What this study adds
While the International Agency for Research on Cancer has classified air pollution as a carcinogen, the studies conducted to date have largely focused on lung cancer. Over the past few years, a small number of papers have emerged linking ambient air pollution to an increased risk of female breast cancer. This study adds to this literature, and our design has notable strengths that include a prospective design, relatively large sample size, ability to control for other risk factors for breast cancer, and the use of remote sensing to assign air pollution concentrations. We found that exposure to air pollution was associated with an increased risk of premenopausal but not postmenopausal breast cancer. Also, our analyses suggest that future studies of air pollution and breast cancer may need to account for screening behaviors.
Exposure to fine particulate matter (PM2.5) is recognized as a leading cause of mortality. The recently published Global Burden of Disease report estimated that in 2010, annual PM2.5 accounted for 3.1 million deaths and around 3.1% of global disability-adjusted life years.1 Although chronic exposure to air pollution is associated with increased rates of mortality and morbidity for cardiovascular and respiratory disease,2 in 2013, the International Agency for Research on Cancer classified air pollution as a human carcinogen.3 The decision was based primarily on the findings of several epidemiological studies that found positive associations with lung cancer.
There have been far fewer studies of associations between past exposures to PM2.5 and other sites of cancer. In Canada, recent work has suggested that female breast cancer and prostate cancer may be associated with traffic-related air pollution.4–8 As the risk factors of female breast cancer vary by menopausal status,9 epidemiological studies of air pollution should investigate their possible modifying effects. Hystad et al.6 observed stronger associations between exposure to ambient to PM2.5 air pollution and incident breast cancer among Canadian premenopausal women.7 These findings of stronger associations in younger women are consistent with a recent meta-analysis of the associations between environmental tobacco smoke and breast cancer, where associations were noted in premenopasual but not in postmenopausal women.10 Despite this, increased air pollution risks of postmenopausal breast cancer have been observed among Montreal women7 and more recently among women enrolled in 15 European cohorts within the European study of cohorts for air pollution effects (ESCAPE) project.11 Analyses of the Danish Nurse Cohort,12 as well as the US Nurses’ Health Study,13 found no association between particulate matter air pollution and breast cancer. Similarly, longitudinal analyses of the Sister Study cohort found no association between PM2.5 air pollution and breast cancer; however, an increased risk was observed between NO2 and estrogen receptor (ER)(+)/progesterone receptor (PR)(+) breast cancer.14 Positive associations between particulate matter air pollution and breast cancer have also been observed for other size fractions such as PM10, PM2.5–<10, and ultrafine particles (UFPs).7,11 The excess risks found in some studies of breast cancer and air pollution are interesting in light of the finding of associations with occupationally generated carbon monoxide (occurring as combustion products) and polycyclic aromatic hydrocarbons from combustion of petroleum liquids, especially among cases with ER+/PR+ status, found in a case–control study of postmenopausal breast cancer in Montreal.15
The purpose of this study was to determine whether there was an association between past exposures to ambient PM2.5 and the incidence of breast cancer in a cohort of approximately 90,000 Canadian women with a 25-year follow-up interval. We have found previously in the same cohort associations between PM2.5 and mortality.16
Material and methods
Our study population included participants of the Canadian National Breast Screening Study, which was a randomized controlled trial of screening for breast cancer.17–19 Briefly, the study included 89,835 women who were recruited from the general population between 1980 and 1985 and were between the ages of 40 and 59 years. Those who were between the ages of 40 and 49 years were randomized to receive either annual mammography screening and a physical examination of the breasts or into the control group where they received a single physical examination of the breasts and annual follow-up through a mailed, self-administered questionnaire. Those in the control group were taught breast self-examination and remained under the care of their family physician. At the time of the study, mammography was not available for screening; however, it could be used for diagnosis if women developed symptoms. Among those between the ages of 50 and 59, for ethical reasons, it was determined that some screening for breast cancer should be provided, so these women were randomized into a group that received annual mammography and breast examination or into a control group that received annual breast examination alone. Women aged 50–59 years were also taught breast self-examination. A detailed breakdown of the numbers of women, by age, who were randomized into mammography and control groups has been published.19 The screening centers were located in teaching hospitals or in cancer centers across the country, and the study center was at the University of Toronto. Women were healthy and free of cancer at the time of enrollment.
