A national prediction model for PM2.5 component exposures and measurement error-corrected health effect inference

Leveraging low-cost sensors to predict nitrogen dioxide for epidemiologic exposure assessment

BACKGROUND

Statistical models of air pollution enable intra-urban characterization of pollutant concentrations, benefiting exposure assessment for environmental epidemiology. The new generation of low-cost sensors facilitate the deployment of dense monitoring networks and can potentially be used to improve intra-urban models of air pollution.

OBJECTIVE

Develop and evaluate a spatiotemporal model for nitrogen dioxide (NO2) in the Puget Sound region of WA, USA for the Adult Changes in Thought Air Pollution (ACT-AP) study and assess the contribution of low-cost sensor data to the model's performance through cross-validation.

METHODS

We developed a spatiotemporal NO2 model for the study region incorporating data from 11 agency locations, 364 supplementary monitoring locations, and 117 low-cost sensor (LCS) locations for the 1996-2020 time period. Model features included long-term time trends and dimension-reduced land use regression. We evaluated the contribution of LCS network data by comparing models fit with and without sensor data using cross-validated (CV) summary performance statistics.

RESULTS

The best performing model had one time trend and geographic covariates summarized into three partial least squares components. The model, fit with LCS data, performed as well as other recent studies (agency cross-validation: CV- root mean square error (RMSE) = 2.5 ppb NO2; CV- coefficient of determination (R 2 ) = 0.85). Predictions of NO2 concentrations developed with LCS were higher at residential locations compared to a model without LCS, especially in recent years. While LCS did not provide a strong performance gain at agency sites (CV-RMSE = 2.8 ppb NO2; CV-R 2  = 0.82 without LCS), at residential locations, the improvement was substantial, with RMSE = 3.8 ppb NO2 andR 2  = 0.08 (without LCS), compared to CV-RMSE = 2.8 ppb NO2 and CV-R 2  = 0.51 (with LCS).

IMPACT

We developed a spatiotemporal model for nitrogen dioxide (NO2) pollution in Washington's Puget Sound region for epidemiologic exposure assessment for the Adult Changes in Thought Air Pollution study. We examined the impact of including low-cost sensor data in the NO2 model and found the additional spatial information the sensors provided predicted NO2 concentrations that were higher than without low-cost sensors, particularly in recent years. We did not observe a clear, substantial improvement in cross-validation performance over a similar model fit without low-cost sensor data; however, the prediction improvement with low-cost sensors at residential locations was substantial. The performance gains from low-cost sensors may have been attenuated due to spatial information provided by other supplementary monitoring data.

Ambient air pollution and rate of spontaneous abortion

Spontaneous abortion (SAB), defined as a pregnancy loss before 20 weeks of gestation, affects up to 30% of conceptions, yet few modifiable risk factors have been identified. We estimated the effect of ambient air pollution exposure on SAB incidence in Pregnancy Study Online (PRESTO), a preconception cohort study of North American couples who were trying to conceive. Participants completed questionnaires at baseline, every 8 weeks during preconception follow-up, and in early and late pregnancy. We analyzed data on 4643 United States (U.S.) participants and 851 Canadian participants who enrolled during 2013-2019 and conceived during 12 months of follow-up. We used country-specific national spatiotemporal models to estimate concentrations of particulate matter <2.5 μm (PM2.5), nitrogen dioxide (NO2), and ozone (O3) during the preconception and prenatal periods at each participant's residential address. On follow-up and pregnancy questionnaires, participants reported information on pregnancy status, including SAB incidence and timing. We fit Cox proportional hazards regression models with gestational weeks as the time scale to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the association of time-varying prenatal concentrations of PM2.5, NO2, and O3 with rate of SAB, adjusting for individual- and neighborhood-level factors. Nineteen percent of pregnancies ended in SAB. Greater PM2.5 concentrations were associated with a higher incidence of SAB in Canada, but not in the U.S. (HRs for a 5 μg/m3 increase = 1.29, 95% CI: 0.99, 1.68 and 0.94, 95% CI: 0.83, 1.08, respectively). NO2 and O3 concentrations were not appreciably associated with SAB incidence. Results did not vary substantially by gestational weeks or season at risk. In summary, we found little evidence for an effect of residential ambient PM2.5, NO2, and O3 concentrations on SAB incidence in the U.S., but a moderate positive association of PM2.5 with SAB incidence in Canada.

Comparison of PM2.5 Air Pollution Exposures and Health Effects Associations Using 11 Different Modeling Approaches in the Women's Health Initiative Memory Study (WHIMS)

BACKGROUND

Many approaches to quantifying air pollution exposures have been developed. However, the impact of choice of approach on air pollution estimates and health-effects associations remains unclear.

OBJECTIVES

Our objective is to compare particulate matter with aerodynamic diameter ≤ 2.5 μ m (PM 2.5 ) concentrations and resulting health effects associations using multiple estimation approaches previously used in epidemiologic analyses.

METHODS

We assigned annual PM 2.5 exposure estimates from 1999 to 2004 derived from 11 different approaches to Women's Health Initiative Memory Study (WHIMS) participant addresses within the contiguous US. Approaches included geostatistical interpolation approaches, land-use regression or spatiotemporal models, satellite-derived approaches, air dispersion and chemical transport models, and hybrid models. We used descriptive statistics and plots to assess relative and absolute agreement among exposure estimates and examined the impact of approach on associations between PM 2.5 and death due to natural causes, cardiovascular disease (CVD) mortality, and incident CVD events, adjusting for individual-level covariates and climate-based region.

RESULTS

With a few exceptions, relative agreement of approach-specific PM 2.5 exposure estimates was high for PM 2.5 concentrations across the contiguous US. Agreement among approach-specific exposure estimates was stronger near PM 2.5 monitors, in certain regions of the country, and in 2004 vs. 1999. Collectively, our results suggest but do not quantify lower agreement at local spatial scales for PM 2.5 . There was no evidence of large differences in health effects associations with PM 2.5 among estimation approaches in analyses adjusted for climate region.

CONCLUSIONS

Different estimation approaches produced similar spatial patterns of PM 2.5 concentrations across the contiguous US and in areas with dense monitoring data, and PM 2.5 -health effects associations were similar among estimation approaches. PM 2.5 estimates and PM 2.5 -health effects associations may differ more in samples drawn from smaller areas or areas without substantial monitoring data, or in analyses with finer adjustment for participant location. Our results can inform decisions about PM 2.5 estimation approach in epidemiologic studies, as investigators balance concerns about bias, efficiency, and resource allocation. Future work is needed to understand whether these conclusions also apply in the context of other air pollutants of interest. https://doi.org/10.1289/EHP12995.

Separating Daily 1 km PM2.5 Inorganic Chemical Composition in China since 2000 via Deep Learning Integrating Ground, Satellite, and Model Data

Fine particulate matter (PM2.5) chemical composition has strong and diverse impacts on the planetary environment, climate, and health. These effects are still not well understood due to limited surface observations and uncertainties in chemical model simulations. We developed a four-dimensional spatiotemporal deep forest (4D-STDF) model to estimate daily PM2.5 chemical composition at a spatial resolution of 1 km in China since 2000 by integrating measurements of PM2.5 species from a high-density observation network, satellite PM2.5 retrievals, atmospheric reanalyses, and model simulations. Cross-validation results illustrate the reliability of sulfate (SO42-), nitrate (NO3-), ammonium (NH4+), and chloride (Cl-) estimates, with high coefficients of determination (CV-R2) with ground-based observations of 0.74, 0.75, 0.71, and 0.66, and average root-mean-square errors (RMSE) of 6.0, 6.6, 4.3, and 2.3 μg/m3, respectively. The three components of secondary inorganic aerosols (SIAs) account for 21% (SO42-), 20% (NO3-), and 14% (NH4+) of the total PM2.5 mass in eastern China; we observed significant reductions in the mass of inorganic components by 40-43% between 2013 and 2020, slowing down since 2018. Comparatively, the ratio of SIA to PM2.5 increased by 7% across eastern China except in Beijing and nearby areas, accelerating in recent years. SO42- has been the dominant SIA component in eastern China, although it was surpassed by NO3- in some areas, e.g., Beijing-Tianjin-Hebei region since 2016. SIA, accounting for nearly half (∼46%) of the PM2.5 mass, drove the explosive formation of winter haze episodes in the North China Plain. A sharp decline in SIA concentrations and an increase in SIA-to-PM2.5 ratios during the COVID-19 lockdown were also revealed, reflecting the enhanced atmospheric oxidation capacity and formation of secondary particles.

Air pollution and fecundability in a North American preconception cohort study

BACKGROUND

Animal and epidemiologic studies indicate that air pollution may adversely affect fertility. However, the level of evidence is limited and specific pollutants driving the association are inconsistent across studies.

