Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Dec 15.
Published in final edited form as: Environ Res. 2023 Oct 9;239(Pt 1):117349. doi: 10.1016/j.envres.2023.117349

Understanding the role of environmental and socioeconomic factors in the geographic variation of breast cancer risk in the US-wide Sister Study

Rachel Carroll a,b,*, Jennifer L Ish c,*, Dale P Sandler c, Alexandra J White c,**, Shanshan Zhao a,**
PMCID: PMC10841999  NIHMSID: NIHMS1938065  PMID: 37821066

Abstract

Objective:

To describe the geographic pattern of breast cancer incidence in a nationwide prospective cohort and investigate whether environmental exposures and/or neighborhood socioeconomic status explain observed geographic disparities.

Methods:

Using accelerated failure time models with a spatial random effect term, we mapped the health region-level association between residential location and breast cancer incidence for 44,707 participants in the Sister Study after controlling for established individual-level breast cancer risk factors. We performed a variable selection process to select environmental exposures [i.e., ambient nitrogen dioxide (NO2) and fine particulate matter (PM2.5), PM2.5 chemical composition, outdoor light at night (LAN), ambient noise, ultraviolet radiation, and greenspace] and neighborhood-level factors [i.e., population density and area deprivation index (ADI)] that predicted breast cancer incidence and quantified the spatial variation explained by the selected factors. We also considered whether the geographic pattern and predictors were similar when restricting to estrogen receptor-positive (ER+) tumors.

Results:

We observed a spatial patterning in the incidence of overall breast cancer (Moran’s I=16.7, p<0.05) and ER+ breast cancer (Moran’s I=13.2, p<0.05), with a lower risk observed in the South and Southeast and a greater risk in the Northwest and certain areas of the Midwest and Northeast. NO2, LAN, and ADI explained 21.4% of the spatial variation in overall breast cancer incidence whereas NO2, PM2.5 chemical composition, LAN, greenspace, and ADI together explained 63.3% of the spatial variation in ER+ breast cancer incidence.

Conclusions:

Our findings provide additional evidence for a role of environmental exposures in breast cancer incidence and suggest that geographic-based risk factors may vary according to breast cancer subtype. Our findings support the need for additional research to quantify the relative contributions of geographic-based risk factors for breast cancer.

Keywords: breast cancer, spatial epidemiology, geographic risk factors

1. Introduction

The incidence of breast cancer—the most diagnosed cancer among women in the United States (US)—varies regionally, with higher rates observed in the Northeast and Midwest compared to the South and Southwest according to national registry statistics (Centers of Disease Control and Prevention, 2022; Laden et al., 1997). Established demographic and reproductive risk factors for breast cancer do not fully explain this disparity (Reynolds et al., 2004). Thus, spatially patterned environmental exposures and neighborhood contextual factors warrant investigation to provide insight into the combined role of these factors in breast cancer etiology (Vieira et al., 2020). The mechanisms by which these factors may influence breast cancer risk include endocrine disruption, oxidative stress and/or inflammation (IOM (Institute of Medicine), 2012). For example, increasing evidence suggests outdoor air pollution may contribute to breast cancer risk (Andersen et al., 2017; White et al., 2023; White et al., 2021; White et al., 2019). Other environmental exposures, such as residential proximity to greenspace and exposure to outdoor light at night, may also contribute to the burden of exogenous endocrine disrupting exposures [e.g., greenspace may improve air quality, which in turn reduces air pollution exposure (Diener and Mudu, 2021)] or interact with endogenous hormones, which are one of the key biologic pathways in breast cancer (Sweeney et al., 2022; Zare Sakhvidi et al., 2022). In addition, neighborhood contextual factors, such as socioeconomic status and disadvantage, may be associated with breast cancer risk independently of individual-level socioeconomic status and other established breast cancer risk factors (Barber et al., 2021). Herein, our objective was to describe the geographic distribution of breast cancer incidence in a nationwide prospective cohort and to investigate whether a select set of environmental and social factors, as a proof of principle, explain observed geographic disparities.