A self-administered questionnaire was used to elicit information on their demographic, reproductive, and lifestyle characteristics. Women’s height, weight, and skinfold thickness were measured by a trained health professional, and body mass index (BMI) was calculated using these weight and height measurements. We modeled BMI as a covariate using the cut points developed for the following categories: <20, 20 to 24, 25 to 29, and ≥30 kg/m2. Residential address information was obtained at baseline, and six-character, Canadian postal codes were collected for all participants. In urban areas, six-character postal codes represent one block face between two intersecting streets.
The questionnaire comprised information on accepted risk factors. Women were asked whether they still had menstrual periods and, if so, to report the date of their last period and if their periods were regular. Women were asked to provide information on past pregnancies, including numbers of stillbirths, miscarriages, live births, and age when their first live child was born. Participants were asked whether any of their female relatives had breast cancer and to indicate their relationship to them. They were also asked about their history of oral contraceptive and estrogen use, as well as whether they had had any x-ray examination of their breasts. Women were also asked to describe their occupation in as much detail. From these details, occupations were classified into broadly based categories (homemaker, clerical, medicine and health, teaching, management of administration, sales, service, arts, retired, social sciences, teaching, unemployed, or other). We did not have information on the consumption of alcohol.
At the time of screening, women indicated in the interview whether they were postmenopausal, regardless of age. This information was not available in a time-dependent fashion throughout the follow-up period, and we thus used attained age as an indicator for menopausal status. Recent data published from the Canadian Longitudinal Study of Aging indicated that the median age of menopause among Canadian women was 51 years of age.20 Therefore, we chose to conduct separate analyses using ages 50, 52, and 54 years as plausible cutoffs for being menopausal. Thus, if a woman declared she was menopausal at the time of entry into the study, she remained in that category throughout follow-up. If she was not menopausal at baseline, she would accumulate person-time in the premenopausal category until she attained one of these cutoffs. For incident breast cancers, we did not have any information on receptor status or other markers.
Ascertainment of vital status and cancer incidence
We ascertained vital status using a probabilistic record linkage to the Canadian Mortality Database21 until the end of 2005. This database provides data on all deaths occurring in Canada, as well as Canadian deaths that occur in approximately 20 US states. The accuracy of identifying deaths has been estimated to be approximately 98%.22 Until 1999, the underlying cause of death was coded using the ninth revision of the International Classification of Diseases; from 2000 onward, the 10th revision was used.
Incident cases of cancer were identified through record linkage to the Canadian Cancer Registry23 between 1968 and 2005. A small number of women were found to have had a previous history of breast cancer from the record linkage and therefore were excluded from further analyses. The date of diagnosis associated with the first diagnosis of breast cancer was used to censor participants; secondary diagnoses of breast cancer were not considered in the analyses.
To assign exposure to PM2.5., we used satellite-based estimates of surface concentrations of PM2.5.24 These satellite-based concentrations were derived from aerosol optical depth data from the Moderate Resolution Imaging Spectroradiometer (MODIS), Multiangle Imaging Spectroradiometer (MISR), and Sea-viewing Wide Field-of-view Sensor (SeaWiFS) instruments. Using satellite data collected from January 1, 1998, to December 31, 2006, long-term average concentrations of PM2.5 were derived at a resolution of approximately 10 × 10 km. These estimates combined the values from models developed in 2010,25 with optimal estimation-based values developed in 201326 to produce an improved representation of PM2.5 with extended temporal range and greater accuracy than either of the earlier models. We used these multiyear, longer-term mean concentrations to overcome limitations and biases associated with persistent cloud cover or presence of forest fires or mineral dust, which prevent accurate sampling over shorter periods of time. These measures cover nearly all of North America except for some coastal and mountain areas where retrieval of images is not possible. These satellite-based average concentrations of PM2.5 have been shown to correlate well with ground-based measurements at fixed-site stations across North America (Pearson correlation coefficient r = 0.76, slope = 0.96, n = 974).24 Similar satellite-based exposure surfaces have been applied previously to examine the associations between long-term exposure to air pollution and mortality.27–29 Residential measures of exposure were determined by mapping the geospatial coordinates of the centroid of each postal code to the PM2.5 surface.
