BS
ra
nged
fro
m
10
(Dub
lin)
to
64
(Ath
ens)
.
The
2nd
ph
ase
of
AP
HE
A
(AP
HE
A
2)
put
emph
asis
on
th
e ef
fect
m
odifi
catio
n by
ci
ty-
spec
ific
fact
ors.
The
firs
t st
age
of c
ity s
peci
fic
regr
essi
ons
used
G
AM
P
oiss
on
mod
el.
The
se
cond
st
age
re
gres
sion
an
alys
is
was
co
nduc
ted
to e
xpla
in a
ny h
eter
ogen
eity
of
air
pollu
tion
effe
cts
usin
g ci
ty-s
peci
fic
vari
able
s.
The
se c
ity-s
pec
ific
varia
ble
s in
clud
ed a
vera
ge
air
pollu
tion
leve
ls,
ave
rage
te
mpe
ratu
re/h
umid
ity,
age
- st
and
ardi
ze
mor
talit
y ra
te, r
egio
n in
dica
tors
, etc
.
The
aut
hors
fou
nd s
ever
al e
ffect
mod
ifier
s.
The
citi
es w
ith h
ighe
r N
O2 l
evel
s sh
ow
ed
larg
er P
M e
ffec
ts.
The
citi
es w
ith w
arm
er
clim
ate
sho
we
d la
rger
PM
eff
ects
. T
he c
ities
w
ith l
ow
sta
ndar
dize
d m
orta
lity
rate
sho
we
d la
rger
PM
effe
cts.
Tot
al m
orta
lity
exce
ss r
isk
per
50
µg/
m3 in
crea
se in
PM
10:
F
ixed
ef
fect
s m
odel
: 3.
5%
(2.9
, 4.
1) R
ando
m e
ffect
s m
odel
: 3.
1%
(2.1
, 4.2
).
report no. 4/11
93
Kat
souy
anni
et
al.
(200
3)
(Kat
souy
anni
et a
l, 20
03)
Re-
anal
ysis
of
ab
ove
st
udy.
Re-
anal
ysis
of
ab
ove
stud
y us
ing
strin
gent
co
nver
genc
e cr
iteria
as
wel
l as
nat
ural
spl
ines
an
d pe
naliz
ed s
plin
es.
The
po
oled
es
timat
e (r
ando
m
effe
cts
estim
ate)
w
as
redu
ced
by
4%
whe
n st
ringe
nt c
onve
rgen
ce c
riter
ia i
n G
AM
wer
e us
ed,
by
34%
w
hen
natu
ral
splin
es
wer
e us
ed,
and
by 1
1% w
hen
pen
aliz
ed s
plin
es
wer
e u
sed.
T
he
patte
rn
of
effe
ct
mod
ifica
tion
orig
inal
ly
rep
orte
d re
mai
ned
the
sam
e.
The
or
igin
al
findi
ngs
wer
e un
chan
ged.
Tot
al m
orta
lity
exce
ss r
isk
per
50
µg/
m3
incr
ease
in
P
M10
us
ing
G
AM
(s
trin
gent
co
nver
genc
e cr
iteria
): f
ixed
effe
cts:
3.3
% (
2.7,
3.
9)
rand
om
effe
cts:
3.
0%
(2.0
, 4.
1)
GLM
/nat
ural
spl
ines
wer
e
fixed
effe
cts:
2.1
% (
1.5,
2.8
)
rand
om e
ffect
s: 2
.1%
(1.
2, 3
.0)
P
enal
ized
spl
ines
fix
ed e
ffect
s: 2
.9%
(2.
3, 3
.6)
ra
ndom
effe
cts:
2.8
% (
1.8,
3.8
).
Tou
loum
i et
al
. (1
997)
(T
oulo
umi e
t al,
1997
) 6
Eur
opea
n ci
ties
(AP
HE
A).
19
77-1
992
(stu
dy
year
s di
ffere
nt
from
city
to c
ity).
M
edia
n B
S le
vels
ran
ged
fr
om
14.6
in
Lo
ndon
to
84
.4 in
Ath
ens.
Res
ults
of
the
shor
t-te
rm e
ffect
s of
am
bien
t N
O2
and/
or
O3
on
daily
de
aths
fr
om
all
caus
es
(exc
lud
ing
acci
dent
s) w
ere
disc
usse
d to
pro
vide
a
basi
s fo
r co
mpa
rison
with
est
imat
ed S
O2
or
BS
ef
fect
s in
A
PH
EA
ci
ties.
P
oiss
on
G
LM
mod
els,
la
g/av
erag
ing
of
pollu
tion,
a
nd
the
com
puta
tion
of
com
bine
d ef
fect
s ac
ross
th
e ci
ties
wer
e do
ne i
n th
e sa
me
way
as
don
e by
K
atso
uyan
ni
et
al.
(199
7)
(Kat
souy
anni
et
al
, 19
97).
Sig
nific
ant
po
sitiv
e as
soci
atio
ns w
ere
fou
nd
betw
een
daily
dea
ths
and
bot
h N
O2
and
O3.
T
here
wa
s a
tend
ency
for
lar
ger
effe
cts
of
NO
2 in
citi
es w
ith h
ighe
r le
vels
of
BS
. W
hen
BS
w
as
incl
uded
in
th
e m
odel
, po
oled
es
timat
e fo
r O
3 ef
fect
w
as
only
sl
ight
ly
redu
ced,
bu
t co
effic
ient
fo
r N
O2
wa
s re
duce
d by
ha
lf.
Aut
hor
s sp
ecul
ated
th
at
shor
t-te
rm e
ffect
s of
NO
2 on
mor
talit
y w
ere
conf
ound
ed
by
othe
r ve
hicl
e-de
rived
po
lluta
nts.
NO
2 an
d/or
O3 e
stim
ates
onl
y.
Zan
obet
ti et
al
., 20
02
(Zan
obet
ti et
al,
2002
) 19
90-1
996
(var
iabl
e fr
om
city
to
ci
ty).
10
E
urop
ean
citie
s (2
/3
of
popu
latio
n)
from
th
e A
PH
EA
2 s
tud
y.
Mea
n P
M10
ran
ged
from
16
(S
tock
holm
) to
76
(P
ragu
e).
Ana
lysi
s of
m
orta
lity
disp
lace
men
t up
to
40
da
ys.
In T
he f
irst
stag
e co
nsis
ted
of c
ity s
pec
ific
regr
essi
ons
usin
g G
AM
Po
isso
n m
odel
s w
ith 4
th
degr
ee
poly
nom
ial
dist
ribut
ed
lag
cons
trai
nts
with
up
to 4
0 da
ys.
In t
he s
econ
d st
age
the
city
-sp
ecifi
c es
timat
es
wer
e co
mbi
ned
usin
g a
ra
ndom
effe
ct m
odel
.
The
ov
eral
l ef
fect
of
P
M10
for
the
4th
dist
ribut
ed
lag
mod
el
was
hi
gher
(1
.61%
in
crea
se i
n da
ily d
eath
s pe
r 10
µg/
m3 )
than
th
e m
odel
con
side
ring
only
lag
0-1
(0.7
0 %
).
The
re
sults
w
ere
unc
han
ged
whe
n a
n un
cons
trai
ned
di
strib
uted
la
g m
odel
w
as
used
. T
his
sugg
ests
tha
t th
e ef
fect
of
PM
on
mor
talit
y is
not
due
prim
arily
to
shor
t-te
rm
mor
talit
y di
spla
cem
ent.
Fo
r 4th
de
gree
po
lyno
mia
l di
strib
uted
la
g m
odel
, to
tal
mor
talit
y e
xces
s pe
r 50
µ
g/m
3 in
crea
se w
as 8
.1%
(5.
1, 1
1.0)
.
report no. 4/11
94
Zan
obet
ti an
d S
chw
artz
(2
003)
(Z
anob
etti
and
Sch
war
tz, 2
003)
R
e-an
alys
is o
f th
e ab
ove
stud
y.
Re-
anal
ysis
of
ab
ove
stud
y us
ing
strin
gent
co
nver
genc
e cr
iteria
as
wel
l as
nat
ural
spl
ines
an
d pe
naliz
ed s
plin
es.
The
po
oled
P
M10
(la
g 0-
1)
mor
talit
y ris
k es
timat
e w
as r
educ
ed b
y 4%
whe
n st
ringe
nt
conv
erge
nce
crite
ria i
n G
AM
wer
e us
ed,
by
18%
whe
n pe
naliz
ed s
plin
es w
ere
used
. F
or
the
4th
degr
ee
poly
nom
ial
dist
ribut
ed
lag
mod
el,
corr
espo
ndin
g re
duct
ions
wer
e 10
%
and
26%
.
Com
bine
d to
tal
mor
talit
y ex
cess
ris
k pe
r 50
µg/
m3 i
ncre
ase
PM
10
(lag
0-1)
w
as
3.4%
(2
.0,
4.8)
us
ing
GA
M
with
st
ringe
nt
conv
erge
nce
crite
ria.
For
4th
de
gree
po
lyno
mia
l di
strib
uted
lag
mod
el,
it w
as 7
.5%
(4
.4,
10.7
).
Cor
resp
ondi
ng
redu
ctio
ns
usin
g pe
naliz
ed
splin
es w
ere
2.9%
(1.
4, 4
.4)
and
5.6%
(1.
5, 9
.8),
res
pect
ivel
y.
Zan
obet
ti et
al
., 20
03
(Zan
obet
ti et
al,
2003
) 10
citi
es o
f th
e A
PH
EA
2
proj
ect,
1990
-199
6.
PM
10
mea
ns
rang
ing
betw
een
15.5
(S
tock
holm
) an
d 76
.2
(Pra
gue)
.
Thi
s st
udy
anal
ysed
the
acu
te a
nd m
ediu
m-t
erm
(w
eeks
to
mon
ths)
exp
osur
e ef
fect
s of
PM
10 o
n
card
iova
scul
ar a
nd r
espi
rato
ry m
orta
lity,
usi
ng a
m
ulti-
city
hi
erar
chic
al
mod
elin
g ap
proa
ch.
A
GA
M
Poi
sson
re
gres
sion
w
as
fitte
d an
d an
un
cons
trai
ned
dist
ribut
ed l
ag a
nd a
4th d
egre
e po
lyno
mia
l di
strib
uted
lag
wer
e us
ed t
o m
odel
th
e ef
fect
of
PM
10 e
xpos
ure
on d
eath
s up
to
40
days
afte
r th
e ex
posu
re.
The
mod
el w
as a
djus
ted
for
wea
ther
, in
fluen
za
epid
emic
s, h
olid
ays,
and
day
of t
he w
eek.
Thi
s st
udy
conf
irmed
tha
t m
ost
of t
he e
ffect
on
mor
talit
y of
air
pollu
tion
was
not
sim
ply
adva
nced
by
a fe
w w
eeks
and
tha
t ef
fect
s pe
rsis
ted
for
mor
e th
an
1 m
onth
af
ter
expo
sure
. T
he e
ffect
siz
e es
timat
e fo
r P
M10
do
uble
d w
hen
cons
ider
ing
long
er-t
erm
ef
fect
s fo
r al
l de
aths
and
for
car
diov
ascu
lar
deat
hs
and
beca
me
five
times
hi
gher
fo
r re
spira
tory
dea
ths.
The
effe
cts
wer
e si
mila
r w
hen
stra
tifyi
ng b
y ag
e gr
oups
.
Con
side
ring
PM
10 t
he d
ay o
f an
d da
y be
fore
the
dea
th,
the
perc
ent
incr
ease
m
orta
lity
for
an
incr
emen
t of
10
µ
g/m
3 of
P
M10
w
as
0.74
%
(-0.
17,
1.66
) fo
r re
spira
tory
de
aths
an
d 0.
69%
(0
.31,
1.
08)
for
CV
D
deat
hs.
Cor
resp
ondi
ng
estim
ates
in
un
rest
ricte
d di
strib
uted
la
g m
odel
s, w
ere
4.20
% (
1.08
, 7.
42)
and
1.97
%
(1.3
8,
2.55
),
resp
ectiv
ely.
report no. 4/11
95
Aga
et
al.,
2003
(A
ga e
t al
, 200
3)
28
Eur
opea
n ci
ties
AP
HE
A 2
pro
ject
. D
aily
m
ean
P
M10
(2
1
citie
s)
betw
een
15
an
d 66
an
d B
S
(14
citie
s)
betw
een
10 a
nd 6
4.
To
stud
y th
e ef
fect
s of
am
bien
t pa
rtic
le
conc
entr
atio
ns
(BS
and
PM
10)
on t
otal
mor
talit
y
amon
g pe
rson
s ag
ed 6
5 ye
ars
or m
ore,
a 2
-st
age
anal
ysis
w
as
app
lied,
as
sess
ing
city
-sp
ecifi
c ef
fect
s fir
st a
nd t
hen
over
all e
ffect
s w
as
used
afte
r ad
just
men
t fo
r se
ason
ality
, lo
ng-t
erm
tr
ends
, te
mpe
ratu
re,
hum
idity
, in
fluen
za
epid
emic
s,
othe
r un
usua
l ev
ents
, da
y of
th
e w
eek,
and
hol
iday
s. L
ag 0
-1 w
as
a p
riori
cho
sen
as e
xpos
ure
mea
sure
.
A s
igni
fican
t as
soci
atio
n w
ith t
otal
mor
talit
y w
as
foun
d fo
r bo
th P
M10
and
BS
am
ong
the
elde
rly,
with
re
lativ
e ris
ks
com
para
ble
or
slig
htly
hig
her
than
tho
se o
bser
ved
for
tota
l m
orta
lity
and
sim
ilar
effe
ct
mod
ifica
tion
patte
rns.
Per
cent
ex
cess
m
orta
lity
for
incr
ease
by
10 µ
g/m
3 of
PM
10 (
lag
0-1)
T
otal
pop
ulat
ion
Fix
ed e
ffect
s:
0.71
% (
0.60
, 0.
83)
R
ando
m
effe
cts:
0.67
%
(0.4
7,
0.87
) E
lder
ly (
65+
) F
ixed
effe
cts:
0.
79%
(0.
66,
0.92
)
Ran
dom
ef
fect
s:
0.
74%
(0
.52,
0.
95)
Per
cent
ex
cess
m
orta
lity
for
incr
ease
by
10
µg/
m3 o
f B
S (
lag
0-
1)
Tot
al p
opul
atio
n F
ixed
effe
cts:
0.
51%
(0.
39,
0.64
)
Ran
dom
ef
fect
s:
0.
58%
(0
.32,
0.
84)
Eld
erly
(65
+)
Fix
ed e
ffect
s:
0.63
% (
0.49
, 0.
78)
R
ando
m
effe
cts:
0.68
%
(0.4
3,
0.92
).
Ana
litis
et
al
., 20
06
(Ana
litis
et a
l, 20
06)
29
Eur
opea
n ci
ties
AP
HE
A2
proj
ect
. M
ean
PM
10
(21
citie
s)
betw
een
9 a
nd
64
and
BS
(1
5 ci
ties)
be
twee
n
14 a
nd 1
66.
To
stud
y th
e ef
fect
s of
am
bien
t pa
rtic
le
conc
entr
atio
ns
(BS
and
PM
10)
on c
ardi
ova
scul
ar
and
resp
irato
ry m
orta
lity,
a 2
-sta
ge h
iera
rchi
cal
mod
elin
g ap
proa
ch
asse
ssin
g ci
ty-s
pec
ific
effe
cts
first
and
the
n ov
eral
l ef
fect
s w
as
used
af
ter
adju
stm
ent
for
seas
onal
ity,
long
-ter
m
tren
ds,
tem
pera
ture
, hu
mid
ity,
influ
enza
ep
idem
ics,
ot
her
unus
ual
even
ts,
day
of
the
wee
k, a
nd h
olid
ays.
Fur
ther
adj
ustm
ent
for
the
daily
leve
ls o
f ot
her
pollu
tant
s w
as p
erfo
rmed
in
2-po
lluta
nt m
odel
s.
A s
ign
ifica
nt e
xce
ss r
isk
wa
s fo
und
for
both
P
M10
and
BS
. T
he g
reat
est
rela
tive
effe
ct
mod
ifica
tion
for
PM
10
on
CV
D
mor
talit
y ca
me
from
m
ean
tem
pera
ture
an
d ci
ty-
aver
age
NO
2 c
once
ntra
tion
s. W
arm
er t
ow
ns
with
mor
e N
O2
sho
wed
larg
er e
ffec
ts.
The
re
wa
s le
ss e
ffect
mod
ifica
tion
by N
O2
for
BS
th
an f
or P
M10
, su
gges
ting
diffe
rent
ial e
ffect
s of
tra
ffic
part
icle
s, f
or w
hich
PM
10 i
s a
less
go
od in
dica
tor
than
BS
.
An
incr
emen
t in
PM
10 o
f 10
µg/
m3
(lag
0-1)
w
as
asso
ciat
ed
with
in
crea
ses
of 0
.76%
(0.
47,
1.05
) in
C
VD
de
aths
an
d 0.
58%
(0
.21,
0.
95)
in
resp
irato
ry
deat
hs.
The
sa
me
incr
emen
t in
B
S
was
as
soci
ated
w
ith
incr
ease
s of
0.
62%
(0
.35,
0.
90)
and
0.84
%
(0.1
1, 1
.57)
, res
pect
ivel
y.
report no. 4/11
96
Tou
loum
i et
al
20
05
(Tou
loum
i et a
l, 20
05)
7 E
urop
ean
citie
s A
PH
EA
-2 p
roje
ct,
1990
-19
97.
Med
ian
PM
10
rang
ing
betw
een
13.7
and
40.
2.
To
inve
stig
ate
the
effe
ct o
f in
fluen
za e
pid
emic
s on
the
ass
ocia
tion
betw
een
PM
10 a
nd t
otal
and
ca
rdio
vasc
ular
m
orta
lity,
a
GA
M
Po
isso
n re
gres
sion
afte
r ad
just
men
t fo
r da
y of
wee
k an
d ho
liday
s w
as
used
. T
o co
ntro
l fo
r in
fluen
za
epid
emic
s, 1
0 m
etho
ds w
ere
com
pare
d.
The
si
gnifi
cant
an
d di
rect
as
soci
atio
n be
twee
n P
M10
and
dai
ly n
um
ber
of d
eath
s re
mai
ned
and
even
in
crea
sed
afte
r co
ntro
lling
for
inf
luen
za,
rega
rdle
ss o
f th
e m
etho
d us
ed f
or c
ontr
ol.
Sim
ple
met
hods
, su
ch a
s 1
or m
ultip
le in
dica
tors
for
influ
enza
ep
idem
ics,
are
rel
ativ
ely
easy
and
effe
ctiv
e w
ays
to
co
ntro
l or
to
ev
alua
te
pote
ntia
l co
nfou
ndin
g ef
fect
of
influ
enza
epi
dem
ics.
R
espi
rato
ry
mor
talit
y da
ta
may
be
us
ed
inst
ead
of i
nflu
enza
cas
es d
ata
if th
e la
tter
ar
e no
t av
aila
ble;
res
pira
tory
mor
talit
y da
ta
may
be
m
ore
appr
opria
te
way
to
co
ntro
l w
hen
the
outc
ome
stud
ied
is
a
mor
talit
y ou
tcom
e. N
one
of t
he m
eth
ods
inve
stig
ated
in
th
is
stu
dy
sugg
este
d ap
prec
iabl
e co
nfou
ndin
g b
y in
fluen
za e
pide
mic
s.
An
incr
emen
t b
y 10
µg/
m3 (
lag
0-1)
in
P
M10
co
ncen
trat
ion
s w
as
asso
ciat
ed
with
a
0.48
%
(0.2
7,
0.70
) in
crea
se
in
daily
m
orta
lity
with
out
adju
stm
ent
for
influ
enza
ep
idem
ics,
a
nd
rang
ed
betw
een
0.
45%
(0
.26,
0.
69)
to
0.67
%
(0.4
6,
0.89
),
afte
r di
ffere
nt
adju
stm
ent
for
influ
enza
ep
idem
ics.
T
he
corr
espo
ndin
g
figur
es
for
CV
D
mor
talit
y w
ere
0.85
%
(0.5
3,
1.18
) w
itho
ut
the
m
etho
ds
of
cont
rol
and
from
0.
86%
(0
.53,
1.
19)
to
1.06
%
(0.7
4, 1
.39)
with
the
met
hod
s of
co
ntro
l.
Nat
iona
l stu
dies
THE
NET
HER
LAN
DS
H
oek
et a
l. (2
000)
(H
oek
et a
l, 20
00)
The
N
ethe
rland
s,
1986
-19
94.
PM
10 m
edia
n: 3
4 B
S m
edia
n: 1
0.
Tot
al,
card
iova
scul
ar,
CO
PD
, an
d pn
eum
onia
m
orta
lity
serie
s w
ere
regr
esse
d on
PM
10,
BS
, su
lfate
, ni
trat
e,
O3,
S
O2,
C
O,
adju
stin
g fo
r se
ason
al
cycl
es,
day
of
wee
k,
influ
enza
, te
mpe
ratu
re,
and
hum
idity
usi
ng P
oiss
on G
AM
m
odel
. D
eath
s oc
curr
ing
insi
de
and
out
side
ho
spita
ls w
ere
als
o e
xam
ine
d.
Par
ticul
ate
air
pollu
tion
w
as
less
co
nsis
tent
ly a
ssoc
iate
d w
ith m
orta
lity
than
w
ere
the
gase
ous
pollu
tant
s S
O2
an
d N
O2.
S
ulfa
te,
nitr
ate,
an
d B
S
wer
e m
ore
cons
iste
ntly
ass
ocia
ted
with
tot
al m
orta
lity
than
was
PM
10.
The
RR
s fo
r al
l po
lluta
nts
w
ere
lar
ger
in t
he s
umm
er m
onth
s th
an i
n th
e w
inte
r m
onth
s.
Mor
talit
y e
xce
ss r
isk
est
ima
te p
er
50 µ
g/m
3 PM
10
(lag
0-6)
: 1.
2% (
0.2,
2.2
) fo
r to
tal m
orta
lity
0.
9% (
-0.8
, 2.7
) fo
r C
VD
5.
9% (
0.9,
11.
2) fo
r C
OP
D
10.1
% (
3.6,
17.
1) fo
r pn
eum
onia
.
Hoe
k (2
003)
(H
oek,
20
03)
Re-
anal
ysis
of
the
abov
e
stud
y.
Re-
anal
ysis
of
ab
ove
stud
y us
ing
strin
gent
co
nver
genc
e cr
iteria
and
nat
ural
spl
ines
. V
ery
little
ch
ange
in
P
M
risk
coef
ficie
nts
(o
ften
slig
htly
inc
reas
ed)
wh
ethe
r G
AM
with
st
ringe
nt
conv
erge
nce
crite
ria
or
GLM
/nat
ural
spl
ines
wer
e us
ed.
Est
imat
es p
er 5
0 µ
g/m
3 PM
10 (
lag
0-
6)
usin
g G
AM
w
ith
strin
gent
co
nver
genc
e cr
iteria
: T
otal
mor
talit
y: 1
.4%
(0.
3, 2
.6)
CV
D: 0
.9%
(-0
.8, 2
.7)
CO
PD
: 6.
1% (
1.0,
11.
4)
Pne
umon
ia: 1
0.3%
(3.
7, 1
7.2)
U
sing
GLM
/na
tura
l spl
ines
wer
e:
Tot
al m
orta
lity:
1.2
% (
-0.1
, 2.5
) C
VD
: 1.6
% (
-0.3
, 3.5
) C
OP
D:
6.0%
(0.
4, 1
1.8)
P
neum
onia
: 10.
7% (
3.5,
18.
3).
report no. 4/11
97
Hoe
k et
al.
(200
1) (
Hoe
k et
al,
2001
) T
he
Net
herla
nds.
19
86-
1994
. P
M10
med
ian:
34
BS
med
ian:
10.
Thi
s st
udy
of
the
who
le
popu
latio
n of
th
e N
ethe
rland
s, w
ith i
ts l
arge
sam
ple
size
(m
ean
daily
tot
al d
eath
s ~
330
), a
llow
ed e
xam
inat
ion
of
spec
ific
card
iova
scul
ar c
ause
of
deat
hs.
GA
M
Poi
sson
re
gre
ssio
n m
odel
s,
adju
stin
g fo
r se
ason
al c
ycle
s, t
empe
ratu
re,
hum
idity
, an
d da
y of
we
ek w
ere
used
.
Dea
ths
due
to
hear
t fa
ilure
, ar
rhyt
hmia
, ce
rebr
ovas
cula
r ca
uses
, an
d th
rom
bocy
tic
caus
es w
ere
mor
e st
rong
ly (
~ 2
.5 t
o 4
times
la
rger
re
lativ
e ris
ks)
asso
ciat
ed
with
ai
r po
llutio
n th
an
the
over
all
CV
D
deat
hs
or
AM
I and
oth
er IH
D d
eath
s.
Fo
r P
M10
(la
g 0-
6),
mor
talit
y pe
r 50
µg/
m3 in
cre
ase:
T
otal
CV
D: 0
.9%
(-0
.8, 2
.7)
AM
I/IH
D: 0
.3%
(-2
.3, 3
.0)
Arr
hyth
mia
2.5
% (
- 4.
3, 9
.9)
Hea
rt fa
ilure
2.2
% (
-2.5
, 7.2
) C
ereb
rova
scul
ar 1
.9%
(-1
.8, 5
.8)
Thr
ombo
cytic
: 0.6
% (
-6.8
, 8.7
).
The
ex
cess
ris
ks
for
BS
w
ere
la
rger
and
mor
e si
gnifi
cant
tha
n th
ose
for
PM
10.
Hoe
k (2
003)
(H
oek,
20
03)
Re-
anal
ysis
of
the
abov
e
stud
y.
Re-
anal
ysis
of
ab
ove
stud
y us
ing
strin
gent
co
nver
genc
e cr
iteria
and
nat
ural
spl
ines
. V
ery
little
ch
ange
in
P
M
risk
coef
ficie
nts
(o
ften
slig
htly
inc
reas
ed)
wh
ethe
r G
AM
with
st
ringe
nt
conv
erge
nce
crite
ria
or
GLM
/nat
ural
spl
ines
wer
e us
ed.
Fo
r P
M10
(la
g 0-
6),
perc
ent
exce
ss r
isk
mor
talit
y pe
r 50
µg/
m3
incr
ease
u
sing
G
AM
w
ith
strin
gent
co
nver
genc
e cr
iteria
w
ere:
T
otal
CV
D: 0
.9%
(-0
.8, 2
.7),
A
MI/I
HD
: 0.4
% (
-2.2
, 3.0
),
Arr
hyth
mia
: 2.7
% (
- 4.
2, 1
0.1)
, H
eart
failu
re: 2
.4%
(-2
.3, 7
.4),
C
ereb
rova
scul
ar: 2
.0%
(-1
.7, 5
.9),
T
hrom
bocy
tic: 0
.7 %
(-6.
8, 8
.8).
T
he R
Rs
for
BS
wer
e la
rger
and
m
ore
sign
ifica
nt
than
th
ose
fo
r P
M10
.
report no. 4/11
98
Fis
cher
et
al
. (2
003)
(F
isch
er e
t al,
2003
) T
he
Net
herla
nds.
19
86-
1994
. P
M10
med
ian:
34
BS
med
ian:
10.
Age
-spe
cific
ana
lysi
s of
the
pre
viou
s st
udy.
The
re
latio
nshi
p b
etw
een
da
ily
(tot
al
and
caus
e-sp
ecifi
c) m
orta
lity
and
air
pollu
tion
(PM
10,
BS
an
d ot
her
co-p
ollu
tant
s, i
nclu
ding
O3,
SO
2, C
O
and
NO
2)
wa
s m
odel
led
usin
g P
oiss
on
regr
essi
on
anal
ysis
. A
ll po
llutio
n m
orta
lity
asso
ciat
ions
wer
e ad
just
ed f
or lo
ng-t
erm
tre
nds,
se
ason
al t
ren
ds,
influ
enza
epi
dem
ics,
am
bien
t te
mpe
ratu
re,
ambi
ent
rela
tive
hum
idity
, da
y of
th
e w
eek
and
holid
ays,
usi
ng
GA
M m
odel
s. T
he
wee
kly
aver
age
conc
entr
atio
n fo
r ot
her
pollu
tant
s (la
g 0
-6)
was
use
d fo
r P
M10
and
BS
.
PM
10 s
light
ly i
ncre
ased
tot
al m
orta
lity,
and
ca
rdio
vasc
ular
mor
talit
y in
the
eld
erly
onl
y (s
ubje
cts
age
d 65
yea
rs o
r m
ore)
. B
oth
BS
an
d P
M10
si
gni
fican
tly
incr
ease
d m
orta
lity
for
pneu
mon
ia i
n th
e el
derly
, on
ly.
Sim
ilar
incr
ease
d ris
ks
wer
e sh
ow
n fo
r th
e ot
her
cons
ider
ed c
o-p
ollu
tant
s.
Per
cent
exc
ess
tota
l m
orta
lity
risk
per
100
µg/
m3 in
crea
se o
f PM
10:
2% (
0, 3
) P
erce
nt
exce
ss
for
80
µg/
m3
incr
ease
of
P
M10
(9
9th-1
st
perc
entil
e):
Car
diov
ascu
lar
mor
talit
y (y
ears
):
<45
: -9.
4 (-
27.2
, 12.
8)
45-6
4: 2
.3 (
-5.5
, 10.
6)
65-7
4: 0
.2 (
-5.5
, 6.2
) 75
+: 1
.6 (
-1.9
, 5.2
) C
OP
D m
orta
lity
(yea
rs):
<
45: 1
5.3
(-41
.3, 1
26.8
) 45
-64:
13.
9 (-
16.1
, 54.
1)
65-7
4: 1
6.6
(-0.
9, 3
7.2)
75
+: 6
.6 (
-3.5
, 17.
8)
Pne
umon
ia m
orta
lity
(yea
rs):
<
45: 4
2.7
(-19
.4, 1
52.5
) 45
-64:
71.
2 (4
.2, 1
81.5
) 65
-74:
24.
0 (-
12.1
, 74.
8)
75+
: 12.
3 (1
.1, 2
4.7)
P
erce
nt
exce
ss
for
40
µg/
m3
incr
ease
of
B
S
(99
th-1
st
perc
entil
e):
Car
diov
ascu
lar
mor
talit
y (y
ears
):
<45
: 5.7
(-6
.4, 1
9.6)
45
-64:
0.0
(-4
.3, 4
.6)
65-7
4: 4
.0 (
0.6,
7.5
) 75
+: 3
.0 (
1.0,
5.1
) C
OP
D m
orta
lity
(yea
rs):
<
45: -
12.8
(-4
0.1,
27.
0)
45-6
4: 2
.5 (
-13.
0, 2
0.9)
65
-74:
20.
4 (9
.5, 3
2.4)
75
+: 1
.2 (
-4.3
, 7.0
) P
neum
onia
mor
talit
y (y
ears
):
<45
:-25
.9 (
-48.
1, 5
.7)
45-6
4: 2
6.4
(-6.
6, 7
1.1)
65
-74:
-4.
9 (-
23.3
, 17.
9)
75+
: 12.
3 (5
.5, 1
9.6)
.
report no. 4/11
99
Subn
atio
nal s
tudi
es
G
REA
T B
RIT
AIN
Bre
mne
r et
al
. (1
999)
(B
rem
ner
et a
l, 19
99)
Lond
on,
UK
, 19
92-1
994.
B
S m
ean:
13
P
M10
mea
n: 2
9.
Tot
al,
card
iova
scul
ar,
and
resp
irato
ry (
by a
ge)
mor
talit
y se
ries
wer
e re
gres
sed
on P
M10
, B
S,
O3,
NO
2, C
O,
and
SO
2, a
djus
ting
for
sea
sona
l cy
cles
, da
y of
w
eek,
in
fluen
za,
holid
ays,
te
mpe
ratu
re,
hum
idity
, an
d au
toco
rrel
atio
n us
ing
Poi
sson
GLM
mod
el.
All
effe
ct s
ize
est
imat
es (
exc
ept
O3)
we
re
posi
tive
for
tota
l de
aths
(t
houg
h no
t si
gnifi
cant
fo
r si
ngle
la
g m
odel
s).
The
ef
fect
s of
O3
foun
d in
198
7-19
92 w
ere
not
re
plic
ated
, ex
cept
in
card
iova
scul
ar d
eath
s.
Mul
tiple
day
ave
ragi
ng (
e.g.
, 0-
1, 0
-2 d
ays)
te
nded
to
give
mor
e si
gnifi
cant
effe
ct s
ize
estim
ates
. T
he e
ffect
siz
e fo
r P
M10
and
BS
w
ere
sim
ilar
for
the
sam
e di
strib
utio
nal
incr
emen
t.
Per
cent
e
xce
ss
risk
1.9%
(0
.0,
3.8)
per
25
µg/
m3 B
S a
t la
g 1;
1.
3%
(-1.
0,
3.6)
pe
r 50
µ
g/m
3 P
M10
at l
ag 1
for
tota
l dea
ths.
R
espi
rato
ry d
eath
s (la
g 3)
: 4.
9%
(0.5
, 9.4
).
CV
D
deat
hs
(lag
1):
3.0%
(0
.3,
5.7)
.
Pre
scot
t et
al
. (1
998)
(P
resc
ott e
t al,
1998
) E
dinb
urgh
, U
K,
1981
-19
95.
PM
10
mea
n:
21,
by
TE
OM
on
ly
for
1992
-19
95;
BS
mea
n: 8
.7.
Bot
h m
orta
lity
(tot
al,
card
iova
scul
ar,
and
resp
irato
ry)
and
emer
genc
y ho
spita
l adm
issi
ons
(car
diov
ascu
lar
and
resp
irato
ry),
in
tw
o ag
e gr
oups
(<
65 a
nd ≥
65
year
s),
wer
e an
alyz
ed f
or
thei
r as
soci
atio
ns w
ith P
M10
, B
S,
SO
2, N
O2,
O3,
an
d C
O,
usi
ng
Po
isso
n
GLM
re
gres
sion
ad
just
ing
for
seas
onal
cy
cles
, da
y-of
-wee
k,
tem
pera
ture
, and
win
d sp
eed.
Am
ong
all
the
pollu
tant
s,
BS
w
as
mos
t si
gnifi
cant
ly
asso
ciat
ed
with
al
l ca
use
, ca
rdio
vasc
ular
, an
d re
spira
tory
m
orta
lity
serie
s.
In
the
su
bset
in
w
hich
P
M10
d
ata
w
ere
av
aila
ble,
th
e R
R
est
imat
es
for
BS
an
d P
M10
for
all
caus
e el
derly
mor
talit
y w
ere
com
para
ble.
O
ther
po
lluta
nts’
m
orta
lity
asso
ciat
ions
wer
e ge
nera
lly in
cons
iste
nt.
Per
cent
e
xce
ss
risk
3.8%
(1
.3,
6.4)
per
25
µg/
m3 i
ncre
ase
in B
S
for
all
caus
e m
orta
lity
in a
ge
65+
gr
oup,
lag
1-3.
Roo
ney
et
al.
(199
8)
(Roo
ney
et a
l, 19
98)
Eng
land
and
Wal
es,
and
G
reat
er L
ondo
n, U
K
PM
10
mea
n:
56
(dur
ing
the
wor
st h
eat
wa
ve);
P
M10
m
ean:
39
(J
uly-
Aug
ust)
.
Exc
ess
deat
hs,
by
age,
sex
, an
d ca
use,
dur
ing
the
1995
hea
t w
ave
wer
e es
timat
ed b
y ta
king
th
e di
ffere
nce
bet
wee
n th
e d
eath
s du
ring
heat
w
ave
an
d th
e 31
-day
m
ovin
g av
erag
es
(for
19
95 a
nd 1
993-
1994
sep
arat
ely)
. T
he p
ollu
tion
effe
cts,
add
itive
ly f
or O
3, P
M10
, an
d N
O2,
wer
e es
timat
ed
base
d on
th
e
publ
ishe
d se
ason
-sp
ecifi
c co
effic
ient
s fr
om t
he 1
987-
199
2 st
udy
(And
erso
n et
al.,
199
6) (
An
ders
on e
t al,
1996
).
Air
pollu
tion
leve
ls a
t al
l th
e lo
catio
ns r
ose
durin
g th
e h
eat
wav
e.
8.9%
an
d
16.1
%
exce
ss d
eath
s w
ere
estim
ated
for
Eng
land
an
d W
ales
, an
d G
reat
er
Lond
on,
resp
ectiv
ely.
Of
thes
e ex
cess
dea
ths,
up
to
62%
an
d
38%
, re
spec
tivel
y fo
r th
ese
lo
catio
ns,
ma
y be
attr
ibut
abl
e to
com
bin
ed
pollu
tion
effe
cts.
2.6%
incr
ease
for
PM
10 in
Gre
ater
Lo
ndon
dur
ing
hea
t w
ave.
Wor
dley
et
al
. (1
997)
(W
ordl
ey e
t al,
1997
) B
irmin
gham
, U
K,
1992
-19
94.
PM
10 m
ean:
26.
Mor
talit
y da
ta
wer
e an
alyz
ed
for
CO
PD
, pn
eum
onia
, al
l re
spira
tory
di
seas
es,
all
circ
ulat
ory
dise
ases
, an
d a
ll ca
uses
. M
ort
ality
as
soci
atio
ns w
ith P
M10
, N
O2,
SO
2, a
nd O
3 w
ere
exam
ine
d u
sing
O
LS
(w
ith
som
e he
alth
ou
tcom
es
log-
or
sq
uare
-roo
t tr
ansf
orm
ed),
ad
just
ing
for
day-
of-w
eek,
mon
th,
linea
r tr
end,
te
mpe
ratu
re
and
rela
tive
hum
idity
. T
he
stud
y al
so a
naly
zed
hosp
ital a
dmis
sion
dat
a.
Tot
al,
circ
ulat
ory,
and
CO
PD
dea
ths
wer
e si
gnifi
cant
ly a
ssoc
iate
d w
ith 1
-day
lag
PM
10.
The
ga
seou
s po
lluta
nts
“did
no
t ha
ve
sign
ifica
nt
ass
ocia
tions
in
depe
nden
t fr
om
that
of
PM
10”,
and
the
res
ults
for
gas
eou
s po
lluta
nts
wer
e no
t pr
esen
ted.
The
im
pact
of
red
ucin
g P
M10
to
belo
w 7
0 µ
g/m
3 was
es
timat
ed
to
be
“sm
all”
(0.2
%
for
tota
l de
aths
), b
ut t
he P
M10
lev
el a
bove
70
µg/
m3
occu
rred
onl
y on
ce d
urin
g th
e st
udy
perio
d.
5.6%
(0
.5,
11.0
) pe
r 50
µ
g/m
3 P
M10
at
1 d
lag
for
tot
al d
eath
s.
CO
PD
(1
da
y la
g)
deat
hs:
27.6
(5
.1,
54.9
). C
ircul
ator
y (1
day
lag)
de
aths
: 8.8
(1.
9, 1
7.1)
.
report no. 4/11
100
And
erso
n et
al
. (1
996)
(A
nder
son
et a
l, 19
96)
Lond
on, U
K, 1
987-
1992
. B
S m
ean
15.
Tot
al,
card
iova
scul
ar,
and
resp
irato
ry m
orta
lity
serie
s w
ere
regr
esse
d on
B
S,
O3,
N
O2,
an
d S
O2,
ad
just
ing
for
sea
sona
l cyc
les,
da
y of
wee
k,
influ
enza
, ho
liday
s, t
empe
ratu
re,
hum
idity
, an
d au
toco
rrel
atio
n us
ing
Poi
sson
GLM
mod
el.
Bot
h O
3 (la
g 0)
an
d B
S
(lag
1)
we
re
sign
ifica
nt p
redi
ctor
s of
tot
al d
eath
s. O
3 w
as
also
pos
itive
ly s
igni
fican
tly a
ssoc
iate
d w
ith
resp
irato
ry a
nd c
ardi
ovas
cula
r de
aths
. T
he
effe
ct
size
es
timat
es
per
the
sam
e di
strib
utio
nal
incr
emen
t (1
0% t
o 90
%)
wer
e la
rger
fo
r O
3
than
fo
r B
S.
The
se
effe
cts
wer
e la
rger
in
war
m s
easo
n. S
O2
an
d N
O2
wer
e
not
cons
iste
ntly
a
ssoc
iate
d w
ith
mor
talit
y.
Per
cent
e
xce
ss
mor
talit
y 2.
8%
(1.4
, 4.
3) p
er 2
5 µ
g/m
3 BS
at
1-d
lag
for
tota
l dea
ths.
C
VD
(1
d): 1
.0%
(-1
.1, 3
.1).
R
espi
rato
ry
(1
d):
1.1%
(-
2.7,
5.
0).
And
erso
n et
al
. (2
001)
(A
nder
son
et a
l, 20
01)
The
W
est
Mid
land
s co
nurb
atio
n,
UK
. 19
94-
1996
. P
M m
eans
: P
M10
: 23
, P
M2.
5: 1
5, P
M10
-2.5:
9,
BS
: 13.
2.
Non
-acc
iden
tal
caus
e,
card
iova
scul
ar,
and
resp
irato
ry
mor
talit
y (a
s w
ell
as
hosp
ital
adm
issi
ons)
w
ere
ana
lyze
d fo
r th
eir
asso
ciat
ions
w
ith
PM
in
dic
es
and
gase
ous
pollu
tant
s us
ing
GA
M P
oiss
on m
odel
s ad
just
ing
for
seas
onal
cyc
les,
day
of
wee
k, a
nd w
eat
her.
Dai
ly
non-
acci
dent
al
mor
talit
y w
as
not
as
soci
ated
w
ith
PM
in
dice
s or
ga
seou
s po
lluta
nts
in t
he a
ll-ye
ar a
nal
ysis
. H
ow
ever
, al
l th
e P
M i
ndic
es (
exc
ept
coar
se p
artic
les)
w
ere
posi
tivel
y an
d si
gnifi
cant
ly a
ssoc
iate
d w
ith n
on-a
ccid
enta
l m
orta
lity
(age
ove
r 65
) in
the
war
m s
easo
n. O
f ga
seou
s po
lluta
nts,
N
O2
and
O3
wer
e po
sitiv
ely
and
sign
ifica
ntly
as
soci
ated
with
non
-acc
ide
ntal
mor
talit
y in
w
arm
sea
son
. T
wo
pollu
tant
mod
els
we
re
not
cons
ider
ed
beca
use
“so
few
as
soci
atio
ns w
ere
foun
d”.
Per
cent
exc
ess
mor
talit
y fo
r P
M10
, P
M2.
5, a
nd P
M10
-2.5 (
lag
0-1)
wer
e:
0.2%
(-1
.8,
2.2
) pe
r 24
.4
µg/
m3
PM
10,
0.6%
(-
1.5,
2.
7)
per
17.7
µ
g/m
3 PM
2.5,
and
-0.
6% (
-4.2
, 2.
3)
per
11.3
µg/
m3 P
M10
-2.5 i
n al
l-yea
r an
alys
is.
The
re
sults
fo
r se
ason
sp
ecifi
c an
alys
is
wer
e gi
ven
only
as
fig
ures
.
Kea
tinge
and
Don
alds
on
(200
1)
(Kea
tinge
an
dDon
alds
on, 2
001)
G
reat
er
Lond
on,
Eng
land
, 197
6-1
995.
B
S m
ean:
17.
7.
The
st
udy
exam
ine
d po
tent
ial
conf
oun
ding
ef
fect
s of
at
ypic
al
cold
w
eat
her
on
air
pollu
tion/
mor
talit
y re
latio
nshi
ps.
Firs
t, ai
r po
llutio
n va
riabl
es
(SO
2,
CO
an
d B
S)
wer
e m
odel
ed
as
a fu
nctio
n of
la
gged
w
eath
er
varia
bles
. T
hese
var
iabl
es w
ere
dese
ason
aliz
ed
by r
egre
ssin
g o
n se
ine
and
cosi
ne v
aria
bles
. M
orta
lity
regr
essi
on
(OLS
) in
clud
ed
vario
us
lagg
ed
and
aver
aged
w
eat
her
and
pollu
tion
varia
bles
. A
naly
ses
wer
e co
nduc
ted
in th
e li
near
ra
nge
of m
orta
lity/
tem
pera
ture
rel
atio
nshi
p (1
5 to
0 d
egre
es C
).
Pol
lute
d da
ys w
ere
foun
d to
be
cold
er a
nd
less
w
indy
a
nd
rain
y th
an
usua
l. In
th
e re
gres
sion
of
m
orta
lity
on
the
mul
tiple
-la
gged
te
mpe
ratu
re,
win
d,
rain
, hu
mid
ity,
suns
hine
, S
O2,
C
O,
and
B
S,
cold
te
mpe
ratu
re w
as
asso
ciat
ed w
ith m
orta
lity
incr
ease
, bu
t no
t S
O2
or C
O.
BS
sho
we
d su
gges
tive
evid
ence
, th
ough
not
sta
tistic
ally
si
gnifi
cant
, of
asso
ciat
ion
at la
g 0
and
lag
1.
Per
cent
exc
ess
mor
talit
y 3%
(N
S)
incr
ease
in
da
ily
mor
talit
y pe
r 17
.7 µ
g/m
3 of
BS
(la
g 0
and
1).
report no. 4/11
101
And
erso
n et
al
. (1
996)
(A
nder
son
et a
l, 19
96)
Lond
on, U
K, 1
987-
1992
. B
S m
ean
15.
Tot
al,
card
iova
scul
ar,
and
resp
irato
ry m
orta
lity
serie
s w
ere
regr
esse
d on
B
S,
O3,
N
O2,
an
d S
O2,
ad
just
ing
for
sea
sona
l cyc
les,
da
y of
wee
k,
influ
enza
, ho
liday
s, t
empe
ratu
re,
hum
idity
, an
d au
toco
rrel
atio
n us
ing
Poi
sson
GLM
mod
el.
Bot
h O
3 (la
g 0)
an
d B
S
(lag
1)
we
re
sign
ifica
nt p
redi
ctor
s of
tot
al d
eath
s. O
3 w
as
also
pos
itive
ly s
igni
fican
tly a
ssoc
iate
d w
ith
resp
irato
ry a
nd c
ardi
ovas
cula
r de
aths
. T
he
effe
ct
size
es
timat
es
per
the
sam
e di
strib
utio
nal
incr
emen
t (1
0% t
o 90
%)
wer
e la
rger
fo
r O
3
than
fo
r B
S.
The
se
effe
cts
wer
e la
rger
in
war
m s
easo
n. S
O2
an
d N
O2
wer
e
not
cons
iste
ntly
a
ssoc
iate
d w
ith
mor
talit
y.
Per
cent
e
xce
ss
mor
talit
y 2.
8%
(1.4
, 4.
3) p
er 2
5 µ
g/m
3 BS
at
1-d
lag
for
tota
l dea
ths.
C
VD
(1
d): 1
.0%
(-1
.1, 3
.1).
R
espi
rato
ry
(1
d):
1.1%
(-
2.7,
5.
0).
And
erso
n et
al
. (2
001)
(A
nder
son
et a
l, 20
01)
The
W
est
Mid
land
s co
nurb
atio
n,
UK
. 19
94-
1996
. P
M m
eans
: P
M10
: 23
, P
M2.
5: 1
5, P
M10
-2.5:
9,
BS
: 13.
2.
Non
-acc
iden
tal
caus
e,
card
iova
scul
ar,
and
resp
irato
ry
mor
talit
y (a
s w
ell
as
hosp
ital
adm
issi
ons)
w
ere
ana
lyze
d fo
r th
eir
asso
ciat
ions
w
ith
PM
in
dic
es
and
gase
ous
pollu
tant
s us
ing
GA
M P
oiss
on m
odel
s ad
just
ing
for
seas
onal
cyc
les,
day
of
wee
k, a
nd w
eat
her.
Dai
ly
non-
acci
dent
al
mor
talit
y w
as
not
as
soci
ated
w
ith
PM
in
dice
s or
ga
seou
s po
lluta
nts
in t
he a
ll-ye
ar a
nal
ysis
. H
ow
ever
, al
l th
e P
M i
ndic
es (
exc
ept
coar
se p
artic
les)
w
ere
posi
tivel
y an
d si
gnifi
cant
ly a
ssoc
iate
d w
ith n
on-a
ccid
enta
l m
orta
lity
(age
ove
r 65
) in
the
war
m s
easo
n. O
f ga
seou
s po
lluta
nts,
N
O2
and
O3
wer
e po
sitiv
ely
and
sign
ifica
ntly
as
soci
ated
with
non
-acc
ide
ntal
mor
talit
y in
w
arm
sea
son
. T
wo
pollu
tant
mod
els
we
re
not
cons
ider
ed
beca
use
“so
few
as
soci
atio
ns w
ere
foun
d”.
Per
cent
exc
ess
mor
talit
y fo
r P
M10
, P
M2.
5, a
nd P
M10
-2.5 (
lag
0-1)
wer
e:
0.2%
(-1
.8,
2.2
) pe
r 24
.4
µg/
m3
PM
10,
0.6%
(-
1.5,
2.
7)
per
17.7
µ
g/m
3 PM
2.5,
and
-0.
6% (
-4.2
, 2.
3)
per
11.3
µg/
m3 P
M10
-2.5 i
n al
l-yea
r an
alys
is.
The
re
sults
fo
r se
ason
sp
ecifi
c an
alys
is
wer
e gi
ven
only
as
fig
ures
.
Kea
tinge
and
Don
alds
on
(200
1)
(Kea
tinge
an
dDon
alds
on, 2
001)
G
reat
er
Lond
on,
Eng
land
, 197
6-1
995.
B
S m
ean:
17.
7.
The
st
udy
exam
ine
d po
tent
ial
conf
oun
ding
ef
fect
s of
at
ypic
al
cold
w
eat
her
on
air
pollu
tion/
mor
talit
y re
latio
nshi
ps.
Firs
t, ai
r po
llutio
n va
riabl
es
(SO
2,
CO
an
d B
S)
wer
e m
odel
ed
as
a fu
nctio
n of
la
gged
w
eath
er
varia
bles
. T
hese
var
iabl
es w
ere
dese
ason
aliz
ed
by r
egre
ssin
g o
n se
ine
and
cosi
ne v
aria
bles
. M
orta
lity
regr
essi
on
(OLS
) in
clud
ed
vario
us
lagg
ed
and
aver
aged
w
eat
her
and
pollu
tion
varia
bles
. A
naly
ses
wer
e co
nduc
ted
in th
e li
near
ra
nge
of m
orta
lity/
tem
pera
ture
rel
atio
nshi
p (1
5 to
0 d
egre
es C
).
Pol
lute
d da
ys w
ere
foun
d to
be
cold
er a
nd
less
w
indy
a
nd
rain
y th
an
usua
l. In
th
e re
gres
sion
of
m
orta
lity
on
the
mul
tiple
-la
gged
te
mpe
ratu
re,
win
d,
rain
, hu
mid
ity,
suns
hine
, S
O2,
C
O,
and
B
S,
cold
te
mpe
ratu
re w
as
asso
ciat
ed w
ith m
orta
lity
incr
ease
, bu
t no
t S
O2
or C
O.
BS
sho
we
d su
gges
tive
evid
ence
, th
ough
not
sta
tistic
ally
si
gnifi
cant
, of
asso
ciat
ion
at la
g 0
and
lag
1.
Per
cent
exc
ess
mor
talit
y 3%
(N
S)
incr
ease
in
da
ily
mor
talit
y pe
r 17
.7 µ
g/m
3 of
BS
(la
g 0
and
1).
report no. 4/11
102
And
erso
n et
al
. (1
996)
(A
nder
son
et a
l, 19
96)
Lond
on, U
K, 1
987-
1992
. B
S m
ean
15.
Tot
al,
card
iova
scul
ar,
and
resp
irato
ry m
orta
lity
serie
s w
ere
regr
esse
d on
B
S,
O3,
N
O2,
an
d S
O2,
ad
just
ing
for
sea
sona
l cyc
les,
da
y of
wee
k,
influ
enza
, ho
liday
s, t
empe
ratu
re,
hum
idity
, an
d au
toco
rrel
atio
n us
ing
Poi
sson
GLM
mod
el.
Bot
h O
3 (la
g 0)
an
d B
S
(lag
1)
we
re
sign
ifica
nt p
redi
ctor
s of
tot
al d
eath
s. O
3 w
as
also
pos
itive
ly s
igni
fican
tly a
ssoc
iate
d w
ith
resp
irato
ry a
nd c
ardi
ovas
cula
r de
aths
. T
he
effe
ct
size
es
timat
es
per
the
sam
e di
strib
utio
nal
incr
emen
t (1
0% t
o 90
%)
wer
e la
rger
fo
r O
3
than
fo
r B
S.
The
se
effe
cts
wer
e la
rger
in
war
m s
easo
n. S
O2
an
d N
O2
wer
e
not
cons
iste
ntly
a
ssoc
iate
d w
ith
mor
talit
y.
Per
cent
e
xce
ss
mor
talit
y 2.
8%
(1.4
, 4.
3) p
er 2
5 µ
g/m
3 BS
at
1-d
lag
for
tota
l dea
ths.
C
VD
(1
d): 1
.0%
(-1
.1, 3
.1).
R
espi
rato
ry
(1
d):
1.1%
(-
2.7,
5.
0).
And
erso
n et
al
. (2
001)
(A
nder
son
et a
l, 20
01)
The
W
est
Mid
land
s co
nurb
atio
n,
UK
. 19
94-
1996
. P
M m
eans
: P
M10
: 23
, P
M2.
5: 1
5, P
M10
-2.5:
9,
BS
: 13.
2.
Non
-acc
iden
tal
caus
e,
card
iova
scul
ar,
and
resp
irato
ry
mor
talit
y (a
s w
ell
as
hosp
ital
adm
issi
ons)
w
ere
ana
lyze
d fo
r th
eir
asso
ciat
ions
w
ith
PM
in
dic
es
and
gase
ous
pollu
tant
s us
ing
GA
M P
oiss
on m
odel
s ad
just
ing
for
seas
onal
cyc
les,
day
of
wee
k, a
nd w
eat
her.
Dai
ly
non-
acci
dent
al
mor
talit
y w
as
not
as
soci
ated
w
ith
PM
in
dice
s or
ga
seou
s po
lluta
nts
in t
he a
ll-ye
ar a
nal
ysis
. H
ow
ever
, al
l th
e P
M i
ndic
es (
exc
ept
coar
se p
artic
les)
w
ere
posi
tivel
y an
d si
gnifi
cant
ly a
ssoc
iate
d w
ith n
on-a
ccid
enta
l m
orta
lity
(age
ove
r 65
) in
the
war
m s
easo
n. O
f ga
seou
s po
lluta
nts,
N
O2
and
O3
wer
e po
sitiv
ely
and
sign
ifica
ntly
as
soci
ated
with
non
-acc
ide
ntal
mor
talit
y in
w
arm
sea
son
. T
wo
pollu
tant
mod
els
we
re
not
cons
ider
ed
beca
use
“so
few
as
soci
atio
ns w
ere
foun
d”.
Per
cent
exc
ess
mor
talit
y fo
r P
M10
, P
M2.
5, a
nd P
M10
-2.5 (
lag
0-1)
wer
e:
0.2%
(-1
.8,
2.2
) pe
r 24
.4
µg/
m3
PM
10,
0.6%
(-
1.5,
2.
7)
per
17.7
µ
g/m
3 PM
2.5,
and
-0.
6% (
-4.2
, 2.
3)
per
11.3
µg/
m3 P
M10
-2.5 i
n al
l-yea
r an
alys
is.
The
re
sults
fo
r se
ason
sp
ecifi
c an
alys
is
wer
e gi
ven
only
as
fig
ures
.
Kea
tinge
and
Don
alds
on
(200
1)
(Kea
tinge
an
dDon
alds
on, 2
001)
G
reat
er
Lond
on,
Eng
land
, 197
6-1
995.
B
S m
ean:
17.
7.
The
st
udy
exam
ine
d po
tent
ial
conf
oun
ding
ef
fect
s of
at
ypic
al
cold
w
eat
her
on
air
pollu
tion/
mor
talit
y re
latio
nshi
ps.
Firs
t, ai
r po
llutio
n va
riabl
es
(SO
2,
CO
an
d B
S)
wer
e m
odel
ed
as
a fu
nctio
n of
la
gged
w
eath
er
varia
bles
. T
hese
var
iabl
es w
ere
dese
ason
aliz
ed
by r
egre
ssin
g o
n se
ine
and
cosi
ne v
aria
bles
. M
orta
lity
regr
essi
on
(OLS
) in
clud
ed
vario
us
lagg
ed
and
aver
aged
w
eat
her
and
pollu
tion
varia
bles
. A
naly
ses
wer
e co
nduc
ted
in th
e li
near
ra
nge
of m
orta
lity/
tem
pera
ture
rel
atio
nshi
p (1
5 to
0 d
egre
es C
).
Pol
lute
d da
ys w
ere
foun
d to
be
cold
er a
nd
less
w
indy
a
nd
rain
y th
an
usua
l. In
th
e re
gres
sion
of
m
orta
lity
on
the
mul
tiple
-la
gged
te
mpe
ratu
re,
win
d,
rain
, hu
mid
ity,
suns
hine
, S
O2,
C
O,
and
B
S,
cold
te
mpe
ratu
re w
as
asso
ciat
ed w
ith m
orta
lity
incr
ease
, bu
t no
t S
O2
or C
O.
BS
sho
we
d su
gges
tive
evid
ence
, th
ough
not
sta
tistic
ally
si
gnifi
cant
, of
asso
ciat
ion
at la
g 0
and
lag
1.
Per
cent
exc
ess
mor
talit
y 3%
(N
S)
incr
ease
in
da
ily
mor
talit
y pe
r 17
.7 µ
g/m
3 of
BS
(la
g 0
and
1).
report no. 4/11
103
Goo
dman
et
al
., 20
04
(Goo
dman
et
al
, 20
04);
D
ublin
, Ir
elan
d,
1980
-19
96, B
S m
ean
40.
Thi
s st
udy
anal
ysed
the
acu
te a
nd m
ediu
m-t
erm
(w
eeks
to
mon
ths)
exp
osur
e ef
fect
s of
BS
and
te
mpe
ratu
re
on
tota
l an
d ca
use-
spec
ific
mor
talit
y. I
n th
e an
alys
is o
f th
e ac
ute
effe
cts
of
expo
sure
(3
-day
m
ean
BS
),
GA
M
Po
isso
n re
gres
sion
mod
els
for
the
log
coun
ts o
f da
ily
age-
stan
dard
ized
mor
talit
y ra
tes
wer
e ad
just
ed
for
met
eoro
logi
cal
mea
sure
s,
incl
udin
g te
mpe
ratu
re,
hum
idity
o
n th
e sa
me
and
prev
ious
day
, pl
us i
ndic
ato
rs f
or d
ay o
f w
eek
and
for
resp
irato
ry e
pide
mic
s.
The
cum
ulat
ive
effe
cts
of
tem
pera
ture
an
d pa
rtic
ulat
e ai
r po
llutio
n ov
er
40
days
w
ere
anal
yzed
si
mul
tane
ousl
y in
a
GA
M
mod
el
incl
udin
g di
strib
uted
la
g fu
nctio
ns
for
min
imum
te
mpe
ratu
re a
nd B
S.
D
aily
B
S
air
pollu
tion
conc
entr
atio
ns
wer
e m
easu
red
at 6
res
iden
tial m
onito
ring
stat
ions
.
The
ef
fect
s of
pa
rtic
ulat
e ai
r po
llutio
n on
m
orta
lity
wer
e s
tron
gest
on
the
day
of a
nd
the
few
day
s af
ter
expo
sure
but
ext
end
ed
out
thro
ugh
abou
t 40
day
s af
ter
expo
sure
. T
his
ext
ende
d a
ir po
llutio
n a
ssoc
iatio
n w
as
mos
t m
arke
d fo
r th
e el
derly
po
pula
tion
grou
ps a
nd f
or r
espi
rato
ry c
ause
s of
dea
th.
The
ext
ende
d f
ollo
w-u
p ef
fect
s w
ere
two
to
thre
e tim
es g
reat
er t
han
the
acut
e ef
fect
s re
port
ed i
n ot
her
stud
ies,
and
app
roac
hed
the
effe
cts
repo
rted
in
long
er-t
erm
sur
viva
l st
udie
s. T
he s
tudi
es o
n th
e ac
ute
effe
cts
of
air
pollu
tion
cou
ld h
ave
un
dere
stim
ated
the
to
tal
effe
ct o
f te
mpe
ratu
re a
nd p
artic
ulat
e ai
r po
llutio
n o
n m
orta
lity.
For
an
incr
emen
t of
10
µg/
m3 o
f B
S,
tota
l m
orta
lity
incr
ease
d by
0.
4% (
0.3,
0.6
) w
hen
only
acu
te
effe
cts
(3-d
ay
mea
n)
wer
e
cons
ider
ed a
nd
by 1
.1%
(0.
8, 1
.3)
whe
n co
nsid
eri
ng
deat
hs
in
the
40
days
af
ter
exp
osur
e.
Cor
resp
ondi
ng
estim
ates
w
ere
0.4%
(0
.2,
0.7)
an
d 1.
1%
(0
.7,
1.5)
for
CV
D m
orta
lity,
and
0.9
%
(0.5
to
1.2)
an
d 3.
6% (
3.0,
4.3
) fo
r re
spira
tory
mor
talit
y.
FIN
LAN
D
P
önkä
et
al
. (1
998)
(P
onka
et a
l, 1
998)
H
elsi
nki,
Fin
land
, 19
87-
1993
. T
SP
med
ian:
64
P
M10
med
ian:
28.
Tot
al a
nd c
ardi
ovas
cula
r d
eath
s, f
or a
ge g
roup
s <
65
and
65 +
, w
ere
rela
ted
to
PM
10,
TS
P,
SO
2,
NO
2,
and
O
3,
usin
g P
oiss
on
GLM
m
odel
ad
just
ing
for
tem
pera
ture
, re
lativ
e hu
mid
ity,
day-
of-w
eek,
te
mpo
ral
patte
rns,
ho
liday
an
d in
fluen
za e
pid
emic
s.
No
pollu
tant
si
gnifi
cant
ly
asso
ciat
ed
with
m
orta
lity
from
al
l ca
rdio
vasc
ular
or
C
VD
ca
uses
in
65+
yea
r ag
e gr
oup.
Onl
y in
age
<
65 y
ear
gro
up,
PM
10 a
ssoc
iate
d w
ith t
otal
an
d C
VD
de
aths
w
ith
4 an
d 5
d la
gs,
re
spec
tivel
y.
The
“s
igni
fican
t”
lags
w
ere
ra
ther
“sp
iky”
. O
3 w
as
also
ass
ocia
ted
with
C
VD
m
orta
lity
<65
ye
ars
grou
p w
ith
inco
nsis
tent
sig
ns a
nd l
ate
and
spik
y la
gs
(neg
ativ
e on
d 5
and
pos
itive
on
d 6)
.
18.8
%
(5.6
, 33
.2)
per
50
µg/
m3
PM
10
4 da
y la
g (o
ther
la
gs
nega
tive
or z
ero
).
report no. 4/11
104
Goo
dman
et
al
., 20
04
(Goo
dman
et
al
, 20
04);
D
ublin
, Ir
elan
d,
1980
-19
96, B
S m
ean
40.