2. Methods

2.1. Study population

We used data from the Sister Study, a prospective cohort of 50,884 U.S. women enrolled between 2003–2009 who had at least one sister who had been diagnosed with breast cancer but had no personal history of breast cancer themselves (Sandler et al., 2017). As part of the baseline questionnaire, participants provided extensive information including their demographics, characteristics of their current residence, and their reproductive and medical history. Participants also completed a home visit with a trained examiner who took height and weight measurements to calculate body mass index (BMI). Annual health updates and more detailed questionnaires were administered throughout follow-up. We identified breast cancer cases by self-report and confirmed using the medical record and pathology reports. Approximately 81% of cases had medical records available. Because agreement between self-reports and medical records was better than 99% for overall and estrogen receptor positive breast cancer, self-reported diagnoses were included when medical records were not obtained.

We first excluded participants who were diagnosed with breast cancer prior to the completion of all baseline data collection, had an uncertain breast cancer diagnosis status, or who did not contribute any follow-up time [n = 457 (0.9%)]. Next, we excluded participants for whom information on individual-level risk factors were not available [n = 3,974, (7.8%)] or who resided outside the contiguous US and thus did not have information available for certain environmental exposures [e.g., air pollution, greenspace, and noise; n = 1,746 (3.4%)]. Thus, the final sample size eligible for our analysis was 44,707. The Sister Study was approved by the institutional review boards of the National Institutes of Health and all participants provided written informed consent. Our analysis considered any breast cancer cases (invasive, ductal carcinoma in situ) and specifically estrogen receptor positive (ER+) cases diagnosed before October 12, 2020 (Data Release 10.1).

2.2. Environmental exposures

We considered the following environmental exposures that may contribute to breast cancer incidence: ambient air pollution [nitrogen dioxide (NO2) and particulate matter < 2.5 microns in aerodynamic diameter (PM2.5)], profiles of PM2.5 chemical composition (Keller et al., 2017), ultraviolet (UV) radiation, greenspace, total and anthropogenic ambient noise, and outdoor light at night (LAN)]. Participants’ geocoded primary residential addresses during the 12 months prior to enrollment (available for > 99% of the study population) were used to assign exposure values. For each exposure, we selected the measurement year that was as close as possible to the mid-point of the enrollment period. Annual average concentrations of NO2 (ppm) and PM2.5 (μg/m3) were estimated for the year 2006 using validated spatiotemporal models (Kirwa et al., 2021). Participants were assigned one of seven profiles of PM2.5 chemical composition previously developed using predictive k-means clustering and 2010 concentrations of 22 PM2.5 components obtained from US Environmental Protection Agency Air Quality System monitoring locations (Keller et al., 2017; White et al., 2019). UV radiation (mW/m2) was assessed using a satellite-based exposure model to predict July 2006 noon-time estimates at 1-km spatial resolution (Gregoire et al., 2022; VoPham et al., 2016). Greenspace was measured using the NASA’s Moderate Resolution Imaging Spectroradiometer to estimate the normalized difference vegetation index (NDVI; https://modis.gsfc.nasa.gov/data/dataprod/mod13.php), with values ranging from −1 to 1, at an approximately 439.5-m resolution. We considered NDVI values from July 2006. Annual average total and anthropogenic ambient noise in decibels (dB) for 2006 were estimated using a land-use regression model to generate a weighted 24-hour equivalent (i.e., median daily sound levels) at a 270-m spatial resolution (Mennitt and Fristrup, 2016). Finally, annual average outdoor LAN in radiance (nW/cm2/sr) for 2006 was estimated at a 1-km resolution based on satellite images (Sweeney et al., 2022; Xiao et al., 2020).

2.3. Neighborhood-level factors

Using data from the US Census (https://www.census.gov/), we assessed population density (population size per square mile) for the year 2000 at the census tract-level and area deprivation index (ADI), a composite index based on 17 census block-level indicators of poverty, education, housing, and employment (Kind et al., 2014). We considered ADI values from the year 2000 which range from 0 (least deprived) to 100 (most deprived).