We excluded women who were diagnosed with an incident breast cancer before they enrolled in the study. This was done on the basis of information provided on the questionnaire, as well as those identified through the record linkage. We censored women at the earliest date of first diagnosis of breast cancer, death, or December 31, 2005. We estimated associations between residential concentrations of PM2.5 and incident breast cancer using the Cox proportional hazards model, with age as the time axis. The use of age as the time axis has the advantage that unlike using calendar time as the axis and adjusting parametrically for age, it does not assume a known functional relationship between age and disease.30 Furthermore, given our interest in generating hazard ratios for time-dependent measures of menopause, this approach facilitated generating these estimates. A continuous measure of PM2.5 was included in this model, and we verified the linear assumption using penalized splines on four degrees of freedom. As all analyses for PM2.5 were consistent with linearity (see results), we computed adjusted hazard ratios and their 95% confidence limits in relation to an increase of 10 μg/m3.
We fitted models that successively included covariates from previous models. First, we fitted a model that contained PM2.5 only (model I) and then another model that also contained individual-level factors for occupation, marital status, and attained education (model II). A third model included the covariates from the previous models, as well as body mass index and cigarette pack-years. Model IV included additional covariables representing reproductive factors (ever pregnant, oral contractive use, hormone replacement therapy, age at menarche), breast self-examination, and family history of breast cancer. Model V included the terms in Model IV, as well as contextual measures of neighborhood characteristics obtained from the 1991 Canadian census data, namely median household income, proportion of individuals with high-school education, percentage of low-income households, and unemployment rate. We followed this approach as it facilitated comparisons for previous risk estimates for all-cause mortality generated in this cohort.16
We conducted additional analyses to determine whether the associations between PM2.5 and incident breast cancer were modified by place of birth (in Canada or elsewhere), whether participants had moved after the annual screens had been completed (between 3 and 5 years after baseline interview), obesity (BMI > 30 kg/m2), residential mobility, and whether they had smoked cigarettes (tobacco). We inspected the estimates associated with these first-order interaction terms, as well as P values from likelihood ratio tests. Finally, we examined associations between air pollution and breast cancer separately for those who were randomized into: (i) the screened and (ii) non-screened arms of the study. This was done to evaluate whether earlier detection of breast cancers may impact the air pollution risk estimates.
There were a total of 89,835 women in the Canadian National Breast Screening Study. We excluded 15 women who, based on the probabilistic record linkage, were diagnosed with an incident breast cancer before entry into the study. Residential measures of PM2.5 were assigned to 99.4% of subjects (n = 89,262). Estimates could not be assigned to 573 women due to errors in postal codes or because the remote sensing was unable to provide estimates of air pollution in some areas of Canada. Thus, there were a total of 89,247 women in our analyses. In total, 9,419 women died during the 25-year follow-up and a total of 6503 women were diagnosed with breast cancer. To facilitate comparisons between models, analyses were restricted to women who were not missing data for any of the key covariates. After applying these exclusions, hazard ratios were generated based on 6,427 incident breast cancers, and using an attained age of menopause of 52 years, 646 of these cases were premenopausal, while the remaining 5781 were postmenopausal.
Figures that depict participants’ places of residence at the time of enrollment and estimates of ambient PM2.5 determined from the satellite observations over the period 1998 to 2006 have been published previously.16 The median exposure to ambient PM2.5 that was assigned to the residences at time of entry into the study using these exposure estimates was 9.1 μg/m3 (SD = 3.4 μg/m3), while the 25th and 75th percentiles were 6.4 and 12.4 μg/m3, respectively. Mean concentrations of PM2.5 at baseline were identical between the women who were randomized between the mammography screening and control groups (9.5 μg/m3).