METHODS

We used data from a web-based preconception cohort study of pregnancy planners enrolled during 2013-2019 (Pregnancy Study Online; PRESTO). Eligible participants self-identified as female, were aged 21-45 years, resided in the United States (U.S.) or Canada, and were trying to conceive without fertility treatments. Participants completed a baseline questionnaire and bi-monthly follow-up questionnaires until conception or 12 months. We analyzed data from 8,747 participants (U.S.: 7,304; Canada: 1,443) who had been trying to conceive for < 12 cycles at enrollment. We estimated residential ambient concentrations of particulate matter < 2.5 µm (PM2.5), nitrogen dioxide (NO2), and ozone (O3) using validated spatiotemporal models specific to each country. We fit country-specific proportional probabilities regression models to estimate the association between annual average, menstrual cycle-specific, and preconception average pollutant concentrations with fecundability, the per-cycle probability of conception. We calculated fecundability ratios (FRs) and 95% confidence intervals (CIs) and adjusted for individual- and neighborhood-level confounders.

RESULTS

In the U.S., the FRs for a 5-µg/m3 increase in annual average, cycle-specific, and preconception average PM2.5 concentrations were 0.94 (95% CI: 0.83, 1.08), 1.00 (95% CI: 0.93, 1.07), and 1.00 (95% CI: 0.93, 1.09), respectively. In Canada, the corresponding FRs were 0.92 (95% CI: 0.74, 1.16), 0.97 (95% CI: 0.87, 1.09), and 0.94 (95% CI: 0.80, 1.09), respectively. Likewise, NO2 and O3 concentrations were not strongly associated with fecundability in either country.

CONCLUSIONS

Neither annual average, menstrual cycle-specific, nor preconception average exposure to ambient PM2.5, NO2, and O3 were appreciably associated with reduced fecundability in this cohort of pregnancy planners.

Impact of COVID-19 restrictions on the concentration and source apportionment of atmospheric ammonia (NH3) across India

The wide spread of the coronavirus disease (COVID-19) has significantly influenced human activities around the world, providing a unique opportunity to investigate the response of air pollution to anthropogenic emission reduction. Compared with numerous studies on conventional air pollutants, atmospheric ammonia (NH₃) that has matched sources from both anthropogenic and natural emissions has been rarely investigated. Here we assess impacts of the COVID-19 lockdown on ambient NH₃ variation across India, one of the most severe NH₃ pollution region in the world. The role of meteorology in shaping emission contribution to NH₃ pollution and respective contribution of each emission source to ambient NH₃ before and after the COVID-19 outbreak were investigated using the XGBoost algorithm coupled with WRF-Chem model. Results showed that ambient NH₃ concentrations in the seven major cities (Hyderabad, Bengaluru, Chennai, Delhi, Lucknow, Kolkata and Mumbai) decreased by 2.1-53.8 % whereas in Ahmedabad increased by 20.3 % during the COVID-19 lockdown period. Obvious decrease in NH₃ in Indo-Gangetic Plain (-17.1 %) was mainly driven by favorable meteorology, whereas the slight decline in NH₃ in South India was mainly driven by epidemic-related emission control (-8.56 %). Source appointment results showed that the contribution of industrial emission (Ind) to ambient NH₃ in most megacities showed a decreasing trend (between 11 % and 26 %) during the lockdown period. However, the reduction effect was mostly compensated by increasing contributions (15-25 %) of residential emission (Res) or agricultural soil emission (Ags). Particularly, in Ahmedabad and Lucknow ambient NH₃ increased by 20.3 % and 12 % during the lockdown period, the reduction effect of Ind on ambient NH₃ (-23 % and -11 %, respectively) was absolutely compensated by enhanced contribution of Res (24 %) and Ags (12 %), respectively. Our results highlight the importance of eliminating residential and agricultural NH₃ emissions especially in North India.

Long-term exposure to ambient particulate matter and stroke etiology: Results from the Women's Health Initiative

BACKGROUND

Ambient particulate matter (PM) air pollution is a leading cause of global disability and accounts for an annual 2.9 million deaths globally. PM is established as an important risk factor for cardiovascular disease, however the evidence supporting a link specifically between long-term exposure to ambient PM and incident stroke is less clear. We sought to evaluate the association of long-term exposure to different size fractions of ambient PM with incident stroke (overall and by etiologic subtypes) and cerebrovascular deaths within the Women's Health Initiative, a large prospective study of older women in the US.

METHODS

We studied 155,410 postmenopausal women without previous cerebrovascular disease enrolled into the study between 1993 and 1998, with follow-up through 2010. We assessed geocoded participant address-specific concentrations of ambient PM (fine [PM2.5], respirable [PM10] and coarse [PM10-2.5]), as well as nitrogen dioxide [NO2] using spatiotemporal models. We classified hospitalization events into ischemic, hemorrhagic, or other/unclassified stroke. Cerebrovascular mortality was defined as death from any stroke etiology. We used Cox proportional hazard models to calculate hazard ratios (HR) and 95% confidence intervals (CI), adjusting for individual and neighborhood-level characteristics.

RESULTS

During a median follow-up time of 15 years, participants experienced 4,556 cerebrovascular events. The hazard ratio for all cerebrovascular events was 2.14 (95% CI: 1.87, 2.44) comparing the top versus bottom quartiles of PM2.5. Similarly, there was a statistically significant increase in events comparing the top versus bottom quartiles of PM10 and NO2 (HR: 1.17; 95% CI: 1.03, 1.33 and HR:1.26; 95% CI: 1.12, 1.42). The strength of association did not vary substantially by stroke etiology. There was little evidence of an association between PMcoarse and incident cerebrovascular events.

CONCLUSIONS

Long-term exposure to fine (PM2.5) and respirable (PM10) particulate matter as well as NO2 was associated with a significant increase of cerebrovascular events among postmenopausal women. Strength of the associations were consistent by stroke etiology.

Effect of long-term exposure to PM2.5 on the risk of type 2 diabetes and arthritis in type 2 diabetes patients: Evidence from a national cohort in China

BACKGROUND

It remains unclear whether type 2 diabetes and the complication of arthritis are causally related to the PM_2.5 pollutant. Therefore, we aimed to investigate the associations of long-term PM_2.5 exposure with type 2 diabetes and with arthritis in type 2 diabetes patients.

MATERIALS AND METHODS

This study used data from the China Health and Retirement Longitudinal Survey (CHARLS) implemented during 2011-2018. The associations were analyzed by Cox proportional hazards regression models, and the population-attributable fraction (PAF) was calculated to assess the burden of type 2 diabetes and arthritis-attributable to PM_2.5.

RESULTS

A total of 21,075 participants were finally included, with 19,121 analyzed for PM_2.5 and type 2 diabetes risk and 12,427 analyzed for PM_2.5 and arthritis risk, of which 1,382 with newly-diagnosed type 2 diabetes and 1,328 with arthritis during the follow-up. Overall, each 10 μg/m³ increment in PM_2.5 concentration was significantly associated with an increase in the risk of type 2 diabetes (HR = 1.26, 95 %CI1.22 to 1.31), and the PAF of type 2 diabetes attributable to PM_2.5 was 13.54 %. In type 2 diabetes patients, each 10 μg/m³ increment in PM_2.5 exposure was associated with an increase in arthritis (HR = 1.42, 95 %CI: 1.28 to 1.57), and the association was significantly greater than that (H = 1.23, 95 %CI: 1.19 to 1.28) in adults without type 2 diabetes. The PAFs of arthritis-attributable to PM_2.5 in participants with and without type 2 diabetes were 18.54 % and 10.69 %, respectively.

CONCLUSION

Long-term exposure to PM_2.5 may increase the risk of type 2 diabetes and make type 2 diabetes patients susceptible to arthritis.

Prediction of Air Pollution Interval Based on Data Preprocessing and Multi-Objective Dragonfly Optimization Algorithm

With the rapid development of global industrialization and urbanization, as well as the continuous expansion of the population, large amounts of industrial exhaust gases and automobile exhaust are released. To better sound an early warning of air pollution, researchers have proposed many pollution prediction methods. However, the traditional point prediction methods cannot effectively analyze the volatility and uncertainty of pollution. To fill this gap, we propose a combined prediction system based on fuzzy granulation, multi-objective dragonfly optimization algorithm and probability interval, which can effectively analyze the volatility and uncertainty of pollution. Experimental results show that the combined prediction system can not only effectively predict the changing trend of pollution data and analyze local characteristics but also provide strong technical support for the early warning of air pollution.