2.4. Statistical analysis

We first fit accelerated failure time (AFT) models to describe the spatial distribution of breast cancer incidence among women in the Sister Study after controlling for established individual-level breast cancer risk factors: self-reported race/ethnicity, educational attainment, household income, body mass index, pack-years of cigarette smoking, alcohol consumption, age at menarche, age at first birth, hormone replacement therapy use, and duration of oral contraceptive use (American Cancer Society, 2022). The AFT models were fit to regress the logarithm of participants’ age at breast cancer diagnosis (timescale) on the fixed effects of individual-level baseline risk factors with a spatial random effect (Carroll et al., 2017). We estimated the spatial random effect at the health region-level to preserve participant anonymity while maintaining sufficient statistical power based on the number of breast cancer cases at various spatial scales. Public health regions are defined by each state and are comprised of a group of neighboring counties. Definitions, shapefiles, and supporting files for the health regions are available on GitHub (user: carrollrm, repo: Shapefiles). Sister Study participants are represented in all of the 382 health regions in the contiguous US, with an average of 117 participants including 10 breast cancer cases and 7.3 ER+ breast cancer cases per health region.

The AFT model for an individual i in health region j was written as: log(Yij) = Xijβ + wj + σϵij where Yij denoted the individual’s age at breast cancer diagnosis, Xijβ denoted the linear predictor of interest, wj denoted the spatial random effect, ϵij denoted the independent Gaussian error, and σ denoted a scale parameter (Carroll et al., 2017; Christensen and Johnson, 1988). Left-truncation due to different enrollment age and right-censoring due to lost to follow-up were accommodated in the AFT likelihood (Ning, Qin and Yu, 2014; Yu, 2010; Zhang and Lawson, 2011). We assumed the correlation structure of the spatial random effects was such that neighboring spatial areas (i.e., areas with shared boundaries) were more alike than non-neighboring areas (i.e., areas without shared boundaries). The health region-level effect can be interpreted as residual spatial variation in breast cancer risk after accounting for the fixed covariates included in the model. All inferences were performed under the Bayesian framework with parametric prior specification of the primary model (Carroll et al., 2017; Carroll and Zhao, 2018; McCoy et al., 2011).

We performed a forward variable selection process (p-value threshold < 0.15) to select environmental exposures to include in the model while forcing all established individual-level risk factors in the model for all breast cancer and ER+ breast cancer, separately. We scaled continuous, non-index environmental and neighborhood-level variables by their interquartile range (IQR). We calculated Moran’s I to quantify and compare the level of spatial autocorrelation in breast cancer incidence both after adjustment for individual-level risk factors and after further adjustment with selected exposures. To quantify the amount of remaining spatial variation in breast cancer incidence explained by the selected environmental and neighborhood-level variables, we calculated the compliment of the ratio of variation in the spatial random effect term (i.e., health region-level association) from the model adjusted for the selected variables and individual-level risk factors to the variation of the spatial random effect term observed in the model adjusted only for individual-level risk factors. Finally, we mapped the association between health region of residence and breast cancer incidence to visualize the observed spatial variation in breast cancer risk after controlling for established individual-level risk factors and after additional adjustment for environmental and neighborhood-level variables. All analyses were performed using R version 4.2.2.

3. Results and Discussion

The characteristics of study participants have been described previously (Sandler et al., 2017). During an average of 11.7 years of follow up, 3,820 breast cancer cases were identified (2,800 ER+ cases). The correlation between environmental exposures was low to moderate (Pearson’s r = −0.54 to 0.57), except for NO2 and noise (NO2 and anthropogenic noise: 0.66; NO2 and total noise: 0.67), NO2 and LAN (0.75), and LAN and noise (LAN and anthropogenic noise: 0.70; LAN and total noise: 0.73) (Figure 1).

Figure 1.

Figure 1.