Table 1 shows the distribution of selected sociodemographic characteristics of the participants, the number of incident cases of breast cancer, and mean concentrations of PM2.5. Women were initially screened between ages 40 and 59 years, with about equal numbers in each 5-year age group, except for 55 to 60 year (18.4%). Most women were born in Canada (82.2%), most were married (79.8%), about 26% had less than a high-school education, and about 55% had college or higher education. There was little variation of concentrations of PM2.5 by age group or educational status. Women born outside of North America had higher concentrations as compared to those born in Canada or the United States, and women who were married or widowed had lower concentrations as compared to those in other categories.
Table 2 shows similar information for accepted or suspected risk factors for breast cancer. About 87% of the population had a body mass index under 30, 30% had a family history of breast cancer, 59.3% started their periods after age 13 years, 27% took hormonal replacement therapy, 60% took oral contraceptive therapy, 88% were pregnant, and 49% smoked cigarettes. There was little variation in exposures to PM2.5 by categories of each of these risk factors.
Table 3 shows adjusted hazards ratios for each of the five models by menopausal status and for the entire cohort. There was no association between PM2.5 and incident breast cancer among postmenopausal women as shown in the fully adjusted model (hazard ratio [HR] was 1.00; 95% confidence interval [CI] = 0.92, 1.09). Among premenopausal women (using 52 years of age as the cutoff), there was evidence of some confounding by the contextual covariates, and the HR was 1.37 (95% CI = 1.09, 1.73) when adjusted for all personal variables (model IV) but decreased to 1.27 (95% CI = 0.99, 1.63) when the contextual-wide variables were added (model V). The HR for all women combined was 1.02 (95% CI = 0.95, 1.10; model IV).
Table 4 shows little variation in the hazard ratios according to the three definitions for premenopausal status; among premenopausal women, the HRs varied between 1.27 and 1.37.
Findings from various stratified analyses are presented in Table 5. We found no statistical evidence of effect modification according to any of the other characteristics examined.
Figure 1 shows the fully adjusted (model V) response pattern of ambient concentrations of PM2.5 for the entire cohort. Hazard ratios (HR) in this analysis and other analyses (data not shown) were consistent with linearity. The adjusted hazard ratios in premenopausal and postmenopausal women, based on whether they were assigned to the screened or control arm of the study, are presented in Figure 2. The adjusted hazard ratio for a 10 μg/m3 increase in PM2.5 for premenopausal breast cancer in the screened arm was 1.54 (95% CI =1.10, 2.15), and in the nonscreened arm, it was 0.99 (95% CI = 0.69, 1.42).
In this cohort study of approximately 88,000 Canadian women, we found excess risks of breast cancer in premenopausal women that were positively associated with satellite measurements of PM2.5 at a resolution of approximately 10 × 10 km, at their residences at time of enrollment. The magnitude of this association was approximately a 26% increased risk per 10 μg/m3 increase of PM2.5. This increase was only evident among those women who were randomized into the screening arm of the study. We found no evidence of associations among postmenopausal women.
In a number case–control studies, increased risks of breast cancer have been reported among those with higher exposures to traffic-related air pollution.4,6,7 In the Sister Study,14 associations were found for PM2.5 and nitrogen dioxide (NO2), a marker for traffic-related air pollution, among cases with positive estrogen receptor and positive progesterone receptor (ER+/PR+) status. No association was found among cases with negative estrogen receptor and negative progesterone receptor status ER(-)/PR(-). Recently, we showed in another population-based case–control study conducted in Montreal no evidence of associations for UFPs across all subjects but found slightly higher associations among cases with positive estrogen and progesterone receptor status.7
Comparison of the present results with findings of other studies is complicated by different exposure metrics used to represent the complex mixture that is air pollution. In our study, we estimated national, spatially resolved concentrations of PM2.5, but in other studies, other pollutants, notably NO2 and UFPs, were measured and at other resolutions. NO2 measured at local scales, such as at street level (e.g., Crouse et al.4,31), is an accepted marker for traffic-related air pollution,32,33 whereas the sources for fine particulates include combustion of fossil fuels, other industrial processes, resuspended dusts, soils, crustal materials from farming, wear of roads and tires, pollen, and also mold spores, and so forth. UFPs, such as used in a case-control study in Montreal,7,34 are secondary particles that form through other processes, and while related to combustion sources, notably mobile diesel sources, appear to have a spatial distributions that is different from NO2, at least in Montreal, and vary considerably.