A novel approach for assessing the spatiotemporal trend of health risk from ambient particulate matter components: Case of Hong Kong

The spatiotemporal assessment of health risk due to exposure to particulate matter (PM) components should be well studied because of the different toxicity among PM components. However, this research topic has long been overlooked. This study aimed to examine the spatiotemporal variability in ambient respirable PM (PM₁₀) components associated inhalation carcinogenic and non-carcinogenic risk (ICR and INCR) in Hong Kong over 2015-2019. The land-use regression (LUR) approach was adopted to predict the spatial distribution of PM₁₀ component concentrations for the period of 2015-2019, whereas the ICR and INCR values of PM₁₀ components were also estimated using the classic health risk assessment method. Both concentration of PM₁₀ and INCR of PM₁₀ components showed a general decreasing trend, while ICR of PM₁₀ components increased slightly over the study period. LUR-model-based spatial maps at 500 m × 500 m resolution revealed the important spatial variability in PM₁₀ and its eleven components, and their associated ICR and INCR values. High pollution levels and high ICR and INCR of studied PM₁₀ components were generally found in developed urban areas and along the road network. Despite the fact that the PM₁₀ concentrations met the Hong Kong annual PM₁₀ air quality objective of 50 μg/m³, there was still significant potential health risk from the studied PM₁₀ components. This study highlights the importance of taking PM component concentrations and associated inhalation health risk as well as PM mass concentrations into account for the perspective of air quality management and protecting public health.

Prenatal Fine Particulate Matter, Maternal Micronutrient Antioxidant Intake, and Early Childhood Repeated Wheeze: Effect Modification by Race/Ethnicity and Sex

Fine particulate matter (PM2.5) potentiates in utero oxidative stress influencing fetal development while antioxidants have potential protective effects. We examined associations among prenatal PM2.5, maternal antioxidant intake, and childhood wheeze in an urban pregnancy cohort (n = 530). Daily PM2.5 exposure over gestation was estimated using a satellite-based spatiotemporally resolved model. Mothers completed the modified Block98 food frequency questionnaire. Average energy-adjusted percentile intake of β-carotene, vitamins (A, C, E), and trace minerals (zinc, magnesium, selenium) constituted an antioxidant index (AI). Maternal-reported child wheeze was ascertained up to 4.1 ± 2.8 years. Bayesian distributed lag interaction models (BDLIMs) were used to examine time-varying associations between prenatal PM2.5 and repeated wheeze (≥2 episodes) and effect modification by AI, race/ethnicity, and child sex. Covariates included maternal age, education, asthma, and temperature. Women were 39% Black and 33% Hispanic, 36% with ≤high school education; 21% of children had repeated wheeze. Higher AI was associated with decreased wheeze in Blacks (OR = 0.37 (0.19-0.73), per IQR increase). BDLIMs identified a sensitive window for PM2.5 effects on wheeze among boys born to Black mothers with low AI (at 33-40 weeks gestation; OR = 1.74 (1.19-2.54), per µg/m3 increase in PM2.5). Relationships among prenatal PM2.5, antioxidant intake, and child wheeze were modified by race/ethnicity and sex.

Multivariate Spatial Prediction of Air Pollutant Concentrations with INLA

Estimates of daily air pollution concentrations with complete spatial and temporal coverage are important for supporting epidemiologic studies and health impact assessments. While numerous approaches have been developed for modeling air pollution, they typically only consider each pollutant separately. We describe a spatial multipollutant data fusion model that combines monitoring measurements and chemical transport model simulations that leverages dependence between pollutants to improve spatial prediction. For the contiguous United States, we created a data product of daily concentration for 12 pollutants (CO, NOx, NO₂, SO₂, O₃, PM₁₀, and PM_2.5 species EC, OC, NO₃, NH₄, SO₄) during the period 2005 to 2014. Out-of-sample prediction showed good performance, particularly for daily PM_2.5 species EC (R² = 0.64), OC (R² = 0.75), NH₄ (R² = 0.84), NO₃ (R2 = 0.73), and SO₄ (R² = 0.80). By employing the integrated nested Laplace approximation (INLA) for Bayesian inference, our approach also provides model-based prediction error estimates. The daily data product at 12km spatial resolution will be publicly available immediately upon publication. To our knowledge this is the first publicly available data product for major PM_2.5 species and several gases at this spatial and temporal resolution.

Long-Term Exposures to Air Pollution and the Risk of Atrial Fibrillation in the Women's Health Initiative Cohort

BACKGROUND

Atrial fibrillation (AF) is associated with substantial morbidity and mortality. Short-term exposures to air pollution have been associated with AF triggering; less is known regarding associations between long-term air pollution exposures and AF incidence.

OBJECTIVES

Our objective was to assess the association between long-term exposures to air pollution and distance to road on incidence of AF in a cohort of U.S. women.

METHODS

We assessed the association of high resolution spatiotemporal model predictions of long-term exposures to particulate matter (PM 10 and PM 2.5 ), sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), and distance to major roads with incidence of AF diagnosis, identified through Medicare linkage, among 83,117 women in the prospective Women's Health Initiative cohort, followed from enrollment in Medicare through December 2012, incidence of AF, or death. Using time-varying Cox proportional hazards models adjusted for age, race/ethnicity, study component, body mass index, physical activity, menopausal hormone therapy, smoking, diet quality, alcohol consumption, educational attainment, and neighborhood socioeconomic status, we estimated the relative risk of incident AF in association with each pollutant.

RESULTS

A total of 16,348 incident AF cases were observed over 660,236 person-years of follow-up. Most exposure-response associations were nonlinear. NO 2 was associated with risk of AF in multivariable adjusted models [Hazard Ratio  ( HR ) = 1.18 ; 95% confidence interval (CI): 1.13, 1.24, comparing the top to bottom quartile, p -for-trend = < 0.0001 ]. Women living closer to roadways were at higher risk of AF (e.g., HR = 1.07 ; 95% CI: 1.01, 1.13 for living within 50 m of A3 roads, compared with ≥ 1,000  m , p -for-trend = 0.02 ), but we did not observe adverse associations with exposures to PM 10 , PM 2.5 , or SO 2 . There were adverse associations with PM 10 (top quartile HR = 1.10 ; 95% CI: 1.05, 1.16, p -for-trend = < 0.0001 ) and PM 2.5 (top quartile HR = 1.09 ; 95% CI: 1.03, 1.14, p -for-trend = 0.002 ) in sensitivity models adjusting for census region.

DISCUSSION

In this study of postmenopausal women, NO 2 and distance to road were consistently associated with higher risk of AF. https://doi.org/10.1289/EHP7683.

Outdoor air pollution exposure and inter-relation of global cognitive performance and emotional distress in older women

The interrelationships among long-term ambient air pollution exposure, emotional distress and cognitive decline in older adulthood remain unclear. Long-term exposure may impact cognitive performance and subsequently impact emotional health. Conversely, exposure may initially be associated with emotional distress followed by declines in cognitive performance. Here we tested the inter-relationship between global cognitive ability, emotional distress, and exposure to PM2.5 (particulate matter with aerodynamic diameter <2.5 μm) and NO2 (nitrogen dioxide) in 6118 older women (aged 70.6 ± 3.8 years) from the Women's Health Initiative Memory Study. Annual exposure to PM2.5 (interquartile range [IQR] = 3.37 μg/m3) and NO2 (IQR = 9.00 ppb) was estimated at the participant's residence using regionalized national universal kriging models and averaged over the 3-year period before the baseline assessment. Using structural equation mediation models, a latent factor capturing emotional distress was constructed using item-level data from the 6-item Center for Epidemiological Studies Depression Scale and the Short Form Health Survey Emotional Well-Being scale at baseline and one-year follow-up. Trajectories of global cognitive performance, assessed by the Modified-Mini Mental State Examination (3MS) annually up to 12 years, were estimated. All effects reported were adjusted for important confounders. Increases in PM2.5 (β = -0.144 per IQR; 95% CI = -0.261; -0.028) and NO2 (β = -0.157 per IQR; 95% CI = -0.291; -0.022) were associated with lower initial 3MS performance. Lower 3MS performance was associated with increased emotional distress (β = -0.008; 95% CI = -0.015; -0.002) over the subsequent year. Significant indirect effect of both exposures on increases in emotional distress mediated by exposure effects on worse global cognitive performance were present. No statistically significant indirect associations were found between exposures and 3MS trajectories putatively mediated by baseline emotional distress. Our study findings support cognitive aging processes as a mediator of the association between PM2.5 and NO2 exposure and emotional distress in later-life.

Air Pollution and the Dynamic Association Between Depressive Symptoms and Memory in Oldest-Old Women

BACKGROUND/OBJECTIVES

Exposure to air pollution may contribute to both increasing depressive symptoms and decreasing episodic memory in older adulthood, but few studies have examined this hypothesis in a longitudinal context. Accordingly, we examined the association between air pollution and changes in depressive symptoms (DS) and episodic memory (EM) and their interrelationship in oldest-old (aged 80 and older) women.