Thi
s st
udy
anal
ysed
the
acu
te a
nd m
ediu
m-t
erm
(w
eeks
to
mon
ths)
exp
osur
e ef
fect
s of
BS
and
te
mpe
ratu
re
on
tota
l an
d ca
use-
spec
ific
mor
talit
y. I
n th
e an
alys
is o
f th
e ac
ute
effe
cts
of
expo
sure
(3
-day
m
ean
BS
),
GA
M
Po
isso
n re
gres
sion
mod
els
for
the
log
coun
ts o
f da
ily
age-
stan
dard
ized
mor
talit
y ra
tes
wer
e ad
just
ed
for
met
eoro
logi
cal
mea
sure
s,
incl
udin
g te
mpe
ratu
re,
hum
idity
o
n th
e sa
me
and
prev
ious
day
, pl
us i
ndic
ato
rs f
or d
ay o
f w
eek
and
for
resp
irato
ry e
pide
mic
s.
The
cum
ulat
ive
effe
cts
of
tem
pera
ture
an
d pa
rtic
ulat
e ai
r po
llutio
n ov
er
40
days
w
ere
anal
yzed
si
mul
tane
ousl
y in
a
GA
M
mod
el
incl
udin
g di
strib
uted
la
g fu
nctio
ns
for
min
imum
te
mpe
ratu
re a
nd B
S.
D
aily
B
S
air
pollu
tion
conc
entr
atio
ns
wer
e m
easu
red
at 6
res
iden
tial m
onito
ring
stat
ions
.
The
ef
fect
s of
pa
rtic
ulat
e ai
r po
llutio
n on
m
orta
lity
wer
e s
tron
gest
on
the
day
of a
nd
the
few
day
s af
ter
expo
sure
but
ext
end
ed
out
thro
ugh
abou
t 40
day
s af
ter
expo
sure
. T
his
ext
ende
d a
ir po
llutio
n a
ssoc
iatio
n w
as
mos
t m
arke
d fo
r th
e el
derly
po
pula
tion
grou
ps a
nd f
or r
espi
rato
ry c
ause
s of
dea
th.
The
ext
ende
d f
ollo
w-u
p ef
fect
s w
ere
two
to
thre
e tim
es g
reat
er t
han
the
acut
e ef
fect
s re
port
ed i
n ot
her
stud
ies,
and
app
roac
hed
the
effe
cts
repo
rted
in
long
er-t
erm
sur
viva
l st
udie
s. T
he s
tudi
es o
n th
e ac
ute
effe
cts
of
air
pollu
tion
cou
ld h
ave
un
dere
stim
ated
the
to
tal
effe
ct o
f te
mpe
ratu
re a
nd p
artic
ulat
e ai
r po
llutio
n o
n m
orta
lity.
For
an
incr
emen
t of
10
µg/
m3 o
f B
S,
tota
l m
orta
lity
incr
ease
d by
0.
4% (
0.3,
0.6
) w
hen
only
acu
te
effe
cts
(3-d
ay
mea
n)
wer
e
cons
ider
ed a
nd
by 1
.1%
(0.
8, 1
.3)
whe
n co
nsid
eri
ng
deat
hs
in
the
40
days
af
ter
exp
osur
e.
Cor
resp
ondi
ng
estim
ates
w
ere
0.4%
(0
.2,
0.7)
an
d 1.
1%
(0
.7,
1.5)
for
CV
D m
orta
lity,
and
0.9
%
(0.5
to
1.2)
an
d 3.
6% (
3.0,
4.3
) fo
r re
spira
tory
mor
talit
y.
FIN
LAN
D
P
önkä
et
al
. (1
998)
(P
onka
et a
l, 1
998)
H
elsi
nki,
Fin
land
, 19
87-
1993
. T
SP
med
ian:
64
P
M10
med
ian:
28.
Tot
al a
nd c
ardi
ovas
cula
r d
eath
s, f
or a
ge g
roup
s <
65
and
65 +
, w
ere
rela
ted
to
PM
10,
TS
P,
SO
2,
NO
2,
and
O
3,
usin
g P
oiss
on
GLM
m
odel
ad
just
ing
for
tem
pera
ture
, re
lativ
e hu
mid
ity,
day-
of-w
eek,
te
mpo
ral
patte
rns,
ho
liday
an
d in
fluen
za e
pid
emic
s.
No
pollu
tant
si
gnifi
cant
ly
asso
ciat
ed
with
m
orta
lity
from
al
l ca
rdio
vasc
ular
or
C
VD
ca
uses
in
65+
yea
r ag
e gr
oup.
Onl
y in
age
<
65 y
ear
gro
up,
PM
10 a
ssoc
iate
d w
ith t
otal
an
d C
VD
de
aths
w
ith
4 an
d 5
d la
gs,
re
spec
tivel
y.
The
“s
igni
fican
t”
lags
w
ere
ra
ther
“sp
iky”
. O
3 w
as
also
ass
ocia
ted
with
C
VD
m
orta
lity
<65
ye
ars
grou
p w
ith
inco
nsis
tent
sig
ns a
nd l
ate
and
spik
y la
gs
(neg
ativ
e on
d 5
and
pos
itive
on
d 6)
.
18.8
%
(5.6
, 33
.2)
per
50
µg/
m3
PM
10
4 da
y la
g (o
ther
la
gs
nega
tive
or z
ero
).
report no. 4/11
105
THE
NET
HER
LAN
DS
R
oem
er a
nd V
an W
ijnen
(2
001)
. (R
oem
er a
ndva
n W
ijnen
, 200
1)
Am
ster
dam
. 198
7-19
98.
BS
and
PM
10 m
eans
in
“b
ackg
roun
d”: 1
0 an
d 39
B
S m
ean
in “
traf
fic”
area
: 21
. (N
o P
M10
m
easu
rem
ents
av
aila
ble
at
traf
fic s
ites)
.
Dai
ly d
eath
s fo
r th
ose
wh
o liv
ed a
long
roa
ds
with
mor
e th
an 1
0,00
0 m
otor
veh
icle
, as
we
ll as
de
aths
for
tot
al p
opul
atio
n, w
ere
ana
lyze
d us
ing
data
fr
om
bac
kgro
und
and
tr
affic
m
onito
rs.
Poi
sson
G
AM
m
odel
w
as
used
ad
just
ing
for
seas
on,
day-
of-w
eek,
and
wea
ther
. B
S,
PM
10,
SO
2, N
O2,
CO
, and
O3
wer
e an
alyz
ed.
Cor
rela
tions
be
twee
n th
e ba
ckgr
oun
d m
onito
rs a
nd t
raffi
c m
onito
rs w
ere
mod
erat
e fo
r B
S (
r =
0.5
5) b
ut h
igh
er f
or N
O2 (
r =
0.
79)
and
O3
(r =
0.8
0).
BS
and
NO
2 w
ere
asso
ciat
ed w
ith m
orta
lity
in b
oth
tota
l an
d tr
affic
pop
ulat
ion.
Est
imat
ed
RR
for
tra
ffic
popu
latio
n u
sing
ba
ckgr
ound
si
tes
was
la
rger
tha
n th
e R
R f
or t
otal
pop
ulat
ion
usin
g ba
ckgr
ound
si
tes.
T
he
RR
fo
r to
tal
popu
latio
n us
ing
traf
fic
site
s w
as
smal
ler
that
R
Rs
for
tota
l po
pula
tion
usin
g ba
ckgr
ound
si
tes.
T
his
is
not
surp
risin
g si
nce
the
mea
n B
S
for
traf
fic
site
s w
ere
la
rger
than
for
back
grou
nd s
ites.
The
exc
ess
risk
per
100
µg
/m3 o
f B
S (
lag
1):
Tot
al
popu
latio
n us
ing
back
grou
nd
site
s:
38.3
%
(15.
3,
65.9
) T
raffi
c po
pula
tion
us
ing
ba
ckgr
ound
si
tes:
88
.7%
(2
0.7,
19
4.9)
T
otal
pop
ulat
ion
usin
g tr
affic
site
s:
12.2
% (
2.3,
23.
1).
Res
ults
for
tra
ffic
popu
latio
n u
sing
tr
affic
site
s no
t re
port
ed.
Hoe
k et
al.
(199
7) (
Hoe
k et
al,
1997
) R
otte
rdam
, th
e N
ethe
rland
s, 1
983-
1991
. T
SP
med
ian:
42,
B
S m
edia
n: 1
3.
Tot
al
mor
talit
y (a
lso
by
age
grou
p)
was
re
gres
sed
on T
SP
, F
e (f
rom
TS
P f
ilter
), B
S,
O3,
S
O2,
CO
, ad
just
ing
for
sea
sona
l cy
cles
, da
y-of
-w
eek,
in
fluen
za,
tem
pera
ture
, an
d hu
mid
ity
usin
g P
oiss
on G
AM
mod
el.
Dai
ly
deat
hs
wer
e m
ost
co
nsis
ten
tly
asso
ciat
ed w
ith T
SP
. T
SP
and
O3
effe
cts
wer
e “
inde
pend
ent”
of
SO
2 an
d C
O.
To
tal
iron
(fro
m T
SP
filt
er)
was
ass
ocia
ted
“less
co
nsis
tent
ly”
with
mor
talit
y th
an T
SP
was
. T
he
estim
ated
R
Rs
for
PM
in
dice
s w
ere
hi
gher
in w
arm
sea
son
tha
n in
col
d se
ason
.
Per
cent
exc
ess
mor
talit
y:
5.5%
(1
.1,
9.9)
pe
r 10
0 µ
g/m
3 T
SP
at l
ag 1
.
report no. 4/11
106
GER
MA
NY
W
ichm
ann
et a
l., (
2000
) (W
ichm
ann
et a
l, 20
00)
Erf
urt,
Ger
man
y.
1995
-19
98.
Num
ber
coun
ts (
NC
) an
d
mas
s co
nce
ntra
tions
(M
C)
of
ultr
afin
e pa
rtic
les
usin
g S
pect
rom
etry
ll M
obile
A
eros
ol
Sp
ectr
omet
ry
(MA
S).
M
AS
M
C
PM
2.5-
0.01
(m
ean
25.8
, m
edia
n
18.8
, IQ
R
19.9
).
Filt
er
mea
sure
men
ts
of
PM
10
(mea
n 38
.2,
med
ian
31
.0,
IQR
27
.7)
and
PM
2.5
(mea
n 26
.3,
med
ian
20.2
, IQ
R 1
8.5)
. M
AS
N
C2.
5-0
.01
(mea
n
17,9
66 p
er c
m3 ,
med
ian
14,7
69, I
QR
13,
269)
.
Tot
al
non-
acci
dent
al,
card
iova
scul
ar,
and
resp
irato
ry d
eath
s (m
ean
4.88
, 2.
87,
1.08
per
da
y, r
espe
ctiv
ely)
wer
e re
late
d to
par
ticle
mas
s co
ncen
trat
ion
and
num
ber
coun
ts i
n ea
ch s
ize
clas
s, a
nd t
o m
ass
conc
ent
ratio
ns o
f g
aseo
us
co-p
ollu
tant
s N
O2,
C
O,
SO
2,
usin
g G
AM
re
gres
sion
mod
els
adju
sted
for
tem
pora
l tre
nds,
da
y of
wee
k, w
eek
ly n
atio
nal
influ
enza
rat
es,
tem
pera
ture
an
d re
lativ
e hu
mid
ity.
Dat
a an
alyz
ed b
y se
ason
, ag
e gr
oup,
and
cau
se o
f de
ath
sepa
rate
ly.
Sin
gle
- da
y la
gs
and
poly
nom
ial
dist
ribut
ed l
ag m
odel
s (P
DL)
use
d.
Par
ticle
indi
ces
and
pollu
tan
ts f
itted
usi
ng li
near
, lo
g- t
rans
form
ed,
and
LOE
SS
tra
nsfo
rmat
ions
. T
wo-
pollu
tant
mod
els
with
a p
artic
le in
dex
and
a ga
seou
s po
lluta
nt w
ere
fitte
d. T
he “
best
” m
odel
w
as
that
ha
ving
th
e hi
ghes
t t-
stat
istic
, si
nce
ot
her
crite
ria
(e.g
., lo
g-lik
elih
ood
for
nest
ed
mod
els)
and
AIC
for
non
-nes
ted
mod
els
coul
d
not
be
appl
ied
due
to
diffe
rent
nu
mbe
rs
of
obse
rvat
ions
in
each
mod
el.
The
re s
houl
d be
lit
tle d
iffer
ence
bet
wee
n t
hes
e ap
proa
ches
and
re
sulti
ng d
iffer
ence
s in
res
ults
sho
uld
be
smal
l in
pr
actic
e.
Sen
sitiv
ity
anal
yses
in
clud
ed
stra
tifyi
ng
data
by
se
ason
, w
inte
r ye
ar,
age,
ca
use
of
deat
h,
or
tran
sfor
mat
ion
of
the
pollu
tion
varia
ble
(non
e,
loga
rithm
ic,
non-
para
met
ric s
moo
th).
Loss
of
stat
istic
al p
ow
er b
y us
ing
a sm
all
city
w
ith
a sm
all
num
ber
of
deat
hs
wa
s of
fset
by
a
dvan
tage
of
ha
ving
go
od
expo
sure
re
pres
enta
tion
fr
om
sing
le
mon
itorin
g si
te.
Sin
ce u
ltraf
ine
part
icle
s ca
n co
agul
ate
into
lar
ger
aggr
egat
es i
n a
few
ho
urs,
ultr
afin
e pa
rtic
le s
ize
and
num
bers
ca
n in
crea
se i
nto
the
fine
par
ticle
cat
egor
y,
resu
lting
in
so
me
ambi
gui
ty.
Sig
nific
ant
asso
ciat
ions
wer
e fo
und
bet
we
en m
orta
lity
and
ultr
afin
e pa
rtic
le N
C,
ultr
afin
e pa
rtic
le
MC
, fin
e pa
rtic
le
mas
s co
ncen
trat
ion,
or
S
O2
conc
entr
atio
n. T
he c
orre
latio
n be
twe
en
MC
0.01
-2.5 a
nd
NC
0.01
-0.1 w
as o
nly
mod
erat
e,
sugg
estin
g it
may
be
poss
ible
to
part
ially
se
para
te
effe
cts
of
ultr
afin
e an
d fin
e pa
rtic
les.
T
he
mos
t pr
edic
tive
sing
le-d
ay
effe
cts
are
eith
er i
mm
edia
te (
lag
0 or
1)
or
dela
yed
(lag
4 or
5),
but
cum
ulat
ive
effe
cts
char
acte
rized
by
PD
L ar
e la
rger
tha
n si
ngle
-da
y ef
fect
s.
The
si
gnifi
can
ce
of
SO
2 w
as
robu
st,
but
hard
to
exp
lain
as
a tr
ue c
ausa
l fa
ctor
si
nce
its
conc
entr
atio
ns
wer
e ve
ry
low
. A
ge w
as
an im
port
ant
mod
ifyin
g fa
ctor
, w
ith l
arge
r ef
fect
s at
age
s <
70
than
> 7
0 ye
ars.
R
espi
rato
ry
mor
talit
y ha
d a
high
er
RR
tha
n C
VD
mor
talit
y. A
lar
ge n
umbe
r of
m
odel
s w
ere
fit
ted,
w
ith
som
e si
gnifi
cant
fin
ding
s of
a
ssoc
iatio
n be
twee
n m
orta
lity
and
part
icle
mas
s or
num
ber
indi
ces.
Tot
al m
orta
lity
exce
ss d
eath
s:
Filt
er
PM
10
(lag
0-4)
: 6.
6 (0
.7,
12.8
) pe
r 50
µg/
m3 .
Filt
er
PM
2.5
(lag
0-1)
: 3.
0 (-
1.7,
7.
9).
MC
for
PM
0.01
-2.5 6
.2%
(1.
4, 1
1.2)
fo
r al
l yea
r W
inte
r: 9
.2%
(3.
0, 1
5.7)
S
prin
g: 5
.2%
(-2
.0, 1
2.8)
S
umm
er: -
4.7%
(-1
8.7,
11.
7)
Fal
l: 9.
7% (
1.9
, 18.
1)
For
ultr
afin
e P
M,
NC
0.01
-0.1 (
lag
0-4)
: A
ll Y
ear:
8.2
% (
0.3,
16.
9)
Win
ter:
9.7
% (
0.3,
19.
9)
Spr
ing:
10.
5%
(-1
.4, 2
3.9)
S
umm
er: -
13.9
% (
-29.
8, 5
.7)
F
all:
12.0
% (
2.1,
22.
7)
Bes
t sin
gle-
day
lag
per
25 µ
g/m
3 : P
M0.
01-0
.1: 3
.6%
(-0
.4, 7
.7)
P
M0.
01-2
.5: 3
.9%
(0.
0, 8
.0)
P
M2.
5: -
4.0%
(-7
.9, 0
)
PM
10: 6
.4%
(0
.3, 1
2.9)
.
report no. 4/11
107
Sto
lzel
et
al
. (2
003)
(S
tölz
el e
t al,
2003
) R
e-an
alys
is o
f th
e ab
ove
st
udy.
Re-
anal
ysis
o
f ab
ove
stud
y us
ing
GA
M
with
st
ringe
nt
con
verg
ence
cr
iteria
as
w
ell
as
GLM
/nat
ural
spl
ines
. T
he p
olyn
omia
l di
strib
uted
la
g m
odel
was
not
re-
anal
yzed
.
Ver
y lit
tle
chan
ge
in
PM
ris
k co
effic
ien
ts
whe
n G
AM
m
odel
s w
ith
strin
gent
co
nver
genc
e
crite
ria
wer
e us
ed.
Wh
en
GLM
/nat
ural
spl
ines
wer
e us
ed,
man
y of
th
e co
effic
ient
s fo
r nu
mbe
r co
ncen
trat
ions
sl
ight
ly
incr
ease
d,
but
the
co
effic
ient
s fo
r m
ass
conc
entr
atio
ns d
ecre
ased
slig
htly
.
Bes
t si
ngle
-day
la
g us
ing
G
AM
(s
trin
gent
) pe
r 25
µg/
m3 :
P
M0.
01-0
.1: 3
.6%
(-0
.4, 7
.7)
P
M0.
01-2
.5: 3
.8%
(-0
.1, 7
.8)
P
M2.
5: -
4.0%
(-7
.8, -
0.1)
P
M10
: 6.2
% (
0.1
, 12.
7)
Bes
t si
ngl
e-d
ay
lag
us
ing
GLM
/nat
ural
spl
ines
:
PM
0.01
-0.1: 3
.1%
(-1
.6, 7
.9)
P
M0.
01-2
.5: 3
.7%
(-0
.9, 8
.4)
P
M2.
5: -
3.4%
(-7
.9, 1
.4)
P
M10
: 5.3
% (
-1.8
, 12.
9).
Sto
lzel
et
al
., 20
06
(Sto
lzel
et a
l, 20
06)
Erf
urt,
Ger
man
y,
1995
-20
01.
To
stud
y th
e as
soci
atio
n b
etw
een
PM
0.01
-0.1 a
nd
PM
0.1-
2.5
and
to
tal
and
card
io-r
espi
rato
ry
mor
talit
y P
oiss
on
GA
M
mod
els
adju
ste
d fo
r tr
end,
sea
sona
lity,
inf
luen
za e
pide
mic
s, d
ay o
f th
e w
eek
, a
nd
met
eoro
logy
us
ing
smoo
th
func
tions
or
indi
cato
r va
riabl
es w
ere
used
.
Thi
s st
udy
show
s th
at
ultr
afin
e pa
rtic
les,
re
pres
entin
g fr
esh
com
bust
ion
part
icle
s,
coul
d be
an
impo
rtan
t co
mpo
nent
of
urba
n ai
r po
llutio
n a
ssoc
iate
d w
ith h
ealth
effe
cts.
In
fac
t, st
atis
tical
ly s
igni
fica
nt a
ssoc
iatio
ns
wer
e fo
und
be
twee
n e
leva
ted
ultr
afin
e pa
rtic
le n
umbe
r co
ncen
trat
ions
and
tot
al a
s w
ell
as c
ardi
o-re
spira
tory
mor
talit
y (la
g 4)
. N
o as
soci
atio
n be
twee
n fin
e pa
rtic
le
MC
an
d m
orta
lity
wa
s fo
und.
The
per
cent
mor
talit
y ex
cess
ris
k (la
g 4)
for
an
inc
rem
ent
of 9
,748
no
/cm
3 in P
M0.
1-2.
5, w
as 2
.9%
(0.
3,
5.5)
for
tot
al m
orta
lity,
and
3.1
%
(0.3
, 6.
0)
for
card
io-r
espi
rato
ry
mor
talit
y.
Rah
lenb
eck
and
Kah
l (1
996)
(R
ahle
nbec
k an
dKah
l, 19
96)
Eas
t B
erlin
, 19
81-1
989.
T
SP
(b
eta
atte
nuat
ion)
: 97
.
Tot
al m
orta
lity
(as
wel
l as
dev
iatio
ns f
rom
lon
g-w
ave
cyc
les)
wa
s re
gres
sed
(OLS
) on
TS
P a
nd
SO
2, a
djus
ting
for
day
of w
eek,
mon
th,
year
, te
mpe
ratu
re,
and
rela
tive
hum
idity
, us
ing
OLS
, w
ith o
ptio
ns t
o lo
g-tr
ansf
orm
pol
lutio
n, a
nd w
/ an
d w
/o d
ays
with
pol
lutio
n ab
ove
150
µg/
m3 .
Bot
h S
P
and
SO
2 w
ere
si
gnifi
cant
ly
asso
ciat
ed w
ith t
otal
mor
talit
y w
ith 2
day
lag
in
sing
le
pollu
tant
m
odel
. W
hen
both
po
lluta
nts
wer
e in
clud
ed,
thei
r co
effic
ien
ts
wer
e re
duce
d by
33%
and
46%
for
TS
P a
nd
SO
2, r
espe
ctiv
ely.
6.1%
per
100
µg/
m3 T
SP
at
lag
2.
report no. 4/11
108
BA
VAR
IA A
ND
CZE
CH
Pet
ers
et
al.
(200
0)
(Pet
ers
et a
l, 20
00)
A
high
ly
pollu
ted
coal
ba
sin
area
in
the
Cze
ch
Rep
ublic
an
d a
rura
l ar
ea
in
Ger
man
y,
nort
heas
t B
avar
ia
dist
ricts
. 198
2-1
994.
T
SP
m
ean:
12
1.1
and
51.6
, re
spec
tivel
y,
for
thes
e tw
o re
gio
ns.
PM
10
and
PM
2.5
wer
e al
so
mea
sure
d in
th
e co
al
basi
n du
ring
1993
-199
4 (m
ean:
65
.9
and
51.0
, re
spec
tivel
y).
Non
-acc
iden
tal
tota
l an
d ca
rdio
vasc
ular
dea
ths
(mea
n: 1
8.2
and
12.0
per
day
, fo
r th
e C
zech
an
d B
avar
ia a
reas
, re
spec
tivel
y).
The
AP
HE
A
appr
oach
(P
oiss
on G
LM m
odel
with
sin
e/co
sine
, te
mpe
ratu
re
as
a qu
adra
tic
func
tion,
re
lativ
e hu
mid
ity,
influ
enza
, da
y-of
-wee
k as
cov
aria
tes)
, as
we
ll as
GL
M w
ith n
atur
al s
plin
es f
or t
empo
ral
tren
ds
and
wea
ther
te
rms
wer
e co
nsid
ered
. Lo
garit
hm o
f T
SP
, S
O2,
NO
2, O
3, a
nd C
O (
and
PM
10 a
nd P
M2.
5 fo
r 19
93-1
994)
wer
e e
xam
ined
at
lags
0 th
roug
h 3
days
.
In t
he c
oal
bas
in (
i.e.,
the
Cze
ch R
epu
blic
po
llute
d ar
ea),
on
the
aver
age,
68%
of
the
TS
P w
as P
M1
0, a
nd m
ost
of P
M10
was
PM
2.5
(75%
).
For
th
e co
al
basi
n,
asso
ciat
ions
w
ere
foun
d b
etw
een
the
lo
garit
hm o
f T
SP
an
d al
l-cau
se m
orta
lity
at l
ag 1
or
2 d
ays
. S
O2
was
al
so
asso
ciat
ed
with
al
l-cau
se
mor
talit
y w
ith
slig
htly
lo
wer
si
gnifi
canc
e.
PM
10 a
nd P
M2.
5 w
ere
both
ass
ocia
ted
with
al
l-cau
se m
orta
lity
in 1
993-
1994
with
a l
ag
of 1
-day
. N
O2,
O3
and
CO
wer
e po
sitiv
ely
but
mor
e w
eak
ly a
ssoc
iate
d w
ith m
orta
lity
than
P
M
indi
ces
or
SO
2.
In
the
Bav
aria
n re
gion
, ne
ither
TS
P n
or S
O2
wa
s as
soci
ate
d w
ith m
orta
lity,
but
CO
(at
la
g 1-
day)
and
O3
(at
lag
0-da
y)
wer
e as
soci
ated
w
ith
all-
caus
e m
orta
lity.
Tot
al m
orta
lity
exce
ss d
eath
s pe
r 10
0 µ
g/m
3 incr
ease
in T
SP
for
the
C
zech
reg
ion:
3.
8% (
0.8,
6.9
) at
lag
2 f
or 1
982-
1994
pe
riod.
F
or
perio
d 19
93-
1994
, 9.
5 %
(1.
2, 1
8.5)
per
100
µ
g/m
3 inc
reas
e in
TS
P a
t la
g 1,
an
d 4.
8% (
0.7
, 9.
0) p
er 5
0 µ
g/m
3 in
crea
se i
n P
M10
; an
d 1.
4 (-
0.5,
3.
4) p
er 2
5 µ
g/m
3 PM
2.5.
Kot
šove
c et
al
. (2
000)
(K
otes
ovec
et
al, 2
000)
N
orth
ern
Boh
emia
, C
zech
R
epu
blic
, 19
82-
1994
. T
SP
mea
n: 1
21.
3.
Tot
al (
excl
udin
g ac
cide
nts
and
child
ren
youn
ger
than
1 y
ear)
, ca
use
spec
ific
(car
diov
ascu
lar
and
canc
er),
age
(65
and
les
s vs
. ot
herw
ise)
, an
d ge
nder
spe
cific
mor
talit
y se
ries
wer
e e
xam
ined
fo
r th
eir
asso
ciat
ions
with
TS
P a
nd S
O2 u
sing
lo
gist
ic
mod
el,
adju
stin
g fo
r se
ason
al
cycl
es,
influ
enza
ep
idem
ics,
lin
ear
an
d qu
adra
tic
tem
pera
ture
ter
ms.
Lag
s 0
thro
ugh
6 da
ys,
as
we
ll as
a 7
da
y m
ean
valu
es w
ere
exa
min
ed.
For
the
tot
al m
orta
lity,
TS
P,
but
not
SO
2,
wa
s as
soci
ated
. A
ppar
ent
diffe
renc
es
in
asso
ciat
ions
wer
e fo
und
bet
we
en m
en a
nd
wom
en.
For
ex
ampl
e,
for
age
belo
w
65
year
s,
card
iova
scul
ar
mor
talit
y w
as
asso
ciat
ed
with
T
SP
in
m
en
but
not
in
wom
en.
Tot
al
mor
talit
y pe
rcen
t ex
cess
de
aths
per
10
0 µ
g/m
3 inc
reas
e in
T
SP
at l
ag 2
wa
s 3.
4% (
0.5
, 6.4
).
FRA
NC
E
Zeg
hnou
n et
al
. (2
001)
(Z
eghn
oun
et a
l, 20
01)
Rou
en
and
Le
Hav
re,
Fra
nce.
199
0-19
95.
PM
13
mea
n:
32.9
fo
r R
ouen
, 36
.4
for
Le
Hav
re.
BS
m
ean:
18
.7
for
Rou
en,
16.3
fo
r Le
H
avre
.
Tot
al,
card
iova
scul
ar,
and
resp
irato
ry m
orta
lity
serie
s w
ere
regr
esse
d on
BS
, P
M13
, S
O2,
NO
2,
and
O3
in 1
- an
d 2-
pollu
tant
mod
els
usin
g G
AM
P
oiss
on m
odel
s ad
just
ing
for
seas
onal
tre
nds,
da
y of
wee
k, a
nd w
eath
er.
In
Rou
en,
O3,
SO
2,
and
NO
2 w
ere
each
si
gnifi
cant
ly
asso
ciat
ed
with
to
tal,
resp
irato
ry,
and
card
iova
scul
ar
mor
talit
y,
resp
ectiv
ely.
In
Le
H
avre
, S
O2
and
PM
13
wer
e as
soci
ated
w
ith
card
iova
scul
ar
mor
talit
y.
Ho
we
ver,
th
e la
ck
of
stat
istic
al
sign
ifica
nce
re
port
ed
for
mos
t of
th
ese
re
sults
may
be
in p
art
due
to t
he r
elat
ive
ly
smal
l po
pula
tion
size
of
th
ese
citie
s (4
30,0
00 a
nd 2
60,0
00, r
espe
ctiv
ely)
.
PM
13 t
otal
mor
talit
y R
Rs
per
IQR
w
ere
0.5
%
(-1.
1,
2.1)
in
R
ouen
(I
QR
=20
.6,
lag
1) a
nd 1
.9%
(-0
.8,
7.4)
in L
e H
avre
(IQ
R=
23.9
, 1-
day
lag
). B
S t
ota
l m
orta
lity
RR
s pe
r IQ
R
wer
e 0.
5%
(-1.
8,
2.9)
in
R
ouen
(IQ
R=
14.2
, 1-
day
lag)
and
0.
3%
(-1.
6,
2.2)
in
Le
H
avre
(I
QR
=11
.5, l
ag
0-1)
.
report no. 4/11
109
Le T
ertr
e et
al.,
200
2 (L
e T
ertr
e et
al,
2002
b)
Nin
e F
renc
h C
ities
, 19
90-1
995.
B
S
med
ian
ra
ngin
g be
twee
n
12.7
-16.
0 μg
/m3 (
base
d on
5 c
ities
, on
ly).
Thi
s st
udy
exa
min
ed t
he s
hort
-ter
m e
ffect
s of
air
pollu
tion
(BS
, S
O2,
NO
2, O
3) o
n m
orta
lity,
usi
ng
Poi
sson
tim
e-se
ries
regr
essi
ons,
con
trol
ling
for
tren
ds
in
seas
ons,
ca
lend
ar
effe
cts,
in
fluen
za
epid
emic
s, t
em
pera
ture
, an
d hu
mid
ity.
Lag
0-1
w
as
used
for
all t
he c
ities
.
Sig
nific
ant
an
d po
sitiv
e as
soci
atio
ns
we
re
foun
d be
twee
n to
tal
daily
de
aths
and
the
fo
ur
air
pollu
tion
indi
cato
rs
cons
ider
ed.
Sim
ilar
resu
lts
wer
e
obta
ined
fo
r ca
rdio
vasc
ular
m
orta
lity.
E
xcep
t fo
r S
O2,
po
sitiv
e bu
t no
n si
gnifi
cant
as
soci
atio
ns
wer
e fo
und
with
res
pira
tory
mor
talit
y.
The
res
ults
wer
e co
nsis
tent
am
ong
all
the
citie
s co
nsid
ered
.
A 5
0 μg
/m3 in
crea
se in
BS
(la
g 0-
1) w
as a
ssoc
iate
d w
ith 2
.9%
(1.
3,
4.4)
in
crea
se
in
tota
l m
ort
ality
, 3.
1% (
0.5,
5.5
) in
car
diov
ascu
lar
mor
talit
y, a
nd 2
.7%
(-2
.6,
8.3)
in
resp
irato
ry
mor
talit
y.
Cor
resp
ondi
ng
estim
ates
af
ter
excl
usio
n of
Par
is w
ere
4.6
% (
1.9,
7.
4),
5.5%
(0.
9, 1
0.4)
and
11.
1%
(1.3
, 21.
9), r
espe
ctiv
ely.
E
ilste
in
et
al.,
2004
(E
ilste
in e
t al,
2004
) N
ine
Fre
nch
Citi
es,
1990
-199
7.
Upd
ate
of
the
abov
e st
udy.
B
S
and
PM
10
leve
l in
dica
tors
not
giv
en.
BS
da
ta w
ere
ava
ilabl
e fo
r 6
ci
ties
and
P
M10
fo
r 3
ci
ties.
Thi
s st
udy
exa
min
ed t
he s
hort
-ter
m e
ffect
s of
air
pollu
tion
(PM
10,
BS
, S
O2,
NO
2, C
O,
O3)
on
tota
l, ca
rdio
vasc
ular
and
res
pira
tory
mor
talit
y, u
sing
P
oiss
on G
AM
, ad
just
ing
for
tren
ds i
n se
aso
ns,
day
of
wee
k,
holid
ays,
in
fluen
za
epid
em
ics,
te
mpe
ratu
re,
and
hum
idity
. E
xces
s ris
ks a
t la
g 0-
1 an
d la
g 0-
5 w
ere
sho
wn.
Sig
nific
ant
di
rect
as
soci
atio
ns
wer
e fo
und
for
all
the
air
pollu
tion
ind
exes
oth
er t
han
C
O.
The
per
cent
exc
ess
ris
ks f
or
a 10
μg
/m3 in
crea
se in
BS
wer
e 0.
8 fo
r to
tal
mor
talit
y,
0.5
fo
r ca
rdio
vasc
ular
an
d 0.
7 (n
on-
sign
ifica
nt,
NS
) fo
r re
spira
tory
m
orta
lity
at
lag
0-
1.
Cor
resp
ondi
ng
estim
ates
at
lag
0-5
wer
e 1.
2,
1.2
and
2.1,
re
spec
tivel
y. T
he p
erce
nt e
xce
ss
risks
for
a 1
0 μg
/m3 i
ncre
ase
in
PM
10 w
ere
0.8
for
tota
l, 0.
3 (N
S)
for
card
iova
scul
ar
and
0.6
(NS
) fo
r re
spira
tory
mor
talit
y at
lag
0-1.
C
orre
spon
din
g es
timat
es a
t la
g 0-
5 w
ere
1.0,
0.
3 (N
S)
and
1.
9,
resp
ectiv
ely.
report no. 4/11
110
ITA
LY
Z
anob
etti
et
al.
(200
0)
(Zan
obet
ti et
al,
2000
) M
ilan,
Ital
y. 1
980
-198
9.
TS
P m
ean:
14
2.
The
foc
us o
f th
is s
tudy
was
to
quan
tify
mo
rtal
ity
disp
lace
men
t us
ing
wha
t th
ey
term
ed
“GA
M
dist
ribut
ed
lag
mod
els”
(s
moo
thin
g te
rm
was
fit
ted
with
pe
naliz
ed
splin
es).
N
on-a
ccid
enta
l to
tal
deat
hs w
ere
regr
esse
d on
sm
ooth
fu
nctio
n of
TS
P d
istr
ibut
ed o
ver
the
sam
e da
y an
d th
e pr
evio
us 4
5 da
ys u
sing
pen
aliz
ed s
plin
es f
or t
he
smoo
th
term
s an
d se
ason
al
cycl
es,
tem
pera
ture
, hu
mid
ity,
day-
of-
wee
k, h
olid
ays,
an
d in
fluen
za
epid
emic
s.
The
m
orta
lity
disp
lace
men
t w
as
mod
elle
d as
th
e
initi
al
posi
tive
incr
eas
e,
nega
tive
re
boun
d (d
ue
to
depl
etio
n),
follo
wed
by
an
othe
r po
sitiv
e co
effic
ient
pe
riod,
an
d th
e su
m
of
the
thre
e ph
ases
wer
e co
nsid
ered
as
the
tota
l cu
mul
ativ
e
effe
ct.
TS
P w
as p
ositi
vely
ass
ocia
ted
with
mor
talit
y up
to
13
da
ys,
follo
wed
by
ne
arly
ze
ro
coef
ficie
nts
bet
we
en 1
4 an
d 20
day
s, a
nd
then
fo
llow
ed
by
smal
ler
but
posi
tive
coef
ficie
nts
up
to t
he 4
5th
day
(max
imum
ex
amin
ed)
. T
he s
um o
f th
ese
coef
ficie
nts
w
as
over
thr
ee t
imes
larg
er t
han
that
for
the
si
ngle
-day
est
imat
e.
Tot
al
mor
talit
y pe
rcen
t in
crea
se
estim
ates
pe
r IQ
R
incr
ease
in
T
SP
: 2.
2%
(1.
4, 3
.1)
for
sing
le-
day
mod
el;
6.7%
(3
.8,
9.6)
fo
r di
strib
uted
lag
mod
el.
Ros
si
et
al.
(199
9)
(Ros
si e
t al,
199
9)
Mila
n, It
aly,
19
80-1
989
T
SP
(P
M13
be
ta
atte
nuat
ion)
: 142
.
Spe
cific
ca
uses
of
de
ath
(res
pira
tory
, re
spira
tory
in
fect
ions
, C
OP
D,
circ
ulat
ory,
ca
rdia
c, h
eart
fai
lure
, an
d A
MI)
wer
e re
late
d to
T
SP
, S
O2,
a
nd
NO
2,
adju
stin
g fo
r se
ason
al
cycl
es,
tem
per
atur
e,
and
hum
idity
, us
ing
P
oiss
on G
AM
mod
el.
All
thre
e po
lluta
nts
wer
e as
soci
ated
with
all
caus
e m
orta
lity.
C
ause
-spe
cific
an
alys
is
wa
s co
nduc
ted
for
TS
P
only
. R
espi
rato
ry
infe
ctio
n an
d he
art
failu
re d
eath
s w
ere
both
as
soci
ated
with
TS
P o
n th
e co
ncur
rent
day
, w
here
as
the
as
soci
atio
ns
for
myo
card
ial
infa
rctio
n an
d C
OP
D d
eath
s w
ere
foun
d fo
r th
e av
erag
e of
3 to
4 d
ay p
rior
TS
P.
3.3%
(2
.4,
4.3)
pe
r 10
0 µ
g/m
3 T
SP
at l
ag 0
.
Mic
helo
zzi
et
al.
(199
8)
(Mic
helo
zzi e
t al
, 199
8)
Rom
e, I
taly
, 19
92-1
995.
T
SP
(P
M13
be
ta
atte
nuat
ion)
: 84.
Tot
al m
orta
lity
was
rel
ated
to
PM
13,
SO
2, N
O2,
C
O,
and
O3,
us
ing
Poi
sson
G
AM
m
odel
, ad
just
ing
for
seas
onal
cy
cles
, te
mpe
ratu
re,
hum
idity
, da
y-of
-wee
k, a
nd h
olid
ay.
Ana
lysi
s of
m
orta
lity
by p
lace
of
resi
denc
e, s
easo
n, a
ge,
plac
e of
dea
th (
in o
r ou
t of
hos
pita
l), a
nd c
ause
w
as
also
con
duc
ted.
PM
13
an
d
NO
2 w
ere
mo
st
cons
iste
ntly
as
soci
ated
w
ith
mor
talit
y.
CO
an
d O
3 co
effic
ient
s w
ere
posi
tive,
SO
2 co
effic
ient
s ne
gativ
e.
RR
es
timat
es
high
er
in
the
war
mer
sea
son.
RR
s si
mila
r fo
r in
- an
d ou
t-of
hos
pita
l de
aths
.
1.9%
(0.
5, 3
.4)
per
50 µ
g/m
3 PM
13
at la
g 0.
report no. 4/11
111
Big
geri
et
al.
(200
5)
(Big
geri
et a
l, 20
05)
Eig
ht
citie
s of
N
orth
, C
entr
e an
d S
out
h of
Ital
y 19
90-1
999.
P
M10
m
ean
daily
co
ncen
trat
ion:
T
urin
, 91-
94: 7
7.6
Tur
in, 9
5-98
: 63.
8 M
ilan,
90-
94: 6
1.8
Mila
n, 9
5-97
: 45.
2 V
eron
a, 9
5-99
: 36.
5 R
aven
na, 9
1-95
: 59.
1 B
olog
na, 9
6-9
8: 4
1.2
Flo
renc
e, 9
6-98
: 40.
3 R
ome,
92-
94: 6
9.7
Pal
erm
o, 9
7-99
: 42.
9.
Met
a-an
alys
is
of
shor
t-te
rm
effe
cts
of
air
pollu
tion
on m
orta
lity
(all
caus
es,
card
iova
scul
ar
caus
es a
nd r
espi
rato
ry c
ause
s) i
n ei
ght
Italia
n ci
ties
from
199
0 to
199
9. D
aily
con
cent
ratio
ns o
f po
lluta
nts
wer
e co
llect
ed.
A g
ener
aliz
ed l
inea
r m
odel
ad
just
ed
for
age,
da
y of
th
e
wee
k,
holid
ays,
in
fluen
za
epid
em
ics,
m
eteo
rolo
gica
l va
riabl
es,
and
sea
sona
lity
pat
tern
wa
s fit
ted
for
PM
10 a
nd o
the
r co
-pol
luta
nt (
SO
2, N
O2,
CO
, an
d O
3).
City
-spe
cific
and
fix
ed-
and
rand
om-e
ffect
s po
oled
es
timat
es
wer
e sh
own.
La
g 0-
1 w
as
used
fo
r m
orta
lity.
F
ixed
- an
d ra
ndom
-eff
ects
m
odel
s w
ere
show
n.
A s
igni
fican
t di
rect
ass
ocia
tion
was
fou
nd
betw
een
PM
10
and
tota
l n
atur
al
mor
talit
y.
The
effe
ct o
f P
M10
was
gre
ater
dur
ing
the
war
m s
easo
n a
nd f
or t
he e
lder
ly.
A N
orth
-S
outh
gra
dien
t in
ris
k w
as o
bser
ved
for
tota
l na
tura
l m
orta
lity.
A
si
gnifi
cant
as
soci
atio
n w
as
also
sh
own
fo
r ca
rdio
vasc
ular
a
nd
resp
irato
ry m
orta
lity.
Sim
ilar
incr
ease
d ris
ks
wer
e fo
und
fo
r th
e ot
her
cons
ider
ed
co-
pollu
tant
s. T
he
effe
cts
of P
M10
sho
wed
the
hi
ghes
t int
er-c
ity v
aria
bilit
y.
Per
cent
ex
cess
m
orta
lity
for
10
µg/
m3 c
hang
e in
PM
10 (
lag
0-1)
: A
ll na
tura
l cau
ses
Fix
ed-e
ffect
s: 0
.85%
(0.
52,
1.18
) R
ando
m-e
ffect
s:
0.98
%
(0.3
5,
1.61
) C
VD
cau
ses
Fix
ed-e
ffect
s: 0
.97%
(0.
45,
1.50
) R
ando
m-e
ffect
s:
1.21
%
(0.3
2,
2.10
) R
espi
rato
ry c
aus
es
Fix
ed-e
ffect
s: 1
.74%
(0.
44,
3.05
) R
ando
m-e
ffect
s:
1.41
%
(-1.
41,
4.32
).
Sta
fogg
ia
et
al.,
2005
(S
tafo
ggia
et
al, 2
005)
R
ome,
Ita
ly,
1998
-200
0 P
M10
mea
n: 5
2.1;
P
NC
m
ean
51,6
69 n
o/cm
3 In
pa
rt,
rean
alys
is
of
abov
e st
udy
as
time
serie
s.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
daily
am
bien
t ai
r po
llutio
n
leve
ls
(PN
C,
as
a
prox
y of
ultr
afin
e pa
rtic
les,
PM
10,
CO
, N
O2
and
O3)
and
the
occ
urre
nce
of f
atal
, ou
t-of
-hos
pita
l co
rona
ry
even
ts
(5,1
44
subj
ects
) an
d fa
tal
hosp
italis
ed c
oron
ary
even
ts (
1,41
1 su
bjec
ts),
us
ing
a G
AM
P
oiss
on
mod
el
adju
stin
g fo
r te
mpo
ral
tren
d,
tem
pera
ture
, hu
mid
ity,
day
of
wee
k,
holid
ays
, an
d po
pula
tion
redu
ctio
n in
su
mm
er m
onth
s. L
ags
0, 1
, 0-
1 an
d 0-
4 w
ere
cons
ider
ed.
Sig
nific
ant
di
rect
as
soci
atio
ns
wer
e fo
und
for
PN
C,
PM
10 a
nd C
O,
and
bot
h ou
t-of
-ho
spita
l de
ath
s an
d fa
tal
coro
nary
eve
nts.
T
he a
ssoc
iatio
ns w
ere
stro
nger
in
subj
ects
ol
der
than
65
year
s.
Incr
ease
in
risk
of o
ut-o
f-h
ospi
tal
coro
nary
de
ath
as
soci
ated
w
ith
29.7
μg/
m3 (
IQR
) P
M10
: 7.
0% (
2.9,
11.
2) a
t lag
0
6.4%
(2.
4, 1
0.6)
at l
ag 1
8.
1% (
3.6,
12.
9) a
t lag
0-1
9.
6% (
4.4,
15.
2) a
t lag
0-4
In
crea
se i
n ris
k fo
r an
inc
rem
ent
in P
NC
of 2
8,00
1 no
/cm
3 (IQ
R):
8.
1% (
3.0,
13.
4) a
t lag
0
4.0%
(-1
.1, 9
.3)
at la
g 1
7.8%
(2.
0, 1
4.0)
at l
ag 0
-1
8.3%
(2.
1, 1
4.9)
at l
ag 0
-4
The
in
crea
ses
in
risk
of
fata
l co
rona
ry
eve
nts
at
lag
0
w
ere
5.0%
(1.
3, 8
.9)
for
PM
10 a
nd 8
.2%
(3
.6, 1
3.1)
for
PN
C.
report no. 4/11
112
For
astie
re
et
al.,
2005
(F
oras
tiere
et a
l, 20
05)
Rom
e, I
taly
, 19
98-2
001,
P
M10
mea
n 51
.0.
Thi
s st
udy
inve
stig
ated
whe
ther
soc
ial
cla
ss i
s an
effe
ct m
odifi
er o
f th
e as
soci
atio
n b
etw
een
P
M10
and
dai
ly m
orta
lity.
Sub
ject
s in
clud
ed
wer
e 83
,253
res
iden
ts o
f R
ome
aged
≥35
yea
rs w
ho
died
for
natu
ral c
ause
s be
twee
n 19
98 a
nd 2
001.
A
ca
se-c
ross
over
de
sign
w
as
used
. C
ontr
ol
days
w
ere
sele
cted
us
ing
th
e tim
e-st
ratif
ied
appr
oach
: al
l sa
me
days
of
the
wee
k w
ithin
tha
t m
onth
(of
the
sam
e ye
ar)
wer
e ch
osen
. La
g 0–
1 w
as
cons
ider
ed.
Con
ditio
nal
log
istic
reg
ress
ion
anal
ysis
was
use
d af
ter
allo
wa
nce
for
appa
rent
te
mpe
ratu
re o
n th
e in
dex
day,
and
its
diffe
renc
e fr
om t
he a
vera
ge o
f th
e pr
evio
us 3
day
s’ v
alue
s,
influ
enza
ep
idem
ics,
ho
liday
s,
and
pop
ulat
ion
re
duct
ion
in s
umm
er m
onth
s. A
s a
surr
ogat
e fo
r in
divi
dual
soc
ioec
onom
ic s
tatu
s (S
ES
) tw
o ar
ea-
base
d in
dica
tors
wer
e us
ed:
inco
me
and
SE
S o
f th
e po
pula
tion
liv
ing
in t
he s
ame
cens
us b
lock
, as
der
ived
from
the
2001
cen
sus.
A
sign
ifica
nt
exce
ss
risk
was
fo
und
fo
r P
M10
. S
ubje
cts
in t
he lo
w,
mid
-low
, m
id-h
igh
cate
gorie
s of
bo
th
inco
me
and
soci
o-ec
onom
ic s
tatu
s sh
ow
ed a
n ef
fect
of
PM
10
very
co
nsis
tent
w
ith
the
over
all
estim
ate
whe
reas
th
ose
in
the
hig
h ca
tego
ry
had
ef
fect
est
imat
es c
lose
to th
e n
ull.
The
in
crea
se
in
mor
talit
y as
soci
ated
w
ith a
n in
crem
ent
of
10 μ
g/m
3 of
PM
10 (
lag
0-1
) w
as
1.1%
(0.
7, 1
.6).
T
here
w
as
a st
atis
tical
ly
sign
ifica
nt
effe
ct
mod
ifica
tion
of
soci
al
clas
s on
a
mul
tiplic
ativ
e sc
ale.
For
the
low
est
and
high
est
quin
tile
of
SE
S
the
incr
ease
s w
ere
1.9
%
and
0.0%
fo
r th
e hi
ghes
t qu
intil
e.
Cor
resp
ondi
ng
valu
es f
or t
he l
owes
t an
d hi
ghes
t qu
intil
es
of
inco
me
wer
e
1.4%
an
d 0.
1%.
San
seba
stia
no
et
al.
(200
3)
(San
seba
stia
no
et
al,
2003
) P
arm
a, It
aly,
199
2-20
01.
mea
n da
ily
conc
entr
atio
n:
PM
10
(199
8-20
01:
from
39
,63
in 1
998
to 5
7.68
in 2
001;
T
SP
: fr
om 4
6.92
in
1996
to
62.
39 in
19
92.
Thi
s st
udy
exam
ined
the
rel
atio
nshi
p be
twee
n T
SP
an
d ot
her
co
-pol
luta
nts
(CO
, N
O2,
an
d S
O2)
, an
d ge
ner
al m
orta
lity
in t
he c
ity o
f P
arm
a (p
opul
atio
n 17
0,00
0).
The
ex
cess
ris
k of
ho
spita
l ad
mis
sion
was
no
t an
alys
ed f
or P
M10
si
nce
info
rmat
ion
for
PM
10 w
as
avai
labl
e fo
r th
e pe
riod
1998
-200
1, o
nly.
A
dire
ct
ass
ocia
tion
has
been
sh
ow
n
betw
een
TS
P,
SO
2 an
d ge
ner
al m
orta
lity.
P
erce
nt e
xce
ss r
isk
(95
% C
I) o
f to
tal
mor
talit
y fo
r 1
µg/
m3
incr
emen
t of
T
SP
: 0.
14%
(0
.08,
0.
23).
SPA
IN
T
obia
s an
d
Cam
pbel
l (1
999)
(T
obia
s an
dCam
pbel
l, 19
99)
Bar
celo
na,
Spa
in.
1991
-19
95.
BS
(no
dat
a di
strib
utio
n
wa
s re
port
ed).
Thi
s st
udy
exa
min
ed t
he s
ens
itivi
ty o
f es
timat
ed
tota
l m
orta
lity
effe
cts
of
BS
to
d
iffer
ent
appr
oach
es
to
mod
elin
g in
fluen
za
epid
emic
s:
(1)
with
a s
ingl
e du
mm
y va
riabl
e; (
2) w
ith t
hree
du
mm
y va
riabl
es;
(3)
usin
g da
ily
num
ber
of
case
s of
in
fluen
za.
Poi
sson
G
LM
regr
essi
on
used
to
mod
el t
otal
dai
ly m
ort
ality
, ad
just
ing
for
wea
ther
, lo
ng-t
erm
tr
end,
an
d se
ason
, ap
pare
ntly
follo
win
g A
PH
EA
pro
toco
l.
Usi
ng
the
re
port
ed
daily
nu
mbe
r of
in
fluen
za c
ase
s re
sulte
d in
a b
ette
r fit
(i.e
., a
low
er
AIC
) th
an
thos
e us
ing
dum
my
varia
bles
. In
th
e “b
ette
r”
mod
el,
the
BS
co
effic
ient
w
as
abou
t 10
%
smal
ler
than
th
ose
in t
he m
odel
s w
ith d
umm
y in
fluen
za
varia
bles
, bu
t re
mai
ned
sign
ifica
nt.
Lags
not
re
port
ed.
Tot
al m
orta
lity
exce
ss d
eath
s pe
r 25
µ
g/m
3 in
crea
se
in
BS
: 1.
37
(0.2
0, 2
.56)
for
mod
el u
sing
the
da
ily
case
of
in
fluen
za;
1.71
(0
.53,
2.9
1) f
or m
odel
with
thr
ee
influ
enza
dum
my
varia
bles
.
report no. 4/11
113
Alb
erdi
O
drio
zola
et
al
. (1
998)
(A
lber
di
Odr
iozo
la e
t al,
1998
) M
adrid
, S
pain
, 19
86-
1992
. T
SP
(b
eta
atte
nuat
ion)
: 47
fo
r av
erag
e of
2
st
atio
ns.
Tot
al,
resp
irato
ry,
and
card
iova
scul
ar
dea
ths
wer
e re
late
d
to
TS
P
and
SO
2.
Mul
tivar
iate
au
tore
gres
sive
in
tegr
ated
m
ovin
g av
erag
e m
odel
s us
ed t
o ad
just
for
sea
son,
tem
pera
ture
, re
lativ
e hu
mid
ity, a
nd in
fluen
za e
pide
mic
s.
TS
P
(lag
1)
and
SO
2 (la
g 3)
w
ere
inde
pend
ently
ass
ocia
ted
with
mor
talit
y.
4.8%
(1
.8,
7.7)
pe
r 10
0 µ
g/m
3 T
SP
at l
ag 1
.
Día
z et
al.
(199
9) (
Dia
z et
al,
1999
) M
adrid
, S
pain
, 19
90-
1992
. T
SP
(no
dat
a di
strib
utio
n
wa
s re
port
ed).
Non
-acc
iden
tal,
resp
irato
ry,
and
card
iova
scul
ar
deat
hs (
mea
n =
62.
4, 6
.3,
and
23.8
pe
r da
y,
resp
ectiv
ely)
. A
uto-
Reg
ress
ive
Inte
grat
ed
Mov
ing
Ave
rage
(A
RIM
A)
mod
els
fit
to
both
de
pend
ent
and
inde
pend
ent
varia
bles
fir
st
to
rem
ove
auto
-cor
rela
tion
and
seas
onal
ity
(i.e.
, pr
e-w
hite
ning
),
follo
wed
by
ex
amin
ing
cros
s-co
rrel
atio
n to
fin
d op
timal
lags
. M
ultiv
aria
te O
LS
mod
els
thus
in
clud
ed
AR
IMA
co
mpo
nent
s,
seas
onal
cy
cles
(s
ine/
cosi
ne),
V
-sha
ped
tem
pera
ture
, an
d op
timal
la
gs
foun
d fo
r po
llutio
n an
d w
eath
er v
aria
bles
. T
SP
, S
O2,
NO
2,
and
O3
wer
e ex
amin
ed.
S
easo
n-sp
ecifi
c an
alys
es w
ere
als
o co
nduc
ted.
TS
P w
as s
ign
ifica
ntly
ass
ocia
ted
with
non
-ac
cide
ntal
mo
rtal
ity a
t la
g 0
for
yea
r ar
ound
an
d w
inte
r, b
ut w
ith a
lag
1 i
n su
mm
er.
A
sim
ilar
patte
rn
wa
s se
en
for
circ
ulat
ory
deat
hs.
For
re
spira
tory
m
orta
lity,
a
sign
ifica
nt a
ssoc
iatio
n w
ith T
SP
was
fou
nd
only
in
sum
mer
(la
g 0)
. S
O2,
NO
x, a
nd
NO
2 sh
ow
ed
sim
ilar
asso
ciat
ions
w
ith
non-
acci
dent
al
dea
ths
at
lag
0 da
y.
O3
asso
ciat
ions
w
ith
non-
acci
dent
al
mor
talit
y w
as
U-s
hape
d, w
ith i
ncon
sist
ent
lags
(1,
4,
and
10).
For
no
n-ac
cide
ntal
m
ort
ality
, ex
cess
de
ath
s w
as
7.4%
(c
onfid
ence
ban
ds n
ot r
epor
ted;
p
< 0
.05)
per
100
µg/
m3 T
SP
at
lag
0.
Stud
ies
in s
ubje
cts
with
CO
PD/a
sthm
a
G
arci
a-A
ymer
ich
et
al.
(200
0) (
Gar
cia-
Aym
eric
h
et a
l, 20
00)
Bar
celo
na,
Spa
in.
1985
-19
89.
BS
(no
dat
a di
strib
utio
n
wa
s re
port
ed).
Dai
ly t
otal
(m
ean
= 1
.8/d
ay)
, re
spira
tory
, an
d ca
rdio
vasc
ular
m
orta
lity
coun
ts
of
a co
hort
(9
,987
su
bjec
ts)
with
C
OP
D
or
asth
ma
w
ere
asso
ciat
ed w
ith B
S (
24-h
r),
SO
2 (2
4- h
r an
d 1-
hr
max
), N
O2
(24-
hr a
nd 1
-hr
max
), O
3 (1
-hr
ma
x),
tem
pera
ture
, an
d re
lativ
e hu
mid
ity.
Poi
sson
G
LM r
egre
ssio
n m
odel
s us
ing
AP
HE
A p
roto
col
wer
e u
sed.
The
res
ultin
g R
Rs
wer
e co
mpa
red
with
thos
e of
the
gen
eral
pop
ulat
ion.
Dai
ly
mor
talit
y in
C
OP
D
patie
nts
was
as
soci
ated
with
all
six
pollu
tion
indi
ces.
Th
is
asso
ciat
ion
was
str
onge
r th
an in
the
gen
era
l po
pula
tion
only
for
dai
ly 1
-hr
max
of
SO
2,
daily
1-h
r m
ax a
nd d
aily
mea
ns o
f N
O2.
BS
an
d da
ily m
eans
of
SO
2 sh
owed
sim
ilar
or
wea
ker
asso
ciat
ions
fo
r C
OP
D
patie
nts
th
an fo
r th
e ge
nera
l pop
ulat
ion.
Tot
al
mor
talit
y pe
rcen
t in
crea
se
per
25 µ
g/m
3 inc
reas
e of
BS
at
lag
0-3:
2.
76%
(1
.31,
4.
23)
in
ge
nera
l po
pula
tion,
and
1.1
4% (
-4.
4, 6
.98)
in th
e C
OP
D c
oho
rt.
report no. 4/11
114
Sun
yer
et
al.
(200
0)
(Sun
yer
et a
l, 20
00)
Bar
celo
na,
Spa
in.
1990
-19
95.
BS
m
eans
: 43
.9
for
case
per
iod,
and
43.
1 fo
r co
ntro
l per
iod.
Tho
se
over
ag
e 35
w
ho
soug
ht
emer
genc
y ro
om
serv
ices
fo
r C
OP
D
exac
erba
tion
durin
g 19
85-1
989
and
died
du
ring
1990
-199
5 w
ere
incl
uded
in
an
alys
is.
Tot
al,
resp
irato
ry,
and
card
iova
scul
ar
deat
hs
wer
e an
alyz
ed
usin
g a
cond
ition
al
logi
stic
re
gres
sion
an
alys
is
with
a
case
-cro
ssov
er
desi
gn,
adju
stin
g fo
r te
mpe
ratu
re,
rela
tive
hum
idity
, an
d in
fluen
za
epid
emic
s.
Bi-d
irect
iona
l co
ntro
l pe
riod
at
7 da
ys w
as u
sed.
Lag
0-2
was
use
d fo
r po
llutio
n ex
posu
re p
erio
d. D
ata
also
str
atifi
ed b
y po
tent
ial
effe
ct m
odifi
ers
(e.g
., ag
e, g
ende
r, s
ever
ity a
nd
num
ber
of e
mer
genc
y ro
om v
isits
, etc
.).
BS
lev
els
wer
e as
soci
ated
with
all
caus
e de
aths
. T
he
asso
ciat
ion
was
st
rong
er
for
resp
irato
ry c
ause
s. O
lder
wom
en,
patie
nts
adm
itted
to
in
tens
ive
care
un
its,
and
patie
nts
with
a
high
er
rate
of
em
erge
ncy
room
vis
its w
ere
at g
reat
er r
isk
of d
eath
s as
soci
ated
with
BS
.
Per
cent
in
crea
se
per
25
µg/
m3
incr
ease
in
3-
day
aver
age
BS
: 14
.2%
(1.
6, 2
8.4)
for
all
caus
es;
9.7%
(-
10.2
, 34
.1)
for
card
iova
scul
ar
deat
hs;
23.2
%
(3.0
, 47.
4) fo
r re
spira
tory
dea
ths.
Sun
yer
and
Bas
agan
a (2
001)
(S
unye
r an
d B
asag
ana,
200
1)
Bar
celo
na,
Spa
in.
1990
- 19
95.
See
Sun
yer
et a
l. (2
000)
(S
unye
r et
al,
2000
) fo
r P
M le
vels
.
The
an
alys
is
asse
ssed
an
y “in
depe
nden
t”
part
icle
ef
fect
s,
afte
r co
ntro
lling
fo
r ga
seou
s po
lluta
nts,
on
a co
hort
of
patie
nts
with
CO
PD
(s
ee t
he s
umm
ary
desc
riptio
n fo
r S
unye
r et
al.
(200
0)
(Sun
yer
et
al,
2000
) fo
r an
alyt
ical
ap
proa
ch).
P
M10
, N
O2,
O
3,
and
CO
w
ere
anal
yzed
.
PM
10,
but
not
gase
ous
pollu
tant
s,
was
as
soci
ated
with
mor
talit
y fo
r al
l ca
uses
. In
th
e tw
o-po
lluta
nt m
odel
s, t
he P
M10
-mor
talit
y as
soci
atio
ns w
ere
not
dim
inis
hed,
whe
reas
th
ose
with
gas
eous
pol
luta
nts
wer
e.
Per
cent
exc
ess
risk
for
all
caus
e m
orta
lity
per
27 µ
g/m
3 (IQ
R)
PM
10
on
the
sam
e-da
y w
as
11%
(0
, 24
). I
n tw
o po
lluta
nt m
odel
s, t
he
PM
10
exce
ss
risks
w
ere
10.5
%,
12.9
%,
and
10.8
% w
ith N
O2,
O3,
an
d C
O, r
espe
ctiv
ely.
report no. 4/11
115
TAB
LE II
– S
HO
RT-
TER
M P
AR
TIC
ULA
TE M
ATT
ER E
XPO
SUR
E A
ND
CA
RD
IOVA
SCU
LAR
HO
SPIT
AL
AD
MIS
SIO
NS
(Der
ived
from
EPA
TA
BLE
8B
-1).
Ref
eren
ce, L
ocat
ion,
Ye
ars,
PM
Inde
x, M
ean
or M
edia
n (µ
g/m
3 ).
Stud
y D
escr
iptio
n. M
ode
ling
met
hods
: lag
s,
smoo
thin
g, c
o-po
lluta
nts
and
cova
riate
s.
Res
ults
and
Com
men
ts.
PM In
dex,
lag,
per
cent
exc
ess
risk%
(95%
LC
L, U
CL)
APH
EA2
Le
T
ertr
e et
al
. (2
002a
) (L
e T
ertr
e et
al,
2002
a)
Eig
ht-C
ity -
AP
HE
A 2
S
tudy
mea
n (S
D)
PM
10
Bar
celo
na -
1/9
4-12
/96
55.7
(18
.4)
Birm
ingh
am -
3/9
2-12
/94
24.8
(13
.1)
Lond
on -
1/9
2-12
/94
28.4
(12
.3)
Mila
n -
No
PM
10
Net
herla
nds
- 1/
92-9
/95
39.5
(19
.9)
Par
is -
1/9
2-9/
96
PM
13 -
22.