Open in a new tab

Correlations between environmental and neighborhood-level exposures among Sister Study participants (n = 44,707). Note: ADI: 2000 census block-level area deprivation index [ADI; percentiles ranging from 0 (least deprived) to 100 (most deprived)]; Particulate matter < 2.5 microns in aerodynamic diameter (PM2.5; μg/m3) and nitrogen dioxide [NO2; ppm); Ultraviolet radiation [UV; milliwatts per square meter (mW/m2)]; NDVI: Normalized difference vegetation index (greenspace), values range from −1 to 1]; Ant Noise and Tot Noise: Anthropogenic and total ambient noise [decibels (dB)]; LAN: Outdoor light at night [radiance (nW/cm2/sr)]; Pop Dens: 2000 census tract-level population density (population per square mile)

After accounting for individual-level breast cancer risk factors, we observed a spatial patterning of overall breast cancer incidence (Moran’s I=16.7, p<0.05), with lower risk in the South and Southeast and greater risk in the Northwest and certain areas of the Midwest and Northeast (Figure 2, Panel A). A similar gradient was observed for ER+ breast cancer incidence (Moran’s I=13.2, p<0.05; Figure 2, Panel C). We present the fixed effect estimates associated with environmental and neighborhood-level factors in Table 1. Because the estimates are directly related to the logarithm of time, negative values correspond to increased risk and vice versa. The estimates can be interpreted as the change in the mean time-to-breast cancer diagnosis [for example, −0.05 for NO2 translates to a 5% decrease (e−0.05) in age at diagnosis]. While the coefficients don’t directly correspond to years, limiting their interpretability, we emphasize the direction of the effect estimates was consistent with existing literature: residential exposure to NO2 was positively associated with overall and ER+ breast cancer risk, while outdoor light at night and neighborhood deprivation as measured by ADI was inversely associated with ER+ breast cancer risk (Barber et al., 2021; Sweeney et al., 2022; White et al., 2019).

Figure 2.

Figure 2.

Open in a new tab

Geographic distribution of incident breast cancer estimated at the health region-level among Sister Study participants (N = 44,707). Smaller effect estimates indicate shorter time-to-diagnosis (i.e., increased risk). Panels A and C: Estimates for all breast cancer (A) and ER+ breast cancer (C) adjusted for race/ethnicity, educational attainment, household income, body mass index, pack-years of smoking, current alcohol use, family history of breast cancer, age at menarche, age at first birth, breastfeeding, hormone therapy use, and oral contraceptive use. Panel B: All breast cancer, additionally adjusted for nitrogen dioxide (NO2), outdoor light at night (LAN), and area deprivation index (ADI) (Variance explained: 21.4%). Panel D: ER+ breast cancer, additionally adjusted for NO2, fine particulate matter (PM2.5) chemical composition, LAN, greenspace, and ADI (Variance explained: 63.3%).

Table 1.

Fixed effect estimates (95% confidence intervals) from accelerated failure time modelsa for the association between environmental exposures and neighborhood-level social factors and incident breast cancer among Sister Study participants (n = 44,707).

Median (IQR) β (95% Confidence Interval)
All breast cancer ER+ breast cancer
Neighborhood-level factors b
Area deprivation index (percentile) 30 (37) 0.001 (0.0, 0.002) 0.002 (0.0, 0.003)
Population density (N/mi2) 1496.16 (3480.43) - -
Environmental exposures c
Total noise (dB) 47.4 (4.77)
Anthropogenic noise (dB) 47.01 (5.23) -
Greenspace (NDVI) 0.65 (0.26) - 0.16 (−0.04, 0.37)
NO2 (ppm) 9.10 (6.26) −0.05 (−0.09, −0.02) −0.09 (−0.17, −0.03)
PM2.5 (μg/m3) 10.81 (3.57) - -
PM2.5 cluster,d n (%)
 1 - - Ref
 2 - - 0.05 (−0.03, 0.14)
 3 - - 0.05 (−0.04, 0.14)
 4 - - 0.11 (−0.03, 0.26)
 5 - - 0.04 (−0.09, 0.17)
 6 - - 0.05 (−0.15, 0.26)
 7 - - 0.06 (−0.1, 0.23)
Outdoor light at night (nW/cm2/sr) 126.03 (202.74) 0.03 (−0.003, 0.07) 0.07 (0.01, 0.13)
Ultraviolet radiation (mW/m2) 173.32 (42.51) - -
Open in a new tab
a