DESIGN

Prospective cohort data from the Women's Health Initiative Memory Study-Epidemiology of Cognitive Health Outcomes.

SETTING

Geographically diverse community-dwelling population.

PARTICIPANTS

A total of 1,583 dementia-free women aged 80 and older.

MEASUREMENTS

Women completed up to six annual memory assessments (latent composite of East Boston Memory Test and Telephone Interview for Cognitive Status) and the 15-item Geriatric Depression Scale (GDS-15). We estimated 3-year average exposures to regional particulate matter with aerodynamic diameter below 2.5 μm (PM_2.5 ) (interquartile range [IQR] = 3.35 μg/m³ ) and gaseous nitrogen dioxide (NO₂ ) (IQR = 9.55 ppb) at baseline and during a remote period 10 years earlier, using regionalized national universal kriging.

RESULTS

Latent change structural equation models examined whether residing in areas with higher pollutant levels was associated with annual changes in standardized EM and DS while adjusting for potential confounders. Remote NO₂ (β = .287 per IQR; P = .002) and PM_2.5 (β = .170 per IQR; P = .019) exposure was significantly associated with larger increases in standardized DS, although the magnitude of the difference, less than 1 point on the GDS-15, is of questionable clinical significance. Higher DS were associated with accelerated EM declines (β = -.372; P = .001), with a significant indirect effect of remote NO₂ and PM_2.5 exposure on EM declines mediated by DS. There were no other significant indirect exposure effects.

CONCLUSION

These findings in oldest-old women point to potential adverse effects of late-life exposure to air pollution on subsequent interplay between DS and EM, highlighting air pollution as an environmental health risk factor for older women.

Application of Bayesian Additive Regression Trees for Estimating Daily Concentrations of PM2.5 Components

Bayesian additive regression tree (BART) is a recent statistical method that combines ensemble learning and nonparametric regression. BART is constructed under a probabilistic framework that also allows for model-based prediction uncertainty quantification. We evaluated the application of BART in predicting daily concentrations of four fine particulate matter (PM_2.5) components (elemental carbon, organic carbon, nitrate, and sulfate) in California during the period 2005 to 2014. We demonstrate in this paper how BART can be tuned to optimize prediction performance and how to evaluate variable importance. Our BART models included, as predictors, a large suite of land-use variables, meteorological conditions, satellite-derived aerosol optical depth parameters, and simulations from a chemical transport model. In cross-validation experiments, BART demonstrated good out-of-sample prediction performance at monitoring locations (R² from 0.62 to 0.73). More importantly, prediction intervals associated with concentration estimates from BART showed good coverage probability at locations with and without monitoring data. In our case study, major PM_2.5 components could be estimated with good accuracy, especially when collocated PM_2.5 total mass observations were available. In conclusion, BART is an attractive approach for modeling ambient air pollution levels, especially for its ability to provide uncertainty in estimates that may be useful for subsequent health impact and health effect analyses.

Probabilistic predictive principal component analysis for spatially misaligned and high-dimensional air pollution data with missing observations

Accurate predictions of pollutant concentrations at new locations are often of interest in air pollution studies on fine particulate matters (PM2.5), in which data is usually not measured at all study locations. PM2.5 is also a mixture of many different chemical components. Principal component analysis (PCA) can be incorporated to obtain lower-dimensional representative scores of such multi-pollutant data. Spatial prediction can then be used to estimate these scores at new locations. Recently developed predictive PCA modifies the traditional PCA algorithm to obtain scores with spatial structures that can be well predicted at unmeasured locations. However, these approaches require complete data, whereas multi-pollutant data tends to have complex missing patterns in practice. We propose probabilistic versions of predictive PCA which allow for flexible model-based imputation that can account for spatial information and subsequently improve the overall predictive performance.

Influence of exposure measurement errors on results from epidemiologic studies of different designs

In epidemiologic studies of health effects of air pollution, measurements or models are used to estimate exposure. Exposure estimates have errors that propagate to effect estimates in exposure-response models. We critically evaluate how types of exposure measurement error influenced bias and precision of effect estimates to understand conditions affecting interpretation of exposure-response models for epidemiologic studies of exposure to PM_2.5, NO₂, and SO₂. We reviewed available literature on exposure measurement error for time-series and long-term exposure epidemiology studies. For time-series studies, time-activity error (daily exposure concentration did not account for variation in exposure due to time-activity during a day) and nonambient (indoor) sources negatively biased the effect estimates and increased standard error, so uncertainty grew with increasing bias while underestimating the true health effect in these studies. Spatial error (deviation between true exposure concentration at an individual's location and concentration at a receptor) was ascribed to negatively biased effect estimates in most cases. Positive bias occurred for spatially variable pollutants when the variance of error correlated with the exposure estimate. For long-term exposure studies, most spatial errors did not bias the effect estimate. For both time-series and long-term exposure studies reviewed, large uncertainties were observed when exposure concentration was modeled with low spatial and temporal resolution for a spatially variable pollutant.

Concentrations of criteria pollutants in the contiguous U.S., 1979 - 2015: Role of prediction model parsimony in integrated empirical geographic regression

National-scale empirical models for air pollution can include hundreds of geographic variables. The impact of model parsimony (i.e., how model performance differs for a large versus small number of covariates) has not been systematically explored. We aim to (1) build annual-average integrated empirical geographic (IEG) regression models for the contiguous U.S. for six criteria pollutants during 1979–2015; (2) explore systematically the impact on model performance of the number of variables selected for inclusion in a model; and (3) provide publicly available model predictions. We compute annual-average concentrations from regulatory monitoring data for PM₁₀, PM_2.5, NO₂, SO₂, CO, and ozone at all monitoring sites for 1979–2015. We also use ~350 geographic characteristics at each location including measures of traffic, land use, land cover, and satellite-based estimates of air pollution. We then develop IEG models, employing universal kriging and summary factors estimated by partial least squares (PLS) of geographic variables. For all pollutants and years, we compare three approaches for choosing variables to include in the PLS model: (1) no variables, (2) a limited number of variables selected from the full set by forward selection, and (3) all variables. We evaluate model performance using 10-fold cross-validation (CV) using conventional and spatially-clustered test data. Models using 3 to 30 variables selected from the full set generally have the best performance across all pollutants and years (median R² conventional [clustered] CV: 0.66 [0.47]) compared to models with no (0.37 [0]) or all variables (0.64 [0.27]). Concentration estimates for all Census Blocks reveal generally decreasing concentrations over several decades with local heterogeneity. Our findings suggest that national prediction models can be built by empirically selecting only a small number of important variables to provide robust concentration estimates. Model estimates are freely available online.

Urban Trees Are Sinks for Soot: Elemental Carbon Accumulation by Two Widespread Oak Species

Urban trees could represent important short- and long-term landscape sinks for elemental carbon (EC). Therefore, we quantified foliar EC accumulation by two widespread oak tree species-Quercus stellata (post oak) and Quercus virginiana (live oak)-as well as leaf litterfall EC flux to soil from April 2017 to March 2018 in the City of Denton, Texas, within the Dallas-Fort Worth metropolitan area. Post oak trees accumulated 1.9-fold more EC (299 ± 45 mg EC m^-2 canopy yr^-1) compared to live oak trees (160 ± 31 mg EC m^-2 canopy yr^-1). However, in the fall, these oak species converged in their EC accumulation rates, with ∼70% of annual accumulation occurring during fall and on leaf surfaces. The flux of EC to the ground via leaf litterfall mirrored leaf-fall patterns, with post oaks and live oaks delivering ∼60% of annual leaf litterfall EC in fall and early spring, respectively. We estimate that post oak and live oak trees in this urban ecosystem potentially accumulate 3.5 t EC yr^-1, equivalent to ∼32% of annual vehicular EC emissions from the city. Thus, city trees are significant sinks for EC and represent potential avenues for climate and air quality mitigation in urban areas.

Genome-wide DNA methylation and long-term ambient air pollution exposure in Korean adults

Background

Ambient air pollution is associated with numerous adverse health outcomes, but the underlying mechanisms are not well understood; epigenetic effects including altered DNA methylation could play a role. To evaluate associations of long-term air pollution exposure with DNA methylation in blood, we conducted an epigenome-wide association study in a Korean chronic obstructive pulmonary disease cohort (N = 100 including 60 cases) using Illumina’s Infinium HumanMethylation450K Beadchip. Annual average concentrations of particulate matter ≤ 10 μm in diameter (PM₁₀) and nitrogen dioxide (NO₂) were estimated at participants’ residential addresses using exposure prediction models. We used robust linear regression to identify differentially methylated probes (DMPs) and two different approaches, DMRcate and comb-p, to identify differentially methylated regions (DMRs).