Another complication is that pre- and postmenopausal breast cancer likely reflect different diseases with differing, albeit overlapping, etiologies. In some studies, no distinction is made between these types of breast cancer. In addition, there may be subtypes of breast cancer, perhaps characterized by morphology but also receptor status in which risks may differ. For example, in a population-based case–control study conducted in Montreal between 2008 and 2011, we did not find evidence of associations for ultrafine particles (UFPs) across all subjects but found slightly higher associations among cases with positive estrogen and progesterone receptor status.7
The positive associations we observed between ambient PM2.5 and premenopausal breast cancer were only apparent among those women who were randomly assigned to the screening group. We were surprised by this result, given the absence of a positive association with post-menopausal breast cancer and given that women’s self-reported indication of breast self-examination did not substantially alter the risk estimates. These findings should be interpreted cautiously as the study had limited statistical power to evaluate effect modification due to the relatively small number of premenopausal breast cancers. In particular, the P value associated with the first-order interaction term to test for effect modification was 0.08. Our finding of a stronger association between air pollution and premenopausal breast cancer among women with mammography screening differs from the finding of Hystad et al.6 who found attenuated risk estimates. More work is needed to better understand the role of screening on associations between air pollution and breast cancer.
Although we adjusted for most accepted and suspected risk factors, residual confounding effects may have occurred because not all risk factors were measured (e.g., alcohol consumption, X- or gamma-radiation, dieldrin, digoxin, ethylene oxide, polychlorinated biphenyls, or shift work).35 From the ones that we did measure, there were no notable differences in concentrations of PM2.5, thus suggesting that any residual confounding would likely be minor.
This study has a number of other strengths. First, we had a large number of cases of incident breast cancer, thereby providing sufficient statistical power to estimate response patterns. Second, the use of remote sensing–derived estimates of PM2.5 allows us to assign exposure to virtually the entire cohort. Previous methods that relied on fixed-site measures of traffic-related pollution are limited in that they are restricted to urban areas where monitoring systems are in place. Another important strength of our study was the availability of individual-level risk factors that allowed us to adjust for smoking behaviors and obesity. Third, we were able to adjust for contextual effects through the use of small-area data obtained from the Canadian census.
A limitation of the study is that because we did not have direct contact with participants after the first six years of following, we could not unambiguously assign menopausal status using other information, such as the WHO criteria for menopausal status that accounts for hormone replacement therapy, hysterectomy, and bilateral oophorectomy).36,37 As the hazard ratios did not vary dramatically according to three different age cutoffs to classify women as postmenopausal, this may not be an important limitation.
Another limitation of our study is the inevitability of some exposure misclassification introduced by relying on assigning exposure to the place of residence at time of entry. Given the length of the follow-up interval, many subjects likely would have moved. Under a classical error model assumption, this would introduce nondifferential exposure measurement error that would serve to understate any true association. Our sensitivity analysis that examined risk differences between those who moved within the first 3 to 5 years of follow-up and those who did not found they were similar. Approximately 19% of individuals had moved during this time frame. This suggests that residential mobility is unlikely to greatly influence the risk estimates.
In addition to misclassification of exposure due to residential mobility, errors in measurement may have also been introduced from the assignment of PM2.5 concentrations using remote sensing data collected from 1998 onwards. As the diagnosis of premenopausal breast cancer would have generally occurred at an earlier stage of the follow-up period than for postmenopausal breast cancer, this potential exposure misclassification would have a greater impact on air pollution risks for premenopausal women. Assuming this exposure measurement error to be nondifferential between those who develop and do not develop breast cancer, this suggest that our risk estimate among those who were diagnosed with premenopausal breast cancer are attenuated to a greater degree than those diagnosed in postmenopausal women. Given the lack of PM2.5 monitoring in Canada before 1990, it is difficult to evaluate the extent of these biases as it is not possible to describe the spatial and temporal changes in PM2.5 concentration during the first 10 years of follow-up of this cohort. However, recent modeling suggests that overall Canadian concentrations of PM2.5 have changed very little since 1990. Specifically, Brauer et al.38 estimated that the population-weighted estimates of PM2.5 in Canada in 1990, 2000, and 2010 were 11.25, 10.47, and 11.89 μg/m3, respectively.