Non-index continuous variables were scaled by their interquartile range (IQR). Positive effect estimates indicate decreased risk and vice versa.

b

2000 census tract-level population density (population per square mile) and 2000 census block-level area deprivation index [ADI; percentiles ranging from 0 (least deprived) to 100 (most deprived)]

c

Total and anthropogenic ambient noise [decibels (dB)]; Greenspace [Normalized difference vegetation index (NDVI), values range from −1 to 1]; Nitrogen dioxide [NO2; ppm) and particulate matter < 2.5 microns in aerodynamic diameter (PM2.5; μg/m3); Outdoor light at night [radiance (nW/cm2/sr)]; Ultraviolet radiation [UV; milliwatts per square meter (mW/m2)]

d

Clusters represent similar profiles of PM2.5 chemical composition (White et al., 2019)

Compared to overall breast cancer, we not only observed that a greater number of environmental factors predicted the spatial variation in ER+ breast cancer but also that the selected variables explained a larger proportion of the spatial disparity (Figure 2). The variable selection process indicated that NO2, LAN, and ADI explained 21.4% of the spatial variation in overall breast cancer incidence whereas NO2, PM2.5 component profile, LAN, NDVI, and ADI together explained 63.3% of the spatial variation in ER+ breast cancer incidence. This is consistent with hypothesized role of environmental exposures in the etiology of hormone receptor positive breast cancer, such that exposures that may act as exogenous endocrine disruptors or interact with endogenous hormone pathways may increase the risk of ER+ breast cancer onset (Althuis et al., 2004). Due to a relatively small number of cases with ER− tumors, we were not able to evaluate the spatial pattern of ER− breast cancer. Nonetheless, our results contribute to evidence that environmental exposures are relevant to ER+ breast cancer risk. A possible direction for future research is to formally evaluate whether geographic risk factors for breast cancer vary according to tumor subtype.

We emphasize that our analysis was not designed to evaluate causal factors but rather to take a data-driven approach identify which exposures may contribute to the observed geographic disparity in breast cancer incidence. In a similar spatial analysis conducted in the Nurses’ Health Study, investigators observed an association between residential location and breast cancer incidence among a cohort of primarily non-Hispanic white women but that spatial patterns of risk were not fully explained by several individual-level risk factors nor socioeconomic factors or environmental exposures (i.e., shift work, light at night, and radon exposure) (Vieira et al., 2020). Whereas in the Sister Study—a somewhat younger and more racially and ethnically diverse population—we considered a larger set of environmental factors, which appeared to play a significant role in geographic disparities in breast cancer risk. Our analysis, however, cannot disentangle the relative contribution of each exposure to explained spatial variance. This is an avenue for future research as well as evaluating the potential interplay between neighborhood context and environmental exposures when examining geographic risk factors for breast cancer. In addition, NO2 and LAN were highly correlated with and possibly capture unmeasured factors related to urbanicity. Our analysis also does not consider local variation in breast cancer risk (Nordsborg et al., 2014; Vieira et al., 2008). Lastly, due in part to historic and contemporary racial residential segregation in the US, racial identity is an important determinant of neighborhood context and exposure to environmental hazards (Morello-Frosch and Lopez, 2006). Thus, geographic disparities and the environmental factors that explain them may be pattered by race, which we were not able to evaluate in our analysis.