Results

After multiple testing correction (false discovery rate < 0.05), there were 12 DMPs and 27 DMRs associated with PM₁₀ and 45 DMPs and 57 DMRs related to NO₂. DMP cg06992688 (OTUB2) and several DMRs were associated with both exposures. Eleven DMPs in relation to NO₂ confirmed previous findings in Europeans; the remainder were novel. Methylation levels of 39 DMPs were associated with expression levels of nearby genes in a separate dataset of 3075 individuals. Enriched networks were related to outcomes associated with air pollution including cardiovascular and respiratory diseases as well as inflammatory and immune responses.

Conclusions

This study provides evidence that long-term ambient air pollution exposure impacts DNA methylation. The differential methylation signals can serve as potential air pollution biomarkers. These results may help better understand the influences of ambient air pollution on human health.

Electronic supplementary material

The online version of this article (10.1186/s13148-019-0635-z) contains supplementary material, which is available to authorized users.

Adaptive predictive principal components for modeling multivariate air pollution

Air pollution monitoring locations are typically spatially misaligned with locations of participants in a cohort study, so to analyze pollution-health associations, exposures must be predicted at subject locations. For a pollution measure like PM_2.5 (fine particulate matter) comprised of multiple chemical components, the predictive principal component analysis (PCA) algorithm derives a low-dimensional representation of component profiles for use in health analyses. Geographic covariates and spatial splines help determine the principal component loadings of the pollution data to give improved prediction accuracy of the principal component scores. While predictive PCA can accommodate pollution data of arbitrary dimension, it is currently limited to a small number of pre-selected geographic covariates. We propose an adaptive predictive PCA algorithm, which automatically identifies a combination of covariates that is most informative in choosing the principal component directions in the pollutant space. We show that adaptive predictive PCA improves the accuracy of multi-pollutant exposure predictions at subject locations.

Space-time trends of PM2.5 constituents in the conterminous United States estimated by a machine learning approach, 2005-2015

Particulate matter with aerodynamic diameter less than 2.5 μm (PM_2.5) is a complex mixture of chemical constituents emitted from various emission sources or through secondary reactions/processes; however, PM_2.5 is regulated mostly based on its total mass concentration. Studies to identify the impacts on climate change, visibility degradation and public health of different PM_2.5 constituents are hindered by limited ground measurements of PM_2.5 constituents. In this study, national models were developed based on random forest algorithm, one of machine learning methods that is of high predictive capacity and able to provide interpretable results, to predict concentrations of PM_2.5 sulfate, nitrate, organic carbon (OC) and elemental carbon (EC) across the conterminous United States from 2005 to 2015 at the daily level. The random forest models achieved high out-of-bag (OOB) R² values at the daily level, and the mean OOB R² values were 0.86, 0.82, 0.71 and 0.75 for sulfate, nitrate, OC and EC, respectively, over 2005-2015. The long-term temporal trends of PM_2.5 sulfate, nitrate, OC and EC predictions agreed well with their corresponding ground measurements. The annual mean of predicted PM_2.5 sulfate and EC levels across the conterminous United States decreased substantially from 2005 to 2015; while concentrations of predicted PM_2.5 nitrate and OC decreased and fluctuated during the study period. The annual prediction maps captured the characterized spatial patterns of the PM_2.5 constituents. The distributions of annual mean concentrations of sulfate and nitrate were generally regional in the extent that sulfate decreased from east to west smoothly with enhancement in California and nitrate had higher concentration in Midwest, Metro New York area, and California. OC and EC had regional high concentrations in the Southeast and Northwest as well as localized high levels around urban centers. The spatial patterns of PM_2.5 constituents were consistent with the distributions of their emission sources and secondary processes and transportation. Hence, the national models developed in this study could provide supplementary evaluations of spatio-temporal distributions of PM_2.5 constituents with full time-space coverages in the conterminous United States, which could be beneficial to assess the impacts of PM_2.5 constituents on radiation budgets and visibility degradation, and support exposure assessment for regional to national health studies at county or city levels to understand the acute and chronic toxicity and health impacts of PM_2.5 constituents, and consequently provide scientific evidence for making targeted and effective regulations of PM_2.5 pollution.

Pollutant composition modification of the effect of air pollution on progression of coronary artery calcium: the Multi-Ethnic Study of Atherosclerosis

Background:

Differences in traffic-related air pollution (TRAP) composition may cause heterogeneity in associations between air pollution exposure and cardiovascular health outcomes. Clustering multipollutant measurements allows investigation of effect modification by TRAP profiles.

Methods:

We measured TRAP components with fixed-site and on-road instruments for two 2-week periods in Baltimore, Maryland. We created representative TRAP profiles for cold and warm seasons using predictive k-means clustering. We predicted cluster membership for 1005 participants in the Multi-Ethnic Study of Atherosclerosis and Air Pollution with follow-up between 2000 and 2012. We estimated cluster-specific relationships between coronary artery calcification (CAC) progression and long-term exposure to fine particulate matter (PM_2.5) and oxides of nitrogen (NO_X).

Results:

We identified two clusters in the cold season, notable for higher ratios of gases and ultrafine particles, respectively. A 5-μg/m³ difference in PM_2.5 was associated with 17.0 (95% confidence interval [CI] = 7.2, 26.7) and 42.6 (95% CI = 25.7, 59.4) Agatston units/year CAC progression among participants in clusters 1 and 2, respectively (effect modification P = 0.006). A 40 ppb difference in NO_X was associated with 22.2 (95% CI = 7.7, 36.7) and 41.9 (95% CI = 23.7, 60.2) Agatston units/year CAC progression in clusters 1 and 2, respectively (P = 0.08). Similar trends occurred using clusters identified from warm season measurements. Clusters correlated highly with baseline pollution level.

Conclusions:

Clustering TRAP measurements identified spatial differences in composition. We found evidence of greater CAC progression rates per unit PM_2.5 exposures among people living in areas characterized by high ratios of ultrafine particle counts relative to NO_X concentrations.

Long-Term Coarse Particulate Matter Exposure Is Associated with Asthma among Children in Medicaid

RATIONALE

Short- and long-term fine particulate matter (particulate matter ≤2.5 μm in aerodynamic diameter [PM2.5]) pollution is associated with asthma development and morbidity, but there are few data on the effects of long-term exposure to coarse PM (PM10-2.5) on respiratory health.

OBJECTIVES

To understand the relationship between long-term fine and coarse PM exposure and asthma prevalence and morbidity among children.

METHODS

A semiparametric regression model that incorporated PM2.5 and PM10 monitor data and geographic characteristics was developed to predict 2-year average PM2.5 and PM10-2.5 exposure during the period 2009 to 2010 at the zip-code tabulation area level. Data from 7,810,025 children aged 5 to 20 years enrolled in Medicaid from 2009 to 2010 were used in a log-linear regression model with predicted PM levels to estimate the association between PM exposure and asthma prevalence and morbidity, adjusting for race/ethnicity, sex, age, area-level urbanicity, poverty, education, and unmeasured spatial confounding.

MEASUREMENTS AND MAIN RESULTS

Exposure to coarse PM was associated with increased asthma diagnosis prevalence (rate ratio [RR] for 1-μg/m3 increase in coarse PM level, 1.006; 95% confidence interval [CI], 1.001-1.011), hospitalizations (RR, 1.023; 95% CI, 1.003-1.042), and emergency department visits (RR, 1.017; 95% CI, 1.001-1.033) when adjusting for fine PM. Fine PM exposure was more strongly associated with increased asthma prevalence and morbidity than coarse PM. The estimates remained elevated across different levels of spatial confounding adjustment.

CONCLUSIONS

Among children enrolled in Medicaid, exposure to higher average coarse PM levels is associated with increased asthma prevalence and morbidity. These results suggest the need for direct monitoring of coarse PM and reconsideration of limits on long-term average coarse PM pollution levels.

Measurement Error Correction for Predicted Spatiotemporal Air Pollution Exposures

BACKGROUND

Air pollution cohort studies are frequently analyzed in two stages, first modeling exposure then using predicted exposures to estimate health effects in a second regression model. The difference between predicted and unobserved true exposures introduces a form of measurement error in the second stage health model. Recent methods for spatial data correct for measurement error with a bootstrap and by requiring the study design ensure spatial compatibility, that is, monitor and subject locations are drawn from the same spatial distribution. These methods have not previously been applied to spatiotemporal exposure data.

METHODS

We analyzed the association between fine particulate matter (PM2.5) and birth weight in the US state of Georgia using records with estimated date of conception during 2002-2005 (n = 403,881). We predicted trimester-specific PM2.5 exposure using a complex spatiotemporal exposure model. To improve spatial compatibility, we restricted to mothers residing in counties with a PM2.5 monitor (n = 180,440). We accounted for additional measurement error via a nonparametric bootstrap.