7 (1
0.8)
R
ome
- N
o P
M10
S
tock
holm
- 3
/94-
12/9
6 15
.5 (
7.2)
.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
PM
10
and
othe
r co
-pol
luta
nts,
an
d ho
spita
l ad
mis
sion
s fo
r ca
rdia
c ca
uses
in
ei
ght
Eur
opea
n ci
ties
with
a c
om
bine
d po
pula
tion
of
38 m
illio
n. A
utho
rs e
xam
ined
age
fac
tors
and
IH
D
and
st
udie
s al
so
stra
tifie
d by
ag
e us
ing
auto
regr
essi
ve
Poi
sson
m
odel
s co
ntro
lled
for
long
-ter
m t
rend
s, s
easo
n, i
nflu
enza
, ep
ide
mic
s,
and
met
eoro
logy
, as
w
ell
as
conf
ound
ing
by
othe
r po
lluta
nts.
A
sign
ifica
nt
effe
ct
of
PM
10
and
BS
on
ad
mis
sion
s fo
r ca
rdia
c ca
uses
(a
ll ag
es)
an
d ca
rdia
c ca
uses
and
IH
D f
or p
eopl
e o
ver
65 y
ears
was
fou
nd.
The
impa
ct o
f P
M1
0 p
er
unit
of p
ollu
tion
was
hal
f th
at f
ound
in
the
Uni
ted
Sta
tes.
P
M10
di
d n
ot
seem
to
be
co
nfou
nded
by
O3
or
SO
2. T
he e
ffect
was
re
duce
d w
hen
CO
was
inc
orp
orat
ed i
n th
e re
gres
sion
m
odel
an
d el
imin
ated
w
hen
cont
rolli
ng
for
NO
2.
The
re
was
lit
tle
evid
ence
of
an
im
pact
of
pa
rtic
les
on
hosp
ital
adm
issi
ons
for
IHD
fo
r pe
opl
e be
low
65
year
s or
str
oke
for
peop
le o
ver
65
year
s.
The
au
thor
s st
ate
re
sults
w
ere
cons
iste
nt
with
a
role
fo
r tr
affic
ex
hau
st/d
iese
l in
Eur
ope.
For
a 1
0 µ
g/m
3 incr
ease
in P
M10
C
ardi
ac a
dmis
sion
s/al
l age
s 0.
5% (
0.2,
0.8
) C
ardi
ac a
dmis
sion
s/ov
er 6
5 ye
ars
0.7%
(0.
4, 1
.0)
IHD
/ove
r 65
yea
rs
0.8%
(0.
3, 1
.2)
For
car
diac
ad
mis
sion
s fo
r p
eopl
e ov
er 6
5 ye
ars:
All
the
city
-sp
ecifi
c es
timat
es
wer
e po
sitiv
e
with
Lo
ndon
, M
ilan,
an
d S
tock
holm
si
gnifi
cant
at t
he 5
% le
vel.
report no. 4/11
116
HEA
PSS
La
nki
et a
l. (2
006)
(La
nki
et
al,
2006
) H
EA
PS
S
Stu
dy:
Fiv
e
Eur
opea
n ci
ties,
199
2-2
000.
P
M10
; P
NC
(1
/cm
3 ) m
edia
n da
ily
conc
entr
atio
n:
Aug
sbur
g,
Ge
rman
y,
95-
99: 4
3.5;
12,
400
Bar
celo
na,
Spa
in,
92-9
5:
57.4
; 76,
300
Hel
sink
i, F
inla
nd,
93-9
9:
21.0
; 13,
600
R
ome,
Ital
y, 9
8-20
00:
48.5
; 46,
000
S
tock
holm
, S
wed
en,
94-
99: 1
2.5;
11,
800.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
mea
sure
s of
dai
ly c
once
ntra
tion
of P
M10
, P
NC
(a
s a
pro
xy
for
ultr
afin
e pa
rtic
les)
, ot
her
co
-po
lluta
nts
(CO
, N
O2,
O
3),
and
hosp
ital
adm
issi
ons
for
first
AM
I in
fiv
e E
urop
ean
citie
s.
AM
I re
gist
ers
(Aug
sbur
g an
d B
arce
lon
a)
or
hosp
ital
disc
harg
e re
gist
ers
(HD
R)
(Hel
sink
i, R
ome
and
Sto
ckho
lm)
wer
e us
ed.
Dur
ing
the
stud
y pe
riod
no
dire
ct
mea
sure
men
t of
P
NC
w
as
avai
lab
le.
City
-spe
cific
m
odel
s fo
r es
timat
ing
PN
C
from
ot
her
pollu
tant
s an
d m
eteo
rolo
gica
l var
iabl
es w
ere
dev
elop
ed f
or t
he
perio
d w
hen
PN
C w
as m
eas
ured
, an
d ap
plie
d th
en
retr
ospe
ctiv
ely
to
pred
ict
daily
P
NC
co
ncen
trat
ion
s du
ring
the
stud
y pe
riod.
G
AM
m
odel
s w
ere
used
afte
r a
step
wis
e se
lect
ion
of
cova
riate
s.
The
st
artin
g m
odel
in
clud
ed
time
tren
d,
appa
rent
te
mpe
ratu
re,
day
of
wee
k,
holid
ay,
popu
latio
n re
duct
ion,
tem
pera
ture
and
ba
rom
etric
pre
ssur
e. L
ag 0
, la
g 1,
lag
2 a
nd la
g 3
wer
e co
nsid
ered
.
Dur
ing
the
stud
y pe
riod,
26,
854
indi
vidu
als
wer
e h
ospi
talis
ed f
or A
MI.
The
num
ber
of
adm
issi
ons
was
hig
her
in t
he c
entr
es w
here
H
DR
s w
ere
used
co
mpa
red
with
ce
ntre
s w
ith A
MI r
egis
ters
. Whe
n co
nsid
erin
g al
l fiv
e ci
ties,
the
sa
me
day
PN
C a
nd P
M10
wer
e no
t si
gnifi
cant
ly
asso
ciat
ed
to
hosp
italis
atio
n.
Lim
iting
an
alys
es
to
citie
s w
ith H
DR
, th
e sa
me
day
PN
C s
how
ed a
n ex
cess
ris
k of
ho
spita
lisat
ion.
N
o sp
ecifi
c pa
ttern
of
risk
was
see
n co
nsid
erin
g la
gs 1
-3.
S
imila
r as
soci
atio
ns
as
for
PN
C
we
re
foun
d fo
r C
O,
but
no e
xces
s ris
k w
as f
ound
fo
r N
O2
and
O3.
In
the
3 ci
ties
with
HD
R,
the
mos
t co
nsis
tent
as
soci
atio
ns
wer
e sh
ow
n am
ong
fata
l ca
ses
aged
<7
5 ye
ars.
Effe
cts
of m
easu
res
of c
once
ntra
tion
of P
M w
ere
m
ore
pron
ounc
ed f
or f
atal
cas
es a
ged
<7
5 ye
ars.
Per
cent
E
xce
ss
Ris
k of
ho
spita
lisat
ion
for
AM
I (95
% C
I):
Poo
led
effe
cts
(fiv
e ci
ties)
co
mpu
ted
for
10 µ
g/m
3 cha
nge
in
PM
10:
Lag
0: 0
.3%
(-0
.5, 1
.1)
Lag
1: 0
.1%
(-1
.0, 1
.1)
Lag
2: 0
.2%
(-0
.6, 1
.0)
Lag
3: 0
.2%
(-0
.9, 1
.3)
citie
s w
ith H
DR
; lag
0:
Ove
rall:
0.3
% (
-0.6
, 1.2
) N
on-f
atal
, A
ge
<75
: -0
.3%
(-1
.4,
0.8)
F
atal
, Age
<7
5: 3
.1%
(0.
0, 6
.4)
Non
-fat
al,
Ag
e 75
+:
1.2%
(-
0.5,
2.
9)
Fat
al, A
ge 7
5+
: 0.9
% (
-1.5
, 3.
4)
Poo
led
effe
cts
(fiv
e ci
ties)
co
mpu
ted
for
10,0
00/c
m3 c
hang
e in
PN
C:
Lag
0: 0
.5%
(-0
.4, 1
.5)
Lag
1: -
0.3%
(-1
.8, 1
.2)
Lag
2: -
0.1%
(-1
.0, 0
.8)
Lag
3: -
0.2%
(-2
.1, 1
.7)
citie
s w
ith H
DR
; lag
0:
Ove
rall:
1.3
% (
0.0,
2.6
) N
on-f
atal
, A
ge
<75
: 0.
1%
(-2.
5,
1.7)
F
atal
, Age
<7
5: 5
.0%
(0.
0, 1
0.1)
N
on-f
atal
, A
ge
75+
: 3.
2%
(0.8
, 5.
6)
Fat
al, A
ge 7
5+
: 1.6
% (
-2.2
, 5.
5).
report no. 4/11
117
von
Klo
t. (
200
5) (
von
Klo
t et
al
, 20
05)
HE
AP
SS
S
tudy
: F
ive
E
urop
ean
citie
s, 1
992-
200
1.
PM
10;
PN
C (
1/c
m3 )
mea
n da
ily c
once
ntra
tion:
A
ugsb
urg,
G
erm
any,
95
-20
00: 4
4.7;
13
,504
B
arce
lona
, S
pain
, 92
-20
00: 5
2.2;
76
,593
H
elsi
nki,
Fin
land
, 93
-20
00: 2
5.3;
14
,283
R
ome,
Ita
ly,
98-2
001:
51
.1;
49,1
95
Sto
ckho
lm,
Sw
eden
, 94
-20
00: 1
4.6;
12
,553
.
22,0
06 s
urvi
vors
of
a fir
st d
iagn
osis
of
AM
I w
ere
recr
uite
d in
5 E
urop
ean
citie
s an
d fo
llow
ed f
or
hosp
ital
read
mis
sion
s.
Dai
ly
conc
entr
atio
n of
P
M10
and
oth
er
co-p
ollu
tant
s (C
O,
NO
2, O
3)
was
m
easu
red
and
PN
C w
as
est
imat
ed a
s a
pro
xy
for
ultr
afin
e p
artic
les.
Sh
ort-
term
effe
cts
of a
ir po
llutio
n on
ho
spita
l re
adm
issi
ons
for
AM
I, an
gina
pec
toris
, an
d ca
rdia
c ca
uses
(m
yoca
rdia
l in
farc
tion,
ang
ina
pect
oris
, d
ysrh
ythm
ia,
or h
eart
fa
ilure
) w
ere
st
udie
d in
ci
ty-s
peci
fic
Poi
sson
re
gres
sion
an
alys
es
with
su
bseq
uent
po
olin
g.
Pot
entia
l co
nfo
unde
rs
wer
e se
lect
ed
thro
ugh
pena
lized
reg
ress
ion
splin
es o
f tr
end
(to
cont
rol
for
long
-ter
m t
rend
s, s
easo
nalit
y, a
nd c
han
ges
in
the
base
line
rate
) an
d m
eteo
rolo
gy
(dai
ly
tem
pera
ture
, re
lativ
e hu
mid
ity,
air
pres
sure
, an
d th
e di
ffere
nce
bet
wee
n cu
rren
t da
y te
mp
erat
ure
and
the
mea
n te
mpe
ratu
re o
f th
e pr
evio
us 3
da
ys),
as
w
ell
as
indi
cato
rs
of
day
of
wee
k,
vaca
tion
perio
ds,
or h
olid
ays.
Tre
nd a
nd o
ne
tem
pera
ture
ter
m r
emai
ned
in t
he m
odel
. La
g 0,
1,
2,
3
wer
e co
nsid
ered
. G
iven
th
e re
sults
ob
tain
ed,
the
ef
fect
s of
la
g 0
air
pollu
tion
conc
entr
atio
ns
wer
e sh
ow
n.
Dur
ing
follo
w-u
p,
6,65
5 ca
rdia
c re
adm
issi
ons
wer
e ob
serv
ed.
C
ardi
ac
read
mis
sion
s in
crea
sed
in
asso
ciat
ion
with
sa
me-
day
conc
entr
atio
ns
of
PM
10
and
P
NC
. N
o a
ssoc
iatio
ns
wer
e fo
und
cons
ider
ing
lags
1
-3.
Effe
cts
of
sim
ilar
stre
ngth
wer
e ob
serv
ed f
or C
O,
NO
2 an
d O
3.
Poo
led
effe
cts
com
pute
d fo
r 10
µ
g/m
3 cha
nge
in P
M10
: A
MI:
2.6%
(-0
.5, 5
.8)
Ang
ina
pect
oris
: 0.8
% (
-1.4
, 3.
2)
Car
diac
re
adm
issi
on:
2.1
%
(0.4
, 3.
9)
Poo
led
effe
cts
com
pute
d fo
r 10
,000
/cm
3 ch
ange
in P
NC
: A
MI:
3.9%
(-0
.2, 8
.2)
Ang
ina
pect
oris
: 2.0
% (
-0.8
, 4.
8)
Car
diac
re
adm
issi
on:
2.6
%
(0.5
, 4.
8).
report no. 4/11
118
Subn
atio
nal s
tudi
es
G
REA
T B
RIT
AIN
Atk
inso
n et
al
. (1
999b
) (A
tkin
son
et a
l, 19
99b)
G
reat
er L
ondo
n, E
ngla
nd
1992
-199
4.
Pol
luta
nt:
me
an,
med
ian,
90
-10
perc
entil
e ra
nge:
P
M10
: 28.
5, 2
4.8
, 30.
7 B
S: 1
2.7,
10.
8, 1
6.1.
Dai
ly e
mer
genc
y ho
spita
l ad
mis
sion
s fo
r to
tal
CV
D
(IC
D9:
39
0-45
9),
and
IHD
(I
CD
9:
410-
414)
, fo
r al
l a
ges,
for
per
sons
les
s th
an 6
5, a
nd
for
pers
ons
65
and
olde
r.
Mea
n da
ily
adm
issi
ons
for
CV
D:
172.
5 a
ll ag
es,
54.5
<65
, 11
7.8
≥ 65
; fo
r IH
D:
24.5
<
65,
37.6
≥
65.
Cov
aria
tes:
N
O2,
O
3,
SO
2,
CO
, te
mpe
ratu
re,
rela
tive
hum
idity
. P
oiss
on
regr
essi
on
usin
g A
PH
EA
met
hodo
logy
; si
ne a
nd c
osin
e fu
nctio
ns
for
seas
onal
co
ntro
l; da
y of
w
eek
dum
my
varia
bles
. La
gs
0-3,
as
wel
l as
cor
resp
ond
ing
mul
ti-da
y av
erag
es
endi
ng
on
lag
0,
wer
e
cons
ider
ed.
In s
ingl
e-po
lluta
nt m
odel
s, b
oth
PM
met
rics
sho
wed
pos
itive
ass
ocia
tion
s w
ith b
oth
CV
D
and
IHD
adm
issi
ons
acro
ss a
ge g
roup
s. I
n T
wo-
po
lluta
nt
mod
els,
th
e B
S e
ffect
, b
ut
not
the
PM
10
effe
ct,
was
ro
bust
. N
o qu
antit
ativ
e re
sults
pr
ovid
ed
for
two-
pollu
tant
m
odel
s.
Thi
s st
udy
does
no
t su
ppor
t a
PM
10 e
ffect
ind
epen
dent
of
co-
pollu
tant
s.
Effe
cts
com
put
ed
for
50
µg/
m3
PM
10 a
nd 2
5 µ
g/m
3 BS
P
M10
(la
g 0)
: A
ll ag
es:
CV
D: 3
.2%
(0.
9, 5
.5)
0- 6
4 ye
ars:
C
VD
: 5.6
% (
2.0,
9.4
) IH
D: 6
.8%
(1.
3, 1
2.7)
65
+ y
ears
: C
VD
: 2.5
% (
-0.2
, 5.3
) IH
D: 5
.0%
(0.
8, 9
.3)
BS
(la
g 0)
: A
ll ag
es:
CV
D: 2
.95
% (
1.00
, 4.9
4)
0-64
yea
rs:
CV
D: 3
.12
% (
0.05
, 6.2
9)
IHD
: 2.7
8%
(-
1.88
, 7.6
3)
65+
yea
rs:
CV
D: 4
.24
% (
1.89
, 6.6
4)
IHD
(la
g 3)
: 4.5
7% (
0.86
, 8.4
2).
Pre
scot
t et
al
. (1
998)
(P
resc
ott e
t al,
1998
) E
dinb
urgh
, Sco
tland
19
81-1
995
(BS
and
SO
2)
1992
-199
5 (P
M10
, N
O2,
O
3, C
O)
Mea
ns f
or l
ong
and
shor
t se
ries:
B
S: 1
2.3,
8.7
P
M10
: NA
, 20.
7.
Dai
ly e
mer
genc
y ho
spita
l ad
mis
sion
s fo
r C
VD
(I
CD
9: 4
10-4
14,
426
-429
, 4
34-4
40)
for
pers
ons
less
tha
n 65
yea
rs a
nd f
or p
erso
ns 6
5 or
old
er.
Sep
arat
e an
alys
es
pres
ente
d fo
r lo
ng
(198
1-19
95)
and
shor
t (1
992-
1995
) se
ries.
M
ean
hosp
ital
adm
issi
ons
for
lon
g an
d sh
ort
serie
s:
<65
, 3.
5, 3
.4;
65+
, 8.
0, 8
.7.
Cov
aria
tes:
SO
2,
NO
2, O
3, C
O,
win
d sp
eed,
tem
pera
ture
, ra
infa
ll.
PM
10
mea
sure
d by
T
EO
M.
Poi
sson
lo
g-li
near
re
gres
sion
; tr
end
and
sea
sons
co
ntro
lled
by
mon
thly
du
mm
y va
riabl
es
over
en
tire
serie
s;
day
of
we
ek
dum
my
varia
bles
; m
in
daily
te
mpe
ratu
re
mod
eled
us
ing
octil
e du
mm
ies.
P
ollu
tant
s e
xpre
ssed
as
cum
ulat
ive
lag
1-3
day
mov
ing
avg.
In
long
se
ries,
ne
ither
B
S
nor
NO
2
we
re
asso
ciat
ed w
ith C
VD
adm
issi
ons
in e
ither
ag
e gr
oup.
In
the
shor
t se
ries,
onl
y 3-
day
mov
ing
aver
age
P
M10
w
as
posi
tivel
y an
d si
gnifi
cant
ly
asso
ciat
ed
with
C
VD
ad
mis
sion
s in
sin
gle-
pol
luta
nt m
odel
s, a
nd
only
for
per
sons
65
or o
lde
r. B
S,
SO
2, a
nd
CO
al
so
sho
wed
po
sitiv
e
asso
ciat
ion
s in
th
is s
ubse
t, b
ut w
ere
not
sign
ifica
nt a
t th
e 0.
05 le
vel.
The
PM
10 e
ffect
rem
aine
d la
rgel
y un
chan
ged
whe
n al
l ot
her
pollu
tant
s w
ere
adde
d to
the
mod
el,
how
ever
qua
ntita
tive
resu
lts w
ere
not
give
n. R
esul
ts a
ppea
r to
sh
ow
an
effe
ct o
f P
M10
ind
epen
dent
of
co-
pollu
tant
s.
Per
cent
Exc
ess
Ris
k (9
5%
CI)
: E
ffect
s co
mp
uted
fo
r 50
µ
g/m
3 ch
ange
in
P
M10
an
d 25
µ
g/m
3 ch
ange
in B
S.
Long
ser
ies:
B
S, l
ag
1-3:
<
65: -
0.5%
(-5
.4, 4
.6)
65+
: -0.
5% (
-3.8
, 2.9
) S
hort
ser
ies:
B
S, l
ag
1-3:
<
65: -
9.5%
(-2
4.6,
8.0
) 65
+: 5
.8%
(-4
.9, 1
7.8)
P
M10
, lag
1-3
: <
65: 2
.0%
(-
12.
5, 1
9.0)
65
+: 1
2.4
% (
4.6,
20.
9).
report no. 4/11
119
Won
g et
al.
(200
2) (
Won
g et
al,
2002
)
Lond
on, U
K, 1
992-
1994
. m
ean
daily
con
cent
ratio
n of
PM
10: 2
8.5.
The
pr
esen
t st
udy
exa
min
ed
the
rela
tion
betw
een
PM
10
and
othe
r co
-pol
luta
nts
(NO
2,
SO
2 an
d O
3),
and
daily
ad
mis
sion
for
car
diac
di
seas
es (
ICD
9: 3
90-4
29)
and
IHD
(IC
D9
: 41
0-41
4) i
n th
e ci
ty o
f Lo
ndon
. T
he s
ame
anal
yses
w
ere
perf
orm
ed
for
the
city
of
H
ong
Kon
g (f
indi
ngs
not
sho
wn
in
the
pres
ent
repo
rt).
P
oiss
on r
egre
ssio
n w
as u
sed
adju
stin
g fo
r no
n-pa
ram
etric
sm
ooth
fu
nctio
ns
of
time,
te
mpe
ratu
re,
hum
idity
, in
fluen
za e
pide
mic
s, d
ay
of t
he w
eek,
hol
iday
s an
d se
ason
s. A
n a
prio
ri la
g 0-
1 w
as u
sed.
Est
imat
es
for
the
best
sin
gle
day
lag
(0,
1,
2,
3)
wer
e al
so
sho
wn.
T
he
exce
ss r
isks
of
hosp
italis
atio
n w
ere
also
sho
wn
afte
r m
utua
l ad
just
men
t fo
r ea
ch
sing
le
co-
pollu
tant
.
PM
10
sign
ifica
ntly
in
crea
sed
the
hosp
italis
atio
n
for
card
iac
dise
ases
an
d IH
D,
part
icul
arly
in c
old
seas
ons,
eve
n af
ter
furt
her
adju
stm
ent
for
O3.
Whe
n ad
just
ing
for
NO
2 an
d S
O2,
no
exc
ess
adm
issi
ons
for
card
iac
dise
ases
wer
e fo
und
with
PM
10.
SO
2 an
d N
O2
incr
ease
d an
d
O3
decr
ease
d ca
rdia
c ho
spita
l adm
issi
on.
R
emar
kabl
y si
mila
r as
soci
atio
ns w
ere
fou
nd
in H
ong
Kon
g.
Per
cent
e
xces
s ris
k fo
r an
in
crem
ent o
f PM
10 b
y 10
µg/
m3 :
Car
diac
dis
ease
s, a
ll ag
es,
lag
0-1:
0.
8% (
0.3,
1.4
) W
arm
sea
son
s: 0
.1%
(-0
.7,
1.0)
C
old
seas
ons:
1.2
% (
0.5,
2.0
) A
djus
ted
for
NO
2: 0
.2%
(-0
.6, 1
.0)
Adj
uste
d fo
r O
3: 0
.8%
(0.
3, 1
.3)
Adj
uste
d fo
r S
O2:
-0
.2%
(-
1.1
, 0.
4)
IHD
, all
ages
, la
g 0-
1:
0.9%
(0.
1, 1
.6)
War
m s
easo
ns:
0.1
% (
-1.1
, 1.
4)
Col
d se
ason
s: 1
.3%
(0.
3, 2
.3)
Wor
dley
et
al
. (1
997)
(W
ordl
ey e
t al,
1997
) B
irmin
gham
, U
K,
4/1/
92-
3/31
/94.
P
M10
mea
n, m
in,
ma
x: 2
6,
3, 1
31.
Dai
ly h
osp
ital
adm
issi
ons
for
IHD
(IC
D9
: 41
0-42
9)
for
all
ages
. M
ean
ho
spita
lizat
ions
: 25
.6/d
ay.
Cov
aria
tes:
tem
pera
ture
and
rel
ativ
e hu
mid
ity.
Lin
ear
regr
essi
on
was
u
sed
af
ter
adju
stm
ent
for
day
of w
eek
and
mon
thly
dum
my
varia
bles
, lin
ear
tren
d te
rm.
Lags
of
0-
3 co
nsid
ered
, as
wel
l as
the
mea
n of
lags
0-2
.
No
stat
istic
ally
si
gnifi
cant
ef
fect
s w
ere
ob
serv
ed f
or P
M10
on
IHD
adm
issi
ons
for
any
lag.
Not
e th
at P
M1
0 w
as
asso
ciat
ed w
ith
resp
irato
ry
adm
issi
ons
an
d
with
ca
rdio
vasc
ular
mor
talit
y in
the
sam
e st
udy
(res
ults
not
sho
wn
here
).
Per
cent
ch
ange
pe
r 50
µ
g/m
3 ch
ange
PM
10 I
HD
adm
issi
on
s:
lag
0: 1
.4%
(-4
.4, 7
.2)
lag
1:-1
.3%
(-7
.1, 4
.4).
report no. 4/11
120
And
erso
n et
al
. (2
001)
(A
nder
son
et a
l, 20
01)
W
est M
idla
nd,
Eng
land
(O
ctob
er 1
994-
Dec
embe
r 19
96)
Pop
ulat
ion:
2.3
mill
ion
P
M10
mea
n: 2
3.3
PM
2.5
mea
n: 1
4.5
PM
10-2
.5: 9
.0.
Car
diov
ascu
lar
hosp
ital
adm
issi
ons
(IC
D9:
390
-45
9)
rela
ted
to
PM
10,
PM
2.5,
P
M10
-2.5,
BS
, su
lfate
, N
O2,
O3,
SO
2, C
O.
GLM
reg
ress
ion
with
qu
asi-
likel
ihoo
d ap
proa
ch,
cont
rolli
ng
for
seas
onal
pa
ttern
s,
tem
pera
ture
, hu
mid
ity,
influ
enza
epi
sode
s, d
ay o
f th
e w
eek.
Adj
uste
d fo
r re
sidu
al
seria
l co
rrel
atio
n an
d
over
-di
sper
sion
. C
ardi
ac
dise
ase
(IC
D9:
39
0-42
9),
IHD
(co
nsid
ered
for
age
≥65
yea
rs,
ICD
9:
410-
414)
and
str
oke
(con
side
red
for
age ≥6
5 ye
ars
ICD
9:
430-
438)
w
ere
co
nsid
ere
d as
su
bdia
gnos
es.
Hos
pita
l ad
mis
sion
s (a
ll a
ges)
fo
r C
VD
, ca
rdia
c di
seas
e an
d IH
D
wer
e no
t as
soci
ated
w
ith
any
cons
ider
ed
pollu
tant
. A
n in
vers
e as
soci
atio
n b
etw
een
P
M10
-2.5
and
adm
issi
on
for
stro
ke w
as
foun
d. T
he
only
sig
nific
ant
seas
onal
int
erac
tion
was
for
B
S,
whi
ch s
how
ed
a po
sitiv
e as
soci
atio
n i
n
the
war
m s
easo
n.
CV
D h
ospi
tal
adm
issi
on;
lag
0-1;
P
M10
In
crem
ent
10-9
0%
(11.
4-38
.3 µ
g/m
3 ):
All
CV
D: -
0.6
(-2.
5, 1
.3)
Car
diac
: 0.3
(-1
.8, 2
.4)
IHD
: 2.1
(-2
.0, 6
.3)
Str
oke:
-3.
3 (-
7.9,
1.4
) P
M2.
5 (6
.0-2
5.8
µg/
m3 )
All
CV
D: -
0.5
(-2.
6, 1
.6)
Car
diac
: -0.
4 (-
2.8,
2.2
) IH
D: -
0.3
(-4.
5, 4
.2)
Str
oke:
-1.
6 (-
6.6,
3.6
) P
M10
-2.5 (
4.1-
15.2
µg/
m3 )
All
CV
D: -
0.7
(-3.
7, 2
.3)
Car
diac
: -0.
9 (-
4.3,
2.7
) IH
D: 1
.1 (
-4.1
, 6.6
) S
trok
e: -
8.2
(-13
.9, -
2.2)
B
S (
5.1-
23.6
µg/
m3 )
All
CV
D: 1
.0 (
-1.0
, 3.1
) C
ardi
ac: 1
.7 (
-0.6
, 3.9
) IH
D: 2
.0 (
-2.3
, 6.4
) S
trok
e: -
2.7
(-7.
6, -
2.4)
.
N
OR
WA
Y
Man
nsak
er
(200
4)
(Man
nsak
er e
t al,
2004
)
Tro
ndhe
im,
Nor
way
, 19
93-2
001.
m
edia
n da
ily
conc
entr
atio
n: P
M10
39.
3;
PM
2.5
13.2
.
Thi
s st
udy
com
pare
d th
e m
ean
daily
num
ber
of
adm
issi
ons
for
CV
D t
o th
e S
t. O
lav
Uni
vers
ity
hosp
ital
in t
he
city
of
Tro
ndh
eim
, on
da
ys w
ith
rela
tivel
y lo
w a
nd h
igh
leve
ls o
f P
M10
, P
M2,
5,
and
othe
r co
-pol
luta
nts,
in
clud
ing
NO
, N
O2,
S
O2,
O3.
A t
ime
serie
s an
alys
is w
as c
arrie
d ou
t to
se
e ho
w
day-
to-d
ay
varia
tion
s in
co
ncen
trat
ion
s of
air
pollu
tant
s co
rrel
ated
with
th
e nu
mbe
r of
ad
mis
sion
s fo
r ca
rdio
vasc
ular
di
seas
e.
Tim
e se
ries
anal
ysis
d
id
not
sho
w
an
y st
atis
tical
ly
sign
ifica
nt
corr
elat
ion
betw
een
da
y-to
-day
va
riatio
ns
in
air
pollu
tion
and
hosp
ital
adm
issi
ons
for
card
iova
scul
ar
dise
ase.
T
he m
ean
daily
num
ber
of h
ospi
taliz
atio
ns
wa
s fo
und
to b
e si
gnifi
cant
ly h
ighe
r on
da
ys
with
hig
h le
vels
of
NO
and
NO
2, b
ut n
ot w
ith
othe
r po
lluta
nts.
No
risk
sho
wn
.
report no. 4/11
121
GER
MA
NY
W
ichm
ann
et
al
(200
4)
(Wic
hman
n, 2
004)
R
hine
-Ruh
r ar
ea,
Ger
man
y,
198
5,
sing
le
epis
ode.
TS
P d
urin
g th
e ep
isod
e in
R
hine
-Ruh
r ar
ea: ~
600
.
In
Janu
ary
1985
an
ex
ten
ded
smog
ep
isod
e oc
curr
ed i
n C
entr
al E
urop
e t
oget
her
with
ver
y lo
w t
empe
ratu
res.
Thi
s pa
per
revi
ew
s th
e he
alth
ef
fect
s, i
nclu
din
g C
VD
adm
issi
on,
inve
stig
ated
du
ring
the
smog
pe
riod
and
a co
ntro
l pe
riod
befo
re a
nd a
fter
the
smog
epi
sode
. T
he R
hine
-Ru
hr a
rea
(Wes
tern
Ger
man
y) w
as
affe
cted
for
5 d
ays
with
ma
xim
um c
once
ntra
tion
of
TS
P
of
600
µ
g/m
3 (2
4 h
aver
age
s).
In
Aug
sbur
g (S
outh
ern
Ger
man
y)
the
smog
ep
isod
e w
as l
ess
seve
re (
100
µg/
m3 T
SP
, 24
h
aver
ages
).
Rhi
ne-R
uhr
hos
pita
l ad
mis
sion
s in
crea
sed
by 5
7% f
or c
ereb
ral
circ
ulat
ion
failu
re,
by
30%
for
cor
ona
ry i
nsuf
ficie
ncy,
by
19%
for
C
VD
, by
14
% f
or h
eart
ins
uffic
ienc
y an
d by
10
% fo
r A
MI.
No
risk
sho
wn
.
Pet
ers
et
al
(200
4)
(Pet
ers
et a
l, 20
04)
Aug
sbur
g,
Ger
man
y,
1999
-200
1.
No
info
rmat
ion
on P
M.
Thi
s is
a
case
-cro
ssov
er
stud
y in
w
hich
69
1 ca
ses
of
non
-fat
al
myo
card
ial
AM
I w
ere
iden
tifie
d in
th
e A
ugsb
urg
Myo
card
ial
Infa
rctio
n R
egis
try
in A
ugsb
urg,
bet
wee
n F
ebru
ary
1999
an
d Ju
ly 2
001.
Cas
es p
rovi
ded
info
rmat
ion
on
fact
ors
that
may
hav
e tr
igge
red
the
myo
card
ial
infa
rctio
n. D
ata
on s
ubje
cts'
act
iviti
es d
urin
g th
e fo
ur
days
pr
eced
ing
the
onse
t of
sy
mpt
oms
wer
e c
olle
cte
d w
ith t
he
use
of p
atie
nt d
iarie
s.
Con
ditio
nal
logi
stic
re
gres
sion
m
odel
s w
ere
used
to
es
timat
e th
e as
soci
atio
n be
twee
n tr
ansi
ent
exp
osu
re t
o tr
affic
and
the
ris
k of
AM
I. T
he “
cont
rol p
erio
d” o
f e
xpo
sure
was
def
ined
as
an e
xpos
ure
to t
raffi
c by
the
sam
e su
bjec
t 24
to
71 h
ours
bef
ore
the
hour
of
the
onse
t of
the
A
MI.
Onl
y su
bjec
ts
for
who
m
ther
e
wer
e di
scor
dant
se
ts
of
data
on
ex
posu
re
wer
e in
clud
ed in
the
ana
lyse
s.
An
asso
ciat
ion
was
fo
und
betw
een
expo
sure
to
traf
fic a
nd t
he o
nset
of
an A
MI
with
in o
ne h
our
afte
rwar
d (O
R=
2.92
; 95
%
CI:
2.22
-3.8
3).
The
tim
e th
e su
bjec
ts s
pen
t in
ca
rs,
on
publ
ic
tran
spor
tatio
n,
or
on
mot
orcy
cles
or
bi
cycl
es
wa
s co
nsis
tent
ly
linke
d w
ith a
n in
crea
sed
risk
of A
MI.
No
data
on
PM
report no. 4/11
122
FRA
NC
E
Eils
tein
et
al
(2
004)
(E
ilste
in e
t al,
2004
)
Nin
e F
renc
h ci
ties,
199
5-19
99.
No
data
di
strib
utio
n w
as
repo
rted
.
Thi
s st
udy
exa
min
ed t
he s
hort
ter
m e
ffect
s of
P
M10
, B
S
an
d ot
her
co-p
ollu
tant
s (C
O,
NO
2,
SO
2,
O3)
on
m
orta
lity
and
hosp
italis
atio
n fo
r ca
rdio
vasc
ular
(a
nd
resp
irato
ry)
diso
rder
s in
ni
ne F
renc
h ci
ties
(Bor
dea
ux,
Le H
avre
, Li
lle,
Lyon
, M
arse
ille,
Par
is,
Rou
en,
Str
asbo
urg
and
Tou
lous
e)
of
the
Sur
veill
ance
A
ir et
S
ante
pr
ogra
m.
The
an
alys
is
asse
ssed
th
e ci
ty-
spec
ific
and
com
bine
d as
soci
atio
ns w
ith v
ario
us
pollu
tant
s us
ing
Poi
sson
GA
M,
afte
r a
djus
ting
for
tren
ds i
n se
ason
s, d
ay o
f w
eek,
hol
iday
s,
influ
enza
ep
idem
ics,
tem
pera
ture
and
hu
mid
ity.
A c
ombi
ned
RR
was
cal
cula
ted
for
all t
he c
ities
. La
g 0-
1 an
d la
g 0-
5 w
ere
used
.
Nei
ther
P
M10
no
r B
S
show
ed
sign
ifica
nt
exce
ss a
dmis
sion
for
CV
D.
CV
D a
dmis
sio
n, e
ffect
s co
mpu
ted
for
10 µ
g/m
3 cha
nge
in e
xpo
sitio
n:
Lag
0-1
PM
10; a
ge 1
5-64
: 0.0
% N
S
PM
10; a
ge 6
5+
: 0.1
% N
S
Lag
0-5
PM
10; a
ge 1
5-64
: 0.0
% N
S
PM
10; a
ge 6
5+
: 0.7
% N
S
Lag
0-1
BS
; age
15-
64: 0
.4%
NS
B
S; a
ge 6
5+:
0.3%
NS
La
g 0-
5 B
S; a
ge 1
5-64
: 0.6
% N
S
BS
; age
65+
: 0.
3% N
S
ITA
LY
B
igge
ri et
al
. (2
005)
(B
igge
ri et
al,
2005
) E
ight
ci
ties
of
Nor
th,
Cen
tre
and
So
uth
of It
aly
1990
-199
9.
PM
10
me
an
daily
co
ncen
trat
ion:
T
urin
, 91-
94: 7
7.6
Tur
in, 9
5-98
: 63.
8 M
ilan,
90-
94: 6
1.8
Mila
n, 9
5-97
: 45.
2 V
eron
a, 9
5-99
: 36.
5 R
aven
na, 9
1-95
: 59.
1 B
olog
na, 9
6-9
8: 4
1.2
Flo
renc
e, 9
6-98
: 40.
3 R
ome,
92-
94: 6
9.7
Pal
erm
o, 9
7-99
: 42.
9.
Met
a-an
alys
is
of
shor
t-te
rm
effe
cts
of
air
pollu
tion
on
mor
talit
y an
d ho
spita
l ad
mis
sion
(c
ardi
ac
cau
ses
– IC
D9:
39
0-45
9-,
and
resp
irato
ry c
ause
s) i
n ei
ght
Italia
n ci
ties
from
19
90 t
o 19
99.
Dai
ly c
once
ntra
tions
of
pollu
tant
s w
ere
colle
cte
d. A
gen
eral
ized
lin
ear
mod
el w
as
used
afte
r ad
just
men
t fo
r a
ge,
day
of t
he w
eek,
ho
liday
s,
influ
enza
ep
ide
mic
s,
met
eoro
logi
cal
varia
bles
, an
d s
easo
nalit
y p
atte
rn w
as
fitte
d fo
r P
M10
and
oth
er
co-p
ollu
tant
(S
O2,
NO
2, C
O,
and
O3)
. C
ity-s
peci
fic a
nd f
ixed
- an
d ra
ndom
-effe
cts
pool
ed e
stim
ate
s w
ere
sho
wn.
La
g 0-
3 w
as
use
d fo
r ho
spita
l adm
issi
on.
A s
igni
fican
t di
rect
ass
ocia
tion
was
fou
nd
betw
een
PM
10
and
card
iac
hosp
ital
adm
issi
on.
The
ot
her
cons
ider
ed
co-
pollu
tant
s, w
ith t
he e
xcep
tion
of O
3, s
how
ed
stro
nger
as
soci
atio
ns.
Th
e ex
cess
ris
ks
wer
e m
odifi
ed b
y de
priv
atio
n sc
ore
and
by
the
NO
2/P
M10
rat
io.
Car
diac
ho
spita
l ad
mis
sion
, ef
fect
s co
mpu
ted
for
10
µg/
m3
chan
ge in
PM
10 (
lag
0-3)
: F
ixed
-effe
cts:
0.7
7% (
0.40
, 1.
15)
Ran
dom
-effe
cts:
0.
82%
(0
.32,
1.
32).
report no. 4/11
123
D’Ip
polit
i (20
03)
(D
'Ippo
liti
et a
l, 20
03)
Rom
e, It
aly,
199
5-19
97.
TS
P
mea
n da
ily
conc
entr
atio
n
from
5
mon
itors
(S
D):
66
.9
(19.
7).
A
case
-cro
ssov
er
anal
ysis
w
as
cond
uct
ed
to
eval
uate
the
rel
atio
n be
twee
n T
SP
and
oth
er
co-p
ollu
tant
s (S
O2,
C
O,
NO
2)
and
hosp
ital
adm
issi
on
for
a fir
st
epis
ode
of
acut
e A
MI
(IC
D9:
410
). I
n th
e pe
riod
of o
bser
vatio
n 6,
531
subj
ects
res
idin
g in
Rom
e w
ere
hosp
italiz
ed f
or
AM
I.
A
time-
stra
tifie
d ca
se-c
ross
over
de
sign
w
as
cons
ider
ed,
usin
g as
con
trol
day
s th
e sa
me
day
of
the
wee
k as
th
e m
yoca
rdia
l in
farc
tion
occu
rred
, in
ot
her
wee
ks
of
the
mon
th.
OR
s w
ere
der
ived
by
cond
ition
al
logi
stic
reg
ress
ion
mod
els
afte
r al
low
ance
fo
r te
mpe
ratu
re
and
hum
idity
. A
n
a pr
iori
lag
0-2
wa
s us
ed.
Est
imat
es f
or l
ags
0, 1
, 2,
3,
and
4 w
ere
also
sh
ow
n.
Pos
itive
as
soci
atio
ns
wer
e
foun
d fo
r to
tal
susp
ende
d p
artic
ulat
e,
NO
2 an
d C
O.
The
st
rong
est
and
m
ost
cons
iste
nt
effe
ct
wa
s fo
und
for
TS
P,
in p
artic
ular
dur
ing
the
war
m
seas
on,
amon
g th
e el
derly
an
d am
ong
peop
le w
ith h
ear
t con
duct
ion
dis
turb
ance
s.
Per
cent
exc
ess
ris
k (9
5% C
I) f
or
an in
crem
ent o
f TS
P b
y 10
µg/
m3 ;
Lag
0: 2
.3 (
0.4,
4.2
) La
g 1:
1.5
(-0
.4, 3
.4)
Lag
2: 1
.7 (
-0.1
, 3.5
) La
g 3:
-0.
9 (-
2.6,
0.3
La
g 4:
0.1
(-1
.3, 1
.6)
Lag
0-2:
O
vera
ll: 2
.8 (
0.5,
5.2
)
Mal
es: 2
.3 (
-0.5
, 5.1
) F
emal
es: 3
.9 (
0.0,
8.0
) A
ge 1
8-64
: 1.3
(-2
.2, 4
.8)
Age
65-
74: 3
.2 (
-1.0
, 7.6
) A
ge 7
5+: 4
.6 (
0.5,
8.9
) C
old
perio
d: 1
.8 (
-1.1
, 4.7
) W
arm
per
iod:
4.6
(0.
8, 8
.7).
S
tafo
ggia
et
al
. (2
005)
(S
tafo
ggia
et
al, 2
005)
R
ome,
Ital
y, 1
998-
2000
. P
M10
; P
NC
(n
o/cm
3 ) m
ean
daily
co
ncen
trat
ion:
52
.1;
51,6
69,
resp
ectiv
ely.
The
stu
dy e
xam
ined
the
rel
atio
nshi
p be
twee
n P
M10
, P
NC
and
oth
er c
o-po
lluta
nts
(CO
, N
O2,
S
O2
and
O3)
and
cor
onar
y ev
ents
for
res
iden
ts
in
Rom
e,
cons
ider
ing
bot
h ou
t-of
-hos
pita
l de
aths
an
d ho
spita
lisat
ions
. G
ener
alis
ed
addi
tive
mod
els
adju
sted
by
te
mpo
ral
tren
d,
tem
pera
ture
, hu
mid
ity,
day
of w
eek,
ho
liday
s,
and
popu
latio
n re
duct
ion
in
sum
mer
m
onth
s w
ere
used
. La
gs
0,
1,
0-1
and
0-4
wer
e co
nsid
ered
.
A
sign
ifica
nt
exce
ss
risk
of
hosp
ital
adm
issi
on
for
coro
nary
ev
ents
w
as
foun
d w
ith
sam
e-d
ay
PN
C
and
C
O.
No
asso
ciat
ion
was
fou
nd w
ith s
ame-
day
PM
10,
NO
2, S
O2
and
O3.
T
he a
ssoc
iatio
ns w
ith P
NC
and
CO
wer
e st
rong
er fo
r el
derly
sub
ject
s.
Per
cent
exc
ess
ris
k (9
5% C
I) f
or
an
incr
emen
t of
P
M10
by
29
.7
µg/
m3 (
IQR
)
Lag
0: 0
.9 (
-2.3
, 4.2
) La
g 1:
-1.
2 (-
4.4,
2.2
) La
g 0-
1: 0
.3 (
-3.2
, 4.0
) La
g 0-
4: -
2.0
(-6.
5, 2
.6)
Lag
0; a
ge 3
5-65
: -2.
5 (-
7.2,
2.4
) La
g 0;
age
65-
74: 1
.5 (
-4.2
, 7.6
) La
g 0;
age
75
+: 3
.2 (
-2.4
, 9.2
) P
erce
nt e
xce
ss r
isk
(95%
CI)
for
an
inc
rem
ent
of P
NC
by
28,0
01
no/c
m3 (
IQR
)
Lag
0: 4
.0 (
0.1,
8.1
) La
g 1:
0.4
(-3
.5, 4
.4)
Lag
0-1:
2.8
(-1
.6, 7
.4)
Lag
0-4:
0.2
(-4
.7, 5
.5)
Lag
0; a
ge 3
5-65
: 2.4
(-3
.5, 8
.6)
Lag
0; a
ge 6
5-74
: -0.
2 (-
6.9,
6.9
) La
g 0;
age
75
+: 9
.0 (
2.2,
16
.2).
report no. 4/11
124
SPA
IN
D
íaz
et a
l. (1
999
) (D
iaz
et
al, 1
999)
M
adrid
, Spa
in,
1994
-199
6 T
SP
by
beta
atte
nuat
ion.
S
umm
ary
stat
istic
s no
t gi
ven.
Dai
ly
emer
genc
y ho
spita
l ad
mis
sion
s fo
r al
l ca
rdio
vasc
ular
cau
ses
(IC
D9:
390
-459
) fo
r th
e G
rego
rio M
aran
on U
nive
rsity
Tea
chin
g H
ospi
tal.
Mea
n ad
mis
sion
s:
9.8/
day.
C
ovar
iate
s:
SO
2,
NO
2,
O3,
te
mpe
ratu
re,
pres
sure
, re
lativ
e hu
mid
ity,
exc
ess
sunl
ight
. B
ox-
Jenk
ins
time-
serie
s m
etho
ds
wer
e us
ed
to
rem
ove
auto
corr
elat
ions
, fo
llow
ed b
y cr
oss-
cor
rela
tion
an
alys
is;
sine
and
cos
ine
term
s fo
r se
ason
ality
; de
tails
unc
lear
.
No
sign
ifica
nt
effe
cts
of
TS
P
on
CV
D
repo
rted
. N
o qu
antit
ativ
e re
sults
pre
sent
ed
for
PM
.
Llor
ca
et
al.
(200
5)
(Llo
rca
et a
l, 20
05)
T
orre
lave
ga,
Spa
in,
1992
-199
5.
No
data
di
strib
utio
n w
as
repo
rted
.
Thi
s st
udy
anal
ysed
the
rel
atio
ns b
etw
een
TS
P
and
othe
r co
-pol
luta
nts
(NO
, N
O2,
S
O2
and
hydr
ogen
sul
fide)
, an
d em
erge
ncy
adm
issi
ons
for
card
iac
dise
ase
(IC
D9:
39
0-4
59)
in
Tor
rela
vega
(6
0,00
0 in
hab
itant
s).
RR
s w
ere
estim
ated
fo
r ea
ch
pollu
tant
by
P
oiss
on
regr
essi
on a
djus
ted
for
influ
enza
, da
y of
wee
k,
win
d sp
eed,
win
d an
d te
mpe
ratu
re.
Lag
0 w
as
appa
rent
ly u
sed.
A s
ign
ifica
nt i
nver
se a
ssoc
iatio
n w
as f
ound
be
twee
n T
SP
an
d ad
mis
sion
fo
r ca
rdia
c di
seas
es;
hyd
roge
n su
lfide
dec
reas
ed a
nd
NO
x in
crea
sed
the
risk
of
card
iac
adm
issi
ons.
Afte
r m
utua
l ad
just
men
t fo
r th
e ot
her
cons
ider
ed
co-p
ollu
tant
s,
no
asso
ciat
ion
em
erge
d.
Per
cent
e
xces
s ris
k fo
r an
in
crem
ent o
f TS
P b
y 1
0 µ
g/m
3 . A
fter
adj
ustm
ent
for
met
eoro
logi
cal
fact
ors,
da
y an
d in
fluen
za:
-1.2
(-1
.9, -
0.5)
A
fter
furt
her
adju
stm
ent
for
all
the
co-p
ollu
tant
s co
nsid
ered
: 0.
2 (-
0.8,
1.2
).
report no. 4/11
125
TAB
LE II
I – S
HO
RT-
TER
M P
AR
TIC
ULA
TE M
ATT
ER E
XPO
SUR
E A
ND
RES
PIR
ATO
RY
HO
SPIT
AL
AD
MIS
SIO
NS
(Der
ived
from
EPA
TA
BLE
8B
-2).
Ref
eren
ce, L
ocat
ion,
Ye
ars,
PM
Inde
x, M
ean
or M
edia
n (µ
g/m
3 ).
Stud
y D
escr
iptio
n. M
ode
ling
met
hods
: lag
s,
smoo
thin
g, c
o-po
lluta
nts
and
cova
riate
s.
Res
ults
and
Com
men
ts.
PM In
dex,
lag,
per
cent
exc
ess
risk%
(95%
LC
L, U
CL)
APH
EA 1
Sun
yer
et
al.
(199
7)
(Sun
yer
et a
l, 19
97)
Bar
celo
na (
86 -
92)
B
S M
edia
n: 4
0 B
S R
ange
: 11
-258
H
elsi
nki (
86 -
92)
B
S M
edia
n: -
B
S R
ange
: -
Par
is (
86 -
92)
B
S M
edia
n: 2
8 B
S R
ange
: 4-1
86
Lond
on (
86 -
92)
B
S M
edia
n: 1
3 B
S R
ange
: 3-9
5.
Thi
s st
udy
exa
min
ed t
he r
elat
ion
betw
een
BS
, S
O2,
NO
2, a
nd
O3,
and
dai
ly c
ount
s of
hos
pita
l ad
mis
sion
s fo
r as
thm
a an
d em
erge
ncy
visi
ts i
n ad
ults
(ag
es 1
5-64
yea
rs:
mea
n/da
y =
3.9
(B
);
0.7
(H);
13.
1 (L
); 7
.3 (
P))
and
chi
ldre
n (a
ges
<
15 y
ears
: m
ean/
day
= 0
.9 (
H);
19.
8 (L
); 4
.6 (
P))
. A
sthm
a (I
CD
9: 4
93)
was
stu
died
in
each
city
, bu
t th
e ou
tcom
e ex
amin
ed
diffe
red
acro
ss
citie
s:
ED
vi
sits
in
B
arce
lona
; em
erge
ncy
hosp
ital
asth
ma
adm
issi
ons
in
Lo
ndon
an
d H
elsi
nki,
and
tot
al a
sthm
a ad
mis
sion
s in
Par
is.
Est
imat
es
fro
m
all
citie
s ob
tain
ed
for
entir
e pe
riod
and
also
by
war
m o
r co
ld s
easo
ns,
usin
g tim
e-se
ries
GLM
re
gres
sion
, co
ntro
lling
fo
r te
mpe
ratu
re
and
RH
, vi
ral
epid
emic
s,
day
of
wee
k ef
fect
s, a
nd s
easo
nal
and
secu
lar
tren
ds
appl
ied
usin
g th
e A
PH
EA
stu
dy a
ppro
ach.
Dai
ly
adm
issi
ons
for
asth
ma
in
adul
ts
incr
ease
d si
gnifi
cant
ly
with
in
crea
sin
g am
bien
t le
vels
of
NO
2, a
nd
posi
tivel
y (b
ut
non-
sign
ifica
ntly
) w
ith B
S.
The
ass
ocia
tion
betw
een
asth
ma
adm
issi
ons
and
pollu
tion
varie
d ac
ross
citi
es,
likel
y du
e to
diff
erin
g as
thm
a ou
tco
mes
con
side
red.
In
child
ren,
da
ily a
dmis
sio
ns in
crea
sed
sign
ifica
ntly
with
S
O2
and
posi
tivel
y (b
ut
non-
sign
ifica
ntly
) w
ith B
S a
nd N
O2,
tho
ugh
the
latte
r on
ly i
n co
ld s
easo
ns.
No
asso
ciat
ion
wa
s ob
serv
ed
in c
hild
ren
for
O3.
Aut
hors
con
clud
ed t
hat
“I
n ad
ditio
n to
par
ticle
s, N
O2
and
SO
2 (b
y th
emse
lves
or
as a
con
stitu
ent
of a
pol
lutio
n m
ixtu
re)
may
be
im
port
ant
in
as
thm
a ex
acer
batio
ns”.
Exc
ess
risk
per
25
µg/
m3 B
S (
24
h A
vera
ge)
Ast
hma
Adm
issi
ons/
Vis
its:
<15
yea
rs.:
Lo
ndon
: 1.5
% (
lg 0
d)
Par
is: 1
.5%
(lg
2d)
T
otal
: 1.5
% (
-1.1
, 4.1
) 15
-64
year
s:
Bar
celo
na: 1
.8%
(lg
3d)
Lo
ndon
: 1.7
% (
lg 0
d)
Par
is: 0
.6%
(lg
0d)
T
otal
: 1.0
% (
-0.8
, 2.9
) T
wo
Pol
luta
nt (
per
25 µ
g/m
3 BS
) A
sthm
a A
dmis
sion
s (2
4 h
Avg
) <
15 y
ears
, (B
S &
NO
2):
Lond
on: 0
.6%
(lg
0d)
P
aris
: 2.9
% (
lg 2
d)
Tot
al: 1
.8%
(-0
.6, 4
.3)
<15
yea
rs, (
BS
& S
O2)
: Lo
ndon
: -1.
1% (
lg 0
d)
Par
is: -
1.4%
(lg
2d)
T
otal
: -1.
3 (-
5.0,
2.5
) 15
-64
year
s, (
BS
& N
O2)
: B
arce
lona
: 1.5
% (
lg 0
d)
Lond
on: -
4.7%
(lg
0d)
P
aris
: -0.
7% (
lg 1
d)
Tot
al: -
0.5%
(-5
.1, 4
.4).
report no. 4/11
126
APH
EA 2
Atk
inso
n
et
al.
(200
1)
(Atk
inso
n et
al,
2001
) E
ight
Eur
opea
n ci
ties
Med
ian/
rang
e B
arce
lona
1/9
4 -
12/9
6 P
M10
53.
3 (1
7.1,
131
.7)
Birm
ingh
am 3
/92
-12/
94
PM
10 2
1.5
(6.5
, 115
) Lo
ndon
1/9
2 -
12/9
4 P
M10
24.
9 (7
.2, 8
0.4)
M
ilan
-No
PM
10
(TS
P)
Net
herla
nds
1/9
2 -
9/95
P
M10
33.
4 (1
1.3,
130
.8)
Par
is 1
/92
- 9/
96
PM
13 2
0.1
(5.8
, 80.
9)
Rom
e -
No
PM
10 (
TS
P)
Sto
ckho
lm 3
/94
- 12
/96
PM
10 1
3.6
(4.3
, 43.
3).
As
part
of
th
e A
PH
EA
2
pr
ojec
t, as
soci
atio
n be
twee
n P
M1
0 an
d da
ily c
ount
s of
em
erg
ency
ho
spita
l adm
issi
ons
for
Ast
hm
a (0
-14
and
15-6
4 ye
ars)
, C
OP
D a
nd a
ll-re
spira
tory
dis
ease
(65
+
year
s)
regr
esse
d us
ing
GA
M,
cont
rolli
ng
for
envi
ronm
enta
l fac
tors
and
tem
pora
l pat
tern
s.
Thi
s st
udy
repo
rts
that
PM
wa
s as
soci
ated
w
ith d
aily
adm
issi
ons
for
resp
irato
ry d
isea
se
in a
sel
ectio
n o
f E
urop
ean
citi
es.
Ave
rage
da
ily
ozon
e
leve
ls
exp
lain
ed
a la
rge
prop
ortio
n of
the
bet
wee
n-ci
ty v
aria
bilit
y in
th
e si
ze o
f th
e pa
rtic
le e
ffect
est
imat
es i
n th
e ov
er 6
5 ye
ar a
ge g
roup
. In
chi
ldre
n, t
he
part
icle
eff
ects
wer
e co
nfo
unde
d w
ith N
O2
on a
day
-to-
day
basi
s.
For
10
µg/
m3 in
crea
se
Ast
hma
Ad
mis
sion
A
ge
0-
14
year
s:
PM
10 f
or c
ities
ran
ged
from
-0.
9%
(-2.
1, 0
.4)
to 2
.8%
(0.
8, 4
.8)
with
an
ove
rall
effe
ct e
stim
ate
of 1
.2%
(0
.2, 2
.3)
Ast
hma
Adm
issi
on
Age
15
-64
year
s:
Ove
rall
PM
1.1
% (
0.3,
1.8
) A
dmis
sion
of
CO
PD
and
Ast
hma
Age
65+
yea
rs:
Ove
rall
PM
1.0
% (
0.4,
1.5
) A
dmis
sion
A
ll R
espi
rato
ry
Dis
ease
Age
65+
yea
rs:
Ove
rall
PM
0.9
% (
0.6,
1.3
).
Subn
atio
nal s
tudi
es
G
REA
T B
RIT
AIN
Atk
inso
n et
al
. (1
999a
) (A
tkin
son
et a
l, 19
99a)
Lo
ndon
(92
- 9
4)
Pop
ulat
ion:
7.2
MM
P
M10
Mea
n: 2
8.5
10th-9
0th IQ
R: 1
5.8-
46.5
B
S m
ean:
12.
7 10
th-9
0th IQ
R: 5
.5-2
1.6
All-
age
resp
irat
ory
(mea
n=15
0.6/
day)
, al
l-age
as
thm
a (3
8.7/
day)
, C
OP
D p
lus
asth
ma
in a
dults
>
64
year
s (2
2.9/
day)
, an
d lo
wer
re
spira
tory
(6
4.1/
day)
in
ad
ults
>
64
year
s (1
6.7/
day)
ho
spita
l ad
mis
sion
s in
Lon
don
hosp
itals
wer
e co
nsid
ered
. C
ount
s fo
r ag
es
0-14
, 15
-64,
and
>
64 y
ears
we
re a
lso
exam
ined
. P
oiss
on G
LM
regr
essi
on w
as
used
, co
ntro
lling
for
sea
son,
day
of
w
eek,
m
eteo
rolo
gy,
auto
corr
elat
ion,
ov
erdi
sper
sio
n, a
nd in
fluen
za e
pide
mic
s.
Pos
itive
ass
oci
atio
ns w
ere
fou
nd b
etw
een
re
spira
tory
- re
late
d em
erge
ncy
hosp
ital
adm
issi
ons
and
PM
10
and
SO
2, b
ut n
ot f
or
O3
or
BS
. W
hen
SO
2
and
PM
10
wer
e in
clud
ed
sim
ulta
neou
sly,
si
ze
and
si
gnifi
canc
e o
f ea
ch w
as r
educ
ed.
Aut
hor
s co
nclu
ded
that
S
O2
and
PM
10
are
both
in
dica
tors
of
the
sam
e po
lluta
nt m
ix i
n th
is
city
. S
O2
and
PM
10 a
naly
ses
by t
empe
ratu
re
tert
ile
sugg
est
that
w
arm
se
ason
ef
fect
s do
min
ate.
O
vera
ll,
resu
lts
cons
iste
nt
with
ea
rlier
an
alys
es
for
Lond
on,
and
com
para
ble
with
th
ose
for
Nor
th
Am
eric
a an
d E
urop
e.
PM
10 (
50 µ
g/m
3 ), n
o co
-pol
luta
nt.
All
Re
spira
tory
Adm
issi
ons
: A
ll ag
e (la
g 1d
):
4.9%
(C
I: 1.
8,
8.1)
0-
14 (
lag
1d):
8.1
% (
CI:
3.5,
12.
9)
15-6
4 (la
g 2
d):
6.9%
(C
I:
2.1,
12
.9)
65+
(la
g 3d
): 4
.9%
(C
I: 0.
8, 9
.3)
Ast
hma
Adm
issi
ons:
A
ll ag
e (la
g 3d
): 3
.4%
(C
I: -1
.8,
8.9)
0-
14
(lag
3d):
5.
4%
(CI:
-1.2
, 12
.5)
15-6
4 (la
g 3
d):
9.4%
(C
I:
1.1,
18
.5)
65+
(la
g 0d
): 1
2% (
CI:
-1.8
, 27
.7)
CO
PD
&
A
sthm
a A
dmis
sion
s (6
5+; l
ag 3
): 8
.6%
(C
I: 2.
6, 1
5)
Low
er
Res
pira
tory
A
dmis
sion
s (6
5+;
lag
3d):
7.
6%
(CI:
0.
9,
14.8
).
report no. 4/11
127
And
erso
n et
al
. (1
998)
(A
nder
son
et a
l, 19
98)
Lond
on, 1
987-
1992
) P
opul
atio
n: 7
.2 M
M
BS
da
ily m
ean
: 14.
6 B
S 2
5-75
th IQ
R: 2
4-38
.
Poi
sson
GLM
log
-line
ar r
egre
ssio
n w
as
used
to
estim
ate
the
RR
of
Lo
ndon
da
ily
asth
ma
hosp
ital
adm
issi
ons
asso
ciat
ed w
ith c
hang
es i
n O
3, S
O2,
NO
2 an
d pa
rtic
les
(BS
) fo
r al
l age
s an
d fo
r 0-
14
year
s (m
ean=
19.5
/day
),
15-6
4 ye
ars
(mea
n=13
.1/d
ay)
and
65
+
year
s (m
ean
=2.
6/da
y).
Ana
lysi
s w
as
cont
rolle
d fo
r tim
e tr
ends
, se
ason
al
fact
ors,
ca
lend
ar
effe
cts,
in
fluen
za
epid
emic
s,
RH
, te
mpe
ratu
re,
and
auto
-cor
rela
tion.
Int
erac
tion
s w
ith c
o-po
lluta
nts
and
aero
alle
rgen
s w
ere
test
ed v
ia 2
po
lluta
nt
mod
els
and
mod
els
with
pol
len
coun
ts (
gras
s,
oak
and
birc
h).
Dai
ly h
ospi
tal
adm
issi
ons
for
asth
ma
we
re
foun
d to
hav
e as
soci
atio
ns
with
O3,
S
O2,
N
O2,
an
d B
S,
but
ther
e w
as
lack
of
co
nsis
tenc
y a
cros
s th
e ag
e gr
oups
in
the
spec
ific
pollu
tant
. B
S
ass
ocia
tion
was
st
rong
est
in t
he 6
5 +
gro
up,
espe
cial
ly i
n w
inte
r.
Pol
lens
w
ere
not
co
nsis
ten
tly
asso
ciat
ed
with
ho
spita
l ad
mis
sion
fo
r as
thm
a,
som
etim
es
bein
g po
sitiv
e,
som
etim
es
nega
tive.
A
ir po
llutio
n as
soci
atio
ns w
ith h
ospi
tal
adm
issi
ons
wer
e no
t ex
pla
ine
d by
ai
rbor
ne
polle
ns
in
sim
ulta
neou
s re
gres
sion
s, a
nd t
here
was
no
cons
iste
nt p
olle
n-po
lluta
nt in
tera
ctio
n.
Ast
hma
Ad
mis
sion
s.
BS
=25
µ
g/m
3 B
S L
ag 0
-3.
All
age:
5.9
8%
(0.
4, 1
1.9)
<
15: 2
.2%
(-4
.6, 9
.5)
15-6
4: 1
.2%
(-5
.3, 8
.1)
65+
: 22.