In summary, our findings suggest that ambient exposure to PM2.5 increases the risk of developing premenopausal breast cancer. Additionally, it suggests that low levels of air pollution, observed in Canadian studies, are relevant from a population health perspective.
Conflicts of interest statement
The authors have no conflicts of interest to declare.
Sources of funding
Funding for this study was provided by Health Canada’s Clean Air Regulatory Agenda program.
Description of process to obtain data: Due to ethics requirements, individual-level data from this study cannot be distributed.
1. Lim SS, Vos T, Flaxman AD, et al. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010.Lancet2012380985922242260
2. Chen H, Goldberg MS, Villeneuve PJ. A systematic review of the relation between long-term exposure to ambient air pollution and chronic diseases.RevEnvironHealth2008234243297
3. Loomis D, Grosse Y, Lauby-Secretan B, et al. The carcinogenicity of outdoor air pollution.Lancet Oncol2013141312621263
4. Crouse DL, Goldberg MS, Ross NA, et al. Postmenopausal breast cancer
is associated with exposure to traffic-related air pollution in Montreal, Canada
: a Case-Control Study.Environ Health Perspect201011815781583
5. Parent ME, Goldberg MS, Crouse DL, et al. Traffic-related air pollution and prostate cancer risk: a case-control study in Montreal, Canada
.Occup Environ Med2013707511518
6. Hystad P, Villeneuve PJ, Goldberg MS, et al. Exposure to traffic-related air pollution and the risk of developing breast cancer
among women in eight Canadian provinces: a case-control study.Environ Int201574240248
7. Goldberg MS, Labrèche F, Weichenthal S, et al. The association between the incidence of postmenopausal breast cancer
and concentrations at street-level of nitrogen dioxide and ultrafine particles2017158715
8. Weichenthal S, Lavigne E, Valois MF, et al. Spatial variations in ambient ultrafine particle concentrations and the risk of incident prostate cancer: a case-control study.Environ Res2017156374380
9. Li C. Breast Cancer
Epidemiology2010New York, NYSpringer Science
10. Lee PN, Hamling JS. Environmental tobacco smoke exposure and risk of breast cancer
in nonsmoking women. An updated review and meta-analysis.Inhal Toxicol20162810431454
11. Andersen ZJ, Stafoggia M, Weinmayr G, et al. Long-term exposure to ambient air pollution and incidence of postmenopausal breast cancer
in 15 European Cohorts within the ESCAPE Project.Environ Health Perspect201712510107005
12. Andersen ZJ, Ravnskjaer L, Andersen KK, et al. Long-term exposure to fine particulate matter
and breast cancer
incidence in the Danish Nurse Cohort Study
.Cancer Epidemiol Biomarkers Prev2017263428430
13. Hart JE, Bertrand KA, DuPre N, et al. Long-term particulate matter exposures during adulthood and risk of breast cancer
incidence in the Nurses’ Health Study II Prospective Cohort.Cancer Epidemiol Biomarkers Prev201625812741276
14. Reding KW, Young MT, Szpiro AA, et al. Breast cancer
risk in relation to ambient air pollution exposure at residences in the Sister Study Cohort.Cancer Epidemiol Biomarkers Prev2015241219071909
15. Labreche F, Goldberg MS, Valois MF, et al. Postmenopausal breast cancer
and occupational exposures.Occup Environ Med2010674263269
16. Villeneuve PJ, Weichenthal SA, Crouse D, et al. Long-term exposure to fine particulate matter
air pollution and mortality among Canadian women.Epidemiology201526536545
17. Miller AB, Baines CJ, To T, et al. Canadian National Breast Screening
Study: 2. Breast cancer
detection and death rates among women aged 50 to 59 years.CMAJ19921471014771488
18. Miller AB, Baines CJ, To T, et al. Canadian National Breast Screening
Study: 1. Breast cancer
detection and death rates among women aged 40 to 49 years.CMAJ19921471014591476