RESULTS

Third trimester PM2.5 exposure was associated with lower birth weight in the uncorrected (-2.4 g per 1 μg/m difference in exposure; 95% confidence interval [CI]: -3.9, -0.8) and bootstrap-corrected (-2.5 g, 95% CI: -4.2, -0.8) analyses. Results for the unrestricted analysis were attenuated (-0.66 g, 95% CI: -1.7, 0.35).

CONCLUSIONS

This study presents a novel application of measurement error correction for spatiotemporal air pollution exposures. Our results demonstrate the importance of spatial compatibility between monitor and subject locations and provide evidence of the association between air pollution exposure and birth weight.

Incorporating Measurement Error from Modeled Air Pollution Exposures into Epidemiological Analyses

PURPOSE OF REVIEW

Outdoor air pollution exposures used in epidemiological studies are commonly predicted from spatiotemporal models incorporating limited measurements, temporal factors, geographic information system variables, and/or satellite data. Measurement error in these exposure estimates leads to imprecise estimation of health effects and their standard errors. We reviewed methods for measurement error correction that have been applied in epidemiological studies that use model-derived air pollution data.

RECENT FINDINGS

We identified seven cohort studies and one panel study that have employed measurement error correction methods. These methods included regression calibration, risk set regression calibration, regression calibration with instrumental variables, the simulation extrapolation approach (SIMEX), and methods under the non-parametric or parameter bootstrap. Corrections resulted in small increases in the absolute magnitude of the health effect estimate and its standard error under most scenarios. Limited application of measurement error correction methods in air pollution studies may be attributed to the absence of exposure validation data and the methodological complexity of the proposed methods. Future epidemiological studies should consider in their design phase the requirements for the measurement error correction method to be later applied, while methodological advances are needed under the multi-pollutants setting.

Methods for Assessing Long-Term Exposures to Outdoor Air Pollutants

Purpose of Review

Epidemiological studies of health effects of long-term exposure to outdoor air pollution rely on different exposure assessment methods. This review discusses widely used methods with a special focus on new developments.

Recent Findings

New data and study designs have been applied, including satellite measurements of fine particles and nitrogen dioxide (NO₂). The methods to apply satellite data for epidemiological studies are improving rapidly and have already contributed significantly to national-, continental- and global-scale models. Spatiotemporal models have been developed allowing more detailed temporal resolution compared to spatial models. The development of hybrid models combining dispersion models, satellite observations, land use and surface monitoring has improved models substantially. Mobile monitoring designs to develop models for long-term UFP exposure have been conducted.

Summary

Methods to assess long-term exposure to outdoor air pollution have improved significantly over the past decade. Application of satellite data and mobile monitoring designs is promising new methods.

Accommodating the ecological fallacy in disease mapping in the absence of individual exposures

In health exposure modeling, in particular, disease mapping, the ecological fallacy arises because the relationship between aggregated disease incidence on areal units and average exposure on those units differs from the relationship between the event of individual incidence and the associated individual exposure. This article presents a novel modeling approach to address the ecological fallacy in the least informative data setting. We assume the known population at risk with an observed incidence for a collection of areal units and, separately, environmental exposure recorded during the period of incidence at a collection of monitoring stations. We do not assume any partial individual level information or random allocation of individuals to observed exposures. We specify a conceptual incidence surface over the study region as a function of an exposure surface resulting in a stochastic integral of the block average disease incidence. The true block level incidence is an unavailable Monte Carlo integration for this stochastic integral. We propose an alternative manageable Monte Carlo integration for the integral. Modeling in this setting is immediately hierarchical, and we fit our model within a Bayesian framework. To alleviate the resulting computational burden, we offer 2 strategies for efficient model fitting: one is through modularization, the other is through sparse or dimension-reduced Gaussian processes. We illustrate the performance of our model with simulations based on a heat-related mortality dataset in Ohio and then analyze associated real data.

Design of an Air Pollution Monitoring Campaign in Beijing for Application to Cohort Health Studies

No cohort studies in China on the health effects of long-term air pollution exposure have employed exposure estimates at the fine spatial scales desirable for cohort studies with individual-level health outcome data. Here we assess an array of modern air pollution exposure estimation approaches for assigning within-city exposure estimates in Beijing for individual pollutants and pollutant sources to individual members of a cohort. Issues considered in selecting specific monitoring data or new monitoring campaigns include: needed spatial resolution, exposure measurement error and its impact on health effect estimates, spatial alignment and compatibility with the cohort, and feasibility and expense. Sources of existing data largely include administrative monitoring data, predictions from air dispersion or chemical transport models and remote sensing (specifically satellite) data. New air monitoring campaigns include additional fixed site monitoring, snapshot monitoring, passive badge or micro-sensor saturation monitoring and mobile monitoring, as well as combinations of these. Each of these has relative advantages and disadvantages. It is concluded that a campaign in Beijing that at least includes a mobile monitoring component, when coupled with currently available spatio-temporal modeling methods, should be strongly considered. Such a campaign is economical and capable of providing the desired fine-scale spatial resolution for pollutants and sources.

Air pollution and subclinical interstitial lung disease: the Multi-Ethnic Study of Atherosclerosis (MESA) air-lung study

We studied whether ambient air pollution is associated with interstitial lung abnormalities (ILAs) and high attenuation areas (HAAs), which are qualitative and quantitative measurements of subclinical interstitial lung disease (ILD) on computed tomography (CT).We performed analyses of community-based dwellers enrolled in the Multi-Ethnic Study of Atherosclerosis (MESA) study. We used cohort-specific spatio-temporal models to estimate ambient pollution (fine particulate matter (PM_2.5), nitrogen oxides (NO_x), nitrogen dioxide (NO₂) and ozone (O₃)) at each home. A total of 5495 participants underwent serial assessment of HAAs by cardiac CT; 2671 participants were assessed for ILAs using full lung CT at the 10-year follow-up. We used multivariable logistic regression and linear mixed models adjusted for age, sex, ethnicity, education, tobacco use, scanner technology and study site.The odds of ILAs increased 1.77-fold per 40 ppb increment in NO_x (95% CI 1.06 to 2.95, p = 0.03). There was an overall trend towards an association between higher exposure to NO_x and greater progression of HAAs (0.45% annual increase in HAAs per 40 ppb increment in NO_x; 95% CI -0.02 to 0.92, p = 0.06). Associations of ambient fine particulate matter (PM_2.5), NO_x and NO₂ concentrations with progression of HAAs varied by race/ethnicity (p = 0.002, 0.007, 0.04, respectively, for interaction) and were strongest among non-Hispanic white people.We conclude that ambient air pollution exposures were associated with subclinical ILD.

Improving modeled air pollution concentration maps by residual interpolation

Models that are used to map air pollutant concentrations are not free of errors. A possible approach for improving the final concentration map is to interpolate the residuals of the initial model concentration estimates. Due to the possible spatial autocorrelation of the residuals of the initial model estimates, Bayesian inference schemes were suggested for this task, since they can correctly adjust the level of fitting of the residuals to the random measurement errors. However, the complexity of Bayesian methods often discourages their use. Here, we present an alternative and simpler approach, using a leave-one-out cross-validation to determine the optimal level of fitting of the residual correction. We show that the optimal correction level is related to the extent of the spatial autocorrelation of the cross-validated residuals. Namely, when the residuals are not autocorrelated residual correction is unnecessary, and if employed may actually degrade the quality of the final concentration map. Moreover, our approach enables to optimize the residual correction based on different target performance measures, with a possibly different optimal correction depending on the performance measure used. Hence, different target performance measures can be chosen to fit best the specific application of interest. The method is demonstrated using output of three different models used for estimating NO_x and NO₂ concentrations over Israel. We show that our approach can be used as an exploratory step, for assessing the potential benefit of residual correction, and as a simple alternative to Bayesian schemes.

Long-term exposure to residential ambient fine and coarse particulate matter and incident hypertension in post-menopausal women

BACKGROUND

Long-term exposure to ambient particulate matter (PM) has been previously linked with higher risk of cardiovascular events. This association may be mediated, at least partly, by increasing the risk of incident hypertension, a key determinant of cardiovascular risk. However, whether long-term exposure to PM is associated with incident hypertension remains unclear.

METHODS

Using national geostatistical models incorporating geographic covariates and spatial smoothing, we estimated annual average concentrations of residential fine (PM2.5), respirable (PM10), and course (PM10-2.5) fractions of particulate matter among 44,255 post-menopausal women free of hypertension enrolled in the Women's Health Initiative (WHI) clinical trials. We used time-varying Cox proportional hazards models to evaluate the association between long-term average residential pollutant concentrations and incident hypertension, adjusting for potential confounding by sociodemographic factors, medical history, neighborhood socioeconomic measures, WHI study clinical site, clinical trial, and randomization arm.