8%
(6.
1, 4
2.5)
B
S=
50
µg/
m3 ,
lag
2 &
co
-po
lluta
nt:
Old
er A
dult
(>64
yea
rs.)
Ast
hma
Vis
its:
BS
alo
ne: 1
4.6%
(2.
7, 2
7.8)
&
O3:
20.
0%
(3.
0, 3
9.8)
&
NO
2: 7
.4%
(-8
.7, 2
6.5)
S
O2:
11.
8%
(-2
.2, 2
7.8)
.
W
ong
et a
l. (2
002)
(W
ong
et a
l, 20
02)
Lond
on, U
K, 1
992-
1994
. m
ean
daily
co
ncen
trat
ion:
of
PM
10: 2
8.5.
The
pr
esen
t st
udy
exa
min
ed
the
rela
tion
betw
een
PM
10
and
othe
r co
-pol
luta
nts
(NO
2,
SO
2 an
d O
3),
and
daily
adm
issi
on f
or a
sthm
a (I
CD
9:
493)
fo
r ag
es
15-6
4 ye
ars,
an
d re
spira
tory
dis
ease
s (I
CD
9:
460-
519)
for
age
s 65
+
year
s in
th
e ci
ty
of
Lond
on.
The
sa
me
anal
yses
wer
e pe
rfor
med
for
the
city
of
Hon
g K
ong
(fin
ding
s no
t sh
ow
n in
the
pre
sent
rep
ort)
. P
oiss
on
regr
essi
on
was
us
ed
cont
rolli
ng
for
non-
para
met
ric
smoo
th
func
tions
of
tim
e,
tem
pera
ture
, hu
mid
ity,
influ
enza
epi
dem
ics,
day
of
the
wee
k, h
olid
ays
and
seas
ons.
An
a pr
iori
aver
age
lag
0-1
wa
s us
ed.
Est
imat
es f
or t
he
best
sin
gle
day
lag
(0,
1, 2
, 3)
wer
e al
so s
how
n.
The
ex
cess
ri
sks
of
hosp
italis
atio
n w
ere
al
so
sho
wn
afte
r m
utua
l ad
just
men
t fo
r ea
ch s
ingl
e co
-pol
luta
nt.
No
sign
ifica
nt
asso
ciat
ion
s ha
ve
bee
n sh
ow
n be
twee
n P
M10
and
hosp
ital
adm
issi
on
for
asth
ma
(15-
64
year
s)
and
resp
irato
ry d
isea
ses
(age
65+
), e
ven
afte
r al
low
ance
for
oth
er c
o-po
lluta
nts,
alth
ough
ex
cess
ris
ks w
ere
syst
ema
tical
ly o
ver
zero
. In
w
arm
se
aso
ns,
the
dire
ct
asso
ciat
ion
reac
hed
sta
tistic
al
sign
ifica
nce.
W
hen
co
nsid
erin
g th
e be
st
sing
le
lag
day,
a
sign
ifica
nt a
ssoc
iatio
n w
as f
ound
bet
we
en
PM
10 a
nd a
dmis
sion
for
bot
h as
thm
a an
d re
spira
tory
dis
ease
s.
NO
2 in
crea
sed
the
adm
issi
on
for
asth
ma
and
O3
the
one
of r
espi
rato
ry d
isea
ses.
Per
cent
exc
ess
ris
k (9
5% C
I) f
or
an
incr
emen
t of
P
M10
b
y 10
µ
g/m
3 ; R
espi
rato
ry
dis
ease
s,
age
65
+,
lag
0-1:
0.4
% (
-0.3
, 1.2
) W
arm
sea
son
s: 1
.8%
(0.
5, 3
.1)
Col
d se
ason
s: -
0.5%
(-1
.5, 0
.5)
Adj
uste
d fo
r N
O2:
0.9
% (
-0.3
, 2.0
) A
djus
ted
for
O3:
0.4
% (
-0.3
, 1.
2)
Adj
uste
d fo
r S
O2:
0.7
% (
-0.5
, 1.8
) La
g 3
(bes
t lag
): 1
.5%
(0.
8-2.
2)
Ast
hma,
age
15-
64, l
ag 0
-1:
1.4%
(-0
.1, 3
.0)
War
m s
easo
ns:
0.6
% (
-1.9
, 3.
1)
Col
d se
ason
s: 1
.6%
(-1
.5, 0
.5)
Lag
2 (b
est l
ag):
2.2
% (
0.8-
3.6)
.
report no. 4/11
128
Pon
ce
de
Leo
n et
al
. (1
996)
(P
once
de
Leon
et
al, 1
996)
Lo
ndon
, 4/1
987-
2/19
92.
Pop
ulat
ion:
7.3
mill
ion
B
S m
ean.
:14.
6 B
S 5
th-9
5th %
=6
– 27
.
Poi
sson
GLM
log
-line
ar r
egre
ssio
n an
alys
is o
f da
ily c
ount
s o
f ho
spita
l adm
issi
ons
(mea
ns/d
ay:
all
ages
=12
5.7
; A
ges
0-14
=45
.4;
Ag
es
15-
64=
33.6
; A
ges
65
+=
46.7
).
Effe
cts
of
tren
d,
seas
on a
nd
othe
r cy
clic
al
fact
ors,
day
of
the
wee
k,
holid
ays,
in
fluen
za
epid
emic
, te
mpe
ratu
re,
hum
idity
, an
d a
utoc
orre
latio
n w
ere
addr
esse
d. H
owe
ver,
tem
pera
ture
mod
eled
as
linea
r,
with
no
R
H
inte
ract
ion.
P
ollu
tion
va
riabl
es w
ere
BS
, S
O2,
O3,
and
NO
2, la
gge
d 0-
3 da
ys.
O3
asso
ciat
ed w
ith in
crea
se in
dai
ly h
ospi
tal
adm
issi
on,
espe
cial
ly in
the
"w
arm
" se
ason
. H
ow
ever
, U
-sha
pe o
f th
e O
3 do
se-r
espo
nse
sugg
ests
th
at
linea
r te
mp
erat
ure
cont
rol
wa
s no
t ad
equa
te.
Few
si
gnifi
cant
as
soci
atio
ns
wer
e fo
und
w
ith
othe
r po
lluta
nts,
but
the
se t
ende
d to
be
posi
tive
(esp
ecia
lly i
n co
ld s
easo
n, O
ct-M
arch
, an
d fo
r ol
der
indi
vidu
als
for
BS
).
Res
pira
tory
ho
spita
l ad
mis
sion
(a
ll ag
es);
BS
: 25
µg/
m3 :
Sin
gle
Pol
luta
nt M
odel
s
Oct
-Mar
; Lag
1: 0
.2%
(–1
.9,
2.3)
A
pr-S
ep; L
ag 1
: -2.
7% (
-6.0
, 0.8
) R
espi
rato
ry
hos
pita
l ad
mis
sion
s (>
65)
Sin
gle
Pol
luta
nt M
odel
s; B
S:
25
µg/
m3
Oct
-Mar
; Lag
2: 1
.2%
(-2
.1, 4
.5)
Apr
-Sep
; La
g 2:
: 4.
5%
(-1.
0,
10.4
).
Wor
dley
et a
l. (1
997)
(W
ordl
ey e
t al,
1997
) B
irmin
gham
, U
K,
4/19
92
–3/1
994.
P
M10
dai
ly v
alue
s:
Mea
n: 2
5.6
rang
e: 2
.8, 1
30.9
P
M10
3 d
ay m
ean:
M
ean:
25.
5 ra
nge:
7.3
, 104
.7.
Rel
atio
n be
twee
n P
M1
0
and
tota
l ho
spita
l ad
mis
sion
fo
r re
spira
tory
(m
ean
=
21.8
/day
),
asth
ma
(mea
n=6.
2/da
y),
bron
chiti
s (m
ean=
2.4/
day)
, pn
eum
onia
(m
ean=
3.4/
day)
, an
d C
OP
D
(mea
n=3.
2/da
y)
was
an
alyz
ed,
usin
g lo
g-lin
ear
reg
ress
ion
afte
r ad
just
men
t fo
r da
y of
w
eek,
m
onth
, lin
ear
tr
end,
R
H,
and
tem
pera
ture
(b
ut
not
tem
pera
ture
-RH
in
tera
ctio
n).
PM
10 p
ositi
vely
ass
ocia
ted
with
all
hosp
ital
adm
issi
ons
for
resp
irato
ry,
asth
ma,
br
onch
itis,
pn
eum
onia
, an
d C
OP
D.
Pne
umon
ia,
all
resp
irato
ry,
and
asth
ma
hosp
ital
adm
issi
ons
also
si
gnifi
cant
ly
posi
tivel
y as
soci
ated
with
the
mea
n of
PM
10
over
the
pas
t th
ree
days
, w
hich
gav
e 1
0 to
20
%
grea
ter
RR
’s
per
10 μg
/m3 ,
as
expe
cted
gi
ven
smal
ler
day
to
day
devi
atio
ns.
Oth
er
air
pollu
tant
s w
ere
exam
ine
d bu
t no
t pr
esen
ted,
as
“the
se d
id
not
have
a
si
gnifi
cant
as
soci
atio
n
with
he
alth
out
com
es i
ndep
ende
nt f
rom
tha
t of
P
M10
”.
50 µ
g/m
3 in P
M10
A
ll R
espi
rato
ry
hosp
ital
adm
issi
ons
(all
ages
) La
g 0:
12.
6% (
5.7,
20)
A
sthm
a ho
spita
l ad
mis
sion
s (a
ll ag
es)
Lag
2: 1
7.6%
(3,
34.
4)
Bro
nchi
tis h
ospi
tal a
dmis
sio
ns (
all
ages
) La
g 0:
32.
6% (
4.4,
68.
3)
Pne
umon
ia
hos
pita
l ad
mis
sion
s (a
ll ag
es)
Lag
3: 3
1.9%
(15
, 51.
4)
CO
PD
ho
spita
l ad
mis
sion
s (a
ll ag
es)
Lag
1: 1
1.5%
(-3
, 28.
2).
McG
rego
r et
al
. (1
999)
(M
cGre
gor
et a
l, 19
99)
Birm
ingh
am,
UK
. M
ean
PM
10: 3
0.0.
A
syno
ptic
cl
imat
olog
ical
ap
proa
ch
used
to
in
vest
igat
e lin
kage
s be
twe
en
air
mas
s ty
pes
(wea
ther
si
tuat
ions
),
PM
10,
and
all
resp
irato
ry
hosp
ital
adm
issi
ons
(mea
n=
19.2
/day
) fo
r th
e B
irmin
gham
are
a.
Thi
s st
udy
resu
lts s
how
dis
tinct
diff
eren
tial
resp
onse
s of
res
pira
tory
adm
issi
on r
ates
to
the
six
win
ter
air
mas
s ty
pes.
Tw
o of
thr
ee
type
s of
air
mas
ses
asso
cia
ted
with
abo
ve-
aver
age
adm
issi
on
rate
s al
so
favo
r hi
gh
PM
10 l
evel
s. T
his
is s
ugg
est
ive
of p
ossi
ble
linka
ge
betw
een
w
eath
er,
air
qual
ity,
and
he
alth
.
Est
imat
es n
ot r
epor
ted
report no. 4/11
129
And
erso
n et
al
. (2
001)
(A
nder
son
et a
l, 20
01)
Wes
t Mid
land
, E
ngla
nd
(Oct
ober
199
4-D
ecem
ber
1996
) P
opul
atio
n: 2
.3 m
illio
n
PM
10 m
ean:
23.
3 P
M2.
5 m
ean:
14.
5 P
M10
-2.5: 9
.0.
Res
pira
tory
hos
pita
l ad
mis
sion
s (m
ean:
66/
day)
re
late
d to
P
M10
, P
M2.
5,
PM
10-2
.5,
BS
, su
lfate
, N
O2,
O3,
SO
2, C
O.
GLM
re
gres
sion
with
qua
si-
likel
ihoo
d ap
proa
ch,
cont
rolli
ng
for
seas
onal
pa
ttern
s,
tem
p,
hum
idity
, in
fluen
za
epis
odes
, da
y of
w
eek.
A
djus
ted
fo
r re
sidu
al
seria
l co
rrel
atio
n an
d ov
er-d
ispe
rsio
n.
Res
pira
tory
adm
issi
ons
(all
ages
) w
ere
not
asso
ciat
ed w
ith a
ny p
ollu
tant
. A
naly
ses
by
age
reve
aled
so
me
asso
ciat
ions
to
P
M10
an
d P
M2.
5 a
nd r
espi
rato
ry a
dmis
sion
s in
the
0-
14
age
grou
p.
The
re
wa
s a
strik
ing
seas
onal
in
tera
ctio
n in
th
e co
ol
seas
on
vers
us
the
war
m
seas
on.
PM
10-2
.5
effe
cts
cann
ot b
e ex
clud
ed.
Tw
o po
lluta
nt m
odel
s ex
amin
ed
par
ticul
ate
mea
sure
s.
PM
2.5
effe
cts
redu
ced
by in
clus
ion
of b
lack
sm
oke.
Res
pira
tory
ho
spita
l ad
mis
sion
(la
g 0-
1);
PM
10 i
ncre
men
t 10
-90%
(1
1.4-
38.3
µg/
m3 ):
A
ll ag
es: 1
.5 (
-0.7
, 3.6
) A
ges
0-14
: 3.9
(0.
6, 7
.4)
Age
s 15
-64:
0.1
(-4
.0, 4
.4)
Age
s 65
: -1.
1 (-
4.3,
2.1
) P
M2.
5 (6
.0-2
5.8)
A
ll ag
es: 1
.2 (
-0.9
, 3.4
) A
ges
0-14
: 3.4
(-0
.1, 7
.0)
Age
s 15
-64:
-2.
1 (-
6.4,
2.4
) A
ges
65: -
1.3
(-4.
7, 2
.2)
PM
10-2
.5 (
4.1,
15.
2)
All
ages
: 0.2
(-2
.5, 3
.0)
Age
s 0-
14: 4
.4 (
-0.3
, 9.4
) A
ges
15-6
4: -
4.9
(-9.
9, 0
.4)
Age
s 65
: -1.
9 (-
6.0,
2.5
) C
OP
D (
ICD
9: 4
90-4
92, 4
94-4
96)
PM
10:
Age
65
+: -
1.8
(-6.
9, 3
.5)
PM
2.5:
Age
65
+: -
3.9
(-9.
0, 1
.6)
PM
10-2
.5: A
ge 6
5+: -
1.7
(-8.
9, 5
.3)
Ast
hma
(IC
D9
: 493
) (la
g 0-
1)
PM
10;
Age
s 0-
14: 8
.3 (
1.7,
15.
3)
PM
10;
Age
s 15
-64:
-2
.3
(-10
.0,
6.1)
P
M2.
5; A
ges
0-14
: 6.0
(-0
.9, 1
3.4)
P
M2.
5;A
ges
15-
64:
-8.4
(-
16.4
, 0.
3)
PM
10-2
.5;
Age
s 0-
14:
7.1
(-
2.1,
17
.2)
PM
10-2
.5;
Age
s 15
-64:
-10
.7 (
-19.
9,
-0.5
)
T
hom
pson
et
al
. (2
001)
(T
hom
pson
et a
l, 20
01)
Bel
fast
, N
orth
ern
Irel
and
1/1/
93 –
12
/31/
95.
PM
10 m
ean
(SD
) M
ay
– O
ctob
er
24.9
(1
3.7)
N
ovem
ber
– A
pril
31.9
(2
4.3)
The
ra
tes
of
acut
e as
thm
a ad
mis
sion
to
ch
ildre
n’s
emer
genc
y w
as s
tudi
ed i
n re
latio
n to
da
y-to
-day
flu
ctua
tion
of
PM
10
and
othe
r po
lluta
nts
usin
g G
LM P
oiss
on
regr
essi
on.
A w
eak
, bu
t si
gnifi
cant
ass
oci
atio
n be
twee
n P
M10
con
cent
ratio
n an
d as
thm
a em
erge
ncy-
de
part
men
t ad
mis
sion
s w
as
seen
. A
fter
adju
stin
g fo
r m
ultip
le
pollu
tant
s on
ly
the
benz
ene
leve
l w
as
inde
pend
ently
as
soci
ated
w
ith
asth
ma
emer
genc
y de
part
men
t ad
mis
sion
. B
enze
ne w
as h
ighl
y co
rrel
ated
to P
M10
, S
O2
and
NO
2 le
vels
.
—
report no. 4/11
130
Pre
scot
t et
al
. (1
998)
(P
resc
ott e
t al,
1998
) E
dinb
urgh
(1
0/19
92-
6/19
95).
P
opul
atio
n: 0
.45
MM
P
M10
mea
n.:2
0.7
PM
10 m
in/m
ax=
5/72
P
M10
90th
-10t
h%
: 20.
Poi
sson
lo
g-lin
ear
regr
essi
on
mod
els
use
d to
in
vest
igat
e re
latio
n of
dai
ly h
ospi
tal
adm
issi
ons
with
NO
2, O
3, C
O,
and
PM
10.
Adj
ustm
ents
wer
e
mad
e fo
r se
ason
al
and
day
of
wee
k,
daily
te
mpe
ratu
re
(usi
ng
8 du
mm
y va
riabl
es),
an
d w
ind
spee
d.
Sep
arat
e a
naly
ses
for
age<
65
year
s (m
ean
resp
irato
ry
hosp
ital
adm
issi
on:
3.4/
day)
and
age
>64
yea
rs (
mea
n re
spira
tory
ho
spita
l ad
mis
sion
: 8.
7/da
y),
and
for
subj
ects
w
ith m
ultip
le h
ospi
tal a
dmis
sion
s.
PM
10 w
as c
onsi
sten
tly p
osi
tivel
y as
soci
ate
d w
ith R
esp
irato
ry h
ospi
tal a
dm
issi
ons
in b
oth
ag
e gr
oups
, w
ith t
he g
reat
est
effe
ct s
ize
in
thos
e >
64,
espe
cial
ly
amo
ng t
hose
with
>4
hosp
ital
adm
issi
ons
duri
ng
1981
-199
5.
Wea
k si
gnifi
canc
es l
ikel
y co
ntrib
uted
to
by
low
pop
ulat
ion
siz
e.
Sin
gle
Pol
luta
nt M
odel
s P
M10
: 50
µg/
m3 , l
ag 1
-3
Res
pira
tory
h
ospi
tal
adm
issi
ons
(age
<65
): 1
.25%
(-1
2.8,
17.
5)
Res
pira
tory
h
ospi
tal
adm
issi
ons
(age
>64
): 5
.33%
(-9
.3, 2
2.3
) R
espi
rato
ry
hos
pita
l ad
mis
sion
s (a
ge>
64,
>4
hos
pita
l adm
issi
ons)
: 7.
93%
(-1
9.0,
43.
7).
SWED
EN
Hag
en
et
al.
(200
0)
(Hag
en e
t al,
2000
) D
ram
men
, S
wed
en
(11/
1994
-12/
1997
) P
opul
atio
n: 1
10,
000
PM
10 m
ean:
16.
8 P
M10
IQR
: 9.
8-2
0.9.
Exa
min
ed
PM
10,
SO
2,
NO
2,
VO
C’s
, an
d O
3 as
soci
atio
ns
with
re
spira
tory
ho
spita
l ad
mis
sion
s fr
om o
ne h
ospi
tal
(mea
n =
2.2
/day
).
Use
d P
oiss
on
GA
M c
ontr
ollin
g fo
r te
mpe
ratu
re
and
RH
(b
ut
not
thei
r in
tera
ctio
n),
lon
g-w
ave
and
seas
onal
ity,
day
of
wee
k,
holid
ays,
an
d in
fluen
za e
pid
emic
s.
As
a si
ngle
pol
luta
nt,
the
PM
10 e
ffect
was
of
sam
e or
der
of
mag
nitu
de
as
repo
rted
in
ot
her
stud
ies.
T
he
PM
10
asso
ciat
ion
decr
ease
d w
hen
othe
r po
lluta
nts
wer
e ad
ded
to t
he m
odel
. H
ow
ever
, th
e V
OC
’s
sho
wed
the
stro
nges
t ass
ocia
tions
.
Res
pira
tory
h
ospi
tal
adm
issi
ons
(all
ages
) fo
r IQ
R=
50 µ
g/m
3
-Sin
gle
Pol
luta
nt M
odel
: P
M10
(la
g 0)
: 18.
3% (
-4.2
, 46
) -T
wo
Pol
luta
nt M
odel
(w
ith O
3):
PM
10 (
lag
0): 1
8.3%
(-4
.2, 4
5.4
) -T
wo
Pol
luta
nt
Mod
el
(with
B
enze
ne):
P
M10
(la
g 0)
: 6.5
% (
-14
, 31.
8).
NO
RW
AY
O
fteda
l et
al
. (2
003)
(O
fteda
l et a
l) D
ram
men
, N
orw
ay,
1995
-20
00.
Pop
ulat
ion:
11
0,00
0 P
M10
mea
n (S
D):
to
tal p
erio
d: 1
6.6
(10.
2).
1995
-199
7: 1
6.8
(10.
2).
1998
-200
0: 1
6.5
(10.
3).
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
PM
10,
and
othe
r co
-pol
luta
nts
(incl
udin
g
SO
2,
NO
2 an
d
O3)
an
d re
spira
tory
h
ospi
tal
adm
issi
ons
in D
ram
men
(m
ean
= 2
.3/d
ay).
T
ime-
serie
s an
alys
is o
f co
unts
was
per
form
ed
by m
eans
of
GA
M w
ith l
og l
ink
and
Poi
sson
di
strib
utio
n,
cont
rolli
ng
for
time
tren
ds.
Lag
0 w
as
appa
rent
ly u
sed.
PM
10 d
id n
ot s
igni
fican
tly i
ncr
ease
hos
pita
l ad
mis
sion
for
res
pira
tory
dis
ease
s. A
dire
ct
asso
ciat
ion
was
fou
nd f
or N
O2
and
SO
2. N
o as
soci
atio
n w
ith O
3. D
ivid
ing
the
tota
l st
udy
perio
d in
to t
wo
3-ye
ar p
erio
ds,
anal
yses
for
th
e se
cond
tim
e pe
riod
show
ed
wea
ker
asso
ciat
ion
bet
we
en
thes
e
pollu
tant
s an
d th
e he
alth
out
com
e.
Per
cent
exc
ess
ris
k of
res
pira
tory
ho
spita
l ad
mis
sion
s fo
r 11
.04
µg/
m3 (
IQR
):
Sin
gle
Po
lluta
nt M
odel
; P
M10
(la
g 0)
T
otal
per
iod:
2.1
% (
-1.0
, 5.3
) 19
94-1
997:
3.5
% (
-1.0
, 8.3
) 19
98-2
000:
-0.
8% (
-5.2
, 3.7
) T
wo
Pol
luta
nt
Mod
el
(with
be
nzen
e): 1
.0%
(-2
.2, 4
.3)
Tw
o P
ollu
tant
Mod
el (
with
NO
2):
-0.3
% (
-4.1
, 3.7
).
report no. 4/11
131
THE
NET
HER
LAN
DS
S
chou
ten
et
al.
(199
6)
(Sch
oute
n et
al,
1996
) A
mst
erda
m/R
otte
rdam
, 19
77-1
989.
A
mst
erda
m
Pop
.: 0.
69
MM
R
otte
rdam
Po
p.: 0
.58
MM
A
mst
erda
m:
BS
mea
n.:1
1
BS
5th
-95t
h%
: 1
- 37
R
otte
rdam
: B
S m
ean.
:26
B
S 5
th-9
5th %
= 6
-61.
Dai
ly
emer
gen
cy
hosp
ital
adm
issi
on
for
resp
irato
ry
dise
ases
(I
CD
9:
460-
519)
, C
OP
D
(490
-492
, 49
4,
496)
, an
d as
thm
a (4
93).
T
he
mea
n H
A/d
(ra
nge)
for
the
se w
ere:
6.7
0 (0
-23)
, 1.
74
(0-9
) an
d 1.
13
(0-7
) re
spec
tivel
y in
A
mst
erda
m
and
4.79
(0
-19)
, 1.
57
(0-9
),
and
0.53
(0
-5)
in
Rot
terd
am.
Hos
pita
l ad
mis
sion
as
soci
atio
ns w
ith B
S,
O3,
NO
2, a
nd S
O2
wer
e an
alyz
ed,
usin
g au
tore
gres
sive
P
oiss
on
GLM
re
gres
sion
al
low
ing
for
ove
rdis
pers
ion
an
d co
ntro
lling
fo
r se
ason
, da
y of
w
eek,
m
eteo
rolo
gica
l fac
tors
, and
influ
enza
epi
dem
ics.
BS
did
not
sho
w a
ny c
onsi
sten
t ef
fect
s in
A
mst
erda
m;
but
in
Rot
terd
am
BS
w
as
posi
tivel
y re
late
d to
ad
mis
sion
s.
Mos
t co
nsis
tent
BS
as
soci
atio
ns
in a
dults
>64
ye
ars.
in
win
ter.
Pos
itive
O3
asso
ciat
ion
in
sum
mer
in
peop
le a
ged
>6
4 in
Am
ster
dam
an
d R
otte
rda
m.
SO
2 a
nd
NO
2 di
d no
t sh
ow
any
clea
r ef
fect
s. R
esul
ts d
id n
ot c
hang
e in
po
lluta
nt i
nter
actio
n an
alys
es.
The
aut
hors
co
nclu
ded
that
sh
ort-
term
ai
r po
llutio
n-
emer
genc
y ad
mis
sion
as
soci
atio
n is
n
ot
alw
ays
cons
iste
nt a
t th
ese
indi
vidu
al c
itie
s’
rela
tivel
y lo
w
coun
ts
of
daily
ho
spita
l ad
mis
sion
s a
nd l
ow l
eve
ls o
f ai
r po
llutio
n.
Ana
lyse
s fo
r al
l ag
es o
f al
l th
e N
ethe
rland
s ga
ve
a st
rong
B
S-h
ospi
tal
adm
issi
on
asso
ciat
ion
in w
inte
r.
Sin
gle
Pol
luta
nt M
odel
s F
or B
S=
25 µ
g/m
3 , lag
2;
For
all
of th
e N
ethe
rland
s:
Res
pira
tory
h
ospi
tal
adm
issi
ons
(all
ages
):
Win
ter:
2.0
% (
-1.5
, 5.7
) S
umm
er: 2
.4%
(0.
6, 4
.3).
GER
MA
NY
W
ichm
ann
(2
004)
(W
ichm
ann,
200
4)
Rhi
ne-R
uhr
area
, G
erm
any,
1
985,
si
ngle
ep
isod
e. T
SP
dur
ing
the
epis
ode
in
Rhi
ne-R
uhr
area
: ~ 6
00.
In
Janu
ary
1985
an
ex
ten
ded
smog
ep
isod
e oc
curr
ed i
n C
entr
al E
urop
e. T
his
pape
r re
vie
ws
the
heal
th
effe
cts,
in
clud
ing
resp
irato
ry
adm
issi
on,
inve
stig
ated
dur
ing
the
smog
per
iod
and
a co
ntro
l pe
riod
befo
re a
nd a
fter
the
smog
ep
isod
e.
The
Rhi
ne-R
uhr
are
a (W
este
rn G
erm
any)
was
af
fect
ed f
or 5
day
s w
ith m
axi
mum
con
cent
ratio
n of
T
SP
of
60
0 µ
g/m
3 (2
4h
aver
age
s).
In
Aug
sbur
g (S
outh
ern
Ger
man
y)
the
smog
ep
isod
e w
as l
ess
seve
re (
100
µg/
m3 T
SP
, 24
h
aver
ages
).
Rhi
ne-R
uhr
hos
pita
l ad
mis
sion
s in
crea
sed
by 3
9% f
or c
hron
ic b
ronc
hitis
, by
7%
(no
t si
gnifi
cant
) fo
r re
spira
tory
di
seas
es,
and
de
crea
sed
by
14%
fo
r as
thm
a (n
ot
sign
ifica
nt).
No
risk
sho
wn
.
report no. 4/11
132
FRA
NC
E
Dab
et
al.
(199
6) (
Dab
et
al, 1
996)
P
aris
, F
ran
ce
(198
7-19
92)
Pop
ulat
ion:
6.1
MM
P
M13
mea
n: 5
0.8
PM
13 5
th-9
5th r
ange
: 19
.0-
137.
3 B
S m
ean:
31.
9 B
S
5th-9
5th
Ran
ge:1
1.0-
123.
3.
Dai
ly m
orta
lity
and
gene
ral
adm
issi
ons
to P
aris
pu
blic
ho
spita
ls
for
resp
irato
ry
caus
es
wer
e
cons
ider
ed
(mea
ns:
all
resp
irato
ry=
79/d
ay,
asth
ma=
14/d
ay,
C
OP
D=
12/d
ay).
T
ime
serie
s an
alys
is u
sed
lin
ear
regr
essi
on m
odel
fo
llow
ed
by a
Po
isso
n re
gres
sion
adj
uste
d fo
r ep
idem
ics
of in
fluen
za A
and
B,
tem
pera
ture
, RH
, ho
liday
s,
day
of w
eek,
tre
nd,
long
-wa
ve v
aria
bilit
y, a
nd
nurs
es’
strik
e.
No
two
po
lluta
nt
mod
els
cons
ider
ed.
For
the
all
resp
irato
ry c
ause
s ca
tego
ry,
the
auth
ors
foun
d “t
he
stro
nge
st
asso
ciat
ion
wa
s ob
serv
ed w
ith P
M13
” fo
r bo
th h
ospi
tal
adm
issi
ons
and
mor
talit
y,
indi
catin
g
a co
here
nce
of a
ssoc
iatio
n ac
ross
out
com
es.
Ast
hma
was
si
gnifi
cant
ly
corr
elat
ed
with
N
O2
leve
ls, b
ut n
ot P
M13
.
Fo
r P
M13
: 5
0 µ
g/m
3 ;
BS
: 25
µ
g/m
3 ; R
espi
rato
ry a
dmis
sion
s (a
ll ag
es)
PM
13 L
ag 0
: 2.2
% (
0.2,
4.3
) B
S L
ag 0
: 1.0
% (
0.2,
1.8
) C
OP
D a
dmis
sion
s (a
ll ag
es)
PM
13 L
ag 2
: 2.3
% (
-6.7
, 2.2
) B
S L
ag 2
: 1.1
% (
-2.9
, 0.6
) A
sthm
a ad
mis
sion
s (a
ll ag
es):
P
M13
Lag
2: 1
.3%
(-4
.6, 2
.2)
BS
Lag
0: 1
.2%
(-0
.5, 2
.9)
Eils
tein
et
al
(2
004)
(E
ilste
in e
t al,
2004
) N
ine
Fre
nch
citie
s, 1
990-
1997
. N
o da
ta
dist
ribut
ion
was
re
port
ed.
Thi
s st
udy
exa
min
ed t
he s
hort
-ter
m e
ffect
s of
P
M10
, B
S
an
d ot
her
co-p
ollu
tant
s (C
O,
NO
2,
SO
2,
O3)
on
m
orta
lity
and
hosp
italis
atio
n fo
r ca
rdio
vasc
ular
or
resp
irato
ry d
isor
ders
in
nine
F
renc
h ci
ties
(Bor
deau
x, L
e H
avre
, Li
lle,
Lyon
, M
arse
ille,
P
aris
, R
ouen
, S
tras
bour
g
and
Tou
lous
e)
of
the
Sur
veill
ance
A
ir et
S
ante
pr
ogra
m.
The
an
alys
is
asse
ssed
th
e ci
ty-
spec
ific
and
com
bine
d as
soci
atio
ns w
ith v
ario
us
pollu
tant
s us
ing
Poi
sson
GA
M,
afte
r ad
just
men
t fo
r tr
ends
in
seas
ons,
day
of
wee
k, h
olid
ays,
in
fluen
za e
pid
emic
s, t
empe
ratu
re a
nd h
um
idity
. A
com
bine
d R
R w
as c
alcu
late
d fo
r al
l the
citi
es.
Lag
0-1
and
Lag
0-5
wer
e u
sed.
At
sele
cted
la
gs,
som
e di
rect
ass
ocia
tion
s w
ere
fou
nd b
etw
een
both
PM
10 a
nd B
S,
and
ad
mis
sion
fo
r re
spira
tory
di
seas
es,
part
icul
arly
in
ch
ildre
n ag
ed
0-14
ye
ars.
S
tron
ger
asso
ciat
ions
wer
e f
ound
for
SO
2,
NO
2 an
d pa
rtic
ular
ly fo
r C
O.
Effe
cts
com
put
ed
for
10
µg/
m3
chan
ge in
exp
ositi
on:
Lag
0-1
PM
10; a
ge 1
5-64
: 0.0
NS
P
M10
; age
65
+: 0
.8 N
S
Lag
0-5
PM
10; a
ge 1
5-64
: 1.7
p<
0.05
P
M10
; age
65
+: 1
.4 p
<0.
05
Lag
0-1
BS
; age
15-
64: 1
.8 p
<0.
05
BS
; age
65+
: 0.
0 N
S
Lag
0-5
BS
; age
15-
64: 4
.7 p
<0.
05
BS
; age
65+
: 0.
0 N
S.
report no. 4/11
133
ITA
LY
Big
geri
et
al.
(200
5)
(Big
geri
et a
l, 20
05)
Eig
ht
citie
s of
N
orth
, C
entr
e an
d S
outh
of
Italy
, 19
90-1
999.
P
M10
m
ean
da
ily
conc
entr
atio
n:
Tur
in, 9
1-94
: 77.
6 T
urin
, 95-
98: 6
3.8
Mila
n, 9
0-94
: 61.
8 M
ilan,
95-
97: 4
5.2
Ver
ona,
95-
99: 3
6.5
Rav
enna
, 91-
95: 5
9.1
Bol
ogna
, 96-
98:
41.
2 F
lore
nce,
96-
98: 4
0.3
Rom
e, 9
2-94
: 69.
7 P
aler
mo,
97-
99: 4
2.9.
Met
a-an
alys
is
of
shor
t-te
rm
effe
cts
of
air
pollu
tion
on m
orta
lity
(all
caus
es,
card
iova
scul
ar
caus
es
and
resp
irato
ry
caus
es)
and
hosp
ital
adm
issi
on
(car
diac
ca
use
s,
and
resp
irato
ry
caus
es –
IC
D9:
460
-519
) in
eig
ht I
talia
n c
ities
fr
om
1990
to
19
99.
Dai
ly
conc
entr
atio
ns
of
pollu
tant
s w
ere
colle
cted
. A
gen
eral
ized
lin
ear
mod
el
adju
sted
fo
r ag
e,
day
of
the
w
eek,
ho
liday
s,
influ
enza
ep
ide
mic
s,
met
eoro
logi
cal
varia
bles
, an
d s
easo
nalit
y p
atte
rn w
as
fitte
d fo
r P
M10
and
oth
er
co-p
ollu
tant
(S
O2,
NO
2, C
O,
and
O3)
. Lag
0-3
was
use
d fo
r ho
spita
l adm
issi
on.
A s
igni
fican
t di
rect
ass
ocia
tion
was
fou
nd
betw
een
PM
10
and
resp
irato
ry
hosp
ital
adm
issi
on.
NO
2 an
d C
O s
how
ed s
tron
ger
as
soci
atio
ns.
Effe
cts
com
put
ed
for
10
µg/
m3
chan
ge in
PM
10 (
lag
0-3
avg
.):
Fix
ed-e
ffect
s: 0
.73%
(0.
27,
1.20
) R
ando
m-e
ffect
s:
0.91
%
(-0.
04,
1.86
).
Vig
otti
et
al.
(199
6)
(Vig
otti
et a
l, 1
996)
M
ilan,
Ialy
, 19
80-1
989
Pop
ulat
ion:
1.5
MM
T
SP
mea
n: 1
39.
0 T
SP
IQR
: 82.
0, 1
75.7
.
Ass
ocia
tion
betw
een
ad
ult
resp
irato
ry
(15-
64
year
s m
ean
=11
.3/d
ay,
and
65 +
yea
rs m
ean
=8.
8/da
y) a
nd a
ir po
llutio
n e
valu
ated
, us
ing
the
AP
HE
A p
roto
col.
Poi
sson
reg
ress
ion
wa
s us
ed
afte
r ad
just
men
t fo
r w
eath
er
and
long
te
rm
tren
d, y
ear,
influ
enza
epi
dem
ics,
and
sea
son.
Incr
ease
d ris
k of
re
spira
tory
ho
spita
l ad
mis
sion
w
as
asso
ciat
ed
with
bo
th
SO
2 an
d T
SP
. T
he r
elat
ive
risks
wer
e si
mila
r fo
r bo
th p
ollu
tant
s. T
here
wa
s no
mod
ifica
tion
of t
he T
SP
eff
ect
by S
O2
leve
l. T
here
was
a
sugg
estio
n o
f a
high
er
TS
P
effe
ct
on
hosp
ital a
dmis
sion
s in
the
cool
mon
ths.
You
ng
Adu
lt (1
5-64
ye
ars.
) R
espi
rato
ry
adm
issi
on
per
100
µg/
m3 in
crea
se in
TS
P:
Lag
2: 5
% (
CI:
0, 1
0)
Old
er
Adu
lt (6
5+
year
s.)
Res
pira
tory
ad
mis
sion
pe
r 10
0 µ
g/m
3 incr
ease
in T
SP
: La
g 1
ER
: 5%
(C
I: -1
, 10)
.
F
usco
et
al
. (2
001)
(F
usco
et a
l, 20
01)
Rom
e, It
aly,
199
5-19
97.
PM
– s
uspe
nded
par
ticle
s m
easu
red.
Dai
ly
cou
nts
of
hosp
ital
adm
issi
ons
for
tota
l re
spira
tory
co
nditi
ons,
ac
ute
resp
irato
ry
infe
ctio
n in
clud
ing
pneu
mon
ia,
CO
PD
, an
d as
thm
a w
as
anal
yzed
in
re
latio
n to
P
M
mea
sure
s a
nd
gase
ous
pollu
tant
s us
ing
gene
raliz
ed a
dditi
ve G
AM
mod
els
cont
rolli
ng f
or
mea
n te
mpe
ratu
re,
influ
enza
, ep
ider
mic
s, a
nd
othe
r fa
ctor
s us
ing
splin
e sm
ooth
s.
No
effe
ct
wa
s fo
und
for
PM
. T
otal
re
spira
tory
ad
mis
sion
w
ere
si
gnifi
cant
ly
asso
ciat
ed w
ith s
ame-
day
leve
l of
NO
2 a
nd
CO
. T
here
wa
s no
indi
catio
n t
hat
the
effe
cts
of a
ir po
llutio
n w
ere
pres
ent
at la
gs >
2 d
ays.
A
mon
g ch
ildre
n,
tota
l re
spira
tory
an
d as
thm
a ad
mis
sion
s w
ere
stro
ngly
as
soci
ated
with
NO
2 an
d C
O.
Mul
tipol
luta
nt
mod
el
anal
ysis
yi
elde
d w
eak
er
and
mo
re
unst
able
res
ults
.
—
report no. 4/11
134
San
seba
stia
no
et
al.
(200
3)
(San
seba
stia
no
et
al,
2003
) P
arm
a, It
aly,
199
2-20
01.
Mea
n da
ily c
onc
entr
atio
n:
PM
10
(199
8-2
001)
: fr
om
39,6
3 in
199
8 t
o 57
.68
in
2001
; T
SP
: fr
om 4
6.92
in
1996
to 6
2.39
in 1
992.
Thi
s st
udy
exam
ined
the
rel
atio
nshi
p be
twee
n T
SP
an
d ot
her
co
-pol
luta
nts
(CO
, N
O2,
an
d S
O2)
, an
d da
ily h
ospi
tal
adm
issi
on f
or s
peci
fic
resp
irato
ry
dise
ases
(in
clud
ing
B
PC
O,
polm
oniti
s,
ple
uriti
s,
asth
ma
and
resp
irato
ry
sym
ptom
s)
in
the
city
of
P
arm
a (p
opul
atio
n 17
0,00
0).
The
exc
ess
risk
of
hosp
ital
adm
issi
on
wa
s no
t an
alys
ed f
or P
M10
sin
ce i
nfor
mat
ion
for
PM
10 w
as a
vaila
ble
for
the
perio
d 19
98-
2001
, on
ly.
A
dire
ct
ass
ocia
tion
has
been
sh
ow
n
betw
een
TS
P,
NO
2 an
d S
O2,
and
adm
issi
on
for
sele
cted
res
pira
tory
dis
eas
es.
Effe
cts
com
put
ed
for
1 µ
g/m
3 in
crem
ent o
f TS
P:
0.14
% (
0.04
, 0.
23).
Mig
liare
tti
et
al.
(200
4)
(Mig
liare
tti
andC
aval
lo,
2004
)
Tur
in,
Italy
, 1/
1997
-12
/199
9.
TS
P l
evel
s fo
r th
e pe
riod
sho
wn
in o
ne f
igur
e.
Thi
s st
udy
exam
ined
the
rel
atio
nshi
p be
twee
n T
SP
and
NO
2, a
nd h
ospi
talis
atio
n fo
r as
thm
a (I
CD
9: 4
93)
sepa
rate
ly f
or c
hild
ren
(<15
yea
rs)
adul
ts (
15-6
4 ye
ars)
and
eld
erly
(65
+ y
ear
s) i
n th
e ci
ty o
f T
urin
usi
ng a
cas
e-co
ntro
l de
sign
. O
vera
ll,
1,40
1 pa
tient
s re
side
nt
in
Tur
in
and
adm
itted
fo
r as
thm
a w
ere
de
fined
as
ca
ses;
20
1,07
1 pa
tient
s ad
mitt
ed f
or c
ause
s ot
her
tha
n re
spira
tory
di
seas
es
or
hear
t di
seas
es
wer
e de
fined
as
cont
rols
. P
erce
nt e
xces
s ris
ks w
ere
de
rived
by
mul
tiple
log
istic
reg
ress
ion
mod
els
adju
sted
fo
r se
x,
age,
ed
ucat
ion,
se
ason
ality
, te
mpe
ratu
re,
hum
idity
, so
lar
radi
atio
n,
day
of
wee
k an
d ho
liday
.
The
num
ber
of e
mer
genc
y ad
mis
sion
s fo
r re
spira
tory
ca
uses
si
gnifi
cant
ly
rose
w
ith
incr
ease
d e
xpos
ure
to
TS
P
and
NO
2,
part
icul
arly
am
ong
youn
g an
d el
derly
pa
tient
s fo
r bo
th
pollu
tant
s.
Afte
r m
utua
l ad
just
men
t fo
r bo
th
pollu
tant
s,
the
asso
ciat
ions
bec
ame
wea
ker
and
no m
ore
sign
ifica
nt.
Effe
cts
com
put
ed
for
10
µg/
m3
incr
emen
t of T
SP
: S
ingl
e-po
lluta
nt a
naly
sis:
T
otal
: 2.3
% (
1.1
, 3.6
) A
ge <
15: 1
.9%
(0.
4, 3
.4)
Age
15-
64: 2
.3%
NS
A
ge 6
5+: 8
.3%
(2.
9, 1
3.7)
T
wo-
pollu
tant
s an
alys
is
Tot
al: 0
.9%
NS
A
ge <
15: -
0.1
% N
S
Age
15-
64: 1
.2%
NS
A
ge 6
5+: 0
.9%
NS
.
Mig
liare
tti
et
al.
(200
5)
(Mig
liare
tti e
t al,
2005
)
Tur
in,
Italy
, 1/
1997
-12
/199
9.
Sam
e st
udy
as
abov
e,
incl
udin
g ad
ults
. T
SP
lev
els
for
the
perio
d sh
ow
n in
one
fig
ure.
Thi
s st
udy
exam
ined
the
rel
atio
nshi
p be
twee
n T
SP
and
NO
2, a
nd h
ospi
talis
atio
n fo
r as
thm
a (I
CD
9: 4
93)
in c
hild
ren
aged
<15
yea
rs i
n th
e ci
ty o
f T
urin
usi
ng a
cas
e-co
ntro
l des
ign.
On
the
basi
s of
the
prim
ary
diag
nos
is,
1,06
0 p
edia
tric
pa
tient
s re
sid
ent
in
Tur
in
and
adm
itted
fo
r as
thm
a w
ere
de
fined
as
ca
ses;
25
,52
3 ag
e-m
atch
ed p
atie
nts
adm
itted
for
cau
ses
oth
er th
an
resp
irato
ry
dise
ases
or
he
art
dise
ases
w
ere
defin
ed a
s co
ntro
ls.
Per
cent
exc
ess
risks
wer
e
deriv
ed b
y m
ultip
le l
ogis
tic r
egre
ssio
n m
odel
s af
ter
adju
stm
ent
for
1)
sex
and
age
, an
d 2
) af
ter
furt
her
allo
wa
nce
for
seas
onal
ity,
tem
pera
ture
, hu
mid
ity,
sola
r ra
diat
ion,
da
y of
w
eek
an
d ho
liday
. Lag
0-3
was
ap
pare
ntly
use
d.
The
num
ber
of e
mer
genc
y ad
mis
sion
s fo
r re
spira
tory
ca
uses
si
gnifi
cant
ly
rose
w
ith
incr
ease
d e
xpos
ure
to
TS
P
and
NO
2.
Con
side
ring
the
mul
tivar
iate
m
odel
, af
ter
mut
ual
adju
stm
ent
for
both
po
lluta
nts,
no
si
gnifi
cant
ass
ocia
tion
was
evi
dent
.
Effe
cts
com
put
ed
for
10
µg/
m3
incr
emen
t of T
SP
: S
ingl
e-po
lluta
nt a
naly
sis:
A
ge-s
ex
adju
sted
mod
el:
Tot
al: 2
.8%
(1
.7, 4
.0)
Age
<4:
2.3
% (
0.2,
3.4
) A
ge 4
-15:
3.1
% (
0.1,
4.8
) M
ultiv
aria
te m
odel
: T
otal
: 1.8
% (
0.3
, 3.2
) A
ge <
4: 1
.8%
(0.
0, 3
.5)
Age
4-1
5: 3
.0%
(0.
1, 5
.8)
Tw
o-po
lluta
nts
anal
ysis
A
ge-s
ex
adju
sted
mod
el:
Tot
al: 1
.4%
(1
.1, 4
.0)
Mul
tivar
iate
mod
el:
Tot
al: 0
.6%
(-0
.2, 3
.0)
report no. 4/11
135
SPA
IN
Ten
ías
et
al
(199
8)
(Ten
ias
et a
l, 19
98)
Stu
dy P
erio
d.:
94 -
95
Val
enci
a, S
pain
H
osp.
C
achm
ent
Pop
.:200
,000
B
S m
ean:
57.
7 B
S IQ
R: 2
5.6-
47.7
Ass
ocia
tions
be
twee
n ad
ult
(14+
ye
ars)
em
erge
ncy
asth
ma
emer
genc
y vi
sits
to
one
city
ho
spita
l (m
ean
=1.
0/da
y) a
nd B
S,
NO
2, O
3, S
O2
anal
yzed
us
ing
GLM
P
oiss
on
auto
-reg
ress
ive
mod
elin
g, a
fter
adju
stm
ent
for
wea
ther
se
ason
an
d tr
ends
usi
ng th
e A
PH
EA
pro
toco
l.
Ass
ocia
tion
with
ast
hma
was
pos
itive
and
m
ore
cons
iste
nt f
or N
O2
and
O3
than
for
BS
or
SO
2. T
he s
tudy
sug
gest
s th
at s
econ
dary
ox
idat
ive-
envi
ronm
ent
pollu
tant
s m
ay
be
mor
e as
thm
a re
leva
nt
than
pr
imar
y re
duct
ion-
envi
ronm
ent
pollu
tant
s (e
.g.,
carb
onac
eous
par
ticle
s).
NO
2 ha
d gr
eate
st
effe
ct o
n B
S i
n co
-pol
luta
nt m
odel
s, b
ut B
S
beca
me
sign
ifica
nt o
nce
1993
was
add
ed,
sh
ow
ing
pow
er
to
be
a lim
itatio
n of
th
is
stud
y.
Adu
lt A
sthm
a, B
S:
25
µg/
m3 ;
Lag
0:
For
199
3-19
95: 1
0.6%
(0.
9, 2
1.1)
F
or 1
994-
1995
: 6.4
% (
-4.8
, 18.
8).
Gal
an e
t al
(2
003)
(G
alan
et
al,
2003
) M
adrid
, S
pain
(1
995-
1998
).
Hos
pita
l C
atch
men
t P
opul
atio
n:55
5,00
0 P
M10
mea
n: 3
2.1.
Thi
s st
udy
show
s th
e as
soci
atio
ns
betw
een
emer
genc
y ro
om a
dmis
sion
for
ast
hma
to o
ne
city
ho
spita
l (m
ean
=3.
3/da
y)
in
Mad
rid,
and
PM
10 a
nd o
the
r co
-pol
luta
nts,
incl
udin
g N
O2,
O3,
an
d S
O2.
T
wo
anal
ytic
al
stra
tegi
es
wer
e
purs
ued:
1)
auto
regr
essi
ve P
oiss
on m
ode
ls a
nd
2)
GA
M
mod
els.
B
oth
the
mod
els
incl
uded
te
rms
for
lon
g-te
rm t
rend
s, s
easo
nalit
y, d
ay o
f w
eek,
hol
iday
, te
mpe
ratu
re,
hum
idity
, in
fluen
za
and
acut
e re
spira
tory
infe
ctio
ns.
Fur
ther
mod
els
incl
uded
var
ious
pol
lens
. La
g 0,
1,
2, 3
, 4
wer
e
cons
ider
ed.
Rel
ativ
e ris
ks o
f a
sthm
a em
erge
ncy
room
ad
mis
sion
s fo
r a
rise
in p
ollu
tant
leve
ls w
ere
syst
emat
ical
ly,
alth
ough
no
t al
way
s si
gnifi
cant
ly,
over
un
ity
for
PM
10
and
the
othe
r po
lluta
nts
, fo
r ea
ch
lag
used
, al
so
afte
r fu
rthe
r al
low
ance
for
var
ious
diff
ere
nt
type
rs o
f po
llen.
The
str
onge
st a
ssoc
iatio
ns
wer
e ob
serv
ed
at 1
day
lag
for
O3
and
3 da
ys
lag
for
the
rem
aini
ng
pollu
tan
ts.
Fur
ther
adj
ustm
ent
for
the
diffe
rent
typ
es o
f po
llen
did
not
sen
sibl
y m
odify
the
est
imat
es.
S
light
diff
eren
ces
wer
e ob
serv
ed b
etw
een
the
estim
ates
ob
tain
ed
with
pa
ram
etric
m
odel
ling
and
thos
e ob
tain
ed
with
G
AM
m
odel
s.
Effe
cts
on a
dm
issi
on f
or a
sthm
a co
mpu
ted
for
10 µ
g/m
3 inc
rem
ent
of P
M10
: La
g 0:
1.1
% (
-2.0
, 4.2
) La
g 1:
0.6
% (
-2.4
, 3.7
) La
g 2:
0.8
% (
-2.2
, 3.8
) La
g 3:
3.9
% (
1.0,
6.8
) La
g 4:
2.7
% (
-0.1
, 5.6
) A
fter
furt
her
adju
stm
ent
for
four
ty
pes
of p
olle
n;
Bes
t lag
(3)
: 4.5
% (
1.6,
7.4
) G
AM
mod
el; 5
.8 (
2.5,
9.1
).
GR
EEC
E
Kon
tos
et
al.
(199
9)
(Kon
tos
et a
l, 19
99)
Pira
eus,
A
the
ns
Gre
ece,
19
87-1
992)
B
S m
ean:
46.5
B
S m
ax:
200.
Rel
atio
n of
re
spira
tory
ho
spita
l ad
mis
sion
fo
r ch
ildre
n (0
-14
year
s.)
(mea
n =
4.3
/day
) to
BS
, S
O2,
N
O2,
an
d O
3 ev
alua
ted
usi
ng
a
nonp
aram
etric
st
ocha
stic
dy
nam
ical
sy
stem
ap
proa
ch a
nd f
requ
ency
do
mai
n an
alys
es.
Long
w
ave
an
d ef
fect
s of
w
eath
er
cons
ider
ed,
but
non-
linea
rity
and
inte
ract
ions
of
te
mp
erat
ure
and
RH
rel
atio
n w
ith H
A’s
no
t add
ress
ed.
Pol
lutio
n w
as
foun
d to
ex
plai
n si
gnifi
cant
po
rtio
n of
th
e H
A
varia
nce.
O
f po
lluta
nts
cons
ider
ed,
BS
w
as
con
sist
ently
am
ong
m
ost
stro
ngly
exp
lana
tory
pol
luta
nts
acro
ss
vario
us r
epor
ted
anal
yses
.
Ris
ks n
ot r
epo
rted
report no. 4/11
136
TAB
LE IV
. AC
UTE
PA
RTI
CU
LATE
MA
TTER
EXP
OSU
RE
AN
D R
ESPI
RA
TOR
Y M
EDIC
AL
VISI
TS (D
eriv
ed fr
om E
PA T
AB
LE 8
B-3
). Ref
eren
ce, L
ocat
ion,
Ye
ars,
PM
Inde
x, M
ean
or M
edia
n (µ
g/m
3 ).
Stud
y D
escr
iptio
n. M
ode
ling
met
hods
: lag
s,
smoo
thin
g, c
o-po
lluta
nts
and
cova
riate
s.
Res
ults
and
Com
men
ts.
PM In
dex,
lag,
per
cent
exc
ess
risk%
(95%
LC
L, U
CL)
Subn
atio
nal s
tudi
es
G
REA
T B
RIT
AIN
Atk
inso
n et
al
. (1
999b
) (A
tkin
son
et a
l, 19
99b)
Lo
ndon
(92
- 9
4)
Pop
ulat
ion:
no
t rep
orte
d P
M10
Mea
n: 2
8.5
10th
-90t
h IQ
R: 1
5.8-
46.5
B
S m
ean:
12.7
10
th-9
0th
IQR
: 5.5
-21.
6.
All-
age
resp
irato
ry
(mea
n=90
/day
),
asth
ma
(25.
9/da
y),
and
othe
r re
spira
tory
(6
4.1/
day)
em
erge
ncy
doc
tor
(ED
) vi
sits
fro
m 1
2 Lo
ndon
ho
spita
ls
we
re
cons
ider
ed,
but
asso
ciat
ed
popu
latio
n si
ze
wa
s no
t re
port
ed.
Cou
nts
for
ages
0-1
4, 1
5-64
, an
d >
64 w
ere
also
exa
min
ed.
Poi
sson
GLM
reg
ress
ion
was
use
d, c
ontr
ollin
g fo
r se
ason
, da
y of
w
eek,
m
ete
orol
ogy,
au
toco
rrel
atio
n,
over
disp
ersi
on,
and
in
fluen
za
epid
emic
s.
PM
10 w
as p
ositi
vely
ass
ocia
ted,
but
not
BS
, fo
r al
l- ag
e/al
l-res
pira
tory
ca
tego
ry.
PM
10
resu
lts
wer
e dr
iven
by
sig
nific
ant
child
ren
and
youn
g ad
ult
asso
ciat
ions
, w
hile
old
er
adul
t vi
sits
ha
d ne
gativ
e (b
ut
non-
sign
ifica
nt)
PM
10-E
D v
isit
rela
tions
hip.
PM
10
wa
s po
sitiv
ely
asso
ciat
ed
for
all
age
s,
child
ren,
and
you
ng a
dults
for
ast
hma
ED
vi
sits
. H
ow
eve
r,
PM
10-a
sth
ma
rela
tions
hip
coul
d no
t be
sep
arat
ed f
rom
SO
2 in
mul
ti-
pollu
tant
reg
ress
ions
. O
lde
r ad
ult
ED
vis
its
mos
t st
rong
ly a
ssoc
iate
d w
ith C
O.
No
O3-
ED
vis
its r
ela
tions
hips
was
fou
nd (
but
no
war
m s
easo
n an
alys
es a
ttem
pted
).
PM
10 (
50 µ
g/m
3 ) N
o co
-pol
luta
nt:
All
Re
spira
tory
ED
vis
its
All
age
(lag
1):
4.9%
(1.
3, 8
.6)
<15
(la
g 2)
: 6.4
% (
1, 1
2.2)
15
-64
(lag
1): 8
.6%
(3.
4, 1
4)
Ast
hma
ED
vis
its
All
age
(lag
1):
8.9%
(3,
15.
2)
<15
(la
g 2)
: 12.
3% (
3.4,
22)
15
-64
(lag
1): 1
3% (
4.6,
22.
1)
PM
10
(50
µg
/m3 );
la
g 2
&
co-
pollu
tant
; ch
ildre
n’s
PM
al
one:
12
.3%
(3.
4, 2
2)
& N
O2:
7.8
% (
-1.2
, 17.
6)
& O
3: 1
0.5
% (
1.6,
20.
1)
& S
O2:
8.1
% (
-1.1
, 18.
2)
& C
O: 1
2.1
% (
3.2,
21.
7).
Haj
at e
t al
. (1
999)
(H
ajat
et
al,
1999
) Lo
ndon
, E
ngla
nd,
1992
-19
94.
Pop
ulat
ion:
28
2,00
0 P
M10
mea
n: 2
8.2
PM
10 1
0-90
th%
=16
.3-4
6.4
B
S m
ean:
10.
1 B
S 1
0t-9
0th%
=4.
5-15
.9
Thi
s st
udy
exa
min
ed a
ssoc
iatio
ns o
f P
M1
0, B
S,
NO
2,
O3,
S
O2,
an
d C
O,
with
pr
imar
y ca
re
gene
ral
prac
titio
ner
asth
ma
and
othe
r lo
wer
re
spira
tory
di
seas
es
(LR
D)
cons
ulta
tions
. A
sthm
a co
nsu
ltatio
n m
eans
per
day
= 3
5.3
(all
ages
); 1
4.(0
-14
year
s,);
17.
7 (1
5-64
yea
rs.)
; 3.
6 (>
64
year
s.).
LR
D
mea
ns
=
155
(all
ages
);
39.7
(0-1
4 ye
ars,
);
73.8
(1
5-64
ye
ars.
);
41.1
(>
64
year
s.).
T
ime-
serie
s an
alys
es
of
daily
nu
mbe
rs o
f co
nsul
tatio
ns p
erfo
rmed
, co
ntro
lling
fo
r tim
e tr
ends
, se
ason
fac
tors
, da
y of
wee
k,
influ
enza
, w
eat
her,
po
llen
leve
ls,
and
seria
l co
rrel
atio
n.
Pos
itive
ass
ocia
tions
, w
eakl
y si
gnifi
cant
and
co
nsis
tent
acr
oss
lags
, ob
serv
ed b
etw
een
asth
ma
cons
ulta
tions
and
NO
2 an
d C
O i
n ch
ildre
n,
and
w
ith
PM
10
in
adul
ts,
and
betw
een
othe
r LR
D c
onsu
ltatio
ns a
nd S
O2
in
child
ren.
A
utho
rs
con
clud
ed
that
th
ere
are
asso
ciat
ions
bet
wee
n a
ir po
llutio
n an
d da
ily c
once
ntra
tions
for
ast
hma
and
oth
er
low
er
resp
irato
ry
dise
ase
in
Lo
ndon
. In
ad
ults
, th
e au
thor
s co
nclu
ded
tha
t th
e on
ly
cons
iste
nt
ass
ocia
tion
was
w
ith
PM
10.
Acr
oss
all
of t
he v
ario
us a
ge,
caus
e, a
nd
seas
on
cate
gor
ies
cons
ide
red,
P
M10
w
as
the
pollu
tant
m
ost
cohe
rent
in
gi
ving
po
sitiv
e po
lluta
nt
RR
es
timat
es
for
both
as
thm
a an
d ot
her
LRD
(11
of
12 c
ateg
orie
s po
sitiv
e)
in
sing
le
pollu
tant
m
odel
s co
nsid
ered
.
Ast
hma
Doc
tor’s
Vis
its
50 µ
g/m
3 PM
10
-Yea
r-ro
und,
Sin
gle
Pol
luta
nt:
All
ages
(la
g 2
): 5
.4%
(-0
.6, 1
1.7)
0-
14 (
lag
1): 6
.4%
(-1
.5, 1
4.6)
15
-64
(lag
0): 9
.2%
(2.
8, 1
5.9)
>
64 (
lag
2): 1
1.7%
(-1
.8, 2
6.9
) -Y
ear-
roun
d, 2
Pol
luta
nt,
Chi
ldre
n (0
-14)
: (la
g 1)
PM
10:
W/ N
O2:
0.8
% (
-8.7
, 11.
4)
W/ O
3: 5
.5%
(-2
.1, 1
3.8)
W
/ SO
2: 3
.2%
(-6
.4, 1
3.7)
O
ther
LR
D; 5
0 µ
g/m
3 PM
10
-Yea
r-ro
und,
Sin
gle
Pol
luta
nt:
All
ages
(la
g 2
): 3
.5%
(0,
7.1
) 0-
14 (
lag
1): 4
.2%
(-1
.2, 9
.9)
15-6
4 (la
g 2)
: 3.7
% (
0.0,
7.6
) >
64yr
s.(la
g 2)
: 6.2
% (
0.5,
12
.9).
report no. 4/11
137
Haj
at e
t al
. (2
001)
(H
ajat
et
al,
2001
) Lo
ndon
,199
2-19
94.
44,4
06-4
9,59
6 re
gist
ered
pa
tient
s <
1 to
14
year
s.
PM
10 m
ean
28.5
(13
.9)
Dai
ly p
hysi
cia
n co
nsul
tatio
ns
(mea
n da
ily 4
.8 f
or
child
ren;
15
.3
for
adul
ts)
for
alle
rgic
rh
initi
s (I
CD
9:
477)
, S
O2,
O
3,
NO
2,
CO
, P
M1
0,
and
polle
n us
ing
gene
raliz
ed a
dditi
ve m
odel
s w
ith
nonp
aram
etric
sm
ooth
er.
SO
2 an
d O
3 sh
ow s
tron
g as
soci
atio
ns w
ith
the
num
ber
of
cons
ulta
tions
fo
r al
lerg
ic
rhin
itis.
Est
imat
es l
arge
st f
or a
lag
of
3 or
4
days
prio
r to
con
sulta
tions
, w
ith c
umul
ativ
e m
easu
res
stro
nger
th
an
sing
le
day
lag
s.
Str
onge
r ef
fect
s w
ere
foun
d fo
r ch
ildre
n th
an a
dults
. T
he t
wo-
po
lluta
nt a
naly
sis
of
the
child
ren’
s m
odel
sho
we
d th
at P
M10
and
N
O2
asso
ciat
ions
dis
appe
are
d on
ce e
ither
S
O2
or O
3 w
as in
corp
orat
ed in
to th
e m
odel
.
PM
10 -
Incr
emen
t (10
-90%
)
(15.
8-46
.5)
Age
<1-
14 y
ear
s la
g 3:
10.
4 (2
.0 to
19.
4)
Cum
0-3
: 17.
4 (6
.8 to
29.
0)
Age
s 15
-64
year
s la
g 2:
7.1
(2.
6 to
11.
7)
Cum
0-6
: 20.
2 (1
4.1
to 2
6.6)
report no. 4/11
138
Haj
at e
t al
. (2
002)
(H
ajat
et
al,
2002
) Lo
ndon
, 199
2-19
94.