The spatial resolution at which we were able to assess the association between residential location (i.e., health region) and breast cancer risk was coarse but was still the smallest possible spatial scale that preserved participant anonymity and afforded sufficient statistical power based on the number of breast cancer cases in each spatial unit. It is possible that a different pattern or more nuanced associations could emerge when considering the spatial variation in breast cancer incidence at a smaller spatial level (i.e., county or census tract) or using a statistical smoothing approach (i.e., generalized additive models) as was done in the previously discussed analysis in the Nurses’ Health Study (Vieira et al., 2020). However, our approach of using accelerated failure time models with a spatial random effect term directly models age at breast cancer diagnosis as a linear function and thus is not subject to the implicit proportional hazards assumptions at each spatial location that is imposed on generalized additive models when modeling time-to-event data. In addition, we did not have residential history information prior to baseline and assumed that exposures assessed at study enrollment were a reasonable proxy of long-term exposure. Further, we assigned exposure values to enrollment residences based on the midpoint of the enrollment period (2006) not the actual year of enrollment (2003–2009). Since air pollutant levels generally declined between 2003 and 2009, exposures assigned to participants before 2006 could be under-estimated and exposures for women who enrolled after 2006 could be over-estimated. However, air pollution levels are highly correlated over this short time period and thus we expect the relative rankings of exposure concentrations among Sister Study participants to have remained stable. Our analysis leveraged fine-scale assessment of exposures assigned to each individual and had the unique advantage of large study population of participants living in all 48 contiguous US states, allowing us to consider nationwide geographic disparities in breast cancer risk. We were also able to leverage prospective time-to-diagnosis information and adjust for a number of established breast cancer risk factors that are available in the Sister Study.

In conclusion, this analysis provides additional support for a role of environmental exposures in breast cancer incidence and that geographic risk factors may vary according to breast cancer subtype. Our findings support the need for additional research that seeks to quantify the relative contribution of geographic risk factors for breast cancer.

Funding:

This work was supported by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences Z01-ES103332 and ZIA-ES103308.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Competing interest: None to declare

The Sister Study was approved by the institutional review board of the National Institutes of Health and all participants provided written informed consent.