RESULTS

During 298,383 person-years of follow-up, 14,511 participants developed incident hypertension. The adjusted hazard ratios per interquartile range (IQR) increase in PM2.5, PM10, and PM10-2.5 were 1.13 (95% CI: 1.08, 1.17), 1.06 (1.03, 1.10), and 1.01 (95% CI: 0.97, 1.04), respectively. Statistically significant concentration-response relationships were identified for PM2.5 and PM10 fractions. The association between PM2.5 and hypertension was more pronounced among non-white participants and those residing in the Northeastern United States.

CONCLUSIONS

In this cohort of post-menopausal women, ambient fine and respirable particulate matter exposures were associated with higher incidence rates of hypertension. These results suggest that particulate matter may be an important modifiable risk factor for hypertension.

Long-term exposure to residential ambient fine and coarse particulate matter and incident hypertension in post-menopausal women

BACKGROUND

Long-term exposure to ambient particulate matter (PM) has been previously linked with higher risk of cardiovascular events. This association may be mediated, at least partly, by increasing the risk of incident hypertension, a key determinant of cardiovascular risk. However, whether long-term exposure to PM is associated with incident hypertension remains unclear.

METHODS

Using national geostatistical models incorporating geographic covariates and spatial smoothing, we estimated annual average concentrations of residential fine (PM_2.5), respirable (PM₁₀), and course (PM_10-2.5) fractions of particulate matter among 44,255 post-menopausal women free of hypertension enrolled in the Women's Health Initiative (WHI) clinical trials. We used time-varying Cox proportional hazards models to evaluate the association between long-term average residential pollutant concentrations and incident hypertension, adjusting for potential confounding by sociodemographic factors, medical history, neighborhood socioeconomic measures, WHI study clinical site, clinical trial, and randomization arm.

RESULTS

During 298,383 person-years of follow-up, 14,511 participants developed incident hypertension. The adjusted hazard ratios per interquartile range (IQR) increase in PM_2.5, PM₁₀, and PM_10-2.5 were 1.13 (95% CI: 1.08, 1.17), 1.06 (1.03, 1.10), and 1.01 (95% CI: 0.97, 1.04), respectively. Statistically significant concentration-response relationships were identified for PM_2.5 and PM₁₀ fractions. The association between PM_2.5 and hypertension was more pronounced among non-white participants and those residing in the Northeastern United States.

CONCLUSIONS

In this cohort of post-menopausal women, ambient fine and respirable particulate matter exposures were associated with higher incidence rates of hypertension. These results suggest that particulate matter may be an important modifiable risk factor for hypertension.

National-scale exposure prediction for long-term concentrations of particulate matter and nitrogen dioxide in South Korea

The limited spatial coverage of the air pollution data available from regulatory air quality monitoring networks hampers national-scale epidemiological studies of air pollution. The present study aimed to develop a national-scale exposure prediction model for estimating annual average concentrations of PM₁₀ and NO₂ at residences in South Korea using regulatory monitoring data for 2010. Using hourly measurements of PM₁₀ and NO₂ at 277 regulatory monitoring sites, we calculated the annual average concentrations at each site. We also computed 322 geographic variables in order to represent plausible local and regional pollution sources. Using these data, we developed universal kriging models, including three summary predictors estimated by partial least squares (PLS). The model performance was evaluated with fivefold cross-validation. In sensitivity analyses, we compared our approach with two alternative approaches, which added regional interactions and replaced the PLS predictors with up to ten selected variables. Finally, we predicted the annual average concentrations of PM₁₀ and NO₂ at 83,463 centroids of residential census output areas in South Korea to investigate the population exposure to these pollutants and to compare the exposure levels between monitored and unmonitored areas. The means of the annual average concentrations of PM₁₀ and NO₂ for 2010, across regulatory monitoring sites in South Korea, were 51.63 μg/m3 (SD = 8.58) and 25.64 ppb (11.05), respectively. The universal kriging exposure prediction models yielded cross-validated R²s of 0.45 and 0.82 for PM₁₀ and NO₂, respectively. Compared to our model, the two alternative approaches gave consistent or worse performances. Population exposure levels in unmonitored areas were lower than in monitored areas. This is the first study that focused on developing a national-scale point wise exposure prediction approach in South Korea, which will allow national exposure assessments and epidemiological research to answer policy-related questions and to draw comparisons among different countries.

Use of mobile and passive badge air monitoring data for NOX and ozone air pollution spatial exposure prediction models

Air pollution exposure prediction models can make use of many types of air monitoring data. Fixed location passive samples typically measure concentrations averaged over several days to weeks. Mobile monitoring data can generate near continuous concentration measurements. It is not known whether mobile monitoring data are suitable for generating well-performing exposure prediction models or how they compare with other types of monitoring data in generating exposure models. Measurements from fixed site passive samplers and mobile monitoring platform were made over a 2-week period in Baltimore in the summer and winter months in 2012. Performance of exposure prediction models for long-term nitrogen oxides (NO_X) and ozone (O₃) concentrations were compared using a state-of-the-art approach for model development based on land use regression (LUR) and geostatistical smoothing. Model performance was evaluated using leave-one-out cross-validation (LOOCV). Models performed well using the mobile peak traffic monitoring data for both NO_X and O₃, with LOOCV R²s of 0.70 and 0.71, respectively, in the summer, and 0.90 and 0.58, respectively, in the winter. Models using 2-week passive samples for NO_X had LOOCV R²s of 0.60 and 0.65 in the summer and winter months, respectively. The passive badge sampling data were not adequate for developing models for O₃. Mobile air monitoring data can be used to successfully build well-performing LUR exposure prediction models for NO_X and O₃ and are a better source of data for these models than 2-week passive badge data.

Multipollutant measurement error in air pollution epidemiology studies arising from predicting exposures with penalized regression splines

Air pollution epidemiology studies are trending towards a multi-pollutant approach. In these studies, exposures at subject locations are unobserved and must be predicted using observed exposures at misaligned monitoring locations. This induces measurement error, which can bias the estimated health effects and affect standard error estimates. We characterize this measurement error and develop an analytic bias correction when using penalized regression splines to predict exposure. Our simulations show bias from multi-pollutant measurement error can be severe, and in opposite directions or simultaneously positive or negative. Our analytic bias correction combined with a non-parametric bootstrap yields accurate coverage of 95% confidence intervals. We apply our methodology to analyze the association of systolic blood pressure with PM2.5 and NO2 in the NIEHS Sister Study. We find that NO2 confounds the association of systolic blood pressure with PM2.5 and vice versa. Elevated systolic blood pressure was significantly associated with increased PM2.5 and decreased NO2. Correcting for measurement error bias strengthened these associations and widened 95% confidence intervals.

A New Technique for Evaluating Land-use Regression Models and Their Impact on Health Effect Estimates

BACKGROUND

Leave-one-out cross-validation that fails to account for variable selection does not properly reflect prediction accuracy when the number of training sites is small. The impact on health effect estimates has rarely been studied. The objective of this study was to develop an improved validation procedure for land-use regression models with variable selection and investigate health effect estimates in relation to land-use regression model performance.

METHODS

We randomly generated 10 training and test sets for nitrogen dioxide and particulate matter. For each training set, we developed models and evaluated them using a cross-holdout validation approach. Cross-holdout validation develops new models for each evaluation compared with refitting the model without variable selection, as in standard leave-one-out cross-validation. We also implemented holdout validation, which evaluates model predictions using independent test sets. We evaluated the relationship between cross-holdout validation and holdout validation R and estimates of the association between air pollution and forced vital capacity in the Dutch birth cohort.

RESULTS

Cross-holdout validation Rs were generally identical to holdout validation Rs, but were notably smaller than leave-one-out cross-validation Rs. Decreases in forced vital capacity in relation to air pollution exposure were larger for land-use regression models that had larger holdout validation and cross-holdout validation Rs rather than leave-one-out cross-validation R.

CONCLUSION

Cross-holdout validation accurately reflects predictive ability of land-use regression models and is a useful validation approach for small datasets. Land-use regression predictive ability in terms of holdout validation and cross-holdout validation rather than leave-one-out cross-validation was associated with the magnitude of health effect estimates in a case study.