268,
718-
295,
740
regi
ster
ed p
atie
nts.
P
M10
mea
n 28
.5 (
13.7
).
BS
mea
n 12
.7 (
7.9)
.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
daily
phy
sici
an
cons
ulta
tion
s (m
ean
daily
4.8
for
ch
ildre
n; 1
5.3
for
adu
lts)
for
uppe
r re
spira
tory
di
seas
es,
exc
ludi
ng a
llerg
ic r
hini
tis (
ICD
9:46
0-46
3,,
465,
47
0-47
5),
and
PM
10,
BS
, S
O2,
O3,
N
O2,
C
O,
usi
ng
GA
Ms
with
no
npar
am
etric
sm
ooth
ers
to c
ontr
ol f
or s
eas
onal
pat
tern
s. T
he
mod
els
wer
e
furt
her
adju
sted
fo
r da
y of
th
e
wee
k,
holid
ay,
in
fluen
za,
wea
ther
, po
llen
conc
entr
atio
ns
and
seria
l co
rrel
atio
n.
Lag
0, l
ag1,
lag
2, l
ag3
we
re c
onsi
dere
d. T
he
mos
t si
gnifi
cant
ly a
ssoc
iate
d la
g, r
egar
dles
s of
w
heth
er
the
as
soci
atio
n
was
po
sitiv
e or
ne
gativ
e, w
as s
how
n.
Hig
her
leve
ls o
f S
O2
incr
ease
d nu
mbe
rs o
f ch
ildho
od
cons
ulta
tion.
S
tron
ger
asso
ciat
ions
wer
e f
ound
with
PM
10
and
BS
in
adu
lts a
ged
15-
64 a
nd i
n a
dults
age
d 65
an
d ov
er,
also
afte
r fu
rthe
r ad
just
men
t fo
r ot
her
co-p
ollu
tant
s. I
n ge
nera
l, as
soci
atio
ns
wer
e st
rong
est
in e
lder
ly p
eopl
e, w
eake
st in
th
e ch
ildre
n, a
nd w
ere
larg
ely
foun
d in
the
w
inte
r m
onth
s fo
r th
ese
two
ag
e gr
oups
, an
d in
the
su
mm
er m
onth
s fo
r ad
ults
ag
ed
15-6
4.
An
ap
pare
nt
decr
ease
in
co
nsul
tatio
ns
was
as
soci
ated
in
co
lder
m
onth
s w
ith O
3 fo
r al
l ag
es
and
with
BS
in
child
ren.
PM
10 -
Incr
emen
t (10
-90%
)
(16-
47 µ
g/m
3 ) A
ll ye
ar:
Age
0-1
4:
2.0%
(-0
.2, 4
.2)
adju
stm
ent
NO
2: 2
.0%
(-0
.2, 4
.2)
adju
stm
ent O
3: 1
.8%
(-0
.4, 3
.9)
adju
stm
ent S
O2:
2.0
% (
-0.6
, 4.6
) A
ge 1
5-64
:
5.7%
(2.
9, 8
.6)
adju
stm
ent
NO
2: 2
.8%
(0.
7, 4
.9)
adju
stm
ent O
3: 4
.8%
(2.
6, 7
.0)
adju
stm
ent S
O2:
4.8
% (
2.2,
7.5
) A
ge 6
5+:
10
.2%
(5.
3, 1
5.3)
ad
just
men
t N
O2:
4.6
% (
0.5,
8.8
) ad
just
men
t O3: 1
0.7%
(5.
7, 1
6.0)
ad
just
men
t S
O2:
10
.6%
(4
.5,
17.1
) W
arm
sea
son:
A
ge 0
-14:
1.1
% (
-2.4
, 4.8
) A
ge 1
5-64
: 6.0
% (
2.7,
9.4
) A
ge 6
5+: 0
.1%
(-7
.7, 8
.5)
Col
d se
ason
A
ge 0
-14:
2.7
% (
-0.1
, 5.5
) A
ge 1
5-64
: 3.6
% (
1.0,
6.4
) A
ge 6
5+: 1
8.9
% (
11.7
, 26.
7)
BS
- In
crem
ent (
10-9
0%)
(6
-22
µg/
m3 )
All
year
: A
ge 0
-14:
-2.
4% (
-4.4
, -0.
3)
Age
15-
64: 3
.5%
(0.
6, 6
.4)
Age
65+
: 8.8
% (
4.6,
13.
3)
War
m s
easo
n:
Age
0-1
4: 3
.5%
(0.
3, 6
.9)
Age
15-
64: 6
.6%
(3.
6, 9
.6)
Age
65+
: 2.8
% (
-3.9
, 10.
1)
Col
d se
ason
A
ge 0
-14:
-3.
3% (
-6.0
, -0.
6)
Age
15-
64: 1
.5%
(-0
.9, 4
.1)
Age
65+
: 9.9
% (
3.7,
16.
3).
report no. 4/11
139
FRA
NC
E
Med
ina
et
al.
(199
7)
(Med
ina
et a
l, 19
97)
Gre
ater
Par
is,
1991
-199
5 P
opul
atio
n: 6
.5 M
M
Mea
n P
M13
: 25
PM
13 m
in/m
ax:
6/9
5 M
ean
BS
: 21
B
S m
in/m
ax:
3/1
30.
Thi
s st
udy
eva
luat
ed
shor
t-te
rm
rela
tions
hips
be
twee
n P
M13
an
d B
S
conc
entr
atio
ns
and
doct
ors'
hou
se c
alls
(m
ean
=8/
day;
20
% o
f ci
ty
tota
l) in
Gre
ater
Par
is.
Poi
sson
reg
ress
ion
used
, w
ith
non-
para
met
ric
smoo
thin
g fu
nct
ions
co
ntro
lling
fo
r tim
e tr
end,
se
ason
al
patt
erns
, po
llen
coun
ts,
influ
enza
ep
idem
ics,
day
of
wee
k,
holid
ays,
and
wea
ther
.
A r
elat
ions
hip
betw
een
all a
ge (
0-64
yea
rs.)
as
thm
a ho
use
calls
an
d P
M13
, B
S,
SO
2,
NO
2,
and
O3
air
pollu
tion,
es
peci
ally
fo
r ch
ildre
n ag
ed 0
-14
year
s (m
ean
= 2
/day
). I
n tw
o-po
lluta
nt
mod
els
incl
udin
g B
S
with
, su
cces
sive
ly,
SO
2, N
O2,
and
O3,
onl
y B
S
and
O3
effe
cts
rem
aine
d st
able
. T
hese
re
sults
al
so
indi
cate
th
at
air
pollu
tant
as
soci
atio
ns n
oted
for
hos
pita
l ED
vis
its a
re
also
app
licab
le t
o a
wid
er p
opul
atio
n th
at
visi
ts th
eir
doct
or.
Doc
tor’s
Ast
hm
a H
ouse
Vis
its:
50 µ
g/m
3 13
P
M Y
ear-
roun
d, S
ingl
e P
ollu
tant
: A
ll ag
es (
lag
2):
12.
7% (
4.1,
21.
9)
0-14
(la
g 0-
3): 4
1.5%
(20
, 66.
8)
15-6
4 (la
g 2)
: 6.3
% (
-4.6
, 18
.5).
ITA
LY
Veg
ni e
t al
. (2
005)
(V
egni
et
al,
2005
) C
omo,
Ital
y 19
95-1
997.
P
opul
atio
n.=
~85
,000
M
ean
TS
P=
62.
4 (2
8.6)
.
Thi
s st
udy
eval
uate
d re
spira
tory
dr
ug
disp
ensi
ng
dat
a as
he
alth
in
dica
tors
fo
r th
e ef
fect
s of
TS
P.
Wee
kly
air
mea
n co
ncen
trat
ions
of
TS
P w
ere
mod
eled
with
: 1)
wee
kly
coun
t of
in
divi
dual
s ha
ving
had
at
leas
t on
e re
spira
tory
dr
ug d
ispe
nse
d (c
ases
) an
d 2
) to
tal w
eek
ly s
um
of
daily
de
fined
do
ses
(DD
D)
of
resp
irato
ry
drug
s di
spen
sed.
A P
ois
son
regr
essi
on m
odel
ad
just
ed
for
long
-ter
m
tren
ds,
seas
onal
va
riatio
ns, h
olid
ays,
and
wea
ther
was
use
d.
Sig
nific
ant
dire
ct a
ssoc
iatio
ns b
etw
een
TS
P
and
both
ca
ses
and
DD
D
wer
e fo
und.
A
utho
rs
conc
lude
d th
at
both
ca
ses
and
DD
D o
f di
spe
nsed
res
pira
tory
dru
gs c
oul
d be
use
ful
for
epid
emio
logi
cal
surv
eilla
nce
of
air
pollu
tant
hea
lth e
ffect
s.
Exc
ess
risk
for
a va
riatio
n fr
om
10th
to
90
th
perc
entil
e of
T
SP
(2
9-92
µg/
m3 ):
C
ases
: 8.2
% (
0.2,
16.
9)
DD
D: 1
3.7
% (
4.4,
23.
8).
SPA
IN
Dam
iá
et
al.
(199
9)
(de
Die
go D
amia
et a
l, 19
99)
Val
enci
a,
Sp
ain,
3/
1994
-3/
1995
).
BS
mea
n: 1
01
BS
ran
ge: 3
4-21
3.
Ass
ocia
tions
of
BS
and
SO
2 w
ith w
eekl
y to
tal
ED
adm
issi
ons
for
asth
ma
pat
ient
s ag
ed >
12
year
s (m
ean
= 1
0/w
eek)
at
one
hosp
ital
over
on
e ye
ar a
sses
sed,
usi
ng l
inea
r st
epw
ise
GLM
re
gres
sion
. S
easo
n-sp
ecifi
c an
alys
es d
one
for
ea
ch
of
4 se
ason
s,
but
no
othe
r lo
ng-
wav
e
cont
rols
. Lin
ear
T, R
H,
BP
, ra
in, a
nd w
ind
spe
ed
incl
uded
as
crud
e w
eath
er c
ontr
ols
in A
NO
VA
m
odel
s.
Bot
h B
S
and
SO
2 co
rrel
ated
w
ith
ED
ad
mis
sion
s fo
r as
thm
a (S
O2:
r=
0.32
; B
S:
r=0.
35),
but
onl
y B
S s
igni
fican
t in
ste
pw
ise
mul
tiple
re
gres
sion
. N
o lin
ear
rela
tions
hip
foun
d w
ith
wea
ther
va
riabl
es.
Str
atifi
ed
AN
OV
A f
ound
str
onge
st B
S-E
D a
ssoc
iatio
n in
the
aut
umn
and
durin
g ab
ove
aver
age
tem
pera
ture
s.
Unc
ontr
olle
d
auto
corr
elat
ion
(e.g
., w
ithin
-sea
son)
an
d w
eath
er
effe
cts
likel
y re
mai
n in
mod
els.
Ast
hma
ED
Vis
its (
all a
ges)
: B
S: 4
0 µ
g/m
3 (si
ngle
pol
luta
nt)
BS
as
a la
g 0
we
ekly
ave
rage
: 41
.5%
(39
.1,
43.9
).
report no. 4/11
140
GR
EEC
E
Pan
tazo
poul
ou
et
al.
(199
5) (
Pan
tazo
poul
ou e
t al
, 199
5)
Ath
ens,
Gre
ece,
198
8.
Win
ter
(1/8
8-3/
88,9
/88-
12/8
8)
BS
mea
n.:7
5
BS
5-9
5th %
=2
6 -
161
Sum
mer
(3
/22/
88-
3/88
,9/2
1/88
) B
S m
ean.
:55
B
S 5
-95th
%=
19
- 90
Thi
s st
udy
exam
ined
effe
cts
of a
ir po
llutio
n on
da
ily
emer
genc
y ou
tpat
ient
vi
sits
an
d ad
mis
sion
s fo
r ca
rdia
c an
d r
espi
rato
ry c
ause
s.
Air
pollu
tant
s in
clud
ed:
BS
, C
O,
and
N
O2.
M
ultip
le
linea
r G
LM
regr
essi
on
mod
els
used
, co
ntro
lling
for
lin
ear
effe
cts
of t
empe
ratu
re a
nd
RH
, da
y of
w
eek,
ho
liday
s,
and
dum
my
varia
bles
fo
r m
onth
to
cr
udel
y co
ntro
l fo
r se
ason
, sep
arat
ely
for
win
ter
and
sum
mer
.
Dai
ly
num
ber
of
emer
genc
y vi
sits
re
late
d po
sitiv
ely
with
eac
h ai
r po
lluta
nt,
but
only
re
ache
d no
min
al
leve
l of
st
atis
tical
si
gnifi
canc
e fo
r N
O2
in w
inte
r. H
owev
er,
the
very
lim
ited
time
for
each
w
ithin
-sea
son
anal
ysis
(6
m
o.)
undo
ubta
bly
limite
d
the
pow
er o
f th
is a
naly
sis
to d
etec
t si
gnifi
cant
ef
fect
s.
Als
o,
poss
ible
la
gged
po
llutio
n ef
fect
s w
ere
ap
pare
ntly
no
t in
vest
igat
ed,
wh
ich
may
ha
ve r
educ
ed e
ffect
est
imat
es.
Sin
gle
Pol
luta
nt M
odel
s F
or W
inte
r (B
S: 2
5 µ
g/m
3 ) O
utpa
tient
Ho
spita
l Vis
its:
1.
1% (
-0.7
, 2.3
) R
espi
rato
ry h
ospi
tal a
dmis
sion
s:
4.3%
(0.
2, 8
.3)
For
Sum
mer
(B
S: 2
5 µ
g/m
3 )
Out
patie
nt H
osp
ital V
isits
0.
6% (
-4.7
, 6.0
) R
espi
rato
ry h
ospi
tal a
dmis
sion
s:
5.5%
(-3
.6, 1
4.7
)
BO
SNIA
H
astin
gs
et
al.
(200
2)
(Has
tings
an
dJar
dine
, 20
02)
Sol
dier
s de
ploy
ed
to
Bos
nia
1997
-199
8.
Mea
n P
M10
m
axim
um
leve
l: 92
.9
Mea
n P
M10
ave
rage
leve
l: 75
.5.
Thi
s ec
olog
ic s
tudy
was
con
duct
ed t
o de
term
ine
the
rela
tion
betw
een
we
ekly
lev
els
of P
M10
and
w
eekl
y up
per
resp
irato
ry
dise
ase
rate
s in
so
ldie
rs d
eplo
yed
to B
osni
a in
199
7-19
98.
PM
10
max
imum
leve
l was
def
ined
as
the
high
est
PM
10
leve
l re
cord
ed a
t ca
mp
durin
g a
give
n w
eek,
P
M10
ave
rage
lev
el w
as d
efin
ed a
s th
e av
erag
e of
all
read
ings
tak
en a
t a
cam
p du
ring
a gi
ven
wee
k.
Whe
n al
l ca
mps
wer
e co
mb
ined
, th
ere
was
a
stat
istic
ally
si
gnifi
cant
as
soci
atio
n be
twee
n th
e
PM
10
ma
xim
um
leve
l an
d up
per
resp
irato
ry
trac
t ra
tes
base
d on
K
rusk
al-W
allis
an
d M
ann-
Whi
tney
U
te
sts
(p=
0.04
7),
and
the
Pea
rson
cor
rela
tion
was
st
atis
tical
ly
sign
ifica
nt
(p=
0.0
41).
A
lthou
gh
the
rela
tion
was
not
sta
tistic
ally
sig
nific
ant
in
anal
yses
co
nduc
ted
on
the
indi
vidu
al
cam
ps,
the
aver
age
rate
in
crea
sed
with
ea
ch q
uart
ile o
f P
M10
ma
xim
um e
xpos
ure.
T
here
w
as
no
stat
istic
ally
si
gnifi
cant
as
soci
atio
n
bet
we
en
PM
10
aver
age
leve
l an
d up
per
resp
irato
ry
dise
ase
rate
s,
alth
ough
th
e av
erag
e ra
te
incr
ease
d w
ith
each
qua
rtile
of P
M10
ave
rage
exp
osu
re.
No
risk
sho
wn
ISR
AEL
G
arty
et
al.
(199
8) (
Gar
ty
et a
l, 19
98)
Tel
Avi
v, Is
rae
l, 19
93.
PM
10 m
ean:
45.
Sev
en d
ay r
unni
ng m
ean
of a
sthm
a E
D v
isits
by
child
ren
(1-1
8 ye
ars)
to
a
pedi
atric
ho
spita
l m
odel
ed in
rel
atio
n to
PM
10 in
Tel
Avi
v, Is
rael
.
No
PM
10 a
sso
ciat
ions
fou
nd w
ith E
D v
isits
. T
he
ER
vi
sits
–pol
luta
nt
corr
elat
ion
incr
ease
d si
gni
fican
tly w
hen
the
Sep
tem
ber
peak
w
as
exc
lude
d.
Use
of
a
wee
k-lo
ng
aver
age
and
asso
ciat
ed u
nco
ntro
lled
long
-w
ave
flu
ctua
tions
(w
ith
resu
ltant
au
toco
rrel
atio
n) l
ikel
y pr
even
ted
mea
ning
ful
anal
yses
of
sh
ort
–ter
m
PM
as
soci
atio
ns
with
ED
vis
its.
No
risks
sho
wn
report no. 4/11
141
TAB
LE V
– L
ON
G-T
ERM
PA
RTI
CU
LATE
MA
TTER
EXP
OSU
RE
AN
D M
OR
TALI
TY.
Ref
eren
ce, L
ocat
ion,
Ye
ars,
PM
Inde
x, M
ean
or M
edia
n (µ
g/m
3 ).
Stud
y D
escr
iptio
n. M
etho
ds, c
o-p
ollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
. PM
Inde
x, R
R (9
5% C
I)
Subn
atio
nal s
tudi
es
N
OR
WA
Y
Nae
ss e
t al
., 20
07 (
Nae
ss
et a
l, 20
07)
Coh
ort
of
all
inha
bita
nts
of O
slo
(143
,842
).
Fol
low
-up
perio
d:
1992
-19
98.
PM
2.5
rang
e: 6
.56-
22.3
4
PM
10 r
ange
: 6.5
7-30
.13
Thi
s st
udy
inve
stig
ated
the
ass
ocia
tion
betw
een
conc
entr
atio
ns
of P
M10
, P
M2.
5 an
d N
O2
in 1
992-
1995
, an
d
caus
e-sp
ecifi
c m
orta
lity.
T
he
popu
latio
n in
clud
ed
all
inha
bita
nts
of
Osl
o N
orw
ay,
aged
51
–90
year
s on
Ja
nuar
y 19
92
with
fol
low
-up
of
deat
hs f
rom
199
2 to
199
8. A
n ai
r di
sper
sion
m
odel
w
as
used
to
es
timat
e le
vels
of
e
xpos
ure
in
all
470
adm
inis
trat
ive
neig
hbou
rhoo
ds.
Co
x pr
opor
tiona
l ha
zard
s re
gres
sion
m
odel
s w
ere
use
d af
ter
adju
stm
ent
for
age,
edu
catio
n an
d oc
cupa
tiona
l cl
ass.
To
mod
el t
he r
elat
ion
betw
een
air
pollu
tant
s an
d m
orta
lity,
G
AM
m
odel
s w
ere
used
.
Som
e di
rect
as
soci
atio
ns
wer
e ob
serv
ed
betw
een
NO
2, P
M10
and
PM
2.5,
and
mor
talit
y fo
r al
l cau
ses,
CV
D,
lung
ca
ncer
, bu
t m
ainl
y C
OP
D.
Incr
eas
ed
risks
fo
r to
tal
mor
talit
y w
ere
str
onge
r fo
r su
bjec
ts a
ged
51-7
0 ye
ars
th
an in
old
er s
ubje
cts.
Am
ong
the
limita
tions
of
th
is
stud
y,
ther
e is
an
in
adeq
uate
al
low
ance
for
cova
riate
s, in
clud
ing
smok
ing
T
he t
hree
pol
luta
nts
shar
ed s
imila
r re
sults
, be
ing
thei
r co
rrel
atio
ns h
igh
(bet
wee
n 0.
88
and
0.95
).
Occ
upat
iona
l cl
ass
and
edu
catio
n ad
just
ed
HR
fo
r th
e 4th
vs
1st
qu
artil
e of
exp
osur
e to
PM
2.5:
All
caus
e m
orta
lity
Men
51
-70
year
s:
1.44
(1
.32,
1.
58)
Wom
en 5
1-70
yea
rs:
1.41
(1.
27,
1.43
) M
en
71-9
0 ye
ars:
1.
18
(1.1
0,
1.26
) W
omen
71-
90 y
ears
: 1.
11 (
1.05
, 1.
17)
CV
D
(qua
rtile
in
crea
se
of
expo
sure
) M
en
51-7
0 ye
ars:
1.
10
(1.0
5,
1.16
) W
omen
51-
70 y
ears
: 1.
14 (
1.06
, 1.
21)
Men
71
-90
year
s:
1.05
(1
.01,
1.
08)
Wom
en 7
1-90
yea
rs:
1.03
(1.
00,
1.05
) C
OP
D
(qua
rtile
in
crea
se
of
expo
sure
) M
en 5
1-70
yea
rs:
1.27
(1
.11,
1.4
7)
Wom
en 5
1-70
yea
rs:
1.09
(0.
94,
1.25
) M
en
71-9
0 ye
ars:
1.
10
(1.0
0,
1.21
) W
omen
71-
90 y
ears
: 1.
05 (
0.96
, 1.
16)
report no. 4/11
142
Lu
ng c
ance
r (q
uart
ile i
ncre
ase
of
expo
sure
) M
en
51-7
0 ye
ars:
1.
07
(0.9
7,
1.18
) W
omen
51-
70 y
ears
: 1.
27 (
1.13
, 1.
43)
Men
71
-90
year
s:
1.07
(0
.97,
1.
18)
Wom
en 7
1-90
yea
rs:
1.16
(1.
02,
1.32
) S
imila
r re
sults
w
ere
show
n fo
r P
M10
.
SW
EDEN
Ros
enlu
nd
et
al.
(200
6)
(Ros
enlu
nd e
t al
, 200
6)
Sto
ckol
m,
Sw
eden
,199
2-19
94.
Cas
e-co
ntro
l stu
dy
Sub
ject
s: 2
72 c
ases
with
fa
tal
myo
card
ial
infa
rctio
n
and
1,87
0 ho
spita
l co
ntro
ls.
30
year
re
side
ntia
l P
M10
ex
posu
re,
med
ian:
2.6
for
ca
ses
(incl
udin
g no
n-fa
tal
case
s)
and
2.4
for
cont
rols
.
Thi
s ca
se-c
ontr
ol
stud
y ex
amin
ed
the
asso
ciat
ion
bet
we
en 3
0 ye
ar a
vera
ge e
xpos
ure
to P
M10
, an
d ot
her
co-p
ollu
tant
s, i
nclu
din
g S
O2,
N
O2
and
CO
, an
d fir
st e
piso
de o
f fa
tal w
ithin
28
days
(an
d no
n-fa
tal)
myo
card
ial
infa
rctio
n (M
I).
The
ann
ual
mea
n le
vel
of P
M10
was
ass
esse
d on
ly f
or t
he y
ear
2000
, th
us a
ssum
ing
cons
tant
le
vels
dur
ing
the
stud
y pe
riod.
The
air
pollu
tion
da
ta f
or e
ach
year
was
lin
ked
to t
he h
isto
ry o
f re
side
nce
of e
ach
subj
ect
for
the
corr
espo
ndin
g ye
ar f
rom
19
60 t
o st
udy
incl
usio
n. O
Rs
wer
e de
rived
by
mul
tiple
log
istic
reg
ress
ion
mod
els
afte
r ad
just
men
t fo
r ag
e,
sex,
h
ospi
tal
catc
hmen
t ar
ea,
smok
ing,
ph
ysic
al
activ
ity,
diab
etes
and
soc
ioec
onom
ic s
tatu
s. E
stim
ates
fo
r P
M2.
5 w
ere
no
t sh
ow
n,
give
n th
e
high
co
rrel
atio
n in
thi
s sa
mpl
e be
twee
n P
M2.
5 an
d P
M10
(r=
0.99
8).
PM
10 w
as n
ot s
igni
fican
tly a
ssoc
iate
d w
ith
risk
of
fata
l M
I. A
bo
rder
line
sign
ifica
nt
exce
ss r
isk
was
fou
nd f
or “
out
of h
ospi
tal
deat
h”.
Sim
ilar
patte
rns
of r
isk
wer
e fo
und
fo
r N
O2,
CO
, an
d S
O2.
RR
for
an
incr
emen
t of
5 μ
g/m
3
(95th
–
5th
perc
entil
e)
of
PM
10.
Mor
talit
y w
ithin
28
da
ys
from
ad
mis
sion
to h
ospi
tal:
Tot
al d
eath
s: 1
.39
(0.9
4, 2
.07)
In
-hos
pita
l de
aths
: 1.
21
(0.7
5,
1.94
) O
ut-o
f-ho
spita
l de
aths
: 1.
84
(1.0
0, 3
.0).
report no. 4/11
143
THE
NET
HER
LAN
DS
H
oek
et a
l. (2
002)
(H
oek
et a
l, 20
02)
The
N
ethe
rland
s,
1986
-19
94
Coh
ort s
tudy
S
ubje
cts:
4,4
92
BS
exp
osur
e, m
ean
(SD
):
back
grou
nd
15.1
(2
.5);
ba
ckgr
ound
an
d lo
cal
15.5
(3.
2).
This
stud
y ex
amin
ed th
e as
soci
atio
n be
twee
n B
S an
d N
O2 i
n a
coho
rt of
4,4
92 re
side
nts i
n va
rious
are
as o
f th
e N
ethe
rland
s and
cau
se-s
peci
fic m
orta
lity.
Lo
ng-t
erm
ex
posu
re
to
tra
ffic-
rela
ted
BS
an
d
NO
2 w
as e
stim
ated
for
the
198
6 ho
me
addr
ess.
T
he
asso
ciat
ion
betw
een
ex
posu
re
to
air
pollu
tion
and
to
tal,
card
iopu
lmon
ary
and
non-
card
iopu
lmon
ary,
non
-lung
can
cer
mor
talit
y w
as
asse
ssed
w
ith
Co
x’s
prop
ortio
nal
haz
ards
m
odel
s,
afte
r ad
just
men
t fo
r ag
e,
sex,
ed
ucat
ion,
BM
I, oc
cupa
tion,
act
ive
and
pass
ive
ci
gare
tte
smok
ing,
an
d ne
ighb
ourh
ood
soci
oeco
nom
ic s
core
. T
wo
mod
els
of e
xpos
ure
for
ever
y po
lluta
nt
we
re
used
: 1
) th
e ba
ckgr
ound
co
ncen
trat
ion
wa
s en
tere
d in
th
e m
odel
with
the
indi
cato
r va
riabl
e fo
r liv
ing
near
a
maj
or r
oad.
2)
the
sum
of
the
back
grou
nd
and
the
estim
ated
co
ntrib
utio
n fr
om
livin
g n
ear
a m
ajor
roa
d to
air
pollu
tion
conc
entr
atio
ns.
Too
fe
w p
eopl
e di
ed f
rom
res
pira
tory
dis
orde
rs a
nd
lung
can
cer
to o
btai
n st
able
est
imat
es f
or t
he
indi
cato
r va
riabl
e fo
r liv
ing
nea
r a
maj
or r
oad.
Ove
rall,
48
9 de
aths
oc
curr
ed
durin
g th
fo
llow
-up
perio
d. C
ardi
opul
mon
ary
mor
talit
y w
as
sign
ifica
ntly
ass
ocia
ted
with
liv
ing
near
a
maj
or
road
, le
ss
cons
iste
ntly
w
ith
the
estim
ated
bac
kgro
und
conc
entr
atio
n. O
ther
ca
uses
of
de
ath,
in
clud
ing
lung
ca
ncer
, w
ere
unre
late
d to
air
pollu
tion.
RR
for
an
incr
emen
t of
10
(95
th –
5th
per
cent
ile)
of B
S.
All
caus
e m
orta
lity
Bac
kgro
und
U
nadj
uste
d: 1
.37
(0.9
5, 1
.97)
A
djus
ted:
1.1
7 (
0.76
, 1.7
8)
Adj
uste
d fo
r in
divi
dual
s liv
ing
10
year
s or
lo
nger
at
th
eir
1986
ad
dres
s: 1
.04
(0.6
5, 1
.64)
B
ackg
roun
d a
nd lo
cal
All
caus
e m
orta
lity
Una
djus
ted:
1.3
7 (1
.06,
1.7
7)
Adj
uste
d: 1
.32
(0.
98, 1
.78)
A
djus
ted
for
indi
vidu
als
livin
g 10
ye
ars
or
long
er
at
thei
r 19
86
addr
ess:
1.3
1 (0
.95,
1.8
0)
Car
diop
ulm
onar
y B
ackg
roun
d: 1
.34
(0.6
8, 2
.64)
B
ackg
roun
d a
nd l
ocal
1.7
1 (1
.10,
2.
67)
Non
-car
diop
ulm
onar
y no
n-lu
ng
canc
er
Bac
kgro
und:
1.1
5 (0
.63,
2.1
0)
Bac
kgro
und
and
loc
al 1
.09
(0.7
1,
1.69
) Lu
ng c
ance
r B
ackg
roun
d a
nd lo
cal:
1.06
(0.
43,
2.63
).
report no. 4/11
144
GER
MA
NY
G
ehrin
g et
al
. (2
006)
(G
ehrin
g et
al,
2006
) 7
citie
s of
N
orth
R
hine
-W
estp
halia
, Ger
man
y
Per
iod
of f
ollo
w-u
p: f
rom
19
85 to
200
2-20
03.
Sub
ject
s= 4
,752
M
ean
TS
P=
62.
4 (2
8.6)
.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
long
-ter
m
expo
sure
to
P
M10
, ca
lcul
ate
d fr
om
TS
P b
y u
sing
a 0
.71
conv
ersi
on f
acto
r, N
O2
and
tota
l, ca
rdio
pulm
onar
y (I
CD
9: 4
00-4
40 a
nd 4
60-
519)
and
non
-car
diop
ulm
onar
y no
n-lu
ng c
ance
r m
orta
lity,
in
a
coho
rt
of
appr
oxim
ate
ly
4,80
0 G
erm
an
mid
dle-
age
wo
men
. T
he
limite
d nu
mbe
r of
de
aths
fo
r lu
ng
canc
er
and
pulm
onar
y ca
uses
did
not
allo
w t
o ob
tain
sta
ble
effe
ct e
stim
ates
, an
d co
rres
pond
ing
RR
s w
ere
no
t sh
ow
n.
RR
s w
ere
est
imat
ed
usin
g C
ox
prop
ortio
nal
haza
rd
regr
essi
on
mod
els
afte
r ad
just
men
t fo
r so
cio-
eco
nom
ic
stat
us
and
smok
ing.
1-y
ear
aver
age
daily
con
cent
ratio
n of
P
M10
for
the
yea
r of
the
bas
elin
e e
xam
inat
ion,
an
d 5-
year
ave
rage
dai
ly c
once
ntra
tion
of P
M10
fo
r th
e ye
ar o
f th
e ba
selin
e ex
amin
atio
n a
nd t
he
prec
edin
g 4
year
s w
ere
con
side
red.
Dur
ing
18 y
ear
follo
w-u
p, 3
99 w
omen
die
d.
One
-yea
r av
erag
e ex
posu
re
to
PM
10
(com
pute
d by
T
SP
) w
as
asso
ciat
ed
to
card
iopu
lmon
ary
mor
talit
y,
but
not
to
all
caus
e an
d
non
-car
diop
ulm
onar
y no
n-lu
ng
canc
er
mor
talit
y.
The
pa
ttern
of
ris
k w
as
sim
ilar
for
5-ye
ar a
vera
ge e
xpos
ure
to P
M10
. Li
ving
with
in a
50-
met
er r
adiu
s of
a m
ajor
ro
ad a
nd N
O2
leve
ls s
how
ed s
imila
r ris
ks.
Mod
estly
str
ong
er a
ssoc
iatio
ns w
ere
fou
nd
for
wom
en
with
a
low
so
cio-
econ
om
ic
stat
us,
for
curr
ent
smok
ers
and
for
non-
smok
ers
with
ex
pos
ure
to
envi
ronm
enta
l to
bacc
o sm
oke
(dat
a no
t sho
wn)
.
Mor
talit
y R
R f
or a
n in
crem
ent
of
7 μg
/m3 o
f P
M10
(co
rres
pond
ing
to
10 μ
g/m
3 of T
SP
):
1-ye
ar a
vera
ge
All
caus
es:
1.0
8 (0
.94-
1.25
) C
ardi
opul
mon
ary:
1.
34
(1.0
6-1.
71)
Non
-car
diop
ulm
onar
y no
n-lu
ng
canc
er: 0
.92
(0.7
6-1.
10)
5-ye
ar a
vera
ge
All
caus
es:
1.1
3 (0
.99-
1.30
) C
ardi
opul
mon
ary:
1.
59
(1.2
3-2.
04)
Non
-car
diop
ulm
onar
y no
n-lu
ng
canc
er: 0
.91
(0.7
6-1.
08).
report no. 4/11
145
FRA
NC
E
Fill
eul e
t al
. (2
005)
(F
illeu
l et
al,
2005
) P
AA
RC
sur
vey
24
area
s of
7
Fre
nch
citie
s pe
riod
of e
xpos
ure
mea
sure
men
t: 19
74-1
976
Sub
ject
s= 1
4,28
4 ad
ults
T
SP
min
/ma
x: 4
5/24
3 B
S m
in/m
ax:
18/
152.
Thi
s st
udy
exam
ined
the
lon
g te
rm e
ffect
s of
T
SP
, B
S,
SO
2,
NO
2 an
d N
O (
daily
ass
esse
d in
19
74-1
976)
on
m
orta
lity
in
the
Pol
lutio
n A
tmos
pher
iqu
e et
A
ffect
ions
R
espi
rato
ires
Chr
oniq
ues
(PA
AR
C)
surv
ey,
cond
ucte
d i
n 24
ar
eas
of
7 F
renc
h ci
ties
on
14,2
84
adul
ts,
follo
wed
for
the
per
iod
1974
-200
1. R
Rs
wer
e es
timat
ed
by
Co
x pr
opor
tiona
l ha
zard
re
gres
sion
m
odel
s,
afte
r ad
just
men
t fo
r ag
e,
sex,
sm
okin
g, B
MI,
educ
atio
n, a
nd o
ccup
atio
nal
expo
sure
. M
orta
lity
for
all n
on-a
ccid
enta
l cau
ses
(IC
D9
<80
0),
card
iopu
lmon
ary
dise
ase
(IC
D9:
40
1-44
0 an
d 46
0-51
9) a
nd l
ung
can
cer
(IC
D9:
16
2)
wer
e co
nsid
ered
. T
he
anal
yses
w
ere
perf
orm
ed f
or a
ll 24
are
as a
nd i
n a
subg
roup
of
18
area
s w
ith
a ra
tio
of
NO
/ N
O2
<3
(w
ith
mon
itors
pot
entia
lly in
fluen
ced
by lo
cal t
raffi
c).
Afte
r ap
prox
imat
ely
25 y
ears
of
follo
w-u
p,
appr
oxim
ate
ly
2,39
6 de
aths
fr
om
non-
acci
dent
al
caus
es,
546
fr
om
card
iopu
lmon
ary
dise
ase
and
178
fro
m lu
ng
canc
er
occu
rred
. C
onsi
deri
ng
all
the
24
area
s, n
o as
soci
atio
n w
as f
ound
bet
we
en
TS
P,
BS
an
d th
e ot
her
cons
ider
ed
co-
pollu
tant
s,
and
m
orta
lity
for
all
non
-ac
cide
ntal
ca
uses
, ca
rdio
pulm
onar
y di
seas
e, o
r lu
ng c
ance
r. A
fter
exc
lusi
on o
f 6
area
s in
fluen
ced
by
loca
l tr
affic
, T
SP
sh
ow
ed
a si
gnifi
cant
as
soci
atio
n w
ith
all
non-
acci
dent
al c
ause
s an
d ca
rdio
pulm
onar
y di
seas
e m
orta
lity,
and
BS
incr
ease
d th
e ris
k of
no
n-ac
cid
enta
l ca
use
s m
orta
lity.
N
O2
sign
ifica
ntly
in
crea
sed
the
mor
talit
y fo
r al
l no
n-ac
cide
nta
l ca
uses
, ca
rdio
pulm
onar
y di
seas
e an
d lu
ng
canc
er.
Aut
hors
co
nclu
ded
tha
t ur
ban
air
pollu
tion
asse
sse
d in
the
197
0s w
as
asso
ciat
ed
with
inc
reas
ed
mor
talit
y ov
er 2
5 ye
ars.
Mor
talit
y R
R f
or a
n in
crem
ent
of
10 μ
g/m
3 of T
SP
: 24
are
as:
Non
-acc
iden
tal:
1.00
(0.
99, 1
.01)
C
ardi
opul
mon
ary:
1.
01
(0.9
9,
1.03
) Lu
ng c
ance
r: 0
.97
(0.9
4, 1
.01)
18
are
as n
ot i
nflu
ence
d b
y lo
cal
traf
fic:
Non
, acc
iden
tal:
1.05
(1.
02, 1
.08)
C
ardi
opul
mon
ary:
1.
06
(1.0
1,
1.12
) Lu
ng c
ance
r: 1
.00
(0.9
2, 1
.10)
M
orta
lity
RR
for
an
incr
emen
t of
10
μg/
m3 o
f BS
: 24
are
as:
Non
-acc
iden
tal:
0.99
(0.
98, 1
.01)
C
ardi
opul
mon
ary:
1.
00
(0.9
7,
1.02
) Lu
ng c
ance
r: 0
.97
(0.9
3, 1
.01)
18
are
as n
ot i
nflu
ence
d b
y lo
cal
traf
fic:
Non
-acc
iden
tal:
1.07
(1.
03, 1
.10)
C
ardi
opul
mon
ary:
1.05
(0
.98,
1.
12)
Lung
can
cer:
1.0
3 (0
.92,
1.1
5).
report no. 4/11
146
TAB
LE V
I - E
CO
LOG
IC S
TUD
IES
ON
LO
NG
TER
M P
AR
TIC
ULA
TE M
ATT
ER E
XPO
SUR
E A
ND
MO
RTA
LITY
. Ref
eren
ce, L
ocat
ion,
Ye
ars,
PM
Inde
x, M
ean
or M
edia
n (µ
g/m
3 ).
Stud
y D
escr
iptio
n. M
etho
ds, c
o-p
ollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
. PM
Inde
x, R
R (9
5% C
I)
Subn
atio
nal s
tudi
es
G
REA
T B
RIT
AIN
Mah
esw
aran
et
al
. (2
005b
) (M
ahes
war
an
et
al, 2
005b
) S
heffi
eld,
UK
, 19
94-1
998
Pop
ulat
ion:
19
9,68
2 su
bjec
ts a
ged ≥
45 y
ears
. P
M10
, med
ian:
18.
8.
To
exam
ine
the
rela
tion
betw
een
area
s w
ith
diffe
rent
air
pollu
tion
leve
ls a
nd c
oron
ary
hear
t di
seas
e (I
CD
9:
410-
414)
m
orta
lity
(and
ho
spita
lizat
ion
), 5
yea
r av
era
ge o
f P
M1
0, N
Ox
and
CO
leve
ls w
ere
inte
rpol
ated
to
1030
cen
sus
enum
erat
ion
dis
tric
ts,
and
lin
ked
to r
esid
ence
of
each
sub
ject
. P
oiss
on r
egre
ssio
n m
odel
s w
ere
us
ed a
fter
adj
ustm
ent
for
age
, se
x, d
epriv
atio
n an
d sm
okin
g pr
eval
ence
. T
o ad
just
for
sm
okin
g pr
eval
ence
, su
rvey
dat
a fr
om a
ran
dom
sam
ple
of a
dults
car
ried
out
in 2
00
0 w
as u
sed.
War
d le
vel
smok
ing
pre
vale
nce
wa
s at
trib
uted
to
all
dist
ricts
with
in e
ach
of t
he 2
9 w
ards
in S
hef
field
. A
naly
ses
wer
e re
run,
sub
stitu
ting
unsm
oot
hed
pollu
tion
vari
able
s w
ith
the
1km
ra
dius
sm
ooth
ed v
aria
bles
.
Dur
ing
the
pe
riod
199
4-19
98,
6,8
57 d
eath
s fo
r co
rona
ry h
eart
dis
ease
occ
urre
d. P
M10
w
as
sign
ifica
ntly
as
soci
ate
d w
ith
mor
talit
y fo
r co
rona
ry h
eart
dis
ease
in
the
age-
and-
sex-
adju
sted
m
odel
on
ly.
Afte
r fu
rthe
r al
low
ance
fo
r ot
her
cova
riate
s,
no
asso
ciat
ion
was
sho
wn.
A s
imila
r pa
ttern
of
risk
w
as
foun
d fo
r C
O.
NO
x in
crea
sed
mor
talit
y fo
r co
rona
ry h
eart
h di
seas
e ev
en
afte
r fu
rthe
r al
low
ance
fo
r sm
okin
g an
d de
priv
atio
n.
RR
for
the
hig
hest
vs
the
low
est
qu
intil
e of
e
xpos
ure
of
PM
10.
Mor
talit
y fo
r co
rona
ry
hear
t di
seas
e:
Adj
uste
d fo
r se
x an
d ag
e:
1.30
(1.
19, 1
.43)
F
urth
er a
djus
ted
for
smok
ing
and
depr
ivat
ion
(not
sm
ooth
ed):
1.
08 (
0.96
, 1.2
0)
Fur
ther
adj
uste
d fo
r sm
okin
g an
d de
priv
atio
n
(1km
ra
dius
sm
ooth
ed):
1.
07 (
0.96
, 1.2
1).
Mah
esw
aran
et
al
. (2
005a
) (M
ahes
war
an
et
al, 2
005a
) S
heffi
eld,
UK
, S
ame
stud
y de
scrib
ed
abov
e.
Thi
s st
udy
use
d th
e sa
me
dat
a an
d m
etho
ds
desc
ribed
ab
ove
to
exa
min
e th
e as
soci
atio
n be
twee
n P
M10
, N
Ox
and
CO
, an
d st
roke
(IC
D9:
43
0-43
8) m
orta
lity
(and
hos
pita
lizat
ion)
.
Dur
ing
the
pe
riod
199
4-19
98,
2,9
79 d
eath
s fo
r st
roke
occ
urre
d. P
M10
wa
s si
gnifi
can
tly
asso
ciat
ed w
ith s
trok
e m
orta
lity
also
aft
er
furt
her
allo
wan
ce
for
othe
r co
varia
tes,
in
clud
ing
smok
ing
prev
alen
ce
and
de
priv
atio
n.
Sim
ilar
patte
rns
of
risk
wer
e fo
und
for
NO
x an
d C
O.
RR
for
the
hig
hest
vs
the
low
est
qu
intil
e of
e
xpos
ure
of
PM
10.
Str
oke
mor
talit
y:
Adj
uste
d fo
r se
x an
d a
ge:
1.
39 (
1.23
, 1.5
8)
Fur
ther
adj
uste
d fo
r sm
okin
g an
d de
priv
atio
n (n
ot s
moo
thed
):
1.33
(1.
14, 1
.56)
F
urth
er a
djus
ted
for
smok
ing
and
depr
ivat
ion
(1
km
radi
us
smoo
thed
):
1.24
(1.
05, 1
.47)
.
report no. 4/11
147
IREL
AN
D
C
lanc
y et
al
. (2
002)
(C
lanc
y et
al,
2002
) D
ublin
, Ir
ela
nd,
1984
-19
96
Inte
rven
tion
stud
y S
ubje
cts:
4,4
92;
B
S
exp
osu
re,
mea
n an
nual
co
ncen
trat
ion
rang
ed
betw
een
10
an
d
120
befo
re t
he b
an,
and
betw
een
10
and
40
ther
eafte
r.
Thi
s st
udy
com
pare
d st
and
ardi
sed
non-
trau
ma
(IC
D9
<80
0),
resp
irato
ry
(IC
D9:
48
0-49
6 an
d 50
7),
and
card
iova
scul
ar (
ICD
9: 3
90-4
48)
deat
h ra
tes
72 m
onth
s be
fore
and
afte
r th
e ba
n of
coa
l sa
les
in D
ublin
(A
ug 1
990)
. T
he e
ffect
of
the
ban
on
age-
stan
dard
ised
de
ath
rate
s w
as
estim
ated
w
ith
an
inte
rrup
ted
time-
serie
s an
alys
is,
adju
stin
g fo
r w
eath
er,
resp
irato
ry
epid
emic
s, a
nd
deat
h ra
tes
in th
e re
st o
f Ire
land
.
Ave
rage
B
S
conc
entr
atio
ns
in
Dub
lin
decl
ined
by
35.
6 µ
g/m
3 (70
%)
afte
r th
e ba
n on
coa
l sa
les.
Adj
uste
d no
n-tr
aum
a de
ath
rate
s de
crea
sed
by 5
·7%
(95
% C
I: 4,
7),
re
spira
tory
dea
ths
by 1
5·5
% (
95%
CI:
12,
19),
an
d ca
rdio
vasc
ular
d
eath
s by
1
0·3%
(9
5%
CI:
8,
13).
R
espi
rato
ry
and
card
iova
scul
ar s
tand
ardi
sed
deat
h ra
tes
fell
coin
cide
nt w
ith th
e ba
n on
coa
l sal
es.
No
RR
sho
wn.
report no. 4/11
148
TAB
LE V
II - L
ON
G T
ERM
PA
RTI
CU
LATE
MA
TTER
EXP
OSU
RE
AN
D C
AR
DIO
VASC
ULA
R M
OR
BID
ITY.
Ref
eren
ce, L
ocat
ion,
Ye
ars,
PM
Inde
x, M
ean
or M
edia
n (µ
g/m
3 ).
Stud
y D
escr
iptio
n. M
etho
ds, c
o-p
ollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
. PM
Inde
x, R
R (9
5% C
I)
Subn
atio
nal s
tudi
es
G
REA
T B
RIT
AIN
Mah
esw
aran
et
al
. (2
005b
) (M
ahes
war
an
et
al, 2
005b
) S
heffi
eld,
UK
, 19
94-1
998.
P
opul
atio
n:
199,
682
subj
ects
age
d ≥4
5 ye
ars.
P
M10
, med
ian:
18.
8.
To
exam
ine
the
rela
tion
betw
een
area
s w
ith
diffe
rent
air
pollu
tion
leve
ls a
nd c
oron
ary
hear
t di
seas
e (I
CD
9:
410-
414)
ho
spita
lizat
ion
(a
nd
mor
talit
y),
5 ye
ar a
vera
ge o
f P
M10
, N
Ox
and
CO
le
vels
w
ere
in
terp
olat
ed
to
1030
ce
nsus
en
umer
atio
n d
istr
icts
, an
d li
nke
d to
res
iden
ce o
f ea
ch s
ubje
ct.
Poi
sson
reg
ress
ion
mod
els
wer
e
used
aft
er a
djus
tmen
t fo
r a
ge,
sex,
dep
rivat
ion
and
smok
ing
prev
alen
ce.
To
adju
st f
or s
mok
ing
prev
alen
ce,
surv
ey d
ata
from
a r
ando
m s
ampl
e of
adu
lts c
arrie
d ou
t in
20
00
was
use
d. W
ard
leve
l sm
okin
g p
reva
lenc
e w
as
attr
ibut
ed t
o al
l di
stric
ts w
ithin
eac
h of
the
29
war
ds in
Sh
effie
ld.
Ana
lyse
s w
ere
reru
n, s
ubst
itutin
g un
smo
othe
d po
llutio
n va
riab
les
with
th
e 1k
m
radi
us
smoo
thed
var
iabl
es.
Dur
ing
the
perio
d 19
94-1
998,
11
,407
ad
mis
sion
s fo
r co
rona
ry
hear
t di
seas
e oc
curr
ed.
PM
10
wa
s as
soci
ated
w
ith
adm
issi
on f
or c
oron
ary
hear
t di
seas
e in
th
e ag
e-an
d-se
x-a
djus
ted
mod
el
only
. A
fter
fu
rthe
r al
low
anc
e fo
r ot
her
cova
riate
s,
no
asso
ciat
ion
was
sho
wn.
A s
imila
r pa
ttern
of
risk
was
foun
d fo
r C
O a
nd N
Ox.
RR
for
the
hig
hest
vs
the
low
est
qu
intil
e of
e
xpos
ure
of
PM
10.
Adm
issi
on
for
coro
nary
he
art
dise
ase:
A
djus
ted
for
sex
and
age:
1.
36 (
1.23
, 1.5
0)
Fur
ther
adj
uste
d fo
r sm
okin
g an
d de
priv
atio
n (n
ot s
moo
thed
):
1.01
(0.
90, 1
.14)
F
urth
er a
djus
ted
for
smok
ing
and
depr
ivat
ion
(1
km
radi
us
smoo
thed
):
1.07
(0.
95, 1
.20)
.
Mah
esw
aran
et
al
. (2
005a
) (M
ahes
war
an
et
al, 2
005a
) S
heffi
eld,
UK
, S
ame
stud
y de
scrib
ed
abov
e.
Thi
s st
udy
use
d th
e sa
me
dat
a an
d m
etho
ds
desc
ribed
ab
ove
to
exa
min
e th
e as
soci
atio
n be
twee
n P
M10
, N
Ox
and
CO
, an
d st
roke
(IC
D9:
43
0-43
8) h
ospi
taliz
atio
n (a
nd m
orta
lity)
.
Dur
ing
the
perio
d 19
94-1
998,
5,
122
adm
issi
ons
for
stro
ke o
ccur
red.
PM
10 w
as
sign
ifica
ntly
ass
ocia
ted
with
hos
pita
lizat
ion
for
stro
ke
in
the
age-
and-
sex-
adju
sted
m
odel
. A
fter
furt
her
allo
wan
ce
for
othe
r co
varia
tes,
a
bord
erlin
e si
gnifi
cant
as
soci
atio
n w
as s
how
n. S
imila
r pa
ttern
s of
ris
k w
ere
foun
d fo
r N
Ox
and
CO
.
RR
for
the
hig
hest
vs
the
low
est
qu
intil
e of
e
xpos
ure
of
PM
10.
Adm
issi
on fo
r st
roke
: A
djus
ted
for
sex
and
age
:
1.40
(1.
26, 1
.55)
F
urth
er a
djus
ted
for
smok
ing
and
depr
ivat
ion
(not
sm
ooth
ed):
1.
13 (
0.99
, 1.2
9)
Fur
ther
adj
uste
d fo
r sm
okin
g an
d de
priv
atio
n
(1km
ra
dius
sm
ooth
ed):
1.
15 (
1.01
, 1.3
1).
report no. 4/11
149
Sol
omon
et
al
. (2
003)
(S
olom
on e
t al,
2003
) 11
el
ecto
ral
war
ds
of
Eng
land
and
Wal
es, U
K.
1,16
6 w
omen
M
ean
BS
le
vel,
rang
e
acro
ss w
ards
: 19
66-1
969:
40/
180
1978
-198
1: 1
4/79
19
94-1
997:
4/1
4.
To
exp
lore
th
e lo
ng-t
erm
influ
ence
of
part
icul
ate
air
pollu
tion
on c
ardi
o-re
spirr
ator
y m
orbi
dity
in
the
UK
, a
cros
s-se
ctio
nal
post
al
surv
ey
was
co
nduc
ted
in 1
1 el
ecto
ral
war
ds o
f U
K.
War
ds
wer
e cl
assi
fied
in h
igh
(BS
>12
0 µ
g/m
3 in
196
6-19
69)
and
low
(B
S<
50
µg/
m3
in
1966
-196
9)
pollu
tion
cate
gor
ies.
W
omen
fro
m t
he t
wo
cate
gorie
s w
ere
com
pare
d in
te
rms
of
self-
repo
rted
hi
stor
y of
is
chae
mic
he
art
dise
ase
(and
as
thm
a an
d pr
oduc
tive
coug
h).
No
exc
ess
risk
of i
scha
emic
hea
rt d
isea
se
wa
s fo
und
in t
he a
reas
with
hig
her
leve
ls o
f B
S.
RR
for
wom
en
livin
g in
the
are
as
with
hig
her
BS
leve
ls:
1.0
(0.7
, 1.4
).
SWED
EN
R
osen
lund
et
al
. (2
006)
(R
osen
lund
et
al, 2
006)
S
tock
olm
, S
wed
en,
1992
-19
94.
Cas
e-co
ntro
l stu
dy
Sub
ject
s: 2
72 c
ases
with
fa
tal
myo
card
ial
infa
rctio
n
and
1,87
0 ho
spita
l co
ntro
ls.
30
year
re
side
ntia
l P
M10
ex
posu
re,
med
ian:
2.6
for
ca
ses
(incl
udin
g no
n-fa
tal
case
s)
and
2.4
for
cont
rols
.
Thi
s ca
se-c
ontr
ol
stud
y ex
amin
ed
the
asso
ciat
ion
bet
we
en 3
0 ye
ar a
vera
ge e
xpos
ure
to P
M10
, an
d ot
her
co-p
ollu
tant
s, i
nclu
din
g S
O2,
N
O2
and
CO
, an
d fir
st e
piso
de o
f no
n-fa
tal (
and
fata
l) m
yoca
rdia
l in
farc
tion
(M
I).
The
an
nual
m
ean
leve
l of
PM
10 w
as
asse
ssed
onl
y fo
r th
e ye
ar 2
000,
thu
s as
sum
ing
co
nsta
nt le
vels
dur
ing
the
stud
y pe
riod.
The
air
pollu
tion
data
for
eac
h ye
ar w
as l
inke
d to
the
his
tory
of
resi
denc
e of
ea
ch s
ubje
ct f
or t
he c
orre
spon
ding
yea
r fr
om
1960
to
stud
y in
clus
ion.
OR
s w
ere
deriv
ed b
y m
ultip
le
logi
stic
re
gres
sion
m
odel
s af
ter
adju
stm
ent
for
age,
se
x,
hosp
ital
catc
hmen
t ar
ea,
smok
ing,
phy
sica
l a
ctiv
ity,
diab
etes
and
so
cioe
cono
mic
sta
tus.
Est
imat
es f
or P
M2.
5 w
ere
no
t sh
own,
gi
ven
the
high
co
rrel
atio
n in
th
is
sam
ple
betw
een
PM
2.5
and
PM
10 (
r=0.
998
).
PM
10 w
as n
ot a
ssoc
iate
d w
ith n
onfa
tal
MI.
Sim
ilar
patt
erns
of
risk
wer
e fo
und
for
NO
2,
CO
, and
SO
2.
RR
for
an
incr
emen
t of
5 μ
g/m
3
(95th
–
5th
perc
entil
e)
of
PM
10.
Hos
pita
l ad
mis
sion
fo
r no
n-fa
tal
MI:
0.
92 (
0.71
, 1.1
9).
report no. 4/11
150
GER
MA
NY
H
offm
ann
et
al.
(200
6)
(Hof
fman
n et
al,
2006
) T
wo
citie
s o
f G
erm
any
(Ess
en
and
M
ulhe
im),
20
00-2
003.
C
ohor
t stu
dy
Sub
ject
s: 3
,39
9 M
ean
(SD
) P
M2.
5:
23.3
(1
.4)
.
Thi
s st
udy
exam
ined
the
rel
atio
nshi
p be
twee
n th
e lo
ng-t
erm
res
iden
tial
exp
osur
e to
tra
ffic
and
self-
repo
rted
his
tory
of
coro
nary
hea
rt d
isea
se,
usin
g ba
selin
e da
ta
from
th
e G
erm
an
Hei
nz
Nix
dorf
R
EC
ALL
st
udy,
a
popu
latio
n-ba
sed
coho
rt
stud
y.
Long
-ter
m
pers
onal
tr
affic
an
d P
M2.
5 e
xpos
ures
w
ere
asse
ssed
fo
r ea
ch
resi
dent
. P
M2.
5 w
as a
sses
sed
for
the
year
200
2.
OR
s w
ere
der
ived
by
m
ultiv
aria
ble
logi
stic
re
gres
sion
m
odel
s.
For
P
M2.
5 fo
ur
diffe
rent
m
odel
s w
ere
cons
ider
ed:
mod
el 1
: ad
just
ed f
or
traf
fic e
xpos
ure
; m
odel
2:
furt
her
adju
sted
for
ag
e an
d se
x;
mod
el
3:
furt
her
adju
sted
fo
r ed
ucat
ion,
di
abet
es,
BM
I,
wai
st-t
o-hi
p ra
tio,
smok
ing,
e
nviro
nmen
tal
toba
cco
smok
e,
phys
ical
ac
tivity
, ci
ty
and
area
of
re
side
nce;
m
odel
4:
furt
her
adj
uste
d fo
r hy
pert
ensi
on
and
lipid
s.
Ove
rall,
242
sub
ject
s re
port
ed a
his
tory
of
coro
nary
hea
rt d
isea
se.
No
asso
ciat
ion
was
fo
und
with
PM
2.5
expo
sure
, bu
t th
ere
was
an
exce
ss r
isk
in m
en o
nly
for
resi
dent
s liv
ing
near
maj
or r
oads
.
OR
fo
r P
M2.
5 (in
crem
ent
appa
rent
ly 1
μg/
m3 ):
M
odel
1: 0
.92
(0.3
6, 2
.39)
M
odel
2: 0
.83
(0.3
1, 2
.27)
M
odel
3: 0
.56
(0.1
6, 2
.01)
M
odel
4: 0
.55
(0.1
4, 2
.11)
.
report no. 4/11
151
TAB
LE V
III -
LON
G T
ERM
PA
RTI
CU
LATE
MA
TTER
EXP
OSU
RE
AN
D R
ESPI
RA
TOR
Y M
OR
BID
ITY
OR
SYM
PTO
MS
(Der
ived
from
EPA
TA
BLE
8B
-8).
Ref
eren
ce, L
ocat
ion,
Ye
ars,
PM
Inde
x, M
ean
or M
edia
n (µ
g/m
3 ).
Stud
y D
escr
iptio
n. M
etho
ds, c
o-p
ollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
. PM
Inde
x, R
R (9
5% C
I)
ECR
HS
I stu
dy
S
unye
r et
al
. (2
006)
(S
unye
r et
al,
2006
) 21
cen
tres
of 1
0 co
untr
ies
Fol
low
-up
perio
d:
2000
-20
02.
Pop
ulat
ion:
6,8
24
PM
2.5
mea
n 1
9.12
S
mea
n 11
67.0
7
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
the
prev
alen
ce a
nd n
ew
on
set
of s
ympt
om
s of
ch
roni
c br
onch
itis
and
urb
an
leve
ls o
f P
M2.
5, S
, N
O2,
in
6,8
24 s
ubje
cts
enro
lled
durin
g 19
91–9
3 in
21
ce
ntre
s of
10
E
urop
ean
coun
trie
s.
Sub
ject
s w
ere
follo
wed
fr
om
2000
to
20
02.
PM
2.5
and
S w
ere
ass
esse
d at
cen
tre
leve
l w
ith
the
sam
e ce
ntra
l m
onito
ring
site
eq
uipm
ent,
whe
reas
NO
2 w
as
mea
sure
d du
ring
a 14
day
pe
riod
for
1,63
4 in
divi
dual
s of
16
ce
ntre
s at
ho
me
(out
door
and
ind
oor)
. T
wo
defin
itio
ns f
or
sym
ptom
s of
ch
roni
c br
onch
itis
wer
e
cons
ider
ed:
1)
prod
uctiv
e ch
roni
c co
ugh
for
chro
nic
coug
h a
nd c
hron
ic p
hleg
m (
mor
e th
an
thre
e m
onth
s ea
ch y
ear,
an
d 2)
chr
onic
phl
egm
al
one.
R
esu
lts
from
th
e la
tter
defin
ition
w
ere
sho
wn.
H
iera
rchi
cal
mo
dels
w
ere
us
ed.
Con
foun
ding
va
riabl
es
(incl
udin
g sm
okin
g,
educ
atio
n,
occu
patio
nal
grou
ps,
occu
patio
nal
expo
sure
s,
resp
irato
ry
infe
ctio
ns
durin
g ch
ildho
od,
rhin
itis,
and
ast
hm
a) w
ere
reta
ined
at
indi
vidu
al l
eve
l if
p <
0.1
0 or
the
coe
ffici
ent
of
the
air
pollu
tion
varia
ble
was
mod
ified
by
10%
or
m
ore.
F
or
PM
2.5,
th
e as
soci
atio
n w
as
mea
sure
d w
ith
inte
rval
o
dds
ratio
(t
hat
is,
cove
ring
80%
of t
he O
Rs)
.
Tra
ffic
inte
nsity
as
w
ell
as
hom
e ou
tdoo
r N
O2
wer
e as
soci
ated
with
(pr
eval
ence
and
ne
w o
nse
t of
= c
hron
ic p
hleg
m,
in w
omen
on
ly.
PM
2.5
and
S c
onte
nt a
t ce
ntre
leve
l did
no
t sh
ow
any
ass
ocia
tion
with
pre
vale
nce
or
new
ons
et o
f ch
roni
c ph
leg
m.
Sim
ilar
resu
lts
wer
e
obta
ine
d w
ith
chro
nic
prod
uctiv
e co
ugh.
Inte
rval
OR
(95
% C
I))
for
PM
2.5
in
µg/
m3
Mal
es: 0
.97
(0.7
0, 1
.35)
F
emal
es: 0
.99
(0.8
5, 1
.17)
.
report no. 4/11
152
CES
AR
stu
dy
Le
onar
di
et
al.
(200
0)
(Leo
nard
i et a
l, 20
00)
17
citie
s of
C
entr
al
Eur
ope
Y
early
av
erag
e co
ncen
trat
ion
(Nov
. 19
95
- O
ct.
1996
) ac
ross
th
e
17
stud
y ar
eas
varie
d
from
41
to 9
6 fo
r P
M10
, fr
om 2
9 to
67
for
PM
2.5,
an
d fr
om
12
to
38
for
PM
10-2
.5.
Cro
ss-s
ectio
nal
st
udy
colle
cted
bl
ood
an
d se
rum
sam
ples
fro
m 1
0-61
sch
ool c
hild
ren
age
d 9
to 1
1 in
ea
ch c
omm
unity
11
Apr
il to
10
May
19
96.
Blo
od a
nd s
erum
sam
ples
exa
min
ed f
or
para
met
ers
in
rela
tion
to
PM
. F
inal
a
naly
sis
grou
p of
36
6 ex
amin
ed
for
perip
hera
l ly
mph
ocyt
e ty
pe
and
tota
l im
mun
oglo
bulin
cl
asse
s. A
sso
ciat
ion
betw
een
PM
and
eac
h lo
g
tran
sfor
med
bi
omar
ker
stud
ied
by
linea
r re
gres
sion
in
two-
stag
e m
odel
with
adj
ust
men
t fo
r co
nfou
ndin
g fa
ctor
s (a
ge,
gend
er,
num
ber
of
smok
ers
in
hous
e,
labo
rato
ry,
and
rece
nt
resp
irato
ry i
llnes
s).
Thi
s su
rvey
was
con
duc
ted
with
in t
he f
ram
e w
ork
of t
he
Cen
tral
Eur
opea
n st
udy
of
Air
Qua
lity
and
Res
pira
tory
H
ealth
(C
EA
SA
R)
stu
dy.
Num
ber
of l
ymph
ocyt
es (
B,
CD
4+
, C
D8
d,
and
NK
) in
crea
sed
with
in
crea
sin
g co
ncen
trat
ion
of
P
M
adju
sted
fo
r co
nfou
nder
s.
The
ad
just
ed
regr
essi
on
slop
es
are
larg
est
and
stat
istic
ally
si
gnifi
cant
for
PM
2.5
as c
om
pare
d to
PM
10,
but
smal
l an
d no
n st
atis
tical
ly s
igni
fican
t fo
r P
M10
-2.5.
Pos
itive
re
latio
nshi
p fo
und
betw
een
conc
entr
atio
n of
Ig
G i
n se
rum
and
P
M2.
5 bu
t no
t fo
r P
M10
or
PM
10-2
.5.
Tw
o ot
her
m
odel
s pr
oduc
ed s
imila
r ou
tcom
es:
a m
ulti-
le
vel l
inea
r re
gres
sion
mod
el a
nd a
n or
dina
l lo
gist
ic r
egre
ssio
n m
odel
.
Adj
uste
d R
egre
ssio
n sl
ope
P
M2.
5
CD
4+
80%
(34
, 143
) p<
0.00
1
Tot
al Ig
G
24%
(2,
52)
p=
0.03
4.
Subn
atio
nal s
tudi
es
G
REA
T B
RIT
AIN
Sol
omon
et
al
. (2
003)
(S
olom
on e
t al,
2003
) 11
el
ecto
ral
war
ds
of
Eng
land
and
Wal
es, U
K.
1,16
6 w
omen
M
ean
BS
le
vel,
rang
e
acro
ss w
ards
: 19
66-1
969:
40/
180
1978
-198
1: 1
4/79
19
94-1
997:
4/1
4.
To
exp
lore
th
e lo
ng-t
erm
influ
ence
of
part
icul
ate
air
pollu
tion
on
ca
rdio
-res
pira
tory
m
orbi
dity
in
th
e U
K,
a cr
oss-
sect
iona
l po
stal
su
rvey
w
as
cond
ucte
d in
11
elec
tora
l war
ds o
f U
K o
n 1
,166
w
omen
. W
ard
s w
ere
clas
sifie
d in
hig
h (B
S>
120
µg/
m3 i
n 19
66-
1969
) an
d lo
w (
BS
<50
µg/
m3 i
n 19
66-1
969)
pol
lutio
n ca
tego
ries.
W
omen
fro
m t
he t
wo
cate
gorie
s w
ere
com
pare
d in
ter
ms
of s
elf-
repo
rted
his
tory
of
asth
ma
and
prod
uctiv
e co
ugh
(and
IHD
).
The
pre
vale
nce
of a
sthm
a w
as
low
er i
n th
e ar
eas
with
hig
her
leve
ls o
f B
S.
No
exc
ess
ris
k of
pro
duct
ive
coug
h w
as f
ound
in
the
area
s w
ith h
ighe
r le
vels
of B
S.
RR
for
wom
en
livin
g in
the
are
as
with
hig
her
BS
leve
ls:
Ast
hma:
0.7
(0.
5, 1
.0)
Pro
duct
ive
coug
h: 1
.0 (
0.7,
1.4
).
report no. 4/11
153
Whe
eler
and
Ben
-Shl
omo
(200
5)
(Whe
eler
an
d B
en-S
hlom
o, 2
005)
Li
nkag
e an
alys
is,
Eng
land
, 199
5-19
97.
Mea
n (S
D)
PM
10:
23.9
5 (3
.58)
; ra
nge:
17
.87-
41.3
7.
Thi
s st
udy
exam
ined
th
e re
latio
n be
twee
n so
cioe
cono
mic
sta
tus
and
loca
l ai
r qu
ality
, an
d
com
bine
d ef
fect
s on
res
pira
tory
hea
lth,
Dat
a on
pe
ople
ta
king
pa
rt
in
the
Hea
lth
Sur
veys
fo
r E
ngla
nd
durin
g 19
95-1
997
wer
e at
trib
uted
with
a s
mal
l are
a in
dex
of a
ir po
llutio
n ba
sed
on a
nnua
l m
ean
conc
entr
atio
ns o
f P
M10
, be
nzen
e,
SO
2 an
d N
O2.
R
egre
ssio
n m
odel
s af
ter
adju
stm
ent
for
smok
ing
stat
us,
pass
ive
smok
ing,
B
MI,
rura
lity,
pr
evio
us
asth
ma
diag
nosi
s an
d in
hale
r us
e w
ere
used
to
m
easu
re a
ssoc
iatio
ns b
etw
een
soci
al c
lass
, ai
r qu
ality
, ag
e-,
sex-
an
d he
ight
-sta
ndar
dize
d F
EV
1, a
nd s
elf r
epor
ted
asth
ma.
Urb
an l
ower
soc
ial
clas
s ho
useh
olds
wer
e m
ore
likel
y to
be
loca
ted
in a
reas
of
poor
air
qual
ity,
but
the
asso
ciat
ion
in
rura
l ar
eas
was
, if
anyt
hing
rev
erse
d. L
ow s
ocia
l cl
ass
and
poor
ai
r qu
ality
w
ere
inde
pend
ently
as
soci
ated
w
ith
decr
ease
d lu
ng
func
tion
(FE
V1)
, bu
t no
t as
thm
a pr
eval
ence
, af
ter
adju
stm
ent
for
a nu
mbe
r of
po
tent
ial
conf
ound
ers.
Soc
ial
clas
s ef
fect
s w
ere
not
atte
nuat
ed
by
adju
stm
ent
for
air
qual
ity.
Aut
hors
co
nclu
ded
that
th
e as
soci
atio
n be
twee
n F
EV
1 an
d lo
cal
air
qual
ity w
as o
f si
mila
r m
agni
tude
to
that
with
soc
ial
clas
s,
and
the
adve
rse
effe
cts
of
air
pollu
tion
seem
ed t
o be
gre
ater
in m
en in
low
er s
ocia
l cl
asse
s.
No
estim
ates
fo
r P
M
wer
e pr
ovid
ed.
Pie
rse
et
al.
(200
6)
(Pie
rse
et a
l, 20
06)
Coh
ort
of 4
,400
chi
ldre
n fr
om L
eice
ster
, UK
. 19
98-2
001.
A
nnua
l m
ean
conc
entr
atio
n of
P
M10
w
as
1.47
in
19
98
and
1.33
in 2
001.
Thi
s st
udy
exam
ined
he
effe
ct o
f pr
imar
y P
M10
(p
artic
les
dire
ctly
em
itted
fro
m lo
cal s
ourc
es)
on
prev
alen
ce
and
inci
denc
e of
re
spira
tory
sy
mpt
oms
in a
ran
dom
sam
ple
coho
rt o
f 4,
400
Leic
este
rshi
re
child
ren
aged
1–
5 ye
ars
surv
eyed
in
1998
and
aga
in i
n 20
01.
Ann
ual
expo
sure
to
prim
ary
PM
10 w
as c
alcu
late
d fo
r th
e ho
me
addr
ess
usin
g th
e A
IVIR
O
disp
ersi
on
mod
el a
nd O
Rs
wer
e ca
lcul
ated
for
eac
h µ
g/m
3 in
crea
se,
afte
r ad
just
men
t fo
r ag
e, s
ex,
fam
ily
hist
ory
of
asth
ma,
co
al
heat
ing
in
the
hom
e,
hous
ehol
d m
embe
r’s
smok
ing
stat
us,
and
mat
erna
l and
pat
erna
l edu
catio
n.
Exp
osur
e to
prim
ary
PM
10 w
as a
ssoc
iate
d w
ith t
he p
reva
lenc
e of
cou
gh w
ithou
t a
cold
in
bo
th
1998
an
d 20
01,
and
with
th
e pr
eval
ence
of
nigh
t tim
e co
ugh
and
curr
ent
whe
eze
in
2001
, on
ly.
An
asso
ciat
ion
betw
een
prim
ary
PM
10
and
inci
denc
e of
co
ugh
with
out
a co
ld a
nd w
heez
e w
as a
lso
foun
d.
Adj
uste
d O
R
(95%
C
I)
for
an
in
crem
ent o
f 1 µ
g/m
3 P
reva
lenc
e:
Cou
gh w
ithou
t a c
old
19
98: 1
.21
(1.0
7, 1
.38)
20
01: 1
.56
(1.3
2, 1
.84)
N
ight
tim
e co
ugh
19
98: 1
.06
(0.9
4, 1
.19)
20
01: 1
.25
(1.0
6, 1
.47)
C
urre
nt w
heez
e
1998
: 0.9
9 (0
.88,
1.1
2)
2001
: 1.2
8 (1
.04,
1.5
8)
Inci
denc
e (n
ew o
nset
sym
ptom
s,
not
pres
ent
in 1
998
surv
ey a
nd
pres
ent
in t
he 2
001
surv
ey v
ersu
s no
sym
ptom
s in
bot
h su
rvey
s)
Cou
gh w
ithou
t a c
old
1.
62 (
1.31
, 2.0
0)
Nig
ht ti
me
coug
h
1.19
(0.
96, 1
.47)
C
urre
nt w
heez
e
1.42
(1.
02, 1
.97)
.
report no. 4/11
154
Kul
karn
i et
al
. (2
006)
(K
ulka
rni e
t al,
2006
) Le
ices
ter,
UK
. N
ov 2
002
to D
ec 2
003.
A
nnua
l m
ean
leve
l of
P
M10
w
as
1.21
(r
ange
: 0.
10-2
.71)
in
he
alth
y ch
ildre
n an
d 1.
81 (
rang
e:
0.17
-2.1
3)
in
asth
mat
ic
child
ren.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
carb
on
cont
ent
ai
rway
m
acro
phag
es,
as
a m
arke
r of
in
divi
dual
e
xpo
sure
to
P
M
deriv
ed
from
fos
sil f
uel,
and
lung
fun
ctio
n in
9 a
sth
mat
ic
and
64
heal
thy
child
ren
from
Le
ices
ter,
U
K.
Air
way
m
acro
phag
es
we
re
obta
ined
fr
om
heal
thy
child
ren
thro
ugh
sput
um i
nduc
tion,
and
th
e ar
ea o
f ai
rway
mac
rop
hage
s oc
cupi
ed b
y ca
rbon
w
as
mea
sure
d.
Lung
fu
nctio
n w
as
mea
sure
d w
ith
the
use
of
spiro
met
ry.
The
ex
posu
re t
o P
M10
was
mo
del
ed a
t or
nea
r ea
ch
child
’s
hom
e
addr
ess.
Li
nea
r re
gres
sio
n w
as
used
to
eval
uate
ass
ocia
tions
bet
we
en c
arbo
n co
nten
t of
alv
eola
r m
acro
phag
es a
nd v
aria
bles
th
at m
ay a
ffect
ind
ivid
ual
expo
sure
. A
ge,
se
x,
race
, w
eig
ht,
heig
ht,
birt
h or
der,
nu
mbe
r of
si
blin
gs,
exe
rcis
e m
easu
res
and
cotin
ine
leve
ls
in
saliv
a w
ere
con
sid
ered
as
po
tent
ial
conf
ound
ing
varia
bles
.
An
incr
ease
in
PM
10 w
as a
ssoc
iate
d w
ith a
si
gnifi
cant
inc
reas
e in
the
ca
rbon
con
tent
of
airw
ay
mac
roph
ages
. C
arb
on c
onte
nt w
as
inve
rsel
y as
soci
ated
w
ith
FE
V1,
F
VC
an
d F
EF
25-7
5.
Incr
ease
d pr
imar
y P
M10
w
as
inve
rsel
y as
soci
ated
with
th
e pe
rcen
tage
of
the
pred
icte
d F
EV
1 (P
=
0.
04)
and
the
FE
F25
–75
(P
=
0.05
).
In
the
mul
tiple
re
gres
sion
ana
lysi
s, e
ven
afte
r ad
just
ing
for
the
carb
on c
onte
nt o
f ai
rwa
y m
acro
phag
es,
PM
10 w
as n
o l
onge
r si
gnifi
cant
ly a
ssoc
iate
d w
ith
lung
fu
nctio
n,
whe
reas
th
e ca
rbon
co
nten
t of
ai
rwa
y m
acro
phag
es
rem
aine
d in
vers
ely
asso
ciat
ed w
ith t
he
perc
enta
ge o
f th
e pr
edic
ted
valu
es
of
FE
V1,
F
VC
, a
nd
FE
F25
–75.
T
he
carb
on
cont
ent
of
airw
ay
mac
roph
ages
w
as
low
er
in
child
ren
with
as
thm
a th
an
in
heal
thy
child
ren.
A
uth
ors
conc
lude
d th
at
whi
le
ther
e w
as
a do
se-
depe
nden
t in
vers
e as
soci
atio
n be
twee
n th
e ca
rbon
con
ten
t of
airw
ay
mac
roph
ages
and
lu
ng
func
tion
in
ch
ildre
n,
ther
e w
as
no
evid
ence
th
at
redu
ced
lun
g fu
nctio
n its
elf
caus
ed a
n in
crea
se in
car
bon
cont
ent.
Exc
ess
carb
on
cont
ent
(µm
2 ) an
d lu
ng f
unct
ion
(%)
asso
ciat
ed
with
an
in
crem
ent
of
one
µg/
m3
of
PM
10.
carb
on
cont
ent
airw
ay
mac
roph
ages
: 0.
10 (
0.01
, 0.1
8)
FE
V1
-4.3
% (
-8.5
, -0.
2)
afte
r al
low
anc
e fo
r ca
rbon
co
nten
t: -2
.9%
(-6
.9, 1
.2)
FV
C
-1.2
% (
-5.6
, 3.2
) af
ter
allo
wa
nce
for
carb
on
cont
ent:
0.1%
(-4
.4, 4
.6)
FE
V1
-8.6
% (
-17.
3, 0
.1)
afte
r al
low
anc
e fo
r ca
rbon
co
nten
t: -5
.5%
(-1
4.2,
3.1
).
report no. 4/11
155
Bur
r et
al.
(200
4) (
Bur
r et
al
, 200
4)
Inte
rven
tion
stud
y,
Nor
th
Wal
es, U
K.
mea
n co
ncen
trat
ion
of
PM
10
Con
gest
ed s
tree
ts:
1996
-199
7: 3
5.2
1998
-199
9:27
.2
Unc
onge
sted
str
eets
: 19
96-1
997:
11.
6 19
98-1
999:
8.2
PM
2.5
C
onge
sted
str
eets
: 19
96-1
997:
21.
2 19
98-1
999:
16.2
U
ncon
gest
ed s
tree
ts:
1996
-199
7: 6
.7
1998
-199
9: 4
.9.
Thi
s st
udy
exam
ine
d w
heth
er
resi
dent
s of
co
nges
ted
stre
ets
had
a hi
gher
pre
vale
nce
of
resp
irato
ry
sym
ptom
s th
an
resi
dent
s of
un
cong
este
d
stre
ets,
an
d w
heth
er
thei
r re
spira
tory
he
alth
im
prov
es
follo
win
g a
redu
ctio
n in
ex
posu
re
to
traf
fic
rela
ted
air
pollu
tant
s,
due
to
the
con
stru
ctio
n of
a
“b
y-pa
ss”.
A r
espi
rato
ry s
urve
y w
as c
ondu
cted
at
base
line
and
ag
ain
a ye
ar
afte
r th
e by
-pas
s op
ened
on
16
5 an
d 28
3 su
bjec
ts
in
the
cong
este
d
and
unco
nges
ted
ar
eas,
re
spec
tivel
y.
Initi
al
conc
entr
atio
ns
of
PM
10
and
PM
2.5
wer
e su
bsta
ntia
lly h
ighe
r in
the
con
gest
ed
than
in
the
unco
nges
ted
stre
ets.
Whe
n th
e by
-pas
s op
ene
d,
the
volu
me
of
heav
y go
ods
traf
fic
fell
by
nea
rly
50%
. P
M10
de
crea
sed
by
23%
(8
.0
mg/
m3 )
in
the
cong
este
d st
reet
s an
d by
29%
(3.
4 m
g/m
3 ) in
th
e un
cong
este
d st
reet
s,
with
si
mila
r pr
opor
tiona
te
falls
in
P
M2
.5.
The
by
-pas
s sh
ow
ed a
ten
denc
y fo
r m
ost
sym
ptom
s to
im
prov
e in
bo
th
area
s.
The
im
prov
emen
t te
nded
to
be
gr
eate
r in
th
e un
cong
este
d ar
ea f
or c
hest
sym
ptom
s an
d fo
r pe
ak f
low
va
riabi
lity,
an
d in
the
con
ges
ted
stre
ets
for
rhin
itis
and
rhin
ocon
junc
tiviti
s T
he b
y-pa
ss
redu
ced
pollu
tant
le
vels
to
a
degr
ee
that
pr
obab
ly
alle
viat
es
rhin
itis
and
rhin
ocon
junc
tiviti
s bu
t h
as
little
ef
fect
on
lo
wer
res
pira
tory
sym
ptom
s.
No
estim
ates
fo
r P
M10
no
r fo
r P
M2.
5 w
ere
prov
ided
.
THE
NET
HER
LAN
DS
B
raue
r et
al
. (2
002)
; (B
raue
r et
al,
2002
) B
irth
coho
rt
of
3,93
4 ch
ildre
n.
PM
2.5
mea
n: 1
6.9
(ran
ge:
13.5
-25.
2).
Thi
s st
udy
exam
ined
the
rel
atio
nshi
p be
twee
n tr
affic
-rel
ated
ai
r po
llutio
n an
d
the
pred
evel
opm
ent
of a
sthm
atic
/alle
rgic
sym
ptom
s an
d re
spira
tory
inf
ectio
ns (
incl
udin
g w
heez
ing,
dr
y ni
ght-
time
coug
h, e
ar,
nose
, th
roat
(E
,N,T
) in
fect
ions
, sk
in r
ash,
and
phy
sici
an-d
iag
nose
d as
thm
a, b
ronc
hitis
, in
fluen
za,
and
ecze
ma
at 2
ye
ars
of a
ge)
in a
coh
ort
of a
ppro
xim
ate
ly 4
,000
bi
rths
. A
va
lidat
ed m
odel
wa
s us
ed t
o as
sign
ou
tdoo
r co
ncen
trat
ions
of
P
M2.
5,
NO
2 at
th
e ho
me
of
each
su
bjec
t o
f th
e co
hort
. T
he
asso
ciat
ion
bet
we
en P
M e
xpos
ure
and
vario
us
outc
omes
w
as
anal
yzed
by
m
ultip
le
logi
stic
re
gres
sion
af
ter
adju
stm
ent
for
vario
us
cova
riate
s,
incl
udin
g se
x,
smok
ing
an
d ed
ucat
ion
of p
aren
ts,
fam
ily h
isto
ry o
f al
lerg
ies,
an
d m
othe
r’s a
ge.
Adj
uste
d od
ds
ratio
s fo
r w
heez
ing,
ph
ysic
ian-
diag
nose
d as
thm
a,
E,N
,T
infe
ctio
ns,
and
flu/s
erio
us
cold
s in
dica
ted
posi
tive
asso
ciat
ions
w
ith
air
pollu
tant
s,
som
e of
whi
ch r
each
ed b
orde
rline
sta
tistic
al
sign
ifica
nce.
N
o as
soci
atio
ns
wer
e ob
serv
ed fo
r th
e ot
her
outc
omes
ana
lyze
d.
Adj
uste
d O
R (
95%
CI)
for
an
IQR
ch
ange
in
co
ncen
trat
ion
of
PM
(i.
e., 3
.2µ
g/m
3 ) W
heez
e: 1
.14
(0.9
8, 1
.34)
A
sthm
a: 1
.12
(0.8
4, 1
.50)
D
ry
coug
h at
ni
ght:
1.04
(0
.88,
1.
23)
Bro
nchi
tis: 1
.04
(0.8
5, 1
.26)
E
,N,T
infe
ctio
ns: 1
.20
(1.0
1, 1
.42)
F
lu/s
erio
us
cold
s:
1.12
(1
.00,
1.
27)
Itchy
ras
h: 1
.01
(0.8
8, 1
.16)
E
czem
a: 0
.95
(0.8
3, 1
.10)
.
report no. 4/11
156
Bra
uer
et
al.
(200
6);
(Bra
uer
et a
l, 20
06)
Birt
h co
hort
s on
3,
700
and
650
infa
nts
fro
m t
he
Net
herla
nds
and
Mun
ich
(Ger
man
y), r
espe
ctiv
ely.
P
erio
d of
re
crui
tmen
t: 19
97-1
998.
P
M2.
5 m
ean:
T
he N
ethe
rland
s: 1
6.9
Mun
ich,
Ger
man
y: 1
3.4.
Thi
s st
udy
anal
ysed
th
e as
soci
atio
n b
etw
een
P
M2.
5 el
emen
tal
carb
on,
and
NO
2, a
nd r
isk
of
otiti
s m
edia
in t
wo
birt
h co
hort
s of
app
roxi
mat
ely
3,70
0 an
d 65
0 in
fant
s fr
om t
he N
ethe
rland
s an
d
Ger
man
y, r
espe
ctiv
ely.
Air
pol
luta
nts
at h
ome
of
child
ren
wer
e m
easu
red
at 4
0 in
divi
dua
l si
tes
in
each
co
untr
y,
and
anal
yzed
in
re
latio
n to
di
agno
sed
hist
ory
of o
titis
med
ia (
diag
nose
d by
a
phys
icia
n)b
y pa
rent
s of
chi
ldre
n in
the
firs
t 2
year
s of
lif
e.
OR
s w
ere
deriv
ed
by
mul
tiple
lo
gist
ic r
egre
ssio
n m
odel
s af
ter
adju
stm
ent
for
pote
ntia
l co
nfou
ndin
g va
riabl
es,
incl
udin
g se
x,
pare
ntal
at
opy
, m
ater
nal
educ
atio
n
and
smok
ing,
en
viro
nmen
tal
toba
cco
smok
e an
d pr
esen
ce o
f pet
at h
ome,
and
sib
lings
.
Som
e si
gnifi
cant
di
rect
as
soci
atio
ns
we
re
obse
rved
bet
wee
n va
rious
mea
sure
s of
air
pollu
tion
and
in
cide
nce
of
otiti
s m
edia
. C
rude
an
d
adju
sted
O
Rs
wer
e sy
stem
atic
ally
ov
er
unity
fo
r al
l th
e co
nsid
ered
po
lluta
nts
and
in
bo
th
the
coho
rts.
Adj
uste
d O
R f
or a
n in
crea
se i
n 3
µg/
m3 o
f PM
2.5:
The
Net
herla
nds
At 1
yea
r of
age
: 1.1
3 (0
.98,
1.3
2)
At
2 ye
ars
of
age:
1.
13
(1.0
0,
1.27
) M
unic
h, G
erm
any
At 1
yea
r of
age
: 1.1
9 (0
.73,
1.9
2)
At
2 ye
ars
of
age:
1.
24
(0.8
4,
1.83
).
GER
MA
NY
H
einr
ich
et
al.
(199
9)
(Hei
nric
h et
al,
1999
) B
itter
feld
, Z
erbs
tand
H
etts
tedt
are
as
of fo
rmer
E
ast G
erm
any,
D
urin
g S
ept.
1992
to
July
19
93 T
SP
ran
ged
from
44
to 6
5;
PM
10
mea
sure
d O
ctob
er
1993
-
Mar
ch
1994
ra
nged
fro
m 3
to
40;
and
BS
ran
ged
fro
m 2
6 to
42.
P
M w
as m
easu
red
with
a
Har
vard
impa
ctor
.
Par
ents
of
247
0 sc
hool
chi
ldre
n (
5-14
yea
r)
com
plet
ed
resp
irato
ry
hea
lth
ques
tionn
aire
. C
hild
ren
exc
lude
d fr
om a
naly
sis
if ha
d liv
ed
< 2
ye
ars
in t
heir
curr
ent
hom
e, y
ield
ing
an a
naly
sis
grou
p of
2,
335
child
ren.
O
utco
mes
st
udie
d:
phys
icia
n d
iagn
osis
fo
r as
thm
a,
bron
chiti
s,
sym
ptom
, br
onch
ial
reac
tivity
, sk
in
pric
k te
st,
spec
ific
IgE
. M
ultip
le
logi
stic
re
gre
ssio
n an
alys
es e
xam
ined
reg
iona
l effe
cts.
Con
trol
ling
for
med
ical
, so
cio-
dem
ogra
phic
, an
d in
door
fac
tors
, ch
ildre
n in
mor
e po
llute
d ar
ea h
ad c
irca
50%
inc
rea
se f
or b
ronc
hitic
sy
mpt
oms
and
phys
icia
n-di
agno
sed
alle
rgie
s co
mpa
red
to c
ontr
ol a
rea
and
circ
a tw
ice
the
re
spira
tory
sy
mp
tom
s (w
heez
e,
shor
tnes
s of
bre
ath
and
coug
h).
Pul
mon
ary
func
tion
test
s su
gges
ted
slig
htly
inc
reas
ed
airw
ay
reac
tivity
to
co
ld
for
child
ren
in
po
llute
d ar
ea.
No
sing
le
pol
luta
nt
coul
d be
se
para
ted
out
as
bein
g
resp
onsi
ble
for
poor
re
spira
tory
he
alth
.
report no. 4/11
157
Hei
nric
h et
al
. (2
000)
(H
einr
ich
et a
l, 20
00)
Thr
ee a
reas
of
form
er E
. G
erm
any
Pol
lutio
n m
easu
res:
SO
2,
TS
P,
and
som
e lim
ited
P
M10
da
ta.
TS
P
decr
ease
d fr
om
65,
48,
and
44 t
o 43
, 39
, an
d 36
in
the
thre
e ar
eas.
P
M w
as m
easu
red
with
a
Har
vard
impa
ctor
.
Cro
ss-s
ectio
nal
st
udy
of
child
ren
(5-1
4 ye
ar).
S
urve
y co
nduc
ted
twic
e,
in
1992
-199
3 an
d 19
95-1
996;
2,
335
child
ren
surv
eyed
in
fir
st
roun
d, a
nd 2
,536
in
seco
nd
roun
d. O
nly
971
child
ren
app
eare
d in
bo
th
surv
eys.
T
he
freq
uenc
y of
br
onch
itis,
ot
itis
med
ia,
fre
quen
t co
lds,
fe
brile
in
fect
ions
st
udie
d.
Bec
ause
ch
ange
s m
easu
red
over
tim
e in
sa
me
area
s,
cova
riate
adj
ust
men
ts n
ot n
eces
sary
.
PM
and
SO
2 le
vels
bot
h d
ecre
ased
in
the
sam
e ar
eas;
so
resu
lts a
re c
onfo
unde
d.
The
pre
vale
nce
of a
ll re
spira
tory
sy
mpt
oms
decr
ease
d si
gnifi
cant
ly
in a
ll th
ree
area
s ov
er ti
me.
Hei
nric
h et
al
. (2
002)
(H
einr
ich
et a
l, 20
02)
Sur
veye
d ch
ildre
n ag
ed
5-14
in 1
992-
3, 1
995-
6,
1998
-9.
An
nual
T
SP
le
vels
ran
ged
from
25-
79.
Sm
all
part
icle
s (N
C0.
01-2
.5
per
103c
m-3
) re
mai
ned
rela
tivel
y co
nsta
nt.
A t
wo-
stag
e l
ogi
stic
reg
ress
ion
mod
el w
as
used
to
an
alyz
e
the
data
w
hich
ad
just
ed
for
age,
ge
nder
, ed
ucat
iona
l lev
el o
f pa
rent
s, a
nd in
door
fa
ctor
s. T
he m
odel
inc
lude
d fix
ed a
rea
effe
cts,
ra
ndom
de
viat
ions
, an
d er
rors
fr
om
the
adju
stm
ents
. P
aram
eter
s w
ere
estim
ated
usi
ng
GE
E m
etho
ds.
The
stu
dy f
ound
bro
nchi
tis a
nd f
requ
ency
of
cold
s w
ere
sig
nific
antly
rel
ated
to T
SP
. A
n in
crem
ent
of
50
µg/
m3
TS
P
wa
s as
soci
ated
w
ith
an
OR
fo
r br
onch
itis
of 3
.02
(1.7
2, 5
.29)
and
an
O
R
of
1.90
(1
.17,
3.
09)
fo
r fr
eque
ncy
of c
olds
.
Fry
e et
al.
(200
3) (
Fry
e et
al
, 200
3)
Sur
veye
d ch
ildre
n in
19
92-3
, 19
95-6
, 19
98-9
. A
nnua
l T
SP
le
vels
ra
nged
from
23-
79.
Thi
s st
udy
exam
ined
the
effe
cts
of i
mpr
oved
air
qual
ity
sinc
e 19
90
in
Eas
t G
erm
any
on
lung
fu
nctio
n in
ch
ildre
n. T
hree
con
secu
tive
cros
s-se
ctio
nal
surv
eys
of s
choo
lchi
ldre
n ag
ed 1
1–14
ye
ars
from
thr
ee c
omm
uniti
es i
n E
ast
Ge
rman
y w
ere
perf
orm
ed
durin
g 19
92-1
999.
Lu
ng
func
tion
test
s w
ere
ava
ilabl
e fr
om
2,49
3 ch
ildre
n.
Per
cent
ch
ange
of
lu
ng
func
tion
para
met
ers
acco
rdin
g to
a d
ecre
ase
of T
SP
by
50
µg/
m3
we
re
deriv
ed
by
linea
r re
gres
sion
m
odel
s af
ter
adj
ustm
ent
for
sex,
hei
ght,
sea
son,
lu
ng
func
tion
eq
uipm
ent,
pare
ntal
ed
ucat
ion,
pa
rent
al
atop
y an
d en
viro
nmen
tal
toba
cco
smok
e.
FV
C
and
FE
V1
of
the
child
ren
incr
ease
d fr
om
1992
–199
3 to
19
98–1
999.
D
ecre
men
ts o
f T
SP
(p
=0.
043)
and
SO
2 (p
=
0.02
9) w
ere
sign
ifica
ntly
ass
ocia
ted
with
an
impr
ovem
ent
of F
VC
. E
ffect
s on
FE
V1
we
re
smal
ler
and
not
sta
tistic
ally
sig
nific
ant.
50 µ
g/m
3 dec
reas
e of
TS
P.
Adj
uste
d p
erce
nt
chan
ge
(95%
C
I):
FV
C
Tot
al: 4
.7%
(0
.2, 9
.5)
Boy
s: 3
.7%
(-1
.2, 8
.8)
Girl
s: 5
.9%
(0.
8, 1
1.1)
F
EV
1
Tot
al: 2
.9%
(-1
.4, 7
.3)
Boy
s: 1
.7%
(-2
.9, 6
.6)
Girl
s: 4
.1%
(-0
.7, 9
.2).
report no. 4/11
158
Krä
mer
et
al
. (1
999)
(K
ram
er e
t al,
1999
) S
ix
Ea
st
and
Wes
t G
erm
any
com
mun
ities
(L
eipz
ig,
Hal
le,
Mad
debu
rg,
Altm
ark,
D
uisb
urg,
Bor
ken)
B
etw
een
199
1 an
d 19
95
TS
P
leve
ls
in
six
com
mun
ities
ran
ged
from
46
to
10
2.
Eac
h E
ast
Ger
man
y co
mm
unity
had
de
crea
se in
TS
P b
etw
een
19
91 a
nd 1
99
5. T
SP
wa
s m
easu
red
usin
g a
low
vo
lum
e sa
mpl
er.
The
stu
dy a
sses
sed
rela
tions
hip
betw
een
TS
P
and
airw
ay d
isea
se a
nd a
llerg
ies
by
pare
ntal
qu
estio
nnai
res
in y
ear
ly s
urve
ys o
f ch
ildre
n (5
-8
year
) be
twe
en
Feb
ruar
y an
d M
ay.
The
qu
estio
ns i
nclu
ded
pneu
mon
ia,
bron
chiti
s ev
er
diag
nose
d b
y ph
ysic
ian,
nu
mbe
r of
co
lds,
fr
eque
nt
coug
h,
alle
rgic
sy
mpt
oms.
In
al
l, 19
,090
chi
ldre
n pa
rtic
ipat
ed.
Ave
rage
res
pon
se
wa
s 87
%.
An
alys
es w
ere
cond
ucte
d on
14,
144
child
ren
for
who
m i
nfor
mat
ion
on a
ll co
varia
tes
wer
e a
vaila
ble
. V
aria
bles
in
clud
ed
gen
der;
pa
rent
ed
ucat
ion,
he
atin
g fu
el,
ET
S.
Log
istic
re
gres
sion
us
ed
to
allo
w
for
time
tren
ds
and
SO
2 an
d T
SP
effe
cts.
Re
gres
sion
coe
ffic
ient
s w
ere
con
vert
ed to
OR
s.
TS
P a
nd S
O2
sim
ulta
neou
sly
incl
uded
in t
he
mod
el.
Bro
nchi
tis e
ver
diag
nose
d sh
ow
ed
a si
gnifi
cant
ass
ocia
tion.
A d
ecre
ase
in r
aw
perc
enta
ge w
as s
een
betw
een
the
star
t of
th
e st
udy
and
th
e en
d
for
bron
chiti
s.
Bro
nchi
tis
seem
ed
to
be
asso
ciat
ed
onl
y w
ith
TS
P
in
spite
of
hu
ge
diffe
renc
es
in
mea
n S
O2
leve
ls.
Bro
nchi
tis e
ver
diag
nose
d
TS
P p
er 5
0 µ
g/m
3 O
R 1
.63
(1.3
7, 1
.93)
H
alle
(E
ast)
C
alen
dar
Ye
ar;
TS
P
(µg/
m3 );
B
ronc
hitis
19
91; 1
02; 6
0.5
1992
; 73;
54.
7
1993
; 62;
49.
6
1994
; 52;
50.
4
1995
; 46;
51.
9
Geh
ring
et
al.
(200
2)
(Geh
ring
et a
l, 20
02)
In M
unic
h, G
erm
any
Dec
embe
r 19
97
- Ja
nuar
y 19
99
Ann
ual
PM
2.5
leve
ls
dete
rmin
ed
by
40
site
s an
d a
GIS
pr
edic
tor
for
mod
el.
Mea
n P
M2.
5 an
nual
av
erag
e of
13
.4
with
ra
nge
of 1
1.90
to 2
1.90
.
Effe
ct o
f tr
affic
-rel
ated
air
pollu
tant
s. P
M2.
5 an
d N
O2
on
resp
irato
ry
heal
th
outc
omes
w
hee
ze,
coug
h,
bron
chiti
s,
resp
irato
ry
infe
ctio
ns,
and
runn
y no
se
wer
e ev
alua
ted
usin
g m
ultip
le
logi
stic
re
gres
sion
an
alys
es
of
l,756
ch
ildre
n du
ring
the
first
and
sec
ond
year
of
life
adju
stin
g fo
r po
tent
ial c
onfo
undi
ng fa
ctor
s.
The
re
wa
s so
me
ind
icat
ion
of
an
asso
ciat
ion
bet
we
en P
M2.
5 a
nd s
ympt
om
s of
co
ugh
but
not
othe
r o
utco
mes
. In
th
e se
cond
ye
ar
of
life
m
ost
ef
fect
s w
ere
atte
nuat
ed.
OR
(9
5%
CI)
fo
r a
chan
ge
in
PM
2.5
conc
entr
atio
n of
1.5
µg/
m3
Whe
eze
A
ge o
f 1 y
ear:
0.9
1 (0
.76,
1.0
9)
Age
of 2
yea
r: 0
.96
(0.8
3, 1
.12)
C
ough
with
out
infe
ctio
n
Age
of 1
yea
r: 1
.34
(1.1
1, 1
.61)
D
ry c
ough
at
nigh
t A
ge o
f 1 y
ear:
1.3
1 (1
.07,
1.6
0)
Age
of 2
yea
r: 1
.20
(1.0
2, 1
.42)
B
ronc
hitis
A
ge o
f 1 y
ear:
0.9
8 (0
.80,
1.2
0)
Age
of 2
yea
r: 0
.92
(0.7
8, 1
.09)
R
espi
rato
ry in
fect
ions
A
ge o
f 1 y
ear:
1.0
4 (0
.91,
1.1
9)
Age
of 2
yea
r: 0
.98
(0.8
0, 1
.20)
S
neez
ing
A
ge o
f 1 y
ear:
1.0
1 (0
.85,
1.2
0)
Age
of 2
yea
r: 0
.96
(0.8
2, 1
.12)
.
report no. 4/11
159
Sch
iko
wsk
i et
al. (
2005
) (S
chik
ow
ski e
t al
, 200
5)
Con
secu
tive
cr
oss-
sect
iona
l st
udie
s.
Rhi
ne-
Rhu
r B
asin
, Ger
man
y 19
85-1
994.
P
M10
ann
ual
mea
n of
43
with
ran
ge o
f 35
to 5
3
PM
10
five
year
m
ean
of
48 w
ith r
ange
of 3
9 to
56.
Thi
s st
udy
exa
min
ed
the
ef
fect
of
lo
ng-t
erm
ex
posu
re t
o P
M10
fro
m i
ndu
stry
and
tra
ffic
and
NO
2 on
CO
PD
, as
def
ined
by
lung
fun
ctio
n, i
n 4,
757
mid
dle-
aged
w
om
en
enro
lled
in
cons
ecut
ive
cros
s se
ctio
nal
stud
ies
cond
ucte
d be
twee
n 19
85
and
1994
in
th
e R
hin
e-R
uhr
Bas
in o
f G
erm
any.
NO
2 a
nd P
M10
exp
osur
e w
as
asse
ssed
by
mea
sure
men
ts d
one
in a
n 8
km
grid
, an
d tr
affic
exp
osur
e b
y di
stan
ce f
rom
the
re
side
ntia
l ad
dres
s to
the
nea
rest
maj
or r
oad
usin
g G
IS d
ata.
Lun
g fu
nctio
n w
as
dete
rmin
ed
and
CO
PD
w
as
defin
ed
by
usin
g
the
GO
LD
crite
ria.
Chr
oni
c re
spira
tory
sy
mpt
oms
and
poss
ible
co
nfou
nder
s w
ere
de
fine
d by
qu
estio
nnai
re
data
. Li
nea
r an
d lo
gist
ic
regr
essi
ons,
in
clud
ing
ran
dom
ef
fect
s w
ere
used
afte
r ad
just
men
t fo
r ag
e, s
ocio
eco
nom
ic
stat
us,
smok
ing,
env
ironm
enta
l to
bacc
o sm
oke,
oc
cupa
tiona
l ex
posu
re t
o te
mpe
ratu
re a
nd d
ust,
and
heat
ing
with
foss
il fu
els.
The
pre
vale
nce
of C
OP
D (
GO
LD s
tage
s 1
–4)
w
as
4.5
%.
CO
PD
an
d pu
lmon
ary
func
tion
wer
e
stro
nges
t af
fect
ed
by
PM
10
and
traf
fic r
elat
ed e
xpo
sure
. W
omen
liv
ing
less
tha
n 10
0 m
fro
m a
bus
y ro
ad a
lso
had
a si
gnifi
cant
ly d
ecre
ased
lun
g fu
nctio
n, a
nd
CO
PD
was
1.7
9 tim
es m
ore
likel
y th
an f
or
thos
e liv
ing
fart
her
awa
y. C
hro
nic
sym
pto
ms
as
base
d
on
ques
tionn
aire
in
form
atio
n sh
ow
ed e
ffect
s in
the
sam
e di
rect
ion,
but
le
ss p
rono
unce
d.
OR
(95
% C
I) f
or 7
µg/
m3 in
crea
se
in P
M10
(IQ
R)
A
nnua
l mea
ns
Chr
onic
br
onch
itis:
1.
00
(0.8
5,
1.18
) C
hron
ic c
ough
: 1.0
3 (0
.87,
1.2
3)
Fre
quen
t cou
gh: 1
.01
(0.9
3, 1
.10)
C
OP
D:
1.37
(0.
98, 1
.92)
F
EV
1: 0
.953
(0.
916,
0.9
89)
FV
C: 0
.966
(0
.940
, 0.9
92)
Fiv
e ye
ar m
eans
C
hron
ic
bron
chiti
s:
1.13
(0
.95,
1.
34)
Chr
onic
cou
gh: 1
.11
(0.9
3, 1
.31)
F
requ
ent c
ough
: 1.0
5 (0
.94,
1.1
7)
CO
PD
: 1.
33 (
1.03
, 1.7
2)
FE
V1:
0.9
49 (
0.92
3, 0
.975
) F
VC
: 0.9
63 (
0.9
45, 0
.982
).
Pol
at e
t al.
(20
02)
(Pol
at e
t al,
2002
) 4
mon
th
cros
s-se
ctio
nal
surv
ey i
n 4
loca
tions
(A
, B
1, B
2, C
) of
3 c
ities
in
Nor
drhe
in-W
est
fale
n,
Ger
man
y F
eb 2
000
to M
ay 2
000.
T
SP
mea
n (
SD
) Lo
catio
n
A: 4
9.1
(23.
9)
Loca
tion
B1:
78.
5 (2
9.8)
Lo
catio
n B
2: 4
6.7
(19.
3)
Loca
tion
C: 3
4.9
(12
.6.)
.
At
4 lo
catio
ns,
in 3
diff
eren
t ci
ties
of N
ordr
hein
-W
estfa
len,
TS
P,
O3,
NO
x an
d S
O2
wer
e d
eriv
ed
from
co
mpl
ianc
e m
easu
rem
ents
by
go
vern
men
tal
offic
es,
and
88
4 su
bjec
ts
(501
m
othe
rs a
nd 3
83 c
hild
ren,
6 ±
7 y
ears
old
) w
ere
scre
ened
usi
ng
nasa
l la
vag
e. A
utho
rs s
tudi
ed a
po
tent
ial a
sso
ciat
ion
betw
een
ambi
ent
exp
osur
e in
the
4 lo
catio
ns a
nd n
asal
infla
mm
atio
n in
a 4
-m
onth
s cr
oss-
sect
iona
l sur
vey.
No
sign
ifica
nt d
iffer
ence
s in
tot
al c
ell c
oun
ts
or
perc
enta
ge
of
neut
rop
hils
w
ere
fo
und
be
twee
n m
othe
rs
or
child
ren
from
th
e 4
diffe
rent
lo
catio
ns,
desp
ite
smal
l b
ut
sign
ifica
nt d
iffer
ence
s in
am
bien
t ex
posu
re
to
TS
P,
SO
2,
O3,
a
nd
N
Ox
duri
ng
this
pe
riod.
No
estim
ates
fo
r T
SP
w
ere
pr
ovid
ed.
report no. 4/11
160
Wol
f (20
02)
(Wol
f, 20
02)
1,43
5 pa
tient
s tr
eate
d fo
r rh
inos
inus
itis
in a
hos
pita
l of
Col
ogne
, 199
0-19
99.
Ann
ual
mea
n of
T
SP
(1
998)
: 39.
The
stu
dy i
nves
tigat
ed t
he a
ssoc
iatio
n b
etw
een
ai
r po
llutio
n an
d ch
roni
c rh
inos
inus
itis
in
Col
ogne
, G
erm
any.
Add
ress
es o
f 1,
435
pat
ient
s tr
eate
d fo
r ch
roni
c rh
inos
inus
itis
at
the
Oto
rhin
olar
yngo
logy
D
epar
tmen
t of
th
e U
nive
rsity
H
osp
ital
of
Col
ogn
e be
twee
n
1990
an
d 19
99 w
ere
assi
gned
to
one
of t
he 8
5 ci
ty
dist
ricts
. T
SP
, S
O2
an
d
NO
x w
ere
lin
ked
to
thes
e ar
eas.
A c
ombi
ned
ind
icat
or o
f ai
r qu
ality
w
as
deriv
ed
from
th
e th
ree
cons
ider
ed
mea
sure
s. R
egr
essi
on a
naly
ses
wer
e co
mpu
ted
adju
stin
g fo
r so
cioe
cono
mic
le
vel
of
the
85
dist
ricts
.
A
wea
k b
ut
cons
iste
nt
rela
tion
betw
een
m
ore
pollu
ted
are
as a
nd t
he
prev
alen
ce o
f ch
roni
c rh
inos
inus
itis
was
foun
d.
No
estim
ates
fo
r T
SP
w
ere
pr
ovid
ed.
FRA
NC
E
Bal
di e
t al
. (1
999)
(B
aldi
et
al,
1999
) 24
are
as o
f se
ven
Fre
nch
tow
ns,
197
4-1
976.
P
ollu
tant
s: T
SP
, B
S,
and
S
O2,
NO
4
3-ye
ar
aver
age
TS
P-
mea
n an
nual
va
lues
ra
ngin
g be
twee
n 45
and
24
3. T
SP
wa
s m
easu
red
by
the
gr
avim
etric
m
etho
d.
Rea
naly
sis
of P
ollu
tion
Atm
osph
eric
of
Aff
ectio
n R
espi
rato
ry C
hron
ique
s (P
AA
RC
) su
rve
y da
ta
to
sear
ch
for
rela
tions
hips
be
twee
n
mea
n an
nual
ai
r po
lluta
nt
leve
ls
and
prev
alen
ce
of
asth
ma
in
1291
ad
ult
(25-
59
year
s)
and
195
child
ren
(5-9
yea
rs)
asth
mat
ics.
Ran
dom
effe
cts
logi
stic
re
gre
ssio
n m
odel
us
ed
and
incl
uded
ag
e, s
mok
ing,
and
edu
catio
n le
vel
in t
he f
inal
m
odel
.
Onl
y an
as
soci
atio
n be
twee
n S
O2
and
asth
ma
in
adul
ts
obse
rved
. N
o ot
her
pollu
tant
w
as
asso
ciat
ed.
N
or
was
re
latio
nshi
p
with
ch
ildre
n se
en.
Met
eoro
logi
cal
varia
bles
an
d O
3 no
t ev
alua
ted.
For
a 5
0 µ
g/m
3 incr
ease
in T
SP
A
dult
asth
ma
prev
alen
ce
OR
1.0
1 (0
.92
, 1.1
1)
Zeg
hnou
n et
al
. (1
999)
(Z
eghn
oun
et a
l, 19
99)
La H
avre
, F
ranc
e, 1
993-
1996
. D
aily
m
ean
B
S
leve
ls
mea
sure
d in
th
ree
stat
ions
ra
nged
be
twee
n
12 a
nd 1
4.
Res
pira
tory
dr
ug
sale
s fo
r m
ucol
ytic
an
d an
ticou
gh
me
dica
tions
(m
ost
pr
escr
ibed
by
a
phys
icia
n) w
ere
eval
uate
d ve
rsus
BS
, S
O2,
and
N
O2
leve
ls.
An
auto
regr
essi
ve
Poi
sson
re
gres
sion
m
odel
pe
rmitt
ing
over
disp
ersi
on
cont
rol w
as u
sed
in th
e an
alys
is.
Res
pira
tory
dru
g sa
les
asso
ciat
ed w
ith B
S,
NO
2,
and
SO
2 le
vels
. B
oth
an
early
re
spon
se (
0 to
3 d
ay l
ag)
and
a lo
nger
one
(la
gs o
f 6 a
nd 9
day
s) w
ere
asso
ciat
ed.
—
report no. 4/11
161
Pen
ard-
Mor
and
et
al.
(200
5);
(Pen
ard-
Mor
and
et a
l, 20
05)
ISA
AC
II
Stu
dy,
6,67
2 sc
hool
child
ren
fr
om
six
Fre
nch
citie
s,
Enr
ollm
ent:
Mar
ch
1999
-O
ctob
er 2
000.
E
xpos
ure:
199
8-20
00.
3-ye
ar-a
vera
ged
conc
entr
atio
n
of
PM
10,
22.
The
pre
sent
cro
ss-s
ectio
nal
stud
y an
alys
ed t
he
asso
ciat
ions
be
twee
n lo
ng-t
erm
ex
posu
re
to
back
grou
nd
air
pollu
tion
an
d at
opic
an
d re
spira
tory
out
com
es i
n 6,
672
sch
oolc
hild
ren
aged
9–1
1 ye
ars
recr
uite
d fr
om 1
08 r
ando
mly
sc
hool
s in
six
Fre
nch
citie
s. C
hild
ren
unde
rwen
t a
clin
ical
exa
min
atio
n in
clu
ding
a s
kin
pric
k te
st
(SP
T)
to
com
mon
al
lerg
ens,
ex
erci
se-in
duce
d br
onch
ial
reac
tivity
(E
IB)
and
skin
exa
min
atio
n fo
r fle
xura
l de
rmat
itis.
T
he
prev
alen
ce
of
asth
ma,
al
lerg
ic
rhin
itis
and
at
opic
de
rmat
itis
wa
s as
sess
ed
by
a st
anda
rdiz
ed
heal
th
ques
tionn
aire
com
plet
ed b
y th
e pa
rent
s. T
hree
-ye
ar-a
vera
ged
conc
entr
atio
ns
of
air
pollu
tant
s (N
O2,
S
O2,
P
M10
an
d O
3)
wer
e
calc
ulat
ed
at
child
ren’
sc
hoo
ls
usin
g m
easu
rem
ents
of
ba
ckgr
ound
m
onito
ring
stat
ions
. O
Rs
wer
e
deriv
ed b
y m
ultip
le l
ogis
tic r
egre
ssio
n m
odel
s af
ter
adju
stm
ent
for
age,
sex
, fa
mily
his
tory
of
alle
rgy,
pa
ssiv
e sm
okin
g an
d pa
rent
eral
ed
ucat
ion.
EIB
, lif
etim
e as
thm
a an
d lif
etim
e al
lerg
ic
rhin
itis
wer
e
dire
ctly
as
soci
ated
to
an
in
crea
se i
n th
e ex
posu
re t
o S
O2,
PM
10 a
nd
O3.
M
oreo
ver,
S
PT
po
sitiv
ity
was
as
soci
ated
with
O3.
A
ssoc
iatio
ns w
ith p
ast
year
sym
ptom
s w
ere
cons
iste
nt,
even
if
not
alw
ays
stat
istic
ally
si
gnifi
cant
. R
esul
ts
pers
iste
d in
lo
ng-t
erm
re
side
nt
(cur
rent
ad
dres
s fo
r at
le
ast
8 ye
ars)
ch
ildre
n.
Aut
hors
co
nclu
ded
that
a
mod
erat
e in
crea
se i
n lo
ng-t
erm
exp
osur
e t
o ba
ckgr
ound
am
bien
t ai
r po
llutio
n w
as
asso
ciat
ed w
ith a
n in
crea
sed
prev
alen
ce o
f re
spira
tory
and
ato
pic
indi
cato
rs in
chi
ldre
n.
OR
(95
% C
I) f
or a
n in
crem
ent
of
PM
10 b
y 10
µg
/m3 :
Dur
ing
the
clin
ical
exa
min
atio
n E
IB: 1
.43
(1.0
2, 2
.01)
F
lexu
ral
derm
atiti
s:
0.79
(0
.59,
1.
07)
Pas
t yea
r W
heez
e: 1
.05
(0.7
2, 1
.54)
A
sthm
a: 1
.23
(0.7
7, 1
.95)
R
hino
conj
unct
iviti
s:
1.17
(0
.86,
1.
59)
Ato
pic
derm
atiti
s:
1.28
(0
.96,
1.
71)
Life
time
Ast
hma:
1.3
2 (0
.96,
1.8
1)
Alle
rgic
rhi
nitis
: 1.3
2 (1
.04,
1.6
8)
Ato
pic
derm
atiti
s:
1.09
(0
.88,
1.
36)
Ato
py: 0
.98
(0.8
0, 1
.22)
P
olle
n: 1
.14
(0.8
5, 1
.53)
In
door
: 0.9
1 (0
.72,
1.1
5)
Mou
lds:
1.0
0 (0
.53,
1.8
8).
SWIT
ZER
LAN
D
A
cker
man
n-Li
ebric
h et
al.
(199
7)
(Ack
erm
ann-
Lieb
rich
et a
l, 19
97)
Eig
ht S
wis
s re
gion
s P
ollu
tant
s:
SO
2,
NO
2,
TS
P, O
3, a
nd P
M10
. P
M w
as m
easu
red
with
a
Har
vard
impa
ctor
. P
M10
ra
nged
fr
om
10
to
53 w
ith a
mea
n of
37.
Long
-ter
m
effe
cts
of
air
pollu
tion
stud
ied
in
cros
s-se
ctio
nal
popu
latio
n-ba
sed
sam
ple
of
adul
ts a
ged
18 t
o 60
yea
rs.
Ran
dom
sam
ple
of
2,50
0 ad
ults
in
ea
ch
regi
on
dra
wn
fr
om
regi
strie
s of
loca
l inh
abita
nts.
Nat
ural
loga
rith
ms
of
FV
C
and
FE
V1
regr
esse
d ag
ains
t n
atur
al
loga
rithm
s of
hei
ght,
wei
ght
, ag
e, g
ende
r, a
topi
c st
atus
, and
po
lluta
nt v
aria
ble
s.
Sig
nific
ant
an
d co
nsis
tent
ef
fect
s on
FV
C
and
FE
V
wer
e fo
und
for
PM
10,
NO
2 a
nd
SO
2.
Est
imat
ed
regr
essi
on
coef
ficie
nt
for
PM
10 v
ersu
s F
VC
: -
0.03
5 (-
0.04
1, -
0.02
8).
Cor
resp
ondi
ng
valu
e fo
r F
EV
1:
-0.0
16
(-0.
023,
-0
.01)
. T
hus,
10
µ
g/m
3 P
M10
inc
reas
e es
timat
ed t
o le
ad
to
estim
ated
3.
4 pe
rcen
t de
crea
se i
n F
VC
and
1.6
per
cent
de
crea
se in
FE
V1.
report no. 4/11
162
Bra
un-F
ahrlä
nder
et
al
. (1
997)
(B
raun
-Fah
rland
er
et a
l, 19
97)
10 S
wis
s co
mm
uniti
es
Pol
luta
nts:
P
M10
, N
O2,
S
O2,
and
O3.
P
M w
as m
easu
red
with
a
Har
vard
impa
ctor
. P
M10
ra
nged
fr
om
10
to
33.
Impa
cts
of l
ong-
term
air
pollu
tion
exp
osur
e on
re
spira
tory
sy
mpt
oms
and
illne
sses
w
ere
ev
alua
ted
in
cros
s-se
ctio
nal
stud
y of
S
wis
s sc
hool
ch
ildre
n,
(age
d 6
to
15
year
s).
Sym
ptom
s an
alyz
ed u
sing
a l
ogis
tic r
egre
ssio
n m
odel
inc
ludi
ng c
ovar
iate
s of
fam
ily h
isto
ry o
f re
spira
tory
an
d al
lerg
ic
dise
ases
, nu
mbe
r of
si
blin
gs,
pare
ntal
ed
ucat
ion,
in
door
fu
els,
pa
ssiv
e sm
okin
g, a
nd o
ther
s.
Res
pira
tory
e
ndpo
ints
of
ch
roni
c co
ugh,
br
onch
itis,
w
heez
e an
d co
njun
ctiv
itis
sym
ptom
s w
ere
al
l re
late
d
to
the
va
rious
po
lluta
nts.
The
col
inea
rity
of t
he p
ollu
tant
s in
clud
ing
NO
2,
SO
2, a
nd O
3,
prev
ente
d an
y ca
usal
sep
ara
tion.
PM
10
Chr
onic
co
ugh
O
R
11.4
(2
.8,
45.5
) B
ronc
hitis
OR
23.
2 (2
.8, 4
5.5)
W
heez
e O
R 1
.41
(0.5
5, 3
.58)
Bay
er-O
gles
by
et
al.
(200
5) (
Bay
er-O
gles
by e
t al
, 200
5)
9,59
1 ch
ildre
n fr
om
9 S
wis
s co
mm
uniti
es,
1992
-200
1.
PM
10
rang
ed
from
10
to
47
.
The
pre
sent
cro
ss-s
ectio
nal
stud
y in
vest
igat
ed
whe
ther
a
rath
er
mod
erat
e de
clin
e
of
air
pollu
tion
leve
ls i
n th
e 19
90s
in S
witz
erla
nd w
as
asso
ciat
ed
with
a
redu
ctio
n in
re
spira
tory
sy
mpt
oms
and
dise
ases
in
9,
591
scho
ol
child
ren
of
nine
S
wis
s co
mm
uniti
es
betw
een
1992
and
200
1. P
aren
ts o
f ch
ildre
n co
mp
lete
d id
entic
al
que
stio
nnai
res
on
heal
th
stat
us
and
cova
riate
s.
We
assi
gned
to
ea
ch
child
an
es
timat
e of
re
gion
al
PM
10
and
dete
rmin
ed
chan
ge
in
PM
10
sinc
e th
e fir
st
surv
ey.
OR
s as
soci
ated
with
a d
eclin
e of
10
µg/
m3 i
n P
M10
w
ere
de
rived
af
ter
a pr
iori
adju
stm
ent
fo
r so
cioe
cono
mic
fa
ctor
s,
heal
th
rela
ted
fact
ors,
in
clud
ing
fam
ily
hist
ory
of
asth
ma,
br
onch
itis
and
atop
y,
indo
or
fact
ors,
in
clud
ing
sm
okin
g st
atus
of
the
mot
her,
and
avo
idan
ce b
ehav
iour
w
ith r
espe
ct to
alle
rgie
s.
PM
10
was
di
rect
ly
asso
ciat
ed
with
pr
eval
ence
o
f ch
roni
c co
ugh,
br
onch
itis,
co
mm
on
cold
, no
ctur
nal
dry
coug
h,
and
conj
unct
iviti
s sy
mpt
oms.
C
hang
es
in
prev
alen
ce
of
snee
zing
du
ring
polle
n se
ason
, as
thm
a,
and
hay
feve
r w
ere
not
asso
ciat
ed w
ith P
M10
.
Adj
uste
d O
R (
95%
CI)
for
a 1
0-µ
g/m
3 dec
line
of P
M10
: C
hron
ic c
ough
: 0.6
5 (0
.54,
0.7
9)
Bro
nchi
tis: 0
.66
(0.5
5, 0
.80)
C
omm
on c
old:
0.7
8 (0
.68,
0.8
9)
Noc
turn
al d
ry c
ough
: 0.
70 (
0.60
, 0.
83)
Con
junc
tiviti
s: 0
.81
(0.7
0, 0
.95)
W
heez
e: 0
.8,
NS
S
neez
ing:
0.9
, N
S
Ast
hma:
1.0
, N
S
Hay
feve
r: 1
.0, N
S.
report no. 4/11
163
Bay
er-O
gles
by
et
al.
(200
6) (
Bay
er-O
gles
by e
t al
, 200
6)
SA
PA
LDIA
co
hort
, S
witz
erla
nd, 1
991-
2002
. P
M10
mea
n: 2
1.4.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
livin
g ne
ar
mai
n st
reet
s an
d re
spira
tory
sy
mpt
oms
usin
g th
e S
wis
s C
ohor
t S
tudy
on
Air
Pol
lutio
n an
d Lu
ng
Dis
ease
s in
A
dults
(S
AP
AL
DIA
) co
hort
. S
AP
ALD
IA
1 w
as
cond
ucte
d in
199
1. I
n 20
02
subj
ects
wer
e re
-in
terv
iew
ed
(SA
PA
LD
IA
2).
The
pr
esen
t an
alys
es
wer
e ba
sed
on
a to
tal
of
12,9
99
obse
rvat
ions
fro
m 8
,555
sub
ject
s. G
eog
raph
ic
info
rmat
ion
syst
em d
ata
wer
e us
ed t
o as
sign
in
divi
dual
tr
affic
e
xpos
ure
es
timat
es
at
each
su
bjec
t’s r
esid
ence
. T
he g
ener
aliz
ed e
stim
atin
g eq
uatio
n re
gres
sion
mod
els
wer
e de
rive
d af
ter
adju
stm
ent
for
vario
us
fact
ors,
in
clud
ing
se
x,
age,
ed
ucat
ion,
ac
tive
and
pa
ssiv
e sm
okin
g,
occu
patio
nal
expo
sure
, re
gion
al
back
grou
nd
PM
10 (
estim
ated
by
a di
sper
sion
mod
el,
usin
g 44
mon
itorin
g si
tes
in S
witz
erla
nd i
n 20
00 )
and
heal
th r
elat
ed f
acto
rs,
Sub
ject
s liv
ing
near
bus
y st
reet
s re
port
ed
mor
e fr
eque
ntly
res
pira
tory
sym
ptom
s. T
his
leav
es
open
th
e qu
estio
n w
heth
er
the
prev
alen
ce o
f br
onch
itis
sym
ptom
s is
due
to
“long
-ter
m
effe
ct”
or
refle
cts
the
acut
e ef
fect
s of
air
pollu
tion
in lo
nger
per
iods
.
No
estim
ates
for
PM
10.
Zem
p et
al.
(199
9) (
Zem
p
et a
l, 19
99)
8 st
udy
site
s in
S
witz
erla
nd.
Pol
luta
nts:
T
SP
, P
M1
0,
SO
2, N
O2,
and
O3.
P
M w
as m
easu
red
with
a
Har
vard
impa
ctor
. P
M10
ra
nged
fr
om
10
to
33 w
ith a
mea
n of
21.
Logi
stic
re
gres
sion
an
alys
is
of
asso
ciat
ions
be
twee
n pr
eval
ence
s of
re
spira
tory
sym
ptom
s in
ran
dom
sam
ple
of a
dults
and
air
pollu
tion.
R
egre
ssio
ns
adju
sted
fo
r ag
e,
BM
I,
gen
der,
pa
rent
al
asth
ma,
ed
ucat
ion,
an
d fo
reig
n ci
tizen
ship
.
Chr
onic
co
ugh
and
chro
nic
phle
gm
and
brea
thle
ssne
ss w
ere
rela
ted
to T
SP
, P
M10
an
d N
O2.
Chr
onic
co
ugh,
ch
roni
c p
hleg
m
and
brea
thle
ssne
ss w
ere
rela
ted
to P
M10
, and
TS
P.
AU
STR
IA
Hor
ak
et
al.
(200
2)
(Hor
ak e
t al,
2002
) F
risch
er
et
al.
(199
9)
(Fris
cher
et a
l, 19
99)
Eig
ht
com
mun
ities
in
lo
wer
A
ustr
ia
betw
een
19
94-1
997.
P
M10
m
ean
su
mm
er
valu
e of
17
.36
an
d w
inte
r va
lue
of 2
1.03
.
Lung
fun
ctio
n as
sess
ed in
975
sch
oolc
hild
ren
in
grad
e 2-
3.
A
seve
ral
step
an
alys
is
incl
uded
G
EE
an
d se
nsiti
vity
ana
lyse
s.
Con
clud
ed t
hat
lon
g te
rm e
xpos
ure
to P
M10
ha
d a
sign
ifica
nt
nega
tive
ef
fect
on
lu
ng
func
tion
with
ad
ditio
nal
evid
ence
fo
r a
furt
her
effe
ct fo
r O
3 an
d N
O2.
Afte
r ad
just
ing
for
con
foun
ders
an
incr
ease
in P
M10
by
10 µ
g/m
3 was
as
soci
ated
w
ith
a de
crea
se
in
FE
V1
gro
wth
at
84 m
L/yr
and
329
m
L/s
year
for
ME
F25
-75.
report no. 4/11
164
GR
EEC
E
Kar
akat
sani
et
al.
(200
3)
(Kar
akat
sani
et a
l, 20
03)
Nes
ted
case
-con
trol
st
udy
of t
he E
PIC
stu
dy,
Ath
ens,
Gre
ece.
B
S f
or t
he m
ost
pollu
ted
area
ra
nged
fr
om
34
to
79.
A c
ase–
cont
rol s
tudy
nes
ted
in t
he E
PIC
coh
ort,
unde
rtak
en i
n A
then
s, G
reec
e, w
as
con
duct
ed
in
orde
r to
ex
amin
e th
e a
ssoc
iatio
n be
twee
n lo
ng-t
erm
exp
osur
e to
am
bien
t ai
r po
llutio
n an
d th
e ris
k of
ch
roni
c br
onch
itis,
em
phys
em
a or
C
OP
D.
Ove
rall,
16
8 pa
rtic
ipan
ts
repo
rtin
g a
hist
ory
of
CO
PD
sy
mpt
om
atol
ogy
and
168
heal
thy
cont
rols
in
divi
dua
lly
mat
ched
fo
r ag
e an
d ge
nder
, w
ere
vis
ited
by
a ph
ysic
ian
at t
heir
hom
es f
or c
ondu
ctin
g sp
irom
etry
and
a m
edic
al
inte
rvie
w.
Indi
vidu
aliz
ed
pers
onal
e
xpos
ure
to
air
pollu
tion
was
ass
esse
d on
the
bas
is o
f lo
ng-
term
res
iden
tial a
nd o
ccup
atio
nal s
ubje
ct h
isto
ry
linke
d w
ith
geog
raph
ical
ai
r po
llutio
n
dist
ribut
ion.
Cas
es w
ere
mor
e ex
pose
d to
air
pollu
tion
com
pare
d to
con
trol
s, w
ith O
Rs
for
sub
ject
s in
th
e h
ighe
st
quar
tile
of
expo
sure
vs
al
l ot
hers
in
clud
ed
betw
een
1.46
an
d 2.
01.
Aut
hors
con
clud
ed t
hat
long
-ter
m e
xpos
ure
to a
ir po
llutio
n w
as a
n im
port
ant
fact
or i
n th
e de
velo
pm
ent
of
chro
nic
resp
irato
ry
dise
ases
.
No
estim
ates
on
PM
.
Sic
hlet
idis
et
al
. (2
005)
(S
ichl
etid
is e
t al
, 200
5)
2000
-200
1. C
hild
ren
from
fiv
e ci
ties
of
Wes
tern
M
aced
onia
, G
reec
e (t
hree
with
inf
orm
atio
n on
T
SP
and
PM
10).
T
SP
mea
n:
Pto
lem
aida
: 132
K
ozan
i: 88
F
lorin
a: 5
6 P
M10
mea
n:
Pto
lem
aida
: 86.
3 K
ozan
i: 64
.2
Flo
rina:
58.
3.
Thi
s st
udy
com
pare
s pr
eval
ence
of
rhin
itis
and
bron
chiti
s in
5
diffe
rent
ci
ties
of
Wes
tern
M
aced
onia
ch
arac
teriz
ed b
y di
ffere
nt t
ype
and
leve
l of
en
viro
nmen
tal
pollu
tion.
D
urin
g
2000
an
d 20
01 3
,559
chi
ldre
n ag
ed 9
-12
year
s w
ere
enro
lled
in
the
stud
y.
Con
cent
ratio
ns
of
TS
P
and
PM
10 w
ere
mea
sure
d in
3 c
ities
onl
y.
Sym
ptom
s fr
om
the
uppe
r an
d lo
wer
re
spira
tory
sy
stem
w
ere
mor
e fr
eque
nt
in
child
ren
livin
g in
mor
e po
llute
d ar
eas.
No
estim
ates
for
PM
report no. 4/11
165
POLA
ND
Jedr
ycho
wsk
i et
al
. (1
999)
(J
edry
cho
wsk
i et
al
, 199
9)
In
Kra
kow
, P
olan
d in
19
95 a
nd 1
997
Spa
cial
di
strib
utio
ns
for
BS
and
SO
2 d
eriv
ed
from
ne
twor
k of
17
ai
r m
onito
ring
stat
ions
. BS
52
.6 µ
g/m
± 5
3.98
in h
igh
ar
ea a
nd
33.2
3 ±
35.9
9 in
low
are
a.
Effe
cts
on
lung
fu
nctio
n
grow
th
stu
died
in
pr
eado
lesc
ent
child
ren.
Lu
ng
func
tion
grow
th
rate
mea
sure
d by
gai
n in
FV
C a
nd F
EV
1 an
d oc
curr
ence
of
slo
w lu
ng f
unct
ion
grow
th (
SLF
G)
over
the
2 y
ear
perio
d de
fined
as
low
est
qui
ntile
of
th
e di
strib
utio
n of
a
gi
ven
test
in
g
ende
r gr
oup.
1,1
29 c
hild
ren
age
9 pa
rtic
ipat
ed i
n fir
st
year
and
1,0
01
in f
ollo
w-u
p 2
yea
rs l
ater
. A
TS
st
anda
rd q
uest
ionn
aire
and
PF
T m
etho
ds u
sed.
In
itial
ly
univ
aria
te
desc
ript
ive
stat
istic
s of
pu
lmon
ary
func
tion
indi
ces
and
SLF
G
wer
e
esta
blis
hed,
fo
llow
ed
by
mul
tivar
iate
lin
ear
regr
essi
on
ana
lyse
s in
clu
ding
ge
nder
, E
TS
, pa
rent
al e
duca
tion,
hom
e he
atin
g sy
ste
m a
nd
mol
d. S
O2
wa
s al
so a
naly
zed.
Sta
tistic
ally
sig
nific
ant
nega
tive
asso
ciat
ion
betw
een
air
pollu
tion
leve
l and
lung
fun
ctio
n gr
owth
(F
VC
and
FE
V1)
ove
r th
e fo
llow
up
in
both
ge
nder
gr
oups
. S
LFG
w
as
sign
ifica
ntly
hi
gher
in
th
e m
ore
pollu
ted
area
s on
ly a
mon
g bo
ys.
In g
irls
ther
e w
as
cons
iste
ncy
in t
he d
irect
ion
of t
he e
ffect
, bu
t no
t st
at.
sign
ifica
nt.
Cou
ld n
ot s
epar
ate
BS
an
d S
O2
effe
cts
on
lung
fu
nctio
n gr
ow
th.
Exc
ludi
ng
asth
ma
subj
ects
su
b-sa
mpl
e (s
ize
917)
pro
vide
d si
mila
r re
sults
.
Boy
s S
LFG
(F
VC
) O
R: 2
.15
(1.2
5, 3
.69)
S
LFG
(F
EV
1)
OR
: 1.9
0 (1
.12,
3.2
5)
Girl
s F
VC
OR
: 1.5
0 (
0.84
, 2.6
8)
FE
V1
OR
: 1.3
9 (0
.78,
2.4
4)
Jedr
ycho
wsk
i an
d F
lak
(199
8)
(Jed
rych
ow
ski
andF
lak,
199
8)
In
Kra
cow
P
olan
d,
in
1991
-199
5 D
aily
24
h co
ncen
trat
ion
of S
PM
(bl
ack
smok
e)
mea
sure
d at
17
ai
r m
onito
ring
stat
ions
. H
igh
area
s ha
d 52
.6
mea
n co
mpa
red
to
low
are
as a
t 33.
2.
Res
pira
tory
he
alth
su
rvey
of
1,
129
scho
ol
child
ren
(age
d 9
year
).
Res
pira
tory
ou
tcom
es
incl
uded
ch
roni
c co
ugh,
ch
roni
c ph
legm
, w
heez
ing,
diff
icul
ty b
reat
hin
g an
d as
thm
a. M
ulti-
varia
ble
logi
stic
re
gres
sion
us
ed
to
calc
ulat
e pr
eval
ence
O
R
for
sym
ptom
s ad
just
ed
for
pote
ntia
l con
foun
ding
.
The
co
mpa
rison
of
a
djus
ted
effe
ct
estim
ates
re
veal
ed
chro
nic
ph
legm
as
un
ique
sym
ptom
rel
ated
nei
ther
to
alle
rgy
nor
to i
ndoo
r va
riabl
e bu
t w
as
asso
ciat
ed
sign
ifica
ntly
w
ith
outd
oor
air
pollu
tion
cate
gory
(A
PL)
. N
o po
tent
ial
conf
ound
ing
varia
ble
had
maj
or e
ffect
.
It w
as
not
poss
ible
to
as
sess
se
para
tely
the
con
trib
utio
n of
the
di
ffere
nt s
ourc
es o
f a
ir po
lluta
nts
to
the
occu
rren
ce
of
resp
irato
ry
sym
ptom
s.
ET
S
and
hous
ehol
d he
atin
g (c
oal
vs.
gas
vs.
cent
ral
heat
ing)
ap
pea
red
to
be
of
min
imal
impo
rtan
ce.
BU
LGA
RIA
T
urno
vska
an
d K
ostir
anev
(1
999)
(T
urno
vska
T
an
dKos
tiane
v S
, 199
9)
Dim
itrov
grad
, B
ulga
ria,
May
199
6 T
SP
m
ean
leve
ls
wer
e
520
± 16
1 in
19
86
and
187
± 9
in
1996
. S
O2,
H
2S,
and
N
O2
also
m
easu
red.
Res
pira
tory
fun
ctio
n of
97
scho
olch
ildre
n (m
ean
age
10.4
± 0
.6 y
ear)
mea
sure
d in
May
199
6 as
a
sam
ple
of
12%
of
al
l fo
ur-g
rade
rs
in
Dim
itrov
grad
. T
he
obta
ined
re
sults
w
ere
co
mpa
red
with
ref
eren
ce v
alue
s fo
r B
ulga
rian
child
ren
aged
7 t
o 14
yea
r, c
alcu
late
d in
the
sa
me
labo
rato
ry
in
1986
. V
aria
tion
ana
lysi
s te
chni
ques
we
re u
sed
to tr
eat t
he d
ata.
Vita
l ca
paci
ty a
nd F
EV
1 w
ere
sign
ifica
ntly
lo
wer
(m
ean
valu
e. =
88.
54%
and
82.
5%
re
spec
tfully
) co
mpa
ring
valu
es
betw
een
19
86
and
1996
. T
SP
ha
d de
crea
sed
by
2.74
tim
es
to
leve
ls
still
hi
gher
th
an
Bul
garia
n an
d W
HO
sta
ndar
ds.
—
report no. 4/11
166
TAB
LE IX
- LO
NG
TER
M P
AR
TIC
ULA
TE M
ATT
ER E
XPO
SUR
E A
ND
CA
NC
ER.
Ref
eren
ce, L
ocat
ion,
Ye
ars,
PM
Inde
x, M
ean
or M
edia
n (µ
g/m
3 ).
Stud
y D
escr
iptio
n. M
etho
ds, c
o-p
ollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
. PM
Inde
x, R
R (9
5% C
I)
EPIC
Vin
eis
et a
l., 2
006
(Vin
eis
et a
l, 20
06);
Vin
eis
et a
l.,
2007
(V
inei
s e
t al,
2007
) N
este
d ca
se-c
ontr
ol
stud
y (2
71
case
s,
737
mat
ched
co
ntro
ls)
in
the
EP
IC
coh
ort;
10
coun
trie
s.
Rec
ruitm
ent
perio
d:
1993
-199
8 M
edia
n fo
llow
-up:
7
ye
ars.
P
M10
be
twee
n 19
.9
(Ile
de
F
ranc
e 19
95-1
999)
an
d 73
.4
(Tur
in
1990
-19
94).
In a
nes
ted
cas
e-co
ntro
l st
udy
of t
he E
uro
pean
P
rosp
ectiv
e S
tudy
in
to
Can
cer
and
Nut
ritio
n (E
PIC
) co
hort
, ba
sed
on
mor
e th
an
500
,000
he
alth
y vo
lunt
eers
ag
ed
35-7
4 fr
om
10
Eur
opea
n co
untr
ies,
27
1 ex
-sm
oker
(s
ince
at
le
ast
10 y
ears
) an
d ne
ver
smok
er c
ases
with
lu
ng c
ance
r o
ccur
red
in a
med
ian
follo
w-u
p of
7
year
s. C
ases
wer
e co
mpa
red
with
737
co
ntro
ls
freq
uenc
y m
atch
ed
by
age,
sm
okin
g st
atus
, co
untr
y an
d tim
e be
twe
en
recr
uitm
ent
and
diag
nosi
s. I
n o
rder
to
asse
ss t
he e
xpos
ure
to
traf
fic-r
elat
ed a
ir po
llutio
n, t
he h
ome
addr
ess
at
the
time
of
enro
llmen
t w
as
cons
ider
ed.
B
oth
dist
ance
s to
a m
ajor
str
eet
and
annu
al a
vera
ge
conc
entr
atio
ns
of P
M1
0 (a
vaila
ble
for
113
cas
es
and
312
cont
rols
) an
d ot
her
co-p
ollu
tant
s,
incl
udin
g N
O2,
O3
and
SO
2,
wer
e co
nsid
ered
. O
Rs
wer
e de
rived
by
co
nditi
onal
lo
gist
ic
regr
essi
on m
odel
s af
ter
furt
her
adju
stm
ent
for
educ
atio
n, B
MI,
phys
ical
act
ivity
, en
ergy
int
ake,
fr
uit,
vege
tabl
es,
mea
t an
d al
coho
l co
nsum
ptio
n.
Fur
ther
m
odel
s in
clu
de
cotin
ine
leve
ls a
nd o
ccup
atio
nal i
nde
x.
No
sign
ifica
nt a
ssoc
iatio
ns w
ere
foun
d fo
r in
crem
ents
by
10
μg
/m3
of
PM
10,
NO
2 (O
R=
1.14
; 95
%
CI:
0.
78-1
.67)
an
d S
O2.
(O
R=
1.08
; 9
5%
CI:
0.
89-
1.30
).
So
me
exce
ss r
isks
wer
e fo
und
whe
n co
nsid
erin
g th
e th
ird
vs
the
othe
r tw
o te
rtile
s of
ex
posu
re
to
NO
2 (O
Rs
rang
ed
betw
een
1.
30 a
nd 1
.62
acc
ordi
ng t
o va
rious
mod
els;
P
AR
%=
5%
), b
ut n
ot t
o P
M10
and
SO
2. T
he
OR
of
livin
g n
earb
y he
avy
traf
fic r
oads
was
1.
46 (
95%
CI:
0.8
9-2.
40),
cor
resp
ondi
ng t
o a
PA
R%
of
7%
.
Adj
uste
d O
R f
or i
ncre
men
ts o
f 10
μg/
m3 o
f P
M10
: 0.
91 (
0.7
0,
1.18
).
OR
for
the
up
per
vs t
he l
ow
est
an
d th
e in
term
edia
te
tert
iles
(i.e.
, ≥27
vs
<27
μg/
m3 ):
A
djus
ted
for
mat
chin
g va
riabl
es: 0
.98
(0.6
6, 1
.45)
F
urth
er a
djus
ted
mod
el:
1.05
(0.
65, 1
.69)
F
urth
er a
djus
ted
for
cotin
ine
: 2.
85 (
0.97
, 8.3
3)
Fur
ther
ad
just
ed
for
occu
patio
nal i
ndex
: 1.
02 (
0.68
, 1.5
1).
report no. 4/11
167
EUR
OPE
AN
EC
OLO
GIC
ST
UD
Y
Na
wro
t et
al
., 20
07
(Naw
rot e
t al,
2007
) E
colo
gica
l st
udy;
15
E
urop
ean
cou
ntrie
s.
Ann
ual
me
an
PM
2.5;
es
timat
ed
for
the
year
20
02)
betw
een
1
(Fin
land
) an
d 22
(B
elgi
um).
Thi
s ec
olog
ical
st
udy
ana
lyze
d th
e re
latio
n be
twee
n ag
e an
d sm
okin
g lu
ng c
ance
r ra
tes
in
men
ac
ross
15
E
urop
ean
co
untr
ies
(obt
aine
d fr
om
the
EU
CA
N
data
base
fo
r th
e199
8),
and
annu
al m
ean
PM
2.5
conc
entr
atio
ns e
stim
ated
for
th
e ye
ar 2
002.
A
sign
ifica
nt
corr
elat
ion
betw
een
coun
trie
s’ag
e
and
smok
ing
stan
dard
ized
m
orta
lity
rate
s fo
r lu
ng c
ance
r in
men
and
an
nual
mea
n P
M2.
5 w
as
show
n (
Spe
arm
an
corr
elat
ion
coef
ficie
nt,
r=0.
57;
p=0.
03).
No
rela
tion
was
ev
iden
t fo
r w
omen
(r
=0.
37;
p=0.
18).
Aut
hor
s co
nclu
ded
sug
gest
ing
that
so
me
of
the
diffe
renc
es
in
lung
ca
ncer
m
orta
lity
in E
urop
e co
uld
be e
xpla
ined
by
fine
part
icul
ate
air
conc
entr
atio
n.
RR
for
an
incr
emen
t of
5 μ
g/m
3
of
PM
2.5
in
men
: 1.
18
(1.0
4,
1.32
).
Subn
atio
nal s
tudi
es
G
REA
T B
RIT
AIN
Kno
x, 2
005
(Kno
x, 2
005)
P
opul
atio
n:
All
child
ren
dy
ing
for
leuk
aem
ia
or
othe
r ca
ncer
s be
twee
n
1955
and
19
80 i
n G
reat
B
ritai
n.
Mea
n va
lue
of
PM
w
as
not g
iven
.
In
this
st
udy,
bi
rth
and
deat
h ad
dres
ses
of
child
ren
(age
d <
16
year
s) d
ying
fro
m c
ance
r be
twee
n 19
55
and
1980
in
Gre
at B
ritai
n w
ere
linke
d to
e
mis
sion
ho
tspo
ts
for
spec
ific
chem
ical
s, i
ncl
udin
g P
M10
, C
O,
NO
x, o
btai
ned
for
the
year
200
1. A
mon
g ch
ildre
n w
ho m
oved
ho
use,
di
stan
ces
from
ea
ch
addr
ess
to
the
near
est
haza
rd w
ere
com
pare
d. T
he a
im o
f th
is
stud
y w
as
to e
xam
ine
the
exc
ess
es o
f cl
ose-
to-
haza
rd b
irth
addr
esse
s co
mpa
red
with
clo
se t
o ha
zard
dea
th a
ddre
sses
, in
ord
er t
o un
ders
tand
w
heth
er p
rena
tal
or e
arly
pos
tnat
al e
xpo
sure
to
oil
com
bust
ion
gase
s in
itiat
es
canc
ers.
E
stim
ates
of
R
R
wer
e b
ased
on
bi
rth/
deat
h ra
tios
for
child
ren
who
m
ove
d ho
uses
w
ithin
va
rious
di
stan
ces.
A
tmos
pher
ic
emis
sion
s ho
tspo
ts
we
re
conv
erte
d to
co
ordi
nate
s.
Dis
tanc
es
to
indu
stria
l si
tes,
in
clud
ing
bu
s st
atio
n,
railw
ays
, ho
spita
ls,
inci
nera
tors
, m
otor
way
s w
ere
als
o co
nsid
ered
.
The
re
wer
e ex
cess
R
R
with
in
1 km
of
ho
tspo
ts f
or P
M10
, C
O,
NO
x, a
nd o
ther
oil
base
d co
mbu
stio
n ga
ses
(par
ticul
arly
1,3
-bu
tadi
ene)
, an
d w
ithin
1.
0 km
of
bu
s st
atio
ns,
hosp
itals
, he
avy
tran
spor
t ce
ntre
s,
railw
ays,
and
oil
inst
alla
tion
s.
The
au
thor
co
nclu
ded
th
at
child
hood
ca
ncer
s w
ere
stro
ngly
de
term
ined
by
pr
enat
al o
r ea
rly p
ostn
atal
exp
osu
res
to o
il ba
sed
com
bust
ion
gase
s,
espe
cial
ly
fro
m
engi
ne e
xhau
st.
Am
ong
child
ren
who
m
ove
d ho
use,
di
stan
ces
with
in
1
km
from
ea
ch
addr
ess
to
the
near
est
emis
sion
of
P
M10
ho
tspo
t: B
irths
= 1
,420
; D
eath
s=77
7
RR
=1.
83 (
1.6
7, 2
.00)
.
report no. 4/11
168
Kno
x, 2
006
(Kno
x, 2
006)
S
ame
stud
y de
sign
as
K
nox,
200
5 (K
nox,
200
5).
Thi
s is
a r
e-an
alys
is o
f th
e p
revi
ous
stu
dy f
or
indu
stria
l si
tes
and
road
s,
cons
ider
ing
mor
e pr
ecis
e an
d ad
equa
te d
ata
on r
oads
, ra
il an
d in
dust
rial s
ites.
Sig
nific
ant
birt
h ex
cess
es w
ere
foun
d w
ithin
sh
ort
dist
ance
s of
bu
s st
atio
ns,
railw
ay
stat
ions
, fe
rrie
s,
railw
ays,
an
d A
, B
cl
ass
ro
ads.
As
in t
he p
revi
ous
stud
y, t
he a
uth
or
conc
lude
d th
at
child
ca
ncer
ini
tiatio
ns w
ere
st
rong
ly
dete
rmin
ed
by
pren
atal
or
ea
rly
post
nata
l e
xpos
ures
to
en
gine
e
xha
ust
gase
s
No
estim
ates
for
PM
.
NO
RW
AY
N
afst
ad
et
al.,
2003
(N
afst
ad
et
al,
2003
);
Naf
stad
et
al
., 20
04
(Naf
stad
et a
l, 20
04)
Coh
ort
of
16,
209
men
fr
om
Osl
o,
Nor
way
, fo
llow
-up
perio
d: 1
972/
73
to 1
998.
N
o in
form
atio
n on
PM
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
long
ter
m e
xpos
ure
to S
O2 a
nd N
Ox
and
lung
ca
ncer
in
cide
nce
in
a co
hort
of
O
slo
men
fo
llow
ed f
or 2
7 ye
ars.
Co
x pr
opor
tiona
l h
azar
ds
regr
essi
on
afte
r ad
just
men
t fo
r sm
okin
g,
educ
atio
n a
nd
age.
A
n
addi
tiona
l m
odel
in
clud
ed th
e tw
o co
-pol
luta
nts
.
Dur
ing
the
follo
w-u
p pe
riod,
lu
ng
canc
er
occu
rred
in 4
18
men
. T
he a
djus
ted
RR
for
a
10
mg/
m3
incr
ease
in
av
erag
e ho
me
addr
ess
expo
sure
bet
we
en 1
974
and
1978
w
as
1.08
(95
% C
I: 1
.02-
1.15
) fo
r N
Ox
and
1.
01 (
95%
CI:
0.94
-1.0
8) fo
r S
O2.
No
estim
ates
for
PM
.
Nae
ss e
t al
., 20
07 (
Nae
ss
et a
l, 20
07)
Coh
ort
of
all
inha
bita
nts
of O
slo
(143
,842
).
Fol
low
-up
perio
d:
1992
-19
98.
PM
2.5
rang
e: 6
.56-
22.3
4 P
M10
ran
ge: 6
.57-
30.1
3
Thi
s st
udy
inve
stig
ated
the
ass
ocia
tion
betw
een
conc
entr
atio
ns
of P
M10
, P
M2.
5 an
d N
O2
in 1
992-
1995
, an
d ca
use-
spec
ific
mor
talit
y,
incl
udin
g lu
ng
canc
er.
The
po
pula
tion
incl
ude
d al
l in
habi
tant
s of
Osl
o N
orw
ay,
aged
51–
90 y
ears
on
Jan
uary
199
2 w
ith f
ollo
w-u
p of
dea
ths
from
19
92 t
o 19
98.
An
air
disp
ersi
on m
odel
was
use
d to
es
timat
e le
vels
of
e
xpos
ure
in
all
470
ad
min
istr
ativ
e ne
ighb
ourh
oods
. C
ox
prop
ortio
nal
haza
rds
regr
essi
on
mod
els
wer
e u
sed
afte
r ad
just
men
t fo
r ag
e, e
duca
tion
and
occu
patio
nal
clas
s. T
o m
odel
the
rel
atio
n be
twee
n ai
r po
lluta
nts
and
mor
talit
y,
GA
M
mod
els
wer
e us
ed.
Dur
ing
follo
w-u
p, 1
,453
lun
g ca
ncer
dea
ths
occu
rred
. S
om
e di
rect
as
soci
atio
ns
we
re
obse
rved
be
twee
n P
M
and
lu
ng
can
cer
mor
talit
y.
The
ef
fect
s w
ere
sm
all
in
the
youn
g ag
e gr
oup
(51-
70 y
ears
) of
men
and
m
oder
ate
to l
arge
in
the
old
age
grou
p (7
1-90
ye
ars)
. W
omen
ha
d pa
rtic
ular
ly
larg
e ef
fect
s in
th
e yo
ung
gro
up.
The
th
ree
pollu
tant
s sh
are
d si
mila
r re
sults
, be
ing
the
ir co
rrel
atio
ns h
igh
(bet
wee
n 0
.88
and
0.95
).
Occ
upat
iona
l cl
ass
and
educ
atio
n ad
just
ed
HR
fo
r qu
artil
e in
crea
se o
f exp
osur
e:
PM
2.5
Men
51
-70
year
s:
1.07
(0
.97-
1.18
) W
omen
51
-70
year
s:
1.27
(1
.13-
1.43
) M
en
71-9
0 ye
ars:
1.
07
(0.9
7-1.
18)
Wom
en
71-9
0 ye
ars:
1.
16
(1.0
2-1.
32)
PM
10
Men
51
-70
year
s:
1.07
(0
.97-
1.18
) W
omen
51
-70
year
s:
1.27
(1
.13-
1.43
) M
en
71-9
0 ye
ars:
1.
08
(0.9
8-1.
18)
Wom
en
71-9
0 ye
ars:
1.
17
(1.0
3-1.
33).
report no. 4/11
169
SWED
EN
N
yber
g et
al
. (2
000)
(N
yber
g et
al,
2000
).
Cas
e-co
ntro
l st
udy
on
1,04
2 ca
ses
died
for
lun
g ca
ncer
and
2,3
64 c
ontr
ols
died
for
oth
er d
isea
ses
in
Sto
ckho
lm.
Stu
dy p
erio
d 1
985-
1990
. N
o in
form
atio
n on
PM
.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
expo
sure
to
NO
x/N
O2
and
SO
2 an
d lu
ng c
ance
r m
orta
lity
in a
cas
e-co
ntro
l stu
dy b
ased
on
1,0
99
case
s di
ed f
or l
ung
canc
er a
nd t
wo
com
pan
ion
grou
ps (
1,27
5 al
ive
cont
rols
and
1,0
90 m
atch
ed
cont
rols
di
ed
for
othe
r d
isea
ses)
. Lo
cal
air
pollu
tion
leve
ls w
ere
estim
ated
usi
ng v
alid
ated
di
sper
sion
mod
els.
OR
s w
ere
deriv
ed b
y lo
gist
ic
regr
essi
on
mod
els
afte
r ad
just
men
t fo
r ag
e,
smok
ing,
exp
osur
e to
occ
upat
iona
l ca
rcin
ogen
s an
d so
cioe
cono
mic
gro
up (
data
obt
aine
d fr
om
each
sub
ject
’s n
ext o
f kin
).
NO
2 e
xpo
sure
ov
er
30
year
s w
as
not
si
gnifi
cant
ly
asso
ciat
ed
to
lung
ca
nce
r m
orta
lity
(OR
for
an
incr
emen
t by
10 μg
/m3
wa
s 1.
05
(95%
C
I: 0.
93,
1.18
).
Whe
n co
nsid
erin
g N
O2
exp
osu
re
20
year
s pr
evio
usly
, a
bord
erlin
e si
gnifi
cant
di
rect
as
soci
atio
n w
as f
ound
with
NO
2 (O
R=
1.10
; 95
%
CI
0.97
, 1.
23)
No
asso
ciat
ion
wa
s ob
serv
ed w
ith S
O2.
No
estim
ates
for
PM
.
DEN
MA
RK
Raa
scho
u-N
iels
en
et
al.
(200
1)
(Raa
scho
u-N
iels
en
et
al,
2001
);
Raa
scho
u-N
iels
en
et
al.
(200
2)
(Raa
scho
u-N
iels
en e
t al,
2002
).
Cas
e-co
ntro
l st
udy
on
1,98
9 ch
ildre
n c
ases
and
5,
506
cont
rols
, D
enm
ark,
19
68-1
991.
N
o in
form
atio
n on
PM
.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
traf
fic-r
elat
ed a
ir po
llutio
n a
nd t
he r
isk
of c
ance
r du
ring
child
hood
. D
urin
g 19
68-1
991,
the
Dan
ish
Can
cer
Reg
istr
y re
port
ed 1
,989
chi
ldre
n w
ith a
di
agno
sis
of
leuk
emia
, ca
ncer
of
th
e ce
ntra
l ne
rvou
s sy
ste
m,
or m
alig
nant
lym
phom
a. T
hese
ch
ildre
n w
ere
com
pare
d
with
5,
506
co
ntro
l ch
ildre
n ra
ndom
ly
sele
cte
d fr
om
the
entir
e ch
ildho
od p
opul
atio
n. T
he r
esid
entia
l his
tori
es o
f th
e ch
ildre
n w
ere
trac
ed f
rom
9 m
onth
s be
fore
bi
rth
until
the
tim
e of
dia
gno
sis
of t
he c
ases
and
a
sim
ilar
perio
d fo
r th
e co
ntro
ls.
For
eac
h of
the
18
,440
id
entif
ied
addr
esse
s,
info
rmat
ion
on
traf
fic
and
the
conf
igur
atio
n of
st
reet
s an
d bu
ildin
gs w
as c
olle
cted
. A
vera
ge c
once
ntra
tions
of
ben
zene
an
d N
O2
(indi
cato
rs o
f tr
affic
-rel
ated
ai
r po
llutio
n)
wer
e
calc
ulat
ed
for
the
rele
vant
pe
riod,
an
d ex
posu
res
to
air
pollu
tion
durin
g pr
egna
ncy
and
durin
g ch
ildho
od
wer
e ca
lcul
ated
sep
arat
ely.
The
ris
ks
of
leuk
emia
, ce
ntra
l ne
rvou
s sy
stem
tu
mor
s,
and
all
sele
cted
ca
ncer
s co
mbi
ned
we
re n
ot l
inke
d to
exp
osur
e t
o be
nzen
e or
N
O2.
T
he
risk
of
lym
pho
mas
in
crea
sed
by 2
5% (
p f
or t
rend
= 0
.06)
and
51
% (
p fo
r tr
end
= 0
.05)
for
a d
oubl
ing
of
the
conc
entr
atio
n of
ben
zene
and
nitr
ogen
di
oxid
e, r
espe
ctiv
ely,
dur
ing
the
pre
gnan
cy.
The
ass
ocia
tion
was
res
tric
ted
to H
odgk
in's
di
seas
e.
No
estim
ates
for
PM
.
report no. 4/11
170
THE
NET
HER
LAN
DS
H
oek
et a
l. (2
002)
(H
oek
et a
l, 20
02)
The
Net
herla
nds
Coh
ort s
tudy
19
86-1
994
Sub
ject
s= 4
,492
; B
S e
xpos
ure,
mea
n (S
D):
ba
ckgr
ound
15
.1
(2.5
);
back
grou
nd
and
loca
l 15
.5 (
3.2)
.
This
stud
y ex
amin
ed th
e as
soci
atio
n be
twee
n B
S an
d N
O2 i
n a
coho
rt of
4,4
92 re
side
nts i
n va
rious
are
as o
f th
e N
ethe
rland
s and
cau
se-s
peci
fic m
orta
lity,
in
clud
ing
lung
can
cer.
Lo
ng-t
erm
ex
posu
re
to
tra
ffic-
rela
ted
BS
an
d
NO
2 w
as e
stim
ated
for
the
198
6 ho
me
addr
ess.
C
ox’
s pr
opor
tiona
l ha
zard
s m
odel
s w
ere
used
af
ter
adju
stm
ent
for
age,
sex
, ed
ucat
ion,
BM
I, oc
cupa
tion,
ac
tive
and
pa
ssiv
e ci
gare
tte
smok
ing,
an
d ne
ighb
ourh
ood
soci
oeco
nom
ic
scor
e.
Tw
o m
odel
s of
ex
posu
re
for
ever
y po
lluta
nt
we
re
used
: 1)
th
e ba
ckgr
ound
co
ncen
trat
ion
wa
s en
tere
d in
the
mod
el w
ith t
he
indi
cato
r va
riabl
e fo
r liv
ing
near
a m
ajor
roa
d. 2
) th
e su
m o
f th
e ba
ckgr
ound
and
the
est
imat
ed
cont
ribut
ion
from
liv
ing
near
a m
ajor
roa
d to
air
pollu
tion
conc
entr
atio
ns.
Too
few
peo
ple
die
d fr
om l
ung
canc
er t
o ob
tain
sta
ble
estim
ates
for
th
e in
dica
tor
varia
ble
for
livin
g ne
ar
a m
ajor
ro
ad.
Ove
rall,
60
lu
ng
canc
er
deat
hs
occu
rred
du
ring
the
follo
w-u
p pe
riod
. N
o si
gnifi
cant
as
soci
atio
n e
mer
ged
betw
een
bot
h B
S a
nd
NO
2 an
d lu
ng c
ance
r m
orta
lity.
RR
fo
r an
in
crem
ent
of
10
μg/m
3 (95
th –
5th p
erce
ntile
) of
B
S.
Bac
kgro
und
and
loca
l: 1.
06
(0.4
3, 2
.63)
.
Vis
ser
et a
l., 2
004
(Vis
ser
et a
l, 20
04)
Eco
logi
c st
udy
Am
ster
dam
, 198
9-19
97;
27,1
57 c
ance
r ca
ses.
N
o in
form
atio
n on
PM
.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
resi
dent
ial t
raffi
c in
tens
ity,
and
can
cer
inci
denc
e in
A
mst
erda
m
in
1989
–199
7.
Dai
ly
traf
fic
inte
nsity
dat
a f
or i
ndiv
idua
l ad
dres
ses
alo
ng t
he
mai
n ro
ads
wer
e lin
ked
with
th
e po
pula
tion-
base
d re
gion
al c
ance
r re
gist
ry.
SIR
w
ere
der
ived
by
co
mpa
rison
b
etw
een
ob
serv
ed n
umbe
rs o
f ca
ncer
for
res
iden
ts a
long
m
ain
road
s w
ith
expe
cte
d nu
mbe
rs
deriv
ed
usin
g th
e ag
e gr
oup-
and
sex
-sp
ecifi
c ca
ncer
in
cide
nce
in t
he p
opul
atio
n n
ot r
esid
ing
alon
g th
e m
ain
road
s.
Ana
lysi
ng
27,1
57 c
ases
, n
o cl
ear
evid
enc
e of
an
asso
ciat
ion
betw
een
resi
denc
e al
ong
mai
n ro
ads
and
the
inci
denc
e of
can
cer
in
adul
ts e
mer
ged
(SIR
=1.
03;
95%
CI:
1.00
-1.
07),
alth
ough
for
sub
ject
s re
side
nts
alon
g th
e m
ain
road
s an
e
xce
ss
risk
of
hem
atol
ogic
al
mal
igna
ncie
s in
fe
mal
es
(SIR
=1.
23;
95
% C
I: 1.
04-
1.44
) an
d ch
ildre
n (S
IR f
or a
cute
lym
phoc
ytic
leu
kaem
ia =
2.5;
95
%
CI:
0.8
-5.9
) w
as
foun
d.
Mor
eove
r,
resi
dent
ial t
raffi
c in
tens
ity i
ncr
ease
d th
e ris
k of
ga
stro
inte
stin
al
canc
er
in
mal
es
(SIR
=1.
16; 9
5C
I: 1.
04-1
.28)
.
No
estim
ates
for
PM
.
report no. 4/11
171
Vis
ser
et a
l., 2
005
(Vis
ser
et a
l, 20
05)
Eco
logi
c st
udy
Sch
ipho
l, A
mst
erda
m,
1988
-200
3;
13,2
07 c
ance
r ca
ses.
P
M10
(20
02),
mea
n: 4
0.
Thi
s st
udy
inve
stig
ated
th
e in
cide
nce
of
lu
ng
canc
er b
etw
een
198
8 an
d 20
03 f
or r
esid
ents
of
the
Am
ster
dam
A
irpor
t S
chip
hol,
whe
re
elev
ated
co
ncen
trat
ions
of
po
tent
ial
carc
inog
enic
co
mpo
unds
, in
clud
ing
ben
zene
, P
M10
, C
O
and
SO
2,
wer
e
obse
rved
. C
ance
r ag
e-st
anda
rdiz
ed
inci
denc
e ra
tio
(SIR
) w
ere
com
pute
d us
ing
natio
nal
inci
denc
e ra
tes
as a
re
fere
nce.
Ana
lysi
ng
13,2
07 c
ases
, n
o cl
ear
evid
enc
e of
an
asso
ciat
ion
betw
een
resi
denc
e ar
ound
S
chip
hol
area
and
the
in
cid
ence
of
canc
er
in a
dults
em
erge
d (S
IR=
1.0
2; 9
5% C
I: 1.
00,
1.03
),
a w
here
as
stat
istic
ally
si
gnifi
can
tly
incr
ease
d in
cide
nce
wer
e fo
und
for
hem
atol
ogic
al
mal
igna
ncie
s (S
IR=
1.1
2;
95%
CI:
1.05
, 1.
19),
the
inc
iden
ce o
f ca
ncer
of
th
e re
spira
tory
sy
stem
w
as
stat
istic
ally
si
gnifi
cant
ly d
ecre
ased
(S
IR=
0.94
, 95
% C
I: 0.
90,
0.99
), b
ecau
se o
f to
the
lo
w r
ate
in
mal
es (
SIR
0.8
9).
No
estim
ates
for
PM
.
FRA
NC
E
Fill
eul e
t al
. (2
005)
(F
illeu
l et
al,
2005
) P
AA
RC
sur
vey
24
area
s of
7
Fre
nch
citie
s pe
riod
of e
xpos
ure
mea
sure
men
t: 19
74-1
976
Sub
ject
s= 1
4,28
4 ad
ults
T
SP
min
/ma
x: 4
5/24
3 B
S m
in/m
ax:
18/
152.
Thi
s st
udy
exam
ined
the
lon
g te
rm e
ffect
s of
T
SP
, B
S,
SO
2,
NO
2 an
d N
O (
daily
ass
esse
d in
19
74-1
976)
on
mor
talit
y, i
nclu
ding
lun
g ca
ncer
m
orta
lity
in
the
Pol
lutio
n
Atm
osph
eriq
ue
et
Affe
ctio
ns
Re
spira
toire
s C
hron
ique
s (P
AA
RC
) su
rvey
, co
nduc
ted
in 2
4 ar
eas
of 7
Fre
nch
citie
s on
14,
284
adul
ts,
follo
wed
for
the
per
iod
1974
-20
01.
RR
s w
ere
estim
ated
by
Co
x pr
opor
tiona
l ha
zard
reg
ress
ion
mod
els,
afte
r ad
just
men
t fo
r ag
e,
sex,
sm
okin
g,
BM
I,
educ
atio
n,
and
occu
patio
nal
expo
sure
. T
he
anal
yses
w
ere
pe
rfor
med
for
all
24 a
reas
and
in
a su
bgro
up o
f 18
ar
eas
with
a
ratio
of
N
O/
NO
2 <
3
(with
m
onito
rs p
oten
tially
influ
ence
d by
loca
l tra
ffic)
.
Afte
r ap
prox
imat
ely
25 y
ears
of
follo
w-u
p,
178
lung
ca
ncer
de
aths
oc
curr
ed.
Con
side
ring
all
the
24
are
as,
and
afte
r ex
clus
ion
of
6 ar
eas
influ
ence
d by
lo
cal
traf
fic,
no a
ssoc
iatio
n w
as
foun
d be
twe
en
TS
P,
BS
an
d th
e ot
her
cons
ider
ed
co-
pollu
tant
s, a
nd
lung
can
cer
mor
talit
y. N
O2
sign
ifica
ntly
in
crea
sed
the
mor
talit
y fo
r lu
ng
canc
er.
Mor
talit
y R
R f
or a
n in
crem
ent
of
10 μ
g/m
3 of T
SP
: 24
are
as: 0
.97
(0.9
4-1.
01)
18 a
reas
not
influ
ence
d by
loca
l tr
affic
: 1.0
0 (0
.92-
1.10
) M
orta
lity
RR
for
an
incr
eme
nt o
f 10
μg/
m3 o
f BS
: 24
are
as: 0
.97
(0.9
3-1.
01)
18 a
reas
not
influ
ence
d by
loca
l tr
affic
: 1.0
3 (0
.92-
1.15
).
ITA
LY
P
arod
i et
al
. (2
005)
(P
arod
i et a
l, 20
05).
C
orni
glia
no,
a di
stric
t of
th
e G
enoa
m
unic
ipal
ity,
nort
hern
Ital
y, 1
986-
1997
. P
M10
, an
nua
l m
ean
be
twee
n 57
an
d 79
be
fore
th
e co
ke
oven
cl
osin
g (2
001
) an
d 43
in
20
02.
Thi
s st
udy
inve
stig
ated
th
e in
cide
nce
of
lu
ng
canc
er b
etw
een
198
6 an
d 19
97 f
or r
esid
ents
of
an a
rea
of n
orth
ern
Italy
nea
r a
coke
ove
n p
lant
, w
here
el
evat
ed
conc
entr
atio
ns
of
pote
ntia
l ca
rcin
ogen
ic
com
poun
ds,
incl
udin
g be
nze
ne,
PM
10,
CO
an
d S
O2,
wer
e ob
serv
ed a
roun
d th
e co
ke
oven
. A
ge-s
tand
ardi
zed
inci
denc
e
ratio
s (S
IR)
betw
een
Cor
nigl
iano
an
d tw
o
sele
cted
re
fere
nce
popu
latio
ns
(Riv
arol
o an
d G
enoa
) w
ere
deriv
ed.
In
Cor
nigl
ian
o,
158
lung
ca
ncer
ca
ses
occu
rred
in
men
, an
d 28
in
wom
en.
Onl
y a
mar
gina
l e
xces
s ris
k of
lu
ng
canc
er
wa
s ob
serv
ed
com
pare
d w
ith
two
sele
cted
co
ntro
l po
pula
tions
whi
le a
gra
dien
t in
the
ar
eas
clos
e to
th
e pl
ant
em
erge
d am
ong
fem
ales
. U
sing
G
enoa
a
s th
e re
fere
nce
popu
latio
n,
the
age-
adju
sted
S
IR
for
Cor
nigl
iano
was
1.1
9 (9
5% C
I: 1.
01-1
.38)
in
men
, an
d 0.
98
(95%
C
I: 0.
65-1
.41)
in
w
omen
.
No
estim
ates
for
PM
.
report no. 4/11
172
Bar
bone
et
al
. (1
995)
(B
arbo
ne e
t al,
1995
).
Cas
e-co
ntro
l st
udy
on
755
case
s di
ed
for
lung
ca
ncer
and
755
mat
ched
co
ntro
ls
died
fo
r ot
her
dise
ases
in T
riest
e.
Stu
dy
perio
d
1979
-198
1 an
d 19
85-1
986.
M
ean
tota
l pa
rtic
ulat
e
depo
sitio
n be
twee
n 0.
210
(rur
al
area
) an
d 0.
721
g/m
2 /day
(in
dust
rial a
rea)
.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
part
icul
ate
depo
sitio
n an
d l
ung
canc
er m
orta
lity
in a
cas
e-co
ntro
l stu
dy b
ase
d on
755
cas
es
died
fo
r lu
ng c
ance
r an
d 75
5 m
atch
ed c
ontr
ols
died
fo
r ot
her
dise
ases
in T
riest
e fr
om 1
976
to 1
981,
an
d fr
om
1985
to
19
86.
Air
pollu
tion
at
the
resi
denc
e of
eac
h su
bjec
t w
as
estim
ate
d fr
om
the
aver
age
valu
e of
par
ticul
ate
depo
sitio
n at
th
e ne
ares
t of
28
mon
itorin
g s
tatio
ns.
OR
s w
ere
de
rived
by
lo
gist
ic
regr
essi
on
mod
els
afte
r ad
just
men
t fo
r ag
e, s
mok
ing
habi
t, ex
posu
re t
o oc
cupa
tiona
l car
cino
gens
and
soc
ial c
lass
(da
ta
obta
ined
from
eac
h su
bjec
t’s n
ext o
f kin
).
The
ris
k of
lu
ng
canc
er
incr
ease
d w
ith
incr
easi
ng l
evel
of
air
pollu
tion
for
all
type
s of
lu
ng
canc
er
com
bine
d (p
pe
r tr
end=
0.02
2),
for
smal
l ce
ll ca
rcin
oma
(p=
0.01
6)
and
for
larg
e ce
ll ca
rcin
oma
(p=
0.04
9).
OR
fo
r to
tal
part
icul
ate
de
posi
tion>
0.29
8 vs
<
0.17
5
g/m
2 /day
: A
ll lu
ng c
ance
rs: 1
.4 (
1.1,
1.8
) S
quam
ous:
1.2
(0.
8, 1
.7)
Sm
all c
ell:
1.7
(1.
1, 2
.5)
Larg
e ce
ll: 1
.7 (
0.9,
3.0
) A
deno
carc
ino
ma:
1.3
(0.
8, 2
.0).
Cro
sign
ani
et
al.
(200
4)
(Cro
sign
ani e
t al
, 200
4).
Cas
e-co
ntro
l st
udy
(120
ca
ses
with
ch
ildho
od
leuk
aem
ia,
and
480
cont
rols
),
Var
ese,
no
rthe
rn It
aly,
197
8-19
97.
No
info
rmat
ion
on P
M.
Thi
s st
udy
exam
ined
the
effe
ct o
f an
nual
mea
n co
ncen
trat
ion
of
be
nzen
e on
th
e
risk
of
child
hood
le
uke
mia
, us
ing
a po
pula
tion-
base
d ca
se-
cont
rol
stud
y in
th
e P
rovi
nce
of V
ares
e,
nort
hern
Ital
y. A
ll 1
20 in
cide
nt c
ases
from
197
8–97
wer
e in
clu
ded
in t
he s
tudy
. C
ontr
ols
wer
e 48
0 se
x-
and
ag
e-m
atch
ed
child
ren
sam
pled
fr
om t
he g
ener
al p
opul
atio
n. T
he c
once
ntra
tion
of
be
nzen
e w
as
mea
sure
d ou
tsid
e th
e
hom
e us
ing
a G
auss
ian
diffu
sion
m
odel
. O
Rs
wer
e de
rived
by
lo
gist
ic
regr
essi
on
cond
ition
ed
by
age
and
sex.
Com
pare
d to
ch
ildre
n w
hos
e ho
mes
w
as
not
expo
sed
to r
oad
traf
fic e
mis
sion
s (<
0.1
μg
/m3 o
f be
nzen
e an
nual
ave
rage
), t
he r
isk
of
child
hood
le
ukae
mia
w
as
sign
ifica
ntly
hi
gher
fo
r he
avily
e
xpos
ed
child
ren
(>10
μg
/m3 ),
the
RR
bei
ng 3
.91
(95%
CI:
1.36
, 11
.27)
.
No
estim
ates
for
PM
.
POLA
ND
Jedr
ycho
wsk
i et
al
. (1
990)
(J
edry
cho
wsk
i et
al
, 199
0)
Cas
e-co
ntro
l st
udy
on
1,09
9 ca
ses
died
for
lun
g ca
ncer
and
1,0
73 c
ontr
ols
freq
uenc
y m
atch
ed
by
age
and
sex
died
fo
r ot
her
dise
ase
s in
Cra
cow
. S
tudy
per
iod
198
0-19
85.
Mea
n T
SP
not
pro
vide
d.
Thi
s st
udy
exa
min
ed t
he a
ssoc
iatio
n b
etw
een
expo
sure
to
ai
r po
llutio
n an
d lu
ng
canc
er
mor
talit
y in
a c
ase-
cont
rol s
tudy
bas
ed o
n 1
,099
ca
ses
died
for
lun
g ca
ncer
and
1,0
73 m
atch
ed
cont
rols
di
ed
for
othe
r di
seas
es
in
Cra
cow
, P
olan
d be
twe
en 1
980
and
1985
. A
ir po
llutio
n at
th
e re
side
nce
of e
ach
subj
ect
wa
s de
fined
on
the
basi
s of
th
e co
mbi
ned
conc
entr
atio
n o
f T
SP
an
d S
O2,
det
erm
ined
by
20 m
onito
ring
stat
ions
fo
r 8
year
s (1
973-
1980
). O
Rs
wer
e d
eriv
ed b
y lo
gist
ic r
egre
ssio
n m
odel
s af
ter
adju
stm
ent
for
age,
sm
okin
g h
abit,
exp
osur
e to
occ
upat
iona
l ca
rcin
ogen
s a
nd e
duca
tion
(da
ta o
btai
ned
from
ea
ch s
ubje
ct’s
nex
t of k
in).
The
ris
k of
lu
ng
canc
er
incr
ease
d w
ith
incr
easi
ng l
evel
of
air
pollu
tion
in m
en (
OR
fo
r th
e hi
ghe
st
vs
the
low
est
leve
l of
ai
r po
llutio
n w
as 1
.48;
95%
CI
1.08
, 2.
01)
but
not
in w
omen
(O
R w
as 1
.17;
95%
CI
0.70
, 1.
96).
No
estim
ates
for
TS
P.
report no. 4/11
173
13. REFERENCES
1. Ackermann-Liebrich U, Leuenberger P, Schwartz J, Schindler C, Monn C, Bolognini G, Bongard JP, Brandli O, Domenighetti G, Elsasser S, Grize L, Karrer W, Keller R, Keller-Wossidlo H, Kunzli N, Martin BW, Medici TC, Perruchoud AP, Schoni MH, Tschopp JM, Villiger B, Wuthrich B, Zellweger JP, Zemp E (1997) Lung function and long term exposure to air pollutants in Switzerland. Study on Air Pollution and Lung Diseases in Adults (SAPALDIA) Team. Am J Respir Crit Care Med 155: 122-129.
2. Aga E, Samoli E, Touloumi G, Anderson HR, Cadum E, Forsberg B, Goodman P, Goren A, Kotesovec F, Kriz B, Macarol-Hiti M, Medina S, Paldy A, Schindler C, Sunyer J, Tittanen P, Wojtyniak B, Zmirou D, Schwartz J, Katsouyanni K (2003) Short-term effects of ambient particles on mortality in the elderly: results from 28 cities in the APHEA2 project. Eur Respir J Suppl 40: 28s-33s.
3. Alberdi Odriozola JC, Diaz Jimenez J, Montero Rubio JC, Miron Perez IJ, Pajares Ortiz MS, Ribera Rodrigues P (1998) Air pollution and mortality in Madrid, Spain: a time-series analysis. Int Arch Occup Environ Health 71: 543-549.
4. Analitis A, Katsouyanni K, Dimakopoulou K, Samoli E, Nikoloulopoulos AK, Petasakis Y, Touloumi G, Schwartz J, Anderson HR, Cambra K, Forastiere F, Zmirou D, Vonk JM, Clancy L, Kriz B, Bobvos J, Pekkanen J (2006) Short-term effects of ambient particles on cardiovascular and respiratory mortality. Epidemiology 17: 230-233.
5. Anderson HR, Bremner SA, Atkinson RW, Harrison RM, Walters S (2001) Particulate matter and daily mortality and hospital admissions in the west midlands conurbation of the United Kingdom: associations with fine and coarse particles, black smoke and sulphate. Occup Environ Med 58: 504-510.
6. Anderson HR, Ponce de Leon A, Bland JM, Bower JS, Emberlin J, Strachan DP (1998) Air pollution, pollens, and daily admissions for asthma in London 1987-92. Thorax 53: 842-848.
7. Anderson HR, Ponce de Leon A, Bland JM, Bower JS, Strachan DP (1996) Air pollution and daily mortality in London: 1987-92. BMJ 312: 665-669.
8. Atkinson RW, Anderson HR, Strachan DP, Bland JM, Bremner SA, Ponce de Leon A (1999a) Short-term associations between outdoor air pollution and visits to accident and emergency departments in London for respiratory complaints. Eur Respir J 13: 257-265.
9. Atkinson RW, Anderson HR, Sunyer J, Ayres J, Baccini M, Vonk JM, Boumghar A, Forastiere F, Forsberg B, Touloumi G, Schwartz J, Katsouyanni K (2001) Acute effects of particulate air pollution on respiratory admissions: results from APHEA 2 project. Air Pollution and Health: a European Approach. Am J Respir Crit Care Med 164: 1860-1866.
10. Atkinson RW, Bremner SA, Anderson HR, Strachan DP, Bland JM, de Leon AP (1999b) Short-term associations between emergency hospital admissions for respiratory and cardiovascular disease and outdoor air pollution in London. Arch Environ Health 54: 398-411.
report no. 4/11
174
11. Baldi I, Tessier JF, Kauffmann F, Jacqmin-Gadda H, Nejjari C, Salamon R (1999) Prevalence of asthma and mean levels of air pollution: results from the French PAARC survey. Pollution Atomospherique et Affections Respiratoires Chroniques. Eur Respir J 14: 132-138.
12. Barbone F, Bovenzi M, Cavallieri F, Stanta G (1995) Air pollution and lung cancer in Trieste, Italy. Am J Epidemiol 141: 1161-1169.
13. Bayer-Oglesby L, Grize L, Gassner M, Takken-Sahli K, Sennhauser FH, Neu U, Schindler C, Braun-Fahrlander C (2005) Decline of ambient air pollution levels and improved respiratory health in Swiss children. Environ Health Perspect 113: 1632-1637.
14. Bayer-Oglesby L, Schindler C, Hazenkamp-von Arx ME, Braun-Fahrlander C, Keidel D, Rapp R, Kunzli N, Braendli O, Burdet L, Sally Liu LJ, Leuenberger P, Ackermann-Liebrich U (2006) Living near main streets and respiratory symptoms in adults: the Swiss Cohort Study on Air Pollution and Lung Diseases in Adults. Am J Epidemiol 164: 1190-1198.
15. Beelen RMJ, Hoek G, Fischer PH, Van den Brandt PA, Bausch-Goldbohm RA, Schouten LJ, Brunekreef B (2007) Long-term effects of trarffic-related air pollution on mortality in a Dutch cohort study (NLCS-AIR Study). Abstract - Health Effects Institute Annual Conference. : Available online at: http://www.healtheffects.org/Pubs/AnnualConferenceProgram2007.pdf
16. Beeson WL, Abbey DE, Knutsen SF (1998) Long-term concentrations of ambient air pollutants and incident lung cancer in California adults: results from the AHSMOG study.Adventist Health Study on Smog. Environ Health Perspect 106: 813-823.
17. Biggeri A, Baccini M, Bellini P, Terracini B (2005) Meta-analysis of the Italian studies of short-term effects of air pollution (MISA), 1990-1999. Int J Occup Environ Health 11: 107-122.
18. Brauer M, Gehring U, Brunekreef B, de Jongste J, Gerritsen J, Rovers M, Wichmann HE, Wijga A, Heinrich J (2006) Traffic-related air pollution and otitis media. Environ Health Perspect 114: 1414-1418.
19. Brauer M, Hoek G, Van Vliet P, Meliefste K, Fischer PH, Wijga A, Koopman LP, Neijens HJ, Gerritsen J, Kerkhof M, Heinrich J, Bellander T, Brunekreef B (2002) Air pollution from traffic and the development of respiratory infections and asthmatic and allergic symptoms in children. Am J Respir Crit Care Med 166: 1092-1098.
20. Braun-Fahrlander C, Vuille JC, Sennhauser FH, Neu U, Kunzle T, Grize L, Gassner M, Minder C, Schindler C, Varonier HS, Wuthrich B (1997) Respiratory health and long-term exposure to air pollutants in Swiss schoolchildren. SCARPOL Team. Swiss Study on Childhood Allergy and Respiratory Symptoms with Respect to Air Pollution, Climate and Pollen. Am J Respir Crit Care Med 155: 1042-1049.
21. Bremner SA, Anderson HR, Atkinson RW, McMichael AJ, Strachan DP, Bland JM, Bower JS (1999) Short-term associations between outdoor air pollution and mortality in London 1992-4. Occup Environ Med 56: 237-244.
22. Brunekreef B, Holgate ST (2002) Air pollution and health. Lancet 360: 1233-1242.
report no. 4/11
175
23. Burr ML, Karani G, Davies B, Holmes BA, Williams KL (2004) Effects on respiratory health of a reduction in air pollution from vehicle exhaust emissions. Occup Environ Med 61: 212-218.
24. Clancy L, Goodman P, Sinclair H, Dockery DW (2002) Effect of air-pollution control on death rates in Dublin, Ireland: an intervention study. Lancet 360: 1210-1214.
25. Crosignani P, Tittarelli A, Borgini A, Codazzi T, Rovelli A, Porro E, Contiero P, Bianchi N, Tagliabue G, Fissi R, Rossitto F, Berrino F (2004) Childhood leukemia and road traffic: A population-based case-control study. Int J Cancer 108: 596-599.
26. D'Ippoliti D, Forastiere F, Ancona C, Agabiti N, Fusco D, Michelozzi P, Perucci CA (2003) Air pollution and myocardial infarction in Rome: a case-crossover analysis. Epidemiology 14: 528-535.
27. Dab W, Medina S, Quenel P, Le Moullec Y, Le Tertre A, Thelot B, Monteil C, Lameloise P, Pirard P, Momas I, Ferry R, Festy B (1996) Short term respiratory health effects of ambient air pollution: results of the APHEA project in Paris. J Epidemiol Community Health 50 Suppl 1: s42-46.
28. de Diego Damia A, Leon Fabregas M, Perpina Tordera M, Compte Torrero L (1999) Effects of air pollution and weather conditions on asthma exacerbation. Respiration 66: 52-58.
29. Diaz J, Garcia R, Ribera P, Alberdi JC, Hernandez E, Pajares MS, Otero A (1999) Modeling of air pollution and its relationship with mortality and morbidity in Madrid, Spain. Int Arch Occup Environ Health 72: 366-376.
30. Dominici F, McDermott A, Daniels M, Zeger SL, Samet JM (2003) Mortality among residents of 90 cities. In Revised analyses of time-series studies of air pollution and health. Special report. Revised analyses of the national morbidity. Mortality and air pollution study. Part II. Health Effects Institute: Boston, MA.
31. Dominici F, McDermott A, Zeger SL, Samet JM (2002) On the use of generalized additive models in time-series studies of air pollution and health. Am J Epidemiol 156: 193-203.
32. Eilstein D, Declercq C, Prouvost H, Pascal L, Nunes C, Filleul L, Cassadou S, Le Tertre A, Zeghnoun A, Medina S, Lefranc A, Saviuc P, Quenel P, Campagna D (2004) The impact of air pollution on health. The "Programme de Surveillance Air et Sante 9 villes" (Air and Health surveillance program in 9 cities). Presse Med 33: 1323-1327.
33. Filleul L, Rondeau V, Vandentorren S, Le Moual N, Cantagrel A, Annesi-Maesano I, Charpin D, Declercq C, Neukirch F, Paris C, Vervloet D, Brochard P, Tessier JF, Kauffmann F, Baldi I (2005) Twenty five year mortality and air pollution: results from the French PAARC survey. Occup Environ Med 62: 453-460.
34. Fischer P, Hoek G, Brunekreef B, Verhoeff A, van Wijnen J (2003) Air pollution and mortality in The Netherlands: are the elderly more at risk? Eur Respir J Suppl 40: 34s-38s.
35. Forastiere F, Stafoggia M, Picciotto S, Bellander T, D'Ippoliti D, Lanki T, von Klot S, Nyberg F, Paatero P, Peters A, Pekkanen J, Sunyer J, Perucci CA (2005) A case-
report no. 4/11
176
crossover analysis of out-of-hospital coronary deaths and air pollution in Rome, Italy. Am J Respir Crit Care Med 172: 1549-1555.
36. Frischer T, Studnicka M, Gartner C, Tauber E, Horak F, Veiter A, Spengler J, Kuhr J, Urbanek R (1999) Lung function growth and ambient ozone: a three-year population study in school children. Am J Respir Crit Care Med 160: 390-396.
37. Frye C, Hoelscher B, Cyrys J, Wjst M, Wichmann HE, Heinrich J (2003) Association of lung function with declining ambient air pollution. Environ Health Perspect 111: 383-387.
38. Fusco D, Forastiere F, Michelozzi P, Spadea T, Ostro B, Arca M, Perucci CA (2001) Air pollution and hospital admissions for respiratory conditions in Rome, Italy. Eur Respir J 17: 1143-1150.
39. Galan I, Tobias A, Banegas JR, Aranguez E (2003) Short-term effects of air pollution on daily asthma emergency room admissions. Eur Respir J 22: 802-808.
40. Garcia-Aymerich J, Tobias A, Anto JM, Sunyer J (2000) Air pollution and mortality in a cohort of patients with chronic obstructive pulmonary disease: a time series analysis. J Epidemiol Community Health 54: 73-74.
41. Garty BZ, Kosman E, Ganor E, Berger V, Garty L, Wietzen T, Waisman Y, Mimouni M, Waisel Y (1998) Emergency room visits of asthmatic children, relation to air pollution, weather, and airborne allergens. Ann Allergy Asthma Immunol 81: 563-570.
42. Gehring U, Cyrys J, Sedlmeir G, Brunekreef B, Bellander T, Fischer P, Bauer CP, Reinhardt D, Wichmann HE, Heinrich J (2002) Traffic-related air pollution and respiratory health during the first 2 yrs of life. Eur Respir J 19: 690-698.
43. Gehring U, Heinrich J, Kramer U, Grote V, Hochadel M, Sugiri D, Kraft M, Rauchfuss K, Eberwein HG, Wichmann HE (2006) Long-term exposure to ambient air pollution and cardiopulmonary mortality in women. Epidemiology 17: 545-551.
44. Goodman PG, Dockery DW, Clancy L (2004) Cause-specific mortality and the extended effects of particulate pollution and temperature exposure. Environ Health Perspect 112: 179-185.
45. Hagen JA, Nafstad P, Skrondal A, Bjorkly S, Magnus P (2000) Associations between outdoor air pollutants and hospitalization for respiratory diseases. Epidemiology 11: 136-140.
46. Hajat S, Anderson HR, Atkinson RW, Haines A (2002) Effects of air pollution on general practitioner consultations for upper respiratory diseases in London. Occup Environ Med 59: 294-299.
47. Hajat S, Haines A, Atkinson RW, Bremner SA, Anderson HR, Emberlin J (2001) Association between air pollution and daily consultations with general practitioners for allergic rhinitis in London, United Kingdom. Am J Epidemiol 153: 704-714.
48. Hajat S, Haines A, Goubet SA, Atkinson RW, Anderson HR (1999) Association of air pollution with daily GP consultations for asthma and other lower respiratory conditions in London. Thorax 54: 597-605.
report no. 4/11
177
49. Hastings DL, Jardine S (2002) The relationship between air particulate levels and upper respiratory disease in soldiers deployed to Bosnia (1997-1998). Mil Med 167: 296-303.