19. Miller AB, To T, Baines CJ, et al. The Canadian National Breast Screening
Study-1: breast cancer
mortality after 11 to 16 years of follow-up. A randomized screening
trial of mammography in women age 40 to 49 years.Ann Intern Med20021375 Pt 1305312
20. Costanian C, McCague H, Tamim H. Age at natural menopause and its associated factors in Canada
: cross-sectional analyses from the Canadian Longitudinal Study on Aging.Menopause201825265272
21. Statistics CanadaVital Statistics – Death Database2017OttawaStatistics CanadaAvailable at: http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=3233
. Accessed 2017
22. Goldberg MS, Carpenter M, Theriault G, et al. The accuracy of ascertaining vital status in a historical cohort study
of synthetic textiles workers using computerized record linkage to the Canadian Mortality Data Base.Can J Public Health199384201224
23. Statistics CanadaCanadian Cancer Registry2017OttawaStatistics CanadaAvailable at: http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=3207
. Accessed 2017
24. van Donkelaar A, Martin RV, Brauer M, et al. Use of satellite observations for long-term exposure assessment of global concentrations of fine particulate matter
.Environ Health Perspect20151232135143
25. van Donkelaar A, Martin RV, Brauer M, et al. Global estimates of ambient fine particulate matter
concentrations from satellite-based aerosol optical depth: development and application.Environ Health Perspect20101186847855
26. Van Donkelaar A, Martin RV, Spurr RJD, et al. Optimal estimation for global ground-level fine particulate matter
concentrations.J Geophys Res Atmos201311856215636
27. Brook RD, Cakmak S, Turner MC, et al. Long-term fine particulate matter
exposure and mortality from diabetes in Canada
28. Crouse DL, Peters PA, van Donkelaar A, et al. Risk of nonaccidental and cardiovascular mortality in relation to long-term exposure to low concentrations of fine particulate matter
: a Canadian national-level cohort study
.Environ Health Perspect20121205708714
29. Crouse DL, Peters PA, Hystad P, et al. Ambient PM2.5, O(3), and NO(2) exposures and associations with mortality over 16 years of follow-up in the Canadian Census Health and Environment Cohort (CanCHEC).Environ Health Perspect20151231111801186
30. Cologne J, Hsu WL, Abbott RD, et al. Proportional hazards regression in epidemiologic follow-up studies: an intuitive consideration of primary time scale.Epidemiology2012234565573
31. Crouse DL, Goldberg MS, Ross NA. A prediction-based approach to modelling temporal and spatial variability of traffic-related air pollution in Montreal, Canada
32. Beckerman B, Jerrett M, Brook JR, et al. Correlation of nitrogen dioxide with other traffic pollutants near a major expressway.Atmos Environ2008422275290
33. HEI Panel on the Health Effects of Traffic-Related Air PollutionTraffic-Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Special Reports2010Boston, MAHealth Effects Institute
34. Weichenthal S, Ryswyk KV, Goldstein A, et al. A land use regression model for ambient ultrafine particles in Montreal, Canada
: a comparison of linear regression and a machine learning approach.Environ Res20161466572
35. International Agency for Research on CancerPainting, firefighting, and shiftwork.In: IARC Working Group on the Evaluation of Carcinogenic Risks to Humans2007Lyon, FranceInternational Agency for Research on Cancer
36. WHOResearch on the Menopause1981Geneva, SwitzerlandWorld Health Organization
37. Research on the Menopause in the 1990s: Report of a WHO Scientific Group1996GenevaWorld Health Organization1107
38. Brauer M, Freedman G, Frostad J, et al. Ambient air pollution exposure estimation for the Global Burden of Disease 2013.Environ Sci Technol20165017988