Advances in Understanding Air Pollution and CVD

The MESA Air (Multi-Ethnic Study of Atherosclerosis and Air Pollution) leveraged the platform of the MESA cohort into a prospective longitudinal study of relationships between air pollution and cardiovascular health. MESA Air researchers developed fine-scale, state-of-the-art air pollution exposure models for the MESA Air communities, creating individual exposure estimates for each participant. These models combine cohort-specific exposure monitoring, existing monitoring systems, and an extensive database of geographic and meteorological information. Together with extensive phenotyping in MESA-and adding participants and health measurements to the cohort-MESA Air investigated environmental exposures on a wide range of outcomes. Advances by the MESA Air team included not only a new approach to exposure modeling, but also biostatistical advances in addressing exposure measurement error and temporal confounding. The MESA Air study advanced our understanding of the impact of air pollutants on cardiovascular disease and provided a research platform for advances in environmental epidemiology.

Prediction of fine particulate matter chemical components with a spatio-temporal model for the Multi-Ethnic Study of Atherosclerosis cohort

Although cohort studies of the health effects of PM2.5 have developed exposure prediction models to represent spatial variability across participant residences, few models exist for PM2.5 components. We aimed to develop a city-specific spatio-temporal prediction approach to estimate long-term average concentrations of four PM2.5 components including sulfur, silicon, and elemental and organic carbon for the Multi-Ethnic Study of Atherosclerosis cohort, and to compare predictions to those from a national spatial model. Using 2-week average measurements from a cohort-focused monitoring campaign, the spatio-temporal model employed selected geographic covariates in a universal kriging framework with the data-driven temporal trend. Relying on long-term means of daily measurements from regulatory monitoring networks, the national spatial model employed dimension-reduced predictors using universal kriging. For the spatio-temporal model, the cross-validated and temporally-adjusted R(2) was relatively higher for EC and OC, and in the Los Angeles and Baltimore areas. The cross-validated R(2)s for both models across the six areas were reasonably high for all components except silicon. Predicted long-term concentrations at participant homes from the two models were generally highly correlated across cities but poorly correlated within cities. The spatio-temporal model may be preferred for city-specific health analyses, whereas both models could be used for multi-city studies.

Spatial measurement errors in the field of spatial epidemiology

Background

Spatial epidemiology has been aided by advances in geographic information systems, remote sensing, global positioning systems and the development of new statistical methodologies specifically designed for such data. Given the growing popularity of these studies, we sought to review and analyze the types of spatial measurement errors commonly encountered during spatial epidemiological analysis of spatial data.

Methods

Google Scholar, Medline, and Scopus databases were searched using a broad set of terms for papers indexed by a term indicating location (space or geography or location or position) and measurement error (measurement error or measurement inaccuracy or misclassification or uncertainty): we reviewed all papers appearing before December 20, 2014. These papers and their citations were reviewed to identify the relevance to our review.

Results

We were able to define and classify spatial measurement errors into four groups: (1) pure spatial location measurement errors, including both non-instrumental errors (multiple addresses, geocoding errors, outcome aggregations, and covariate aggregation) and instrumental errors; (2) location-based outcome measurement error (purely outcome measurement errors and missing outcome measurements); (3) location-based covariate measurement errors (address proxies); and (4) Covariate-Outcome spatial misaligned measurement errors. We propose how these four classes of errors can be unified within an integrated theoretical model and possible solutions were discussed.

Conclusion

Spatial measurement errors are ubiquitous threat to the validity of spatial epidemiological studies. We propose a systematic framework for understanding the various mechanisms which generate spatial measurement errors and present practical examples of such errors.

Current Methods and Challenges for Epidemiological Studies of the Associations Between Chemical Constituents of Particulate Matter and Health

Epidemiological studies have been critical for estimating associations between exposure to ambient particulate matter (PM) air pollution and adverse health outcomes. Because total PM mass is a temporally and spatially varying mixture of constituents with different physical and chemical properties, recent epidemiological studies have focused on PM constituents. Most studies have estimated associations between PM constituents and health using the same statistical methods as in studies of PM mass. However, these approaches may not be sufficient to address challenges specific to studies of PM constituents, namely assigning exposure, disentangling health effects, and handling measurement error. We reviewed large, population-based epidemiological studies of PM constituents and health and describe the statistical methods typically applied to address these challenges. Development of statistical methods that simultaneously address multiple challenges, for example, both disentangling health effects and handling measurement error, could improve estimation of associations between PM constituents and adverse health outcomes.

A novel principal component analysis for spatially misaligned multivariate air pollution data

We propose novel methods for predictive (sparse) PCA with spatially misaligned data. These methods identify principal component loading vectors that explain as much variability in the observed data as possible, while also ensuring the corresponding principal component scores can be predicted accurately by means of spatial statistics at locations where air pollution measurements are not available. This will make it possible to identify important mixtures of air pollutants and to quantify their health effects in cohort studies, where currently available methods cannot be used. We demonstrate the utility of predictive (sparse) PCA in simulated data and apply the approach to annual averages of particulate matter speciation data from national Environmental Protection Agency (EPA) regulatory monitors.

Spatial measurement error and correction by spatial SIMEX in linear regression models when using predicted air pollution exposures

Spatial modeling of air pollution exposures is widespread in air pollution epidemiology research as a way to improve exposure assessment. However, there are key sources of exposure model uncertainty when air pollution is modeled, including estimation error and model misspecification. We examine the use of predicted air pollution levels in linear health effect models under a measurement error framework. For the prediction of air pollution exposures, we consider a universal Kriging framework, which may include land-use regression terms in the mean function and a spatial covariance structure for the residuals. We derive the bias induced by estimation error and by model misspecification in the exposure model, and we find that a misspecified exposure model can induce asymptotic bias in the effect estimate of air pollution on health. We propose a new spatial simulation extrapolation (SIMEX) procedure, and we demonstrate that the procedure has good performance in correcting this asymptotic bias. We illustrate spatial SIMEX in a study of air pollution and birthweight in Massachusetts.

Exposure to sub-chronic and long-term particulate air pollution and heart rate variability in an elderly cohort: the Normative Aging Study

BACKGROUND

Short-term particulate air pollution exposure is associated with reduced heart rate variability (HRV), a risk factor for cardiovascular morbidity and mortality, in many studies. Associations with sub-chronic or long-term exposures, however, have been sparsely investigated. We evaluated the effect of fine particulate matter (PM2.5) and black carbon (BC) exposures on HRV in an elderly cohort: the Normative Aging Study.

METHODS

We measured power in high frequency (HF) and low frequency (LF), standard deviation of normal-to-normal intervals (SDNN), and the LF:HF ratio among participants from the Greater Boston area. Residential BC exposures for 540 men (1161 study visits, 2000-2011) were estimated using a spatio-temporal land use regression model, and residential PM2.5 exposures for 475 men (992 visits, 2003-2011) were modeled using a hybrid satellite based and land-use model. We evaluated associations between moving averages of sub-chronic (3-84 day) and long-term (1 year) pollutant exposure estimates and HRV parameters using linear mixed models.

RESULTS

One-standard deviation increases in sub-chronic, but not long-term, BC were associated with reduced HF, LF, and SDNN and an increased LF:HF ratio (e.g., 28 day BC: -2.3% HF [95% CI:-4.6, -0.02]). Sub-chronic and long-term PM2.5 showed evidence of relations to an increased LF and LF:HF ratio (e.g., 1 year PM: 21.0% LF:HF [8.6, 34.8]), but not to HF or SDNN, though the effect estimates were very imprecise and mostly spanned the null.

CONCLUSIONS

We observed some evidence of a relation between longer-term BC and PM2.5 exposures and changes in HRV in an elderly cohort. While previous studies focused on short-term air pollution exposures, our results suggest that longer-term exposures may influence cardiac autonomic function.

Using High-Resolution Satellite Aerosol Optical Depth To Estimate Daily PM2.5 Geographical Distribution in Mexico City

Recent advances in estimating fine particle (PM2.5) ambient concentrations use daily satellite measurements of aerosol optical depth (AOD) for spatially and temporally resolved exposure estimates. Mexico City is a dense megacity that differs from other previously modeled regions in several ways: it has bright land surfaces, a distinctive climatological cycle, and an elevated semi-enclosed air basin with a unique planetary boundary layer dynamic. We extend our previous satellite methodology to the Mexico City area, a region with higher PM2.5 than most U.S. and European urban areas. Using a novel 1 km resolution AOD product from the MODIS instrument, we constructed daily predictions across the greater Mexico City area for 2004-2014. We calibrated the association of AOD to PM2.5 daily using municipal ground monitors, land use, and meteorological features. Predictions used spatial and temporal smoothing to estimate AOD when satellite data were missing. Our model performed well, resulting in an out-of-sample cross-validation R(2) of 0.724. Cross-validated root-mean-squared prediction error (RMSPE) of the model was 5.55 μg/m(3). This novel model reconstructs long- and short-term spatially resolved exposure to PM2.5 for epidemiological studies in Mexico City.