50. HEI (2003) Revised analyses of time-series studies of air pollution and health. Special report.
51. Heinrich J, Hoelscher B, Frye C, Meyer I, Pitz M, Cyrys J, Wjst M, Neas L, Wichmann HE (2002) Improved air quality in reunified Germany and decreases in respiratory symptoms. Epidemiology 13: 394-401.
52. Heinrich J, Hoelscher B, Wichmann HE (2000) Decline of ambient air pollution and respiratory symptoms in children. Am J Respir Crit Care Med 161: 1930-1936.
53. Heinrich J, Hoelscher B, Wjst M, Ritz B, Cyrys J, Wichmann H (1999) Respiratory diseases and allergies in two polluted areas in East Germany. Environ Health Perspect 107: 53-62.
54. Hoek G (2003) Daily mortality and air pollution in The Netherlands. In Revised analyses of time-series studies of air pollution and health. Special report pp. 133-142: Boston, MA.
55. Hoek G, Brunekreef B, Fischer P, van Wijnen J (2001) The association between air pollution and heart failure, arrhythmia, embolism, thrombosis, and other cardiovascular causes of death in a time series study. Epidemiology 12: 355-357.
56. Hoek G, Brunekreef B, Goldbohm S, Fischer P, van den Brandt PA (2002) Association between mortality and indicators of traffic-related air pollution in the Netherlands: a cohort study. Lancet 360: 1203-1209.
57. Hoek G, Brunekreef B, Verhoeff A, van Wijnen J, Fischer P (2000) Daily mortality and air pollution in The Netherlands. J Air Waste Manag Assoc 50: 1380-1389.
58. Hoek G, Schwartz JD, Groot B, Eilers P (1997) Effects of ambient particulate matter and ozone on daily mortality in Rotterdam, The Netherlands. Arch Environ Health 52: 455-463.
59. Hoffmann B, Moebus S, Stang A, Beck EM, Dragano N, Mohlenkamp S, Schmermund A, Memmesheimer M, Mann K, Erbel R, Jockel KH (2006) Residence close to high traffic and prevalence of coronary heart disease. Eur Heart J 27: 2696-2702.
60. Horak F, Jr., Studnicka M, Gartner C, Spengler JD, Tauber E, Urbanek R, Veiter A, Frischer T (2002) Particulate matter and lung function growth in children: a 3-yr follow-up study in Austrian schoolchildren. Eur Respir J 19: 838-845.
61. Jedrychowski W, Becher H, Wahrendorf J, Basa-Cierpialek Z (1990) A case-control study of lung cancer with special reference to the effect of air pollution in Poland. J Epidemiol Community Health 44: 114-120.
62. Jedrychowski W, Flak E (1998) Separate and combined effects of the outdoor and indoor air quality on chronic respiratory symptoms adjusted for allergy among preadolescent children. Int J Occup Med Environ Health 11: 19-35.
report no. 4/11
178
63. Jedrychowski W, Flak E, Mroz E (1999) The adverse effect of low levels of ambient air pollutants on lung function growth in preadolescent children. Environ Health Perspect 107: 669-674.
64. Karakatsani A, Andreadaki S, Katsouyanni K, Dimitroulis I, Trichopoulos D, Benetou V, Trichopoulou A (2003) Air pollution in relation to manifestations of chronic pulmonary disease: a nested case-control study in Athens, Greece. Eur J Epidemiol 18: 45-53.
65. Katsouyanni K, Schwartz J, Spix C, Touloumi G, Zmirou D, Zanobetti A, Wojtyniak B, Vonk JM, Tobias A, Ponka A, Medina S, Bacharova L, Anderson HR (1996) Short term effects of air pollution on health: a European approach using epidemiologic time series data: the APHEA protocol. J Epidemiol Community Health 50 Suppl 1: S12-18.
66. Katsouyanni K, Touloumi G, Samoli E, Gryparis A, Le Tertre A, Monopolis Y, Rossi G, Zmirou D, Ballester F, Boumghar A, Anderson HR, Wojtyniak B, Paldy A, Braunstein R, Pekkanen J, Schindler C, Schwartz J (2001) 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 12: 521-531.
67. Katsouyanni K, Touloumi G, Samoli E, Petasakis Y, Analitis A, Le Tertre A, Rossi G, Zmirou D, Ballester F, Boumghar A, Anderson HR, Wojtyniak B, Paldy A, Braunstein R, Pekkanen J, Schindler C, Schwartz J (2003) Sensitivity analysis of various models of short-term effects of ambient particles on total mortality in 29 cities in APHEA2. In Revised analyses of time-series studies of air pollution and health. Special report pp. 157-164. Health Effects Institute: Boston, MA.
68. 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 (1997) 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 314: 1658-1663.
69. Keatinge WR, Donaldson GC (2001) Mortality related to cold and air pollution in London after allowance for effects of associated weather patterns. Environ Res 86: 209-216.
70. Knox EG (2005) Oil combustion and childhood cancers. J Epidemiol Community Health 59: 755-760.
71. Knox EG (2006) Roads, railways, and childhood cancers. J Epidemiol Community Health 60: 136-141.
72. Kontos AS, Fassois SD, Deli MF (1999) Short-term effects of air pollution on childhood respiratory illness in Piraeus, Greece, 1987-1992: nonparametric stochastic dynamic analysis. Environ Res 81: 275-296.
73. Kotesovec F, Skorkovsky J, Brynda J, Peters A, Heinrich J (2000) Daily mortality and air pollution in northern Bohemia: different effects for men and women. Cent Eur J Public Health 8: 120-127.
74. Kramer U, Behrendt H, Dolgner R, Ranft U, Ring J, Willer H, Schlipkoter HW (1999) Airway diseases and allergies in East and West German children during the first 5
report no. 4/11
179
years after reunification: time trends and the impact of sulphur dioxide and total suspended particles. Int J Epidemiol 28: 865-873.
75. Kulkarni N, Pierse N, Rushton L, Grigg J (2006) Carbon in airway macrophages and lung function in children. N Engl J Med 355: 21-30.
76. Laden F, Schwartz J, Speizer FE, Dockery DW (2006) Reduction in fine particulate air pollution and mortality: Extended follow-up of the Harvard Six Cities study. Am J Respir Crit Care Med 173: 667-672.
77. Lanki T, Pekkanen J, Aalto P, Elosua R, Berglind N, D'Ippoliti D, Kulmala M, Nyberg F, Peters A, Picciotto S, Salomaa V, Sunyer J, Tiittanen P, von Klot S, Forastiere F (2006) Associations of traffic-related air pollutants with hospitalisation for first acute myocardial infarction. The HEAPSS study. Occup Environ Med 63: 844-851.
78. Le Tertre A, Medina S, Samoli E, Forsberg B, Michelozzi P, Boumghar A, Vonk JM, Bellini A, Atkinson R, Ayres JG, Sunyer J, Schwartz J, Katsouyanni K (2002a) Short-term effects of particulate air pollution on cardiovascular diseases in eight European cities. J Epidemiol Community Health 56: 773-779.
79. Le Tertre A, Quenel P, Eilstein D, Medina S, Prouvost H, Pascal L, Boumghar A, Saviuc P, Zeghnoun A, Filleul L, Declercq C, Cassadou S, Le Goaster C (2002b) Short-term effects of air pollution on mortality in nine French cities: a quantitative summary. Arch Environ Health 57: 311-319.
80. Leonardi GS, Houthuijs D, Steerenberg PA, Fletcher T, Armstrong B, Antova T, Lochman I, Lochmanova A, Rudnai P, Erdei E, Musial J, Jazwiec-Kanyion B, Niciu EM, Durbaca S, Fabianova E, Koppova K, Lebret E, Brunekreef B, van Loveren H (2000) Immune biomarkers in relation to exposure to particulate matter: a cross-sectional survey in 17 cities of Central Europe. Inhal Toxicol 12 Suppl 4: 1-14.
81. Llorca J, Salas A, Prieto-Salceda D, Chinchon-Bengoechea V, Delgado-Rodriguez M (2005) Nitrogen dioxide increases cardiorespiratory admissions in Torrelavega (Spain). J Environ Health 68: 38867.
82. Maheswaran R, Haining RP, Brindley P, Law J, Pearson T, Fryers PR, Wise S, Campbell MJ (2005a) Outdoor air pollution and stroke in Sheffield, United Kingdom: a small-area level geographical study. Stroke 36: 239-243.
83. Maheswaran R, Haining RP, Brindley P, Law J, Pearson T, Fryers PR, Wise S, Campbell MJ (2005b) Outdoor air pollution, mortality, and hospital admissions from coronary heart disease in Sheffield, UK: a small-area level ecological study. Eur Heart J 26: 2543-2549.
84. Mannsaker B, Vikan T, Holme J (2004) Air pollution and cardiovascular disease in Trondheim. Tidsskr Nor Laegeforen 124: 1381-1383.
85. McGregor GR, Walters S, Wordley J (1999) Daily hospital respiratory admissions and winter air mass types, Birmingham, UK. Int J Biometeorol 43: 21-30.
86. Medina S, Le Tertre A, Quenel P, Le Moullec Y, Lameloise P, Guzzo JC, Festy B, Ferry R, Dab W (1997) Air pollution and doctors' house calls: results from the ERPURS system for monitoring the effects of air pollution on public health in Greater Paris, France, 1991-1995. Evaluation des Risques de la Pollution Urbaine pour la Sante. Environ Res 75: 73-84.
report no. 4/11
180
87. Michelozzi P, Forastiere F, Fusco D, Perucci CA, Ostro B, Ancona C, Pallotti G (1998) Air pollution and daily mortality in Rome, Italy. Occup Environ Med 55: 605-610.
88. Migliaretti G, Cadum E, Migliore E, Cavallo F (2005) Traffic air pollution and hospital admission for asthma: a case-control approach in a Turin (Italy) population. Int Arch Occup Environ Health 78: 164-169.
89. Migliaretti G, Cavallo F (2004) Urban air pollution and asthma in children. Pediatr Pulmonol 38: 198-203.
90. Naess O, Nafstad P, Aamodt G, Claussen B, Rosland P (2007) Relation between concentration of air pollution and cause-specific mortality: Four-year exposures to nitrogen dioxide and particulate matter pollutants in 470 neighborhoods in Oslo, Norway. Am J Epidemiol 165: 435-443.
91. Nafstad P, Haheim LL, Oftedal B, Gram F, Holme I, Hjermann I, Leren P (2003) Lung cancer and air pollution: a 27 year follow up of 16 209 Norwegian men. Thorax 58: 1071-1076.
92. Nafstad P, Haheim LL, Wisloff T, Gram F, Oftedal B, Holme I, Hjermann I, Leren P (2004) Urban air pollution and mortality in a cohort of Norwegian men. Environ Health Perspect 112: 610-615.
93. Nawrot TS, Nackaerts K, Hoet PH, Nemery B (2007) Lung cancer mortality and fine particulate air pollution in Europe. Int J Cancer 120: 1825-1826; author reply 1827.
94. Nyberg F, Gustavsson P, Jarup L, Bellander T, Berglind N, Jakobsson R, Pershagen G (2000) Urban air pollution and lung cancer in Stockholm. Epidemiology 11: 487-495.
95. Oftedal B, Nafstad P, Magnus P, Bjorkly S, Skrondal A (2003) Traffic related air pollution and acute hospital admission for respiratory diseases in Drammen, Norway 1995-2000. Eur J Epidemiol 18: 671-675.
96. Paatero P, Aalto P, Picciotto S, et a (2005) Estimating time series of aerosol particle number concentrations in the five HEAPSS cities on the basis of measured air pollution and meteorological variables. Atmos Environ 39: 2261-2273.
97. Pantazopoulou A, Katsouyanni K, Kourea-Kremastinou J, Trichopoulos D (1995) Short-term effects of air pollution on hospital emergency outpatient visits and admissions in the greater Athens, Greece area. Environ Res 69: 31-36.
98. Parodi S, Stagnaro E, Casella C, Puppo A, Daminelli E, Fontana V, Valerio F, Vercelli M (2005) Lung cancer in an urban area in Northern Italy near a coke oven plant. Lung Cancer 47: 155-164.
99. Penard-Morand C, Charpin D, Raherison C, Kopferschmitt C, Caillaud D, Lavaud F, Annesi-Maesano I (2005) Long-term exposure to background air pollution related to respiratory and allergic health in schoolchildren. Clin Exp Allergy 35: 1279-1287.
100. Peters A, Skorkovsky J, Kotesovec F, Brynda J, Spix C, Wichmann HE, Heinrich J (2000) Associations between mortality and air pollution in central Europe. Environ Health Perspect 108: 283-287.
report no. 4/11
181
101. Peters A, von Klot S, Heier M, Trentinaglia I, Hormann A, Wichmann HE, Lowel H (2004) Exposure to traffic and the onset of myocardial infarction. N Engl J Med 351: 1721-1730.
102. Pierse N, Rushton L, Harris RS, Kuehni CE, Silverman M, Grigg J (2006) Locally generated particulate pollution and respiratory symptoms in young children. Thorax 61: 216-220.
103. Polat D, Eberwein G, Becker A, Weishaupt C, Schins RP, Ranft U, Borm PJ (2002) Ambient exposure and nasal inflammation in adults and children--a preliminary analysis. Int J Hyg Environ Health 205: 229-234.
104. Ponce de Leon A, Anderson HR, Bland JM, Strachan DP, Bower J (1996) Effects of air pollution on daily hospital admissions for respiratory disease in London between 1987-88 and 1991-92. J Epidemiol Community Health 50 Suppl 1: s63-70.
105. Ponka A, Savela M, Virtanen M (1998) Mortality and air pollution in Helsinki. Arch Environ Health 53: 281-286.
106. Pope CA, 3rd, Burnett RT, Thun MJ, Calle EE, Krewski D, Ito K, Thurston GD (2002) Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA 287: 1132-1141.
107. Pope CA, 3rd, Burnett RT, Thurston GD, Thun MJ, Calle EE, Krewski D, Godleski JJ (2004) Cardiovascular mortality and long-term exposure to particulate air pollution: epidemiological evidence of general pathophysiological pathways of disease. Circulation 109: 71-77.
108. Prescott GJ, Cohen GR, Elton RA, Fowkes FG, Agius RM (1998) Urban air pollution and cardiopulmonary ill health: a 14.5 year time series study. Occup Environ Med 55: 697-704.
109. Raaschou-Nielsen O, Hertel O, Thomsen BL, Olsen JH (2001) Air pollution from traffic at the residence of children with cancer. Am J Epidemiol 153: 433-443.
110. Raaschou-Nielsen O, Hertel O, Thomsen BL, Olsen JH (2002) Traffic-related air pollution at the place of residence and risk of cancer among children. Ugeskr Laeger 164: 2283-2287.
111. Rahlenbeck SI, Kahl H (1996) Air population and mortality in East Berlin during the winters of 1981-1989. Int J Epidemiol 25: 1220-1226.
112. Roemer WH, van Wijnen JH (2001) Daily mortality and air pollution along busy streets in Amsterdam, 1987-1998. Epidemiology 12: 649-653.
113. Rooney C, McMichael AJ, Kovats RS, Coleman MP (1998) Excess mortality in England and Wales, and in Greater London, during the 1995 heatwave. J Epidemiol Community Health 52: 482-486.
114. Rosenlund M, Berglind N, Pershagen G, Hallqvist J, Jonson T, Bellander T (2006) Long-term exposure to urban air pollution and myocardial infarction. Epidemiology 17: 383-390.
115. Ross Anderson H, Atkinson R, Peacock J, Marston L, Konstantinou K (2004) Meta-analysis of time-series studies ad panel studies of Particulate Matter (PM) and
report no. 4/11
182
Ozone (O3). Report of a WHO task group. World Health Organization: Denmark. Availabe online at: http://www.euro.who.int/document/e82792.pdf.
116. Rossi G, Vigotti MA, Zanobetti A, Repetto F, Gianelle V, Schwartz J (1999) Air pollution and cause-specific mortality in Milan, Italy, 1980-1989. Arch Environ Health 54: 158-164.
117. Samoli E, Schwartz J, Analitis A, Petasakis Y, Wojtyniak B, Touloumi G, Spix C, Balducci F, Medina S, Rossi G, Sunyer J, Anderson HR, Katsouyanni K (2003) Sensitivity analyses of regional differences in short-term effects of air pollution on daily mortality in APHEA cities. In Revised analyses of time-series studies of air pollution and health. Special report pp. 205-210. Health Effects Institute: Boston, MA.
118. Samoli E, Schwartz J, Wojtyniak B, Touloumi G, Spix C, Balducci F, Medina S, Rossi G, Sunyer J, Bacharova L, Anderson HR, Katsouyanni K (2001) Investigating regional differences in short-term effects of air pollution on daily mortality in the APHEA project: a sensitivity analysis for controlling long-term trends and seasonality. Environ Health Perspect 109: 349-353.
119. Sansebastiano G, Zoni R, Veronesi L, Stano MC, Tanzi ML, Caminiti C, Impallomeni M (2003) Air pollution and mortality and hospital admission for respiratory disease in Parma on 1992-2001 period. Ann Ig 15: 965-974.
120. Schikowski T, Sugiri D, Ranft U, Gehring U, Heinrich J, Wichmann HE, Kramer U (2005) Long-term air pollution exposure and living close to busy roads are associated with COPD in women. Respir Res 6: 152.
121. Schouten JP, Vonk JM, de Graaf A (1996) Short term effects of air pollution on emergency hospital admissions for respiratory disease: results of the APHEA project in two major cities in The Netherlands, 1977-89. J Epidemiol Community Health 50 Suppl 1: s22-29.
122. Sichletidis L, Tsiotsios I, Gavriilidis A, Chloros D, Gioulekas D, Kottakis I, Pataka A (2005) The effects of environmental pollution on the respiratory system of children in western Macedonia, Greece. J Investig Allergol Clin Immunol 15: 117-123.
123. Solomon C, Poole J, Jarup L, Palmer K, Coggon D (2003) Cardio-respiratory morbidity and long-term exposure to particulate air pollution. Int J Environ Health Res 13: 327-335.
124. Stafoggia M, Picciotto S, Forastiere F, D'Ippoliti D, Cattani G, Marconi A, Perucci CA (2005) Air pollution and fatal and non fatal coronary events in Rome. Epidemiol Prev 29: 40-47.
125. Stolzel M, Breitner S, Cyrys J, Pitz M, Wolke G, Kreyling W, Heinrich J, Wichmann HE, Peters A (2006) Daily mortality and particulate matter in different size classes in Erfurt, Germany. J Expo Sci Environ Epidemiol (Epub ahead of print).
126. Stölzel M, Peters A, Wichmann H-E (2003) Daily mortality and fine and ultrafine particles in Erfurt, Germany. In Revised analyses of time-series studies of air pollution and health. Special report pp. 231-240. Health Effects Institute: Boston, MA.
report no. 4/11
183
127. Sunyer J, Basagana X (2001) Particles, and not gases, are associated with the risk of death in patients with chronic obstructive pulmonary disease. Int J Epidemiol 30: 1138-1140.
128. Sunyer J, Jarvis D, Gotschi T, Garcia-Esteban R, Jacquemin B, Aguilera I, Ackerman U, de Marco R, Forsberg B, Gislason T, Heinrich J, Norback D, Villani S, Kunzli N (2006) Chronic bronchitis and urban air pollution in an international study. Occup Environ Med 63: 836-843.
129. Sunyer J, Schwartz J, Tobias A, Macfarlane D, Garcia J, Anto JM (2000) Patients with chronic obstructive pulmonary disease are at increased risk of death associated with urban particle air pollution: a case-crossover analysis. Am J Epidemiol 151: 50-56.
130. Sunyer J, Spix C, Quenel P, Ponce-de-Leon A, Ponka A, Barumandzadeh T, Touloumi G, Bacharova L, Wojtyniak B, Vonk J, Bisanti L, Schwartz J, Katsouyanni K (1997) Urban air pollution and emergency admissions for asthma in four European cities: the APHEA Project. Thorax 52: 760-765.
131. Tenias JM, Ballester F, Rivera ML (1998) Association between hospital emergency visits for asthma and air pollution in Valencia, Spain. Occup Environ Med 55: 541-547.
132. Thompson AJ, Shields MD, Patterson CC (2001) Acute asthma exacerbations and air pollutants in children living in Belfast, Northern Ireland. Arch Environ Health 56: 234-241.
133. Tobias A, Campbell MJ (1999) Modelling influenza epidemics in the relation between black smoke and total mortality. A sensitivity analysis. J Epidemiol Community Health 53: 583-584.
134. Touloumi G, Katsouyanni K, Zmirou D, Schwartz J, Spix C, de Leon AP, Tobias A, Quennel P, Rabczenko D, Bacharova L, Bisanti L, Vonk JM, Ponka A (1997) Short-term effects of ambient oxidant exposure on mortality: a combined analysis within the APHEA project. Air Pollution and Health: a European Approach. Am J Epidemiol 146: 177-185.
135. Touloumi G, Samoli E, Quenel P, Paldy A, Anderson RH, Zmirou D, Galan I, Forsberg B, Schindler C, Schwartz J, Katsouyanni K (2005) Short-term effects of air pollution on total and cardiovascular mortality: the confounding effect of influenza epidemics. Epidemiology 16: 49-57.
136. Turnovska T, Kostianev S (1999) Effects of reduced air pollution on children's pulmonary function. Cent Eur J Public Health 7: 77-79.
137. US EPA (1997) Air quality criteria for particulate matter: Vol II. US Environmental Protection Agency, Research Triangle Park, NC. Document EPA/600/P-95/001cF. 3v.
138. US EPA (2004) Air quality criteria for particulate matter. Vol. II: Vol II. US Environmental Protection Agency, Research Triangle Park, NC. Document EPA/600/P-99/002aF.
139. Vegni FE, Castelli B, Auxilia F, Wilkinson P (2005) Air pollution and respiratory drug use in the city of Como, Italy. Eur J Epidemiol 20: 351-358.
report no. 4/11
184
140. Vigotti MA, Rossi G, Bisanti L, Zanobetti A, Schwartz J (1996) Short term effects of urban air pollution on respiratory health in Milan, Italy, 1980-89. J Epidemiol Community Health 50 Suppl 1: s71-75.
141. Vineis P, Hoek G, Krzyzanowski M, Vigna-Taglianti F, Veglia F, Airoldi L, Autrup H, Dunning A, Garte S, Hainaut P, Malaveille C, Matullo G, Overvad K, Raaschou-Nielsen O, Clavel-Chapelon F, Linseisen J, Boeing H, Trichopoulou A, Palli D, Peluso M, Krogh V, Tumino R, Panico S, Bueno-De-Mesquita HB, Peeters PH, Lund EE, Gonzalez CA, Martinez C, Dorronsoro M, Barricarte A, Cirera L, Quiros JR, Berglund G, Forsberg B, Day NE, Key TJ, Saracci R, Kaaks R, Riboli E (2006) Air pollution and risk of lung cancer in a prospective study in Europe. Int J Cancer 119: 169-174.
142. Vineis P, Hoek G, Krzyzanowski M, Vigna-Taglianti F, Veglia F, Airoldi L, Overvad K, Raaschou-Nielsen O, Clavel-Chapelon F, Linseisen J, Boeing H, Trichopoulou A, Palli D, Krogh V, Tumino R, Panico S, Bueno-De-Mesquita HB, Peeters PH, Lund EE, Agudo A, Martinez C, Dorronsoro M, Barricarte A, Cirera L, Quiros JR, Berglund G, Manjer J, Forsberg B, Day NE, Key TJ, Kaaks R, Saracci R, Riboli E (2007) Lung cancers attributable to environmental tobacco smoke and air pollution in non-smokers in different European countries: a prospective study. Environ Health 6: 7.
143. Visser O, van Wijnen JH, van Leeuwen FE (2004) Residential traffic density and cancer incidence in Amsterdam, 1989-1997. Cancer Causes Control 15: 331-339.
144. Visser O, van Wijnen JH, van Leeuwen FE (2005) Incidence of cancer in the area around Amsterdam Airport Schiphol in 1988-2003: a population-based ecological study. BMC Public Health 5: 127.
145. von Klot S, Peters A, Aalto P, Bellander T, Berglind N, D'Ippoliti D, Elosua R, Hormann A, Kulmala M, Lanki T, Lowel H, Pekkanen J, Picciotto S, Sunyer J, Forastiere F (2005) Ambient air pollution is associated with increased risk of hospital cardiac readmissions of myocardial infarction survivors in five European cities. Circulation 112: 3073-3079.
146. Wheeler BW, Ben-Shlomo Y (2005) Environmental equity, air quality, socioeconomic status, and respiratory health: a linkage analysis of routine data from the Health Survey for England. J Epidemiol Community Health 59: 948-954.
147. Wichmann H-E, Spix C, Tuch T, Wolke G, Peters A, Heinrich J, Kreyling WG, Heyder J (2000) Daily mortality and fine and ultrafine particles in Erfurt, Germany. Part I: role of particle number and particle mass. Research Report no. 98. Health Effects Institute: Cambridge, MA.
148. Wichmann HE (2004) What can we learn today from the Central European smog episode of 1985 (and earlier episodes)? Int J Hyg Environ Health 207: 505-520.
149. Wolf C (2002) Urban air pollution and health: an ecological study of chronic rhinosinusitis in Cologne, Germany. Health Place 8: 129-139.
150. Wong CM, Atkinson RW, Anderson HR, Hedley AJ, Ma S, Chau PY, Lam TH (2002) A tale of two cities: effects of air pollution on hospital admissions in Hong Kong and London compared. Environ Health Perspect 110: 67-77.
151. Wordley J, Walters S, Ayres JG (1997) Short term variations in hospital admissions and mortality and particulate air pollution. Occup Environ Med 54: 108-116.
report no. 4/11
184
140. Vigotti MA, Rossi G, Bisanti L, Zanobetti A, Schwartz J (1996) Short term effects of urban air pollution on respiratory health in Milan, Italy, 1980-89. J Epidemiol Community Health 50 Suppl 1: s71-75.
141. Vineis P, Hoek G, Krzyzanowski M, Vigna-Taglianti F, Veglia F, Airoldi L, Autrup H, Dunning A, Garte S, Hainaut P, Malaveille C, Matullo G, Overvad K, Raaschou-Nielsen O, Clavel-Chapelon F, Linseisen J, Boeing H, Trichopoulou A, Palli D, Peluso M, Krogh V, Tumino R, Panico S, Bueno-De-Mesquita HB, Peeters PH, Lund EE, Gonzalez CA, Martinez C, Dorronsoro M, Barricarte A, Cirera L, Quiros JR, Berglund G, Forsberg B, Day NE, Key TJ, Saracci R, Kaaks R, Riboli E (2006) Air pollution and risk of lung cancer in a prospective study in Europe. Int J Cancer 119: 169-174.
142. Vineis P, Hoek G, Krzyzanowski M, Vigna-Taglianti F, Veglia F, Airoldi L, Overvad K, Raaschou-Nielsen O, Clavel-Chapelon F, Linseisen J, Boeing H, Trichopoulou A, Palli D, Krogh V, Tumino R, Panico S, Bueno-De-Mesquita HB, Peeters PH, Lund EE, Agudo A, Martinez C, Dorronsoro M, Barricarte A, Cirera L, Quiros JR, Berglund G, Manjer J, Forsberg B, Day NE, Key TJ, Kaaks R, Saracci R, Riboli E (2007) Lung cancers attributable to environmental tobacco smoke and air pollution in non-smokers in different European countries: a prospective study. Environ Health 6: 7.
143. Visser O, van Wijnen JH, van Leeuwen FE (2004) Residential traffic density and cancer incidence in Amsterdam, 1989-1997. Cancer Causes Control 15: 331-339.
144. Visser O, van Wijnen JH, van Leeuwen FE (2005) Incidence of cancer in the area around Amsterdam Airport Schiphol in 1988-2003: a population-based ecological study. BMC Public Health 5: 127.
145. von Klot S, Peters A, Aalto P, Bellander T, Berglind N, D'Ippoliti D, Elosua R, Hormann A, Kulmala M, Lanki T, Lowel H, Pekkanen J, Picciotto S, Sunyer J, Forastiere F (2005) Ambient air pollution is associated with increased risk of hospital cardiac readmissions of myocardial infarction survivors in five European cities. Circulation 112: 3073-3079.
146. Wheeler BW, Ben-Shlomo Y (2005) Environmental equity, air quality, socioeconomic status, and respiratory health: a linkage analysis of routine data from the Health Survey for England. J Epidemiol Community Health 59: 948-954.
147. Wichmann H-E, Spix C, Tuch T, Wolke G, Peters A, Heinrich J, Kreyling WG, Heyder J (2000) Daily mortality and fine and ultrafine particles in Erfurt, Germany. Part I: role of particle number and particle mass. Research Report no. 98. Health Effects Institute: Cambridge, MA.
148. Wichmann HE (2004) What can we learn today from the Central European smog episode of 1985 (and earlier episodes)? Int J Hyg Environ Health 207: 505-520.
149. Wolf C (2002) Urban air pollution and health: an ecological study of chronic rhinosinusitis in Cologne, Germany. Health Place 8: 129-139.
150. Wong CM, Atkinson RW, Anderson HR, Hedley AJ, Ma S, Chau PY, Lam TH (2002) A tale of two cities: effects of air pollution on hospital admissions in Hong Kong and London compared. Environ Health Perspect 110: 67-77.
151. Wordley J, Walters S, Ayres JG (1997) Short term variations in hospital admissions and mortality and particulate air pollution. Occup Environ Med 54: 108-116.
report no. 4/11
185
152. Zanobetti A, Schwartz J (2003) Multicity assessment of mortality displacement within the APHEA2 project. In Revised analyses of time-series studies of air pollution and health. Special report pp. 249-254. Health Effects Institute: Boston, MA.
153. Zanobetti A, Schwartz J, Dockery DW (2000) Airborne particles are a risk factor for hospital admissions for heart and lung disease. Environ Health Perspect 108: 1071-1077.
154. Zanobetti A, Schwartz J, Samoli E, Gryparis A, Touloumi G, Atkinson R, Le Tertre A, Bobros J, Celko M, Goren A, Forsberg B, Michelozzi P, Rabczenko D, Aranguez Ruiz E, Katsouyanni K (2002) The temporal pattern of mortality responses to air pollution: a multicity assessment of mortality displacement. Epidemiology 13: 87-93.
155. Zanobetti A, Schwartz J, Samoli E, Gryparis A, Touloumi G, Peacock J, Anderson RH, Le Tertre A, Bobros J, Celko M, Goren A, Forsberg B, Michelozzi P, Rabczenko D, Hoyos SP, Wichmann HE, Katsouyanni K (2003) The temporal pattern of respiratory and heart disease mortality in response to air pollution. Environ Health Perspect 111: 1188-1193.
156. Zeghnoun A, Beaudeau P, Carrat F, Delmas V, Boudhabhay O, Gayon F, Guincetre D, Czernichow P (1999) Air pollution and respiratory drug sales in the City of Le Havre, France, 1993-1996. Environ Res 81: 224-230.
157. Zeghnoun A, Czernichow P, Beaudeau P, Hautemaniere A, Froment L, Le Tertre A, Quenel P (2001) Short-term effects of air pollution on mortality in the cities of Rouen and Le Havre, France, 1990-1995. Arch Environ Health 56: 327-335.
158. Zemp E, Elsasser S, Schindler C, Kunzli N, Perruchoud AP, Domenighetti G, Medici T, Ackermann-Liebrich U, Leuenberger P, Monn C, Bolognini G, Bongard JP, Brandli O, Karrer W, Keller R, Schoni MH, Tschopp JM, Villiger B, Zellweger JP (1999) Long-term ambient air pollution and respiratory symptoms in adults (SAPALDIA study). The SAPALDIA Team. Am J Respir Crit Care Med 159: 1257-1266.
159. Zmirou D, Schwartz J, Saez M, Zanobetti A, Wojtyniak B, Touloumi G, Spix C, Ponce de Leon A, Le Moullec Y, Bacharova L, Schouten J, Ponka A, Katsouyanni K (1998) Time-series analysis of air pollution and cause-specific mortality. Epidemiology 9: 495-503.
report no. 4/11
186
APP
END
IX 1
.1 -
SHO
RT-
TER
M P
AR
TIC
ULA
TE M
ATT
ER (P
M) A
ND
MO
RTA
LITY
(DER
IVED
FR
OM
EPA
199
7).
Ref
eren
ce, L
ocat
ion,
Yea
rs, P
M In
dex,
M
ean
or M
edia
n (µ
g/m
3 ). St
udy
Des
crip
tion.
Mod
elin
g m
etho
ds: l
ags,
sm
ooth
ing,
co-
pollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
.
G
REA
T B
RIT
AIN
T
hurs
ton
et a
l., (
1989
) (T
hurs
ton
et a
l, 19
89)
B
S m
ean:
90.
1; 2
4-h
avg
. d
aily
max
: 709
. D
aily
tot
al m
orta
lity
ana
lyze
d fo
r as
soci
atio
ns w
ith B
S,
SO
2, a
nd H
2S
O4
in L
ond
on,
Eng
land
, du
ring
196
3 to
197
2 w
inte
rs. M
ean
dai
ly t
empe
ratu
re a
nd R
H a
lso
cons
ider
ed.
PM
, S
O2,
and
H2S
O4
all f
ound
to h
ave
sta
tistic
al
sign
ifica
nt a
ssoc
iatio
ns w
ith m
orta
lity
(0, 1
da
y la
g).
Tem
pera
ture
als
o co
rre
late
d (n
egat
ive
ly)
with
mor
talit
y,
but
with
2-d
ay
lag.
Se
ason
ality
add
ress
ed b
y st
udyi
ng
only
win
ters
and
by
appl
yin
g hi
gh-p
ass
filte
r to
the
se
ries
and
anal
yzin
g re
sidu
als.
Ito e
t al
., 1
993
(It
o et
al,
1993
) B
S m
ean:
90.
1; r
ang
e: 0
-350
. F
urth
er a
nal
ysis
of
Lond
on,
Eng
land
dat
a (
196
5 to
19
72)
ex
amin
ed b
y T
hurs
ton
et a
l. 19
89
(Thu
rsto
n et
al,
198
9).
Spe
ctra
l and
adv
ance
d tim
e se
ries
met
hods
use
d, e
.g.
pre
wh
iten
ing
and
aut
o-re
gres
sive
mov
ing
aver
age
m
etho
ds. V
aria
bles
con
sid
ered
incl
ude
d w
ere
BS
, SO
2,
H2S
O4,
tem
pera
ture
an
d R
H.
Est
imat
ed p
ollu
tion
mea
n e
ffec
t of
2 to
7%
of
all
Lon
don
win
ter
deat
hs (
mea
n =
281
/da
y),
but
vari
ous
pollu
tant
s’
effe
cts
not s
epar
ated
. In
depe
nden
t mod
el te
st o
n 1
962
epis
ode
con
firm
ed a
ppro
pria
ten
ess
of s
uch
met
hods
. Lo
ng-w
ave
add
ress
ed b
y co
nsid
erin
g w
inte
rs o
nly
and
b
y pr
ew
hite
nin
g th
e d
ata.
GER
MAN
Y
S
pix
et a
l. (1
993)
(S
pix
et a
l, 1
993)
T
SP
ran
ge: 1
0 to
650
.
Dai
ly t
ota
l mor
talit
y in
Erf
urt,
Eas
t Ger
man
y, d
urin
g 1
980
to 1
989
(m
edia
n =
6/d
ay)
rel
ated
to
SO
2, S
P, T
, RH
, and
pr
ecip
itatio
n. S
P o
nly
mea
sure
d in
198
8-1
989.
A
utor
egre
sssi
ve P
oiss
on m
ode
ls u
sed
(due
to
low
de
aths
/da
y) a
lso
incl
uded
indi
cato
r va
riabl
es f
or e
xtre
me
tem
pera
ture
an
d ad
just
men
ts f
or tr
end,
sea
son
, an
d in
flue
nza
ep
idem
ics.
Bot
h S
O2
and
SP
fou
nd to
be
sig
nific
antly
ass
ocia
ted
with
incr
ease
d m
orta
lity.
In
a si
mul
tane
ous
regr
essi
on,
S
P r
emai
ned
sig
nific
ant
wh
ile S
O2
did
not
. Cor
rela
tion
s of
thes
e co
effic
ient
s no
t pro
vide
d, h
ow
ever
. Pol
lutio
n ef
fect
siz
e si
mila
r to
that
for
met
eoro
log
y.
FR
AN
CE
Der
rien
nic
et a
l. (1
989
) (D
err
ienn
ic e
t al,
1989
) m
ean
TS
P (
OE
CD
Met
hod)
: L
yons
, Fra
nce:
87
(3 y
ear
) M
arse
illes
, Fra
nce:
126
(3
ye
ar).
Dai
ly t
ota
l, re
spira
tory
, an
d ca
rdia
c m
orta
lity
for
pers
ons
65 y
ears
of
age
test
ed
for
ass
ocia
tions
with
SO
2 an
d T
SP
du
ring
197
4 to
197
6 in
Lyo
ns a
nd M
arse
illes
, Fra
nce.
T
empe
ratu
re a
lso
cons
ider
ed in
ana
lyse
s.
No
sig
nific
ant m
orta
lity
asso
ciat
ions
fou
nd
with
TS
P,
but
S
O2
repo
rted
as
asso
ciat
ed w
ith to
tal e
lder
ly d
eath
s in
bo
th c
ities
. Se
ason
ality
ad
dre
ssed
by
anal
yzin
g de
viat
ions
fro
m 3
-ye
ar a
vera
ge o
f 31
-da
y ru
nni
ng
mea
ns o
f va
riabl
es, b
ut t
em
pera
ture
lags
not
con
side
red
and
prob
abl
e se
ason
al d
iffer
ence
s in
win
ter/
sum
mer
te
mpe
ratu
re-m
orta
lity
rela
tions
hip
not a
ddr
esse
d.
report no. 4/11
187
GR
EEC
E
K
atso
uya
nni e
t al
. 19
90a
(K
atso
uya
nni e
t al,
1990
a)
BS
.
Dai
ly t
ota
l mor
talit
y in
Ath
ens
, Gre
ece,
and
sur
rou
ndin
g bo
rou
ghs
(197
5 to
19
87)
rela
ted
to B
S,
SO
2, N
O2,
O3
, an
d C
O2
usin
g m
ultip
le r
egr
essi
on.
Dur
ing
win
ter
mon
ths
1983
to
1987
, th
e da
ily n
umbe
r of
de
aths
was
pos
itive
ly a
nd
stat
istic
ally
sig
nifi
cant
ly
asso
ciat
ed w
ith a
ll po
lluta
nts
, but
the
asso
ciat
ion
was
st
rong
est
with
BS
.
Kat
sou
yann
i et
al. 1
990
b (
Kat
sou
yann
i et a
l, 19
90b)
B
S a
nnua
l mea
n ra
nge:
51.
6 to
73.
3;
max
imum
dai
ly v
alue
: 79
0.
For
197
5 to
198
2 in
Ath
ens
, Gre
ece
199
da
ys w
ith h
igh
SO
2 (>
150
μg/m
3 ) ea
ch m
atch
ed o
n te
mp.
, ye
ar, s
easo
n,
day
of w
eek,
and
hol
ida
ys w
ith t
wo
low
SO
2 d
ays
. M
orta
lity
by-
caus
e co
mpa
red
bet
we
en g
roup
s b
y A
NO
VA
b
y ra
ndom
ize
d b
lock
s. B
S c
orre
late
d w
ith S
O2
at r
= 0
.73,
bu
t not
dire
ctly
use
d in
ana
lysi
s.
Mor
talit
y w
as g
ener
ally
hig
her
on h
igh
SO
2 d
ays
, w
ith
the
diff
eren
ce b
eing
mos
t pro
nou
nced
for
res
pira
tory
co
nditi
ons
. BS
leve
ls f
or e
ach
gro
up n
ot p
rovi
ded
, an
d B
S-S
O2
conf
oun
din
g n
ot a
ddr
esse
d, li
miti
ng
inte
rpre
tab
ility
of
resu
lts.
K
atso
uya
nni e
t al
. 19
93 (
Kat
sou
yan
ni e
t al,
1993
) B
S r
ang
e: 5
0 to
250
.
Dai
ly t
ota
l mor
talit
y in
Ath
ens
, Gre
ece,
dur
ing
July
, 19
87
(wh
en a
maj
or h
eat
wa
ve o
ccur
red)
com
pare
d to
dea
ths
in J
uly
for
pre
viou
s 6
yr.
Va
riabl
es c
onsi
dere
d in
clud
ed:
BS
, S
O2,
tem
pera
ture
, dis
com
fort
inde
x (D
I). E
ffec
ts o
f da
y-of
-wee
k, m
onth
, and
long
-ter
m tr
ends
add
ress
ed v
ia
dum
my
vari
able
s in
OL
S r
egr
essi
on m
odel
s.
Mea
n d
aily
tem
pera
ture
ab
ove
30
C f
ound
to
be
sign
ifica
ntly
ass
ocia
ted
with
mor
talit
y. T
he m
ain
effe
cts
of a
ll ai
r p
ollu
tant
s w
ere
non-
sign
ifica
nt, b
ut t
he
inte
ract
ion
betw
een
hig
h a
ir p
ollu
tion
and
tem
pera
ture
w
as s
ign
ifica
nt f
or S
O2
and
sugg
estiv
e (
p <
0.2
0) f
or O
3 an
d B
S.
T
oulo
umi e
t al.
(19
94)
(Tou
loum
i et a
l, 1
994)
B
S m
ean:
83;
ra
nge:
18
to 3
58.
Dai
ly t
ota
l mor
talit
y in
Ath
ens
, Gre
ece,
dur
ing
1984
-19
88
(mea
n =
38/
da
y) r
ela
ted
to B
S,
SO
2, C
O, T
, and
RH
. A
utor
egre
ssiv
e O
LS
mod
els
empl
oye
d al
so in
clud
ed
ind
icat
or v
aria
ble
s fo
r se
ason
, da
y of
wee
k, a
nd y
ear
.
BS
, S
O2,
and
CO
eac
h in
divi
dua
lly s
igni
fica
ntly
as
soci
ated
with
incr
ease
d m
orta
lity.
The
siz
e of
all
coef
ficie
nts
decl
ine
d in
sim
ulta
neo
us r
egre
ssio
ns,
with
S
O2
still
sig
nifi
cant
and
BS
app
roac
hin
g si
gni
fican
ce.
CO
was
no
long
er s
ign
ifica
nt, b
ut h
igh
ly c
orre
late
d w
ith
BS
(r
= 0
.74)
.
report no. 4/11
188
APP
END
IX 1
.2 -
SHO
RT-
TER
M P
AR
TIC
ULA
TE M
ATT
ER (P
M) A
ND
RES
PIR
ATO
RY
MO
RB
IDIT
Y.
Ref
eren
ce, L
ocat
ion,
Yea
rs, P
M In
dex,
M
ean
or M
edia
n (µ
g/m
3 ). St
udy
Des
crip
tion.
Mod
elin
g m
etho
ds: l
ags,
sm
ooth
ing,
co-
pollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
.
G
ERM
ANY
Sch
war
tz e
t al
., 1
991
(Sch
war
tz e
t al,
199
1)
Stu
dy
of a
cute
res
pira
tory
illn
ess
in c
hild
ren
in 5
Ger
man
com
mun
ities
, 19
83-1
985.
T
SP
med
ians
17
to 5
6.
Thi
s st
udy
exam
ined
the
impa
ct o
f sh
ort-
term
exp
osur
e to
ai
r po
llutio
n (T
SP
, S
O2
and
NO
2) o
n re
spira
tory
illn
ess
in
child
ren.
In
5 G
erm
an c
ities
dai
ly c
oun
ts o
f ch
ildre
n's
visi
ts
for
crou
p sy
mpt
oms
and
obst
ruct
ive
bron
chiti
s w
ere
reco
rded
. Dat
a w
ere
colle
cted
for
at l
east
2 y
ear
s in
ea
ch
loca
tion.
Au
tore
gres
sive
Po
isso
n re
gres
sio
n us
ing
GE
E
was
use
d. T
o fo
cus
the
ana
lysi
s on
sho
rt-t
erm
cor
rela
tions
an
d av
oid
sea
son
al c
onf
ound
ing,
bia
nnu
al, a
nnu
al
(sea
sona
l), a
nd s
ix s
hort
er t
erm
cyc
les
wer
e co
ntro
lled
for
in
the
reg
ress
ion
mod
els.
A to
tal o
f 6,
330
case
s of
cro
up a
nd 4
,755
cas
es o
f ob
stru
ctiv
e br
onch
itis
wer
e ob
serv
ed
dur
ing
the
stud
y.
Aft
er c
ontr
ollin
g fo
r sh
ort-
term
wea
ther
fac
tors
, TS
P
and
NO
2 w
ere
ass
ocia
ted
with
cro
up
case
s. A
n in
crea
se in
TS
P le
vels
fro
m 1
0 to
70
µg/
m3 w
as
asso
ciat
ed w
ith a
27%
incr
ease
in c
ases
of
crou
p; t
he
sam
e in
crea
se in
NO
2 le
vels
res
ulte
d in
a 2
8%
incr
ease
in c
ases
. No
pol
luta
nt w
as a
ssoc
iate
d w
ith
daily
cas
es o
f ob
stru
ctiv
e b
ronc
hitis
.
SP
AIN
S
unye
r et
al.
(19
93)
(Sun
yer e
t al,
1993
) A
dults
in B
arce
lona
, S
pai
n, 1
985-
198
9.
15 m
onito
ring
sta
tions
mea
surin
g B
S,
mea
n:
win
ter:
33%
tile
: 49
; 67
% t
ile =
77;
sum
mer
: 33
% t
ile =
36;
67%
tile
= 5
5.
In o
rder
to
exam
ine
the
asso
ciat
ion
betw
een
BS
and
SO
2 an
d th
e da
ily n
umbe
r of
em
erge
ncy
room
adm
issi
on in
B
arce
lona
, S
pain
, for
the
per
iod
198-
198
9, a
n au
tore
gres
sive
lin
ear
regr
ess
ion
anal
ysis
(0-
d la
g be
st)
afte
r ad
just
men
t for
tem
pera
ture
, da
y of
wee
k an
d ye
ar,
w
as u
sed.
An
incr
ease
of
25 µ
g/m
3 in S
O2
(24-
hou
r a
vera
ge)
prod
uce
d ad
just
ed
cha
nge
s of
6%
and
9%
in
emer
genc
y ro
om a
dmis
sion
s fo
r ch
roni
c ob
stru
ctiv
e
pulm
onar
y di
seas
e du
ring
win
ter
and
sum
mer
, re
spec
tivel
y. F
or B
S, a
sim
ilar
chan
ge w
as f
ound
du
ring
win
ter
(RR
for
50
µg
/m3 in
crea
se w
as 1
.15,
95
% C
I: 1
.09-
1.2
1), a
lthou
gh
the
cha
nge
was
sm
alle
r in
sum
mer
(R
R f
or 5
0 µ
g/m
3 incr
ease
was
1.0
5, 9
5%
CI:
0.98
-1.1
2). T
he a
ssoc
iatio
n o
f ea
ch p
ollu
tant
with
ch
roni
c o
bstr
uctiv
e p
ulm
onar
y d
isea
se a
dmis
sion
s re
mai
ned
sign
ifica
nt a
fter
con
trol
for
the
oth
er
pollu
tant
.
report no. 4/11
189
APP
END
IX 1
.3 -
LON
G-T
ERM
PA
RTI
CU
LATE
MA
TTER
(PM
) AN
D M
OR
TALI
TY.
Ref
eren
ce, L
ocat
ion,
Yea
rs, P
M In
dex,
M
ean
or M
edia
n (µ
g/m
3 ). St
udy
Des
crip
tion.
Mod
elin
g m
etho
ds: l
ags,
sm
ooth
ing,
co-
pollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
.
C
ZEC
H R
EPU
BLI
C
Bob
ak &
Leo
n, 1
992
(Bo
bak
andL
eon,
19
92)
Eco
logi
c st
ud
y, C
zech
Rep
ublic
, 19
86-1
988.
P
M1
0 m
ean:
68.
5.
An
eco
logi
cal s
tud
y of
neo
nat
al m
orta
lity
(age
s le
ss th
an
1 m
onth
) an
d po
st-n
eon
ata
l mor
talit
y (a
ges
1 t
o 1
2 m
onth
s)
and
air
pol
lutio
n w
as c
ondu
cted
in th
e C
zech
Rep
ubl
ic.
Dat
a on
infa
nt m
orta
lity
and
PM
10, S
O2
and
NO
x w
ere
colle
cted
in t
he p
erio
d 1
986
-88
for
46 o
f th
e 8
5 di
stric
ts in
th
e re
pub
lic. R
Rs
wer
e de
rive
d b
y lo
gist
ic r
egre
ssio
n m
odel
s ad
just
ed f
or d
istr
ict
soci
oeco
nom
ic c
hara
cter
istic
s,
such
as
inco
me,
car
ow
ners
hip,
and
abo
rtio
n ra
te.
A w
eak
pos
itive
ass
ocia
tion
bet
wee
n ne
ona
tal
mor
talit
y a
nd q
uint
ile o
f P
M10
and
SO
2 w
as f
ound
. S
tron
ger
adju
sted
eff
ects
wer
e se
en f
or p
ostn
eona
tal
mor
talit
y, w
ith a
con
sist
ent
incr
ease
in r
isk
for
incr
easi
ng q
uint
iles
of P
M10
(p <
0.0
01)
. Wea
ker
and
less
con
sist
ent e
vide
nce
of
a po
sitiv
e as
soci
atio
n w
ith
NO
x (p
= 0
.061
) w
as o
bser
ved
. The
str
onge
st e
ffec
ts
wer
e se
en
for
post
neon
ata
l res
pira
tory
mor
talit
y,
whi
ch in
crea
sed
cons
iste
ntly
with
incr
easi
ng q
uint
iles
of P
M10
(p
= 0
.013
). R
Rs
for
post
neon
ata
l res
pira
tory
m
orta
lity
for
the
high
est
vs t
he lo
wes
t qu
intil
e w
ere
2.41
(9
5% C
l: 1
.10-
5.2
8) f
or
PM
10,
3.9
1 (0
.90-
16.9
) fo
r S
O2,
and
1.2
0 (0
.37-
3.9
1) f
or N
Ox.
report no. 4/11
190
APP
END
IX 1
.4 -
LON
G-T
ERM
PA
RTI
CU
LATE
MA
TTER
(PM
) AN
D R
ESPI
RA
TOR
Y M
OR
BID
ITY.
Ref
eren
ce, L
ocat
ion,
Yea
rs, P
M In
dex,
M
ean
or M
edia
n (µ
g/m
3 ). St
udy
Des
crip
tion.
Mod
elin
g m
etho
ds: l
ags,
sm
ooth
ing,
co-
pollu
tant
s an
d co
varia
tes.
R
esul
ts a
nd C
omm
ents
.
G
REA
T B
RIT
AIN
Lu
nn e
t al
., 1
967
(L
unn
et a
l, 1
967)
S
heff
ield
, E
ngla
nd.
A s
tud
y a
naly
zed
the
asso
ciat
ion
betw
een
mea
sure
s of
P
M a
nd S
O2
and
res
pira
tory
illn
esse
s in
5-
and
6-y
ear
old
sc
hool
ch
ildre
n liv
ing
in f
our
area
s of
She
ffie
ld, E
ngla
nd.
Pos
itive
ass
ocia
tions
wer
e f
ound
be
twee
n a
ir po
llutio
n co
ncen
trat
ions
and
bot
h up
per
and
low
er r
esp
irato
ry
illne
ss. L
ow
er r
esp
irato
ry il
lnes
s w
as 3
3 to
56%
mor
e fr
eque
nt in
the
hig
her
pol
lutio
n ar
eas
than
in t
he lo
w-
pollu
tion
are
a (
p <
0.0
05).
Als
o, d
ecre
men
ts in
lung
fu
nctio
n, m
easu
red
by
spiro
met
ry te
sts,
wer
e cl
osel
y as
soci
ated
with
res
pira
tory
dis
ease
sym
ptom
rat
es.
Lu
nn e
t al
., 1
970
(L
unn
et a
l, 1
970)
E
ngla
nd,
196
3-19
64.
A
stu
dy
con
duct
ed
in E
ngla
nd b
etw
een
196
3 a
nd
196
4
ana
lyze
d th
e as
soci
atio
n b
etw
een
BS
and
SO
2 a
nd
resp
irato
ry il
lnes
ses
in 1
1-ye
ar-o
ld c
hild
ren
.
Res
ults
wer
e si
mila
r to
tho
se f
ound
for
the
yo
unge
r gr
oup.
On
the
basi
s of
the
resu
lts r
epor
ted,
it
appe
ared
tha
t in
cre
ased
fre
quen
cy o
f lo
wer
re
spira
tory
sym
ptom
s an
d de
crea
sed
lung
fun
ctio
n in
ch
ildre
n co
uld
occ
ur w
ith lo
ng-t
erm
exp
osur
es to
an
nual
BS
leve
ls in
the
ran
ge o
f 23
0 to
301
µg/
m3 a
nd
SO
2 le
vels
of
181
to 2
75 µ
g/m
3 . How
eve
r, it
was
no
ted
that
thes
e w
ere
only
ver
y ap
prox
imat
e ob
serv
ed-
effe
ct le
vels
bec
aus
e of
unc
erta
intie
s as
soci
ated
with
est
imat
ing
PM
mas
s ba
sed
on B
S
read
ings
.
SW
ITZE
RLA
ND
A
cker
man
n-Li
ebric
h et
al.
(199
6)
(Ack
erm
ann-
Lieb
rich
et a
l, 19
97)
eig
ht d
iffer
ent a
reas
fro
m S
witz
erla
nd.
The
aim
of
this
stu
dy
was
to
exam
ine
the
effe
cts
of lo
ng
term
exp
osur
e to
air
pollu
tant
s on
lun
g fu
nctio
n in
adu
lts,
taki
ng a
dva
ntag
e fr
om a
sam
ple
of 9
,651
su
bjec
ts a
ged
18 to
60
from
eig
ht d
iffer
ent a
reas
of
Sw
itzer
land
. F
VC
an
d F
EV
1 w
ere
regr
esse
d a
gain
st th
e n
atur
al lo
gari
thm
s of
hei
ght
, w
eig
ht,
age,
ag
e sq
uare
d, g
end
er,
educ
atio
nal
le
vel,
natio
nalit
y, a
nd w
ork
plac
e ex
posu
re.
The
res
ults
sug
gest
ed t
hat
a 10
µg/
m3 in
crea
se in
an
nual
ave
rag
e P
M10
was
ass
ocia
ted
with
a 3
.4
perc
ent d
ecre
ase
in F
VC
for
hea
lthy
neve
r sm
oker
s.
Res
ults
wer
e a
lso
cons
iste
nt a
nd
sig
nific
ant
for
NO
2 an
d S
O2,
but
less
for
O3.
report no. 4/11
191
IT
ALY
Aro
ssa
et a
l. (1
987)
(A
ross
a et
al,
198
7)
Tur
in, I
taly
, 19
80-
1983
.
In th
is s
tud
y, lu
ng
func
tion
of 1
,880
sch
ool
ch
ildre
n w
as
mea
sure
d in
Tur
in, I
taly
, du
ring
a ti
me
peri
od w
hen
both
T
SP
and
SO
2 w
ere
bei
ng
redu
ced
. Thr
ee a
reas
of
Tur
in
(cen
tral
city
, pe
riph
era
l are
a, a
nd
sub
urba
n a
rea)
wer
e st
udie
d d
urin
g th
e w
inte
rs o
f 19
80 to
19
81 a
nd 1
982
to
1983
. Eac
h ch
ild's
res
pira
tory
he
alth
was
ass
esse
d at
the
be
gin
nin
g a
nd
end
of
the
stud
y us
ing
a qu
estio
nna
ire
whi
ch a
lso
obta
ined
dem
ogra
phic
info
rmat
ion.
Lu
ng
func
tion
mea
sure
men
ts in
clud
ed F
VC
, F
EV
1, F
EF
25-7
5,
and
ME
F50
. Da
ily S
O2
and
TS
P m
easu
rem
ents
wer
e av
aila
ble
fro
m s
even
mon
itorin
g si
tes
in t
he a
rea.
A
GLM
an
alys
is a
djus
ted
for
sex
and
ant
hro
pom
etric
va
riab
les
was
use
d to
est
imat
e lu
ng f
unct
ion
valu
es.
The
pol
lutio
n da
ta c
onfir
med
tha
t th
e la
rge
SO
2 di
ffer
ence
s ac
ross
are
as in
198
0 to
198
1 w
ere
re
duce
d su
bsta
ntia
lly b
y 19
82 to
19
83. T
he
diff
eren
ces
in T
SP
rem
aine
d sm
all b
ut c
ons
tant
du
ring
the
time
peri
od.
Ave
rage
slo
pes
wer
e si
gnifi
can
tly h
igh
er w
ithin
the
city
of
Tur
in w
hen
com
pare
d w
ith t
he s
ubur
ban
area
, su
gges
ting
to th
e
auth
ors
that
a d
ecre
ase
in p
ollu
tion
(pr
imar
ily S
O2)
re
sulte
d in
an
impr
ovem
ent o
f lu
ng f
unct
ion
.
report no. 4/11
192
APP
END
IX 1
.5 -
LON
G-T
ERM
PA
RTI
CU
LATE
MA
TTER
(PM
) AN
D L
UN
G C
AN
CER
.
Ref
eren
ce, L
ocat
ion,
Yea
rs, P
M In
dex,
Mea
n or
Med
ian
(µg/
m3 ).
Stud
y D
escr
iptio
n. M
odel
ing
met
hods
: lag
s,
smoo
thin
g, c
o-po
lluta
nts
and
cova
riate
s.
Res
ults
and
Com
men
ts.
K
atso
uya
nni e
t al
., 1
991
(Kat
sou
yan
ni e
t al,
199
1).
A h
osp
ital-b
ase
d ca
se-c
ontr
ol s
tud
y w
as
unde
rtak
en in
Ath
ens
bet
wee
n 19
87
and
198
9
on 1
89 w
omen
with
lun
g ca
ncer
and
89
fem
ale
cont
rols
with
fra
ctur
es o
r ot
her
orth
ope
dic
cond
itio
ns. L
ifetim
e ex
posu
re to
air
pollu
tion
was
ass
esse
d b
y lin
king
blin
dly
lifel
ong
re
side
ntia
l and
em
plo
ymen
t ad
dres
ses
of a
ll su
bjec
ts w
ith o
bjec
tive
ly e
stim
ated
or
pres
umed
ai
r po
llutio
n le
vels
. RR
s w
ere
der
ive
d b
y m
ultip
le lo
gist
ic r
egr
essi
on m
odel
s ad
just
ed f
or
age.
Air
pol
lutio
n le
vels
wer
e a
ssoc
iate
d w
ith
incr
ease
d r
isk
for
lung
ca
ncer
but
the
age
-ad
just
ed R
R w
as s
mal
l an
d n
ot s
tatis
tica
lly
sign
ifica
nt.
Whe
reas
ther
e w
as n
o ef
fect
of
air
pollu
tion
amon
g no
n-sm
oker
s, a
n in
tera
ctio
n w
ith s
mok
ing
dura
tion
has
bee
n sh
ow
n. T
he in
tera
ctio
n w
as a
lmos
t ex
clus
ivel
y ac
cou
nte
d fo
r b
y th
e n
on-
aden
ocar
cino
ma
lung
can
cers
.
report no. 4/11
193
APPENDIX 1.6 - REFERENCES TO APPENDICES
1. Ackermann-Liebrich U, Leuenberger P, Schwartz J, Schindler C, Monn C, Bolognini G, Bongard JP, Brandli O, Domenighetti G, Elsasser S, Grize L, Karrer W, Keller R, Keller-Wossidlo H, Kunzli N, Martin BW, Medici TC, Perruchoud AP, Schoni MH, Tschopp JM, Villiger B, Wuthrich B, Zellweger JP, Zemp E (1997) Lung function and long term exposure to air pollutants in Switzerland. Study on Air Pollution and Lung Diseases in Adults (SAPALDIA) Team. Am J Respir Crit Care Med 155: 122-129.
2. Arossa W, Spinaci S, Bugiani M, Natale P, Bucca C, de Candussio G (1987) Changes in lung function of children after an air pollution decrease. Arch Environ Health 42: 170-174.
3. Bobak M, Leon DA (1992) Air pollution and infant mortality in the Czech Republic, 1986-88. Lancet 340: 1010-1014.
4. Derriennic F, Richardson S, Mollie A, Lellouch J (1989) Short-term effects of sulphur dioxide pollution on mortality in two French cities. Int J Epidemiol 18: 186-197.
5. Ito K, Thurston GD, Hayes C, Lippmann M (1993) Associations of London, England, daily mortality with particulate matter, sulfur dioxide, and acidic aerosol pollution. Arch Environ Health 48: 213-220.
6. Katsouyanni K, Hatzakis A, Kalandidi A, Trichopoulos D (1990a) Short-term effects of atmospheric pollution on mortality in Athens. Arch Hellen Med 7: 126-132.
7. Katsouyanni K, Karakatsani A, Messari I, Touloumi G, Hatzakis A, Kalandidi A, Trichopoulos D (1990b) Air pollution and cause specific mortality in Athens. J Epidemiol Community Health 44: 321-324.
8. Katsouyanni K, Pantazopoulou A, Touloumi G, Tselepidaki I, Moustris K, Asimakopoulos D, Poulopoulou G, Trichopoulos D (1993) Evidence for interaction between air pollution and high temperature in the causation of excess mortality. Arch Environ Health 48: 235-242.
9. Katsouyanni K, Trichopoulos D, Kalandidi A, Tomos P, Riboli E (1991) A case-control study of air pollution and tobacco smoking in lung cancer among women in Athens. Prev Med 20: 271-278.
10. Lunn JE, Knowelden J, Handyside AJ (1967) Patterns of respiratory illness in Sheffield infant schoolchildren. Br J Prev Soc Med 21: 7-16.
11. Lunn JE, Knowelden J, Roe JW (1970) Patterns of respiratory illness in Sheffield junior schoolchildren. A follow-up study. Br J Prev Soc Med 24: 223-228.
12. Schwartz J, Spix C, Wichmann HE, Malin E (1991) Air pollution and acute respiratory illness in five German communities. Environ Res 56: 1-14.
13. Spix C, Heinrich J, Dockery D, Schwartz J, Volksch G, Schwinkowski K, Collen C, Wichmann HE (1993) Air pollution and daily mortality in Erfurt, east Germany, 1980-1989. Environ Health Perspect 101: 518-526.
report no. 4/11
194
14. Sunyer J, Saez M, Murillo C, Castellsague J, Martinez F, Anto JM (1993) Air pollution and emergency room admissions for chronic obstructive pulmonary disease: a 5-year study. Am J Epidemiol 137: 701-705.
15. Thurston GD, Ito K, Lippmann M, Hayes C (1989) Reexamination of London, England, mortality in relation to exposure to acidic aerosols during 1963-1972 winters. Environ Health Perspect 79: 73-82.
16. Touloumi G, Pocock SJ, Katsouyanni K, Trichopoulos D (1994) Short-term effects of air pollution on daily mortality in Athens: a time-series analysis. Int J Epidemiol 23: 957-967.