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References

  1. Althuis MD, et al. , 2004. Etiology of hormone receptor-defined breast cancer: a systematic review of the literature. Cancer Epidemiol Biomarkers Prev. 13, 1558–68. [PubMed] [Google Scholar]
  2. American Cancer Society, Breast Cancer Facts and Figures 2022–2024. Atlanta, GA, 2022. [Google Scholar]
  3. Andersen ZJ, et al. , 2017. Long-Term Exposure to Ambient Air Pollution and Incidence of Postmenopausal Breast Cancer in 15 European Cohorts within the ESCAPE Project. Environmental Health Perspectives. 125, 107005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Barber LE, et al. , 2021. Neighborhood disadvantage and individual-level life stressors in relation to breast cancer incidence in US Black women. Breast Cancer Research. 23, 108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Carroll R, et al. , 2017. Assessment of spatial variation in breast cancer-specific mortality using Louisiana SEER data. Social Science & Medicine. 193, 1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Carroll R, Zhao S, 2018. Gaining relevance from the random: Interpreting observed spatial heterogeneity. Spat Spatiotemporal Epidemiol. 25, 11–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Centers of Disease Control and Prevention, United States Cancer Statistics. Vol. 2023, 2022. [Google Scholar]
  8. Christensen R, Johnson W, 1988. Modelling Accelerated Failure Time with a Dirichlet Process. Biometrika. 75, 693–704. [Google Scholar]
  9. Diener A, Mudu P, 2021. How can vegetation protect us from air pollution? A critical review on green spaces’ mitigation abilities for air-borne particles from a public health perspective - with implications for urban planning. Sci Total Environ. 796, 148605. [DOI] [PubMed] [Google Scholar]
  10. Gregoire AM, et al. , 2022. Residential ultraviolet radiation and breast cancer risk in a large prospective cohort. Environ Int. 159, 107028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. IOM (Institute of Medicine), Breast cancer and the environment: A life course approach. Washington, DC, 2012. [Google Scholar]
  12. Keller JP, et al. , 2017. Covariate-adaptive clutering of exposures for air pollution epidemiology cohorts. The Annals of Applied Statistics. 11, 93–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Kind AJ, et al. , 2014. Neighborhood socioeconomic disadvantage and 30-day rehospitalization: a retrospective cohort study. Ann Intern Med. 161, 765–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kirwa K, et al. , 2021. Fine-Scale Air Pollution Models for Epidemiologic Research: Insights From Approaches Developed in the Multi-ethnic Study of Atherosclerosis and Air Pollution (MESA Air). Current Environmental Health Reports. 8, 113–126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Laden F, et al. , 1997. Geographic Variation in Breast Cancer Incidence Rates in a Cohort of U.S. Women. JNCI: Journal of the National Cancer Institute. 89, 1373–1378. [DOI] [PubMed] [Google Scholar]
  16. McCoy AC, et al. , 2011. Serum Mullerian inhibiting substance levels are lower in premenopausal women with breast precancer and cancer. BMC Res Notes. 4, 152. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Mennitt DJ, Fristrup KM, 2016. Influence factors and spatiotemporal patterns of environmental sound levels in the contiguous United States. Noise Control Engineering Journal. 64, 342–353. [Google Scholar]
  18. Morello-Frosch R, Lopez R, 2006. The riskscape and the color line: examining the role of segregation in environmental health disparities. Environmental Research. 102, 181–196. [DOI] [PubMed] [Google Scholar]
  19. Ning J, Qin J, Yu Y, 2014. Semiparametric accelerated failure time model for length-biased data with application to dementia study. Statistica Sinica. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Nordsborg RB, et al. , 2014. Space-time clusters of breast cancer using residential histories: A Danish case–control study. BMC Cancer. 14, 255. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Reynolds P, et al. , 2004. Regional variations in breast cancer among california teachers. Epidemiology. 15, 746–54. [DOI] [PubMed] [Google Scholar]
  22. Sandler DP, et al. , 2017. The Sister Study Cohort: Baseline Methods and Participant Characteristics. Environmental Health Perspectives. 125, 127003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Sweeney MR, et al. , 2022. Light at night and the risk of breast cancer: Findings from the Sister study. Environment International. 169, 107495. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Vieira VM, et al. , 2020. Contribution of socioeconomic and environmental factors to geographic disparities in breast cancer risk in the Nurses’ Health Study II. Environ Epidemiol. 4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Vieira VM, et al. , 2008. Spatial-temporal analysis of breast cancer in upper Cape Cod, Massachusetts. International Journal of Health Geographics. 7, 46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. VoPham T, et al. , 2016. Spatiotemporal exposure modeling of ambient erythemal ultraviolet radiation. Environ Health. 15, 111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. White AJ, et al. , 2023. Ambient fine particulate matter and breast cancer incidence in a large prospective US cohort. JNCI: Journal of the National Cancer Institute. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. White AJ, et al. , 2021. Air pollution and breast cancer risk in the Black Women’s Health Study. Environmental Research. 194, 110651. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. White AJ, et al. , 2019. Air Pollution, Clustering of Particulate Matter Components, and Breast Cancer in the Sister Study: A U.S.-Wide Cohort. Environmental Health Perspectives. 127, 107002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Xiao Q, et al. , 2020. Outdoor light at night and postmenopausal breast cancer risk in the NIH-AARP diet and health study. International journal of cancer. 147, 2363–2372. [DOI] [PubMed] [Google Scholar]
  31. Yu B, 2010. A Bayesian MCMC approach to survival analysis with doubly-censored data. Comput Stat Data Anal. 54, 1921–1929. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Zare Sakhvidi MJ, et al. , 2022. Exposure to greenspace and cancer incidence, prevalence, and mortality: A systematic review and meta-analyses. Sci Total Environ. 838, 156180. [DOI] [PubMed] [Google Scholar]
  33. Zhang J, Lawson AB, 2011. Bayesian Parametric Accelerated Failure Time Spatial Model and its Application to Prostate Cancer. Journal of Applied Statistics. 38, 591–603. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES