Modeling individual exposures to ambient PM2.5 in the diabetes and the environment panel study (DEPS)

K-means cluster analysis of cooperative effects of CO, NO2, O3, PM2.5, PM10, and SO2 on incidence of type 2 diabetes mellitus in the US

Air pollution (AP) has been shown to increase the risk of type 2 diabetes mellitus, as well as other cardiometabolic diseases. AP is characterized by a complex mixture of components for which the composition depends on sources and metrological factors. The US Environmental Protection Agency (EPA) monitors and regulates certain components of air pollution known to have negative consequences for human health. Research assessing the health effects of these components of AP often uses traditional regression models, which might not capture more complex and interdependent relationships. Machine learning has the capability to simultaneously assess multiple components and find complex, non-linear patterns that may not be apparent and could not be modeled by other techniques. Here we use k-means clustering to assess the patterns associating PM2.5, PM10, CO, NO2, O3, and SO2 measurements and changes in annual diabetes incidence at a US county level. The average age adjusted annual decrease in diabetes incidence for the entire US populations is -0.25 per 1000 but the change shows a significant geographic variation (range: -17.2 to 5.30 per 1000). In this paper these variations were compared with the local daily AP concentrations of the pollutants listed above from 2005 to 2015, which were matched to the annual change in diabetes incidence for the following year. A total of 134,925 daily air quality observations were included in the cluster analysis, representing 125 US counties and the District of Columbia. K-means successfully clustered AP components and indicated an association between exposure to certain AP mixtures with lower decreases on T2D incidence.

A review and analysis of personal and ambient PM2.5 measurements: Implications for epidemiology studies

BACKGROUND

In epidemiology studies, ambient measurements of PM_2.5 are often used as surrogates for personal exposures. However, it is unclear the degree to which ambient PM_2.5 reflects personal exposures.

OBJECTIVE

In order to examine potential sources of bias in epidemiology studies, we conducted a review and meta-analysis of studies to determine the extent to which short-term measurements of ambient PM_2.5 levels are related to short-term measurements of personal PM_2.5 levels.

METHODS

We conducted a literature search of studies reporting both personal and ambient measurements of PM_2.5 published in the last 10 years (2009-2019) and incorporated studies published prior to 2009 from reviews.

RESULTS

Seventy-one studies were identified. Based on 17 studies reporting slopes, a meta-analysis revealed an overall slope of 0.56 μg/m³ (95% CI: [0.39, 0.73]) personal PM_2.5 per μg/m³ increase in ambient PM_2.5. Slopes for summer months were higher (slope = 0.73, 95% CI: [0.64, 0.81]) than for winter (slope = 0.46, 95% CI: [0.36, 0.57]). Based on 44 studies reporting correlations, we calculated an overall personal-ambient PM_2.5 correlation of 0.63 (95% CI: [0.55, 0.71]). Correlations were stronger in studies conducted in Canada (r = 0.86, 95% CI: [0.67, 0.94]) compared to the USA (r = 0.60, 95% CI: [0.49, 0.70]) and China (r = 0.60, 95% CI: [0.46, 0.71]). Correlations also were stronger in urban areas (r = 0.53, 95% CI: [0.43, 0.62]) vs. suburban areas (r = 0.36, 95% CI: [0.21, 0.49]).

SIGNIFICANCE

Our results suggest a large degree of variability in the personal-ambient PM_2.5 association and the potential for exposure misclassification and measurement error in PM_2.5 epidemiology studies.

Integrating Personal Air Sensor and GPS to Determine Microenvironment-Specific Exposures to Volatile Organic Compounds

Personal exposure to volatile organic compounds (VOCs) from indoor sources including consumer products is an understudied public health concern. To develop and evaluate methods for monitoring personal VOC exposures, we performed a pilot study and examined time-resolved sensor-based measurements of geocoded total VOC (TVOC) exposures across individuals and microenvironments (MEs). We integrated continuous (1 min) data from a personal TVOC sensor and a global positioning system (GPS) logger, with a GPS-based ME classification model, to determine TVOC exposures in four MEs, including indoors at home (Home-In), indoors at other buildings (Other-In), inside vehicles (In-Vehicle), and outdoors (Out), across 45 participant-days for five participants. To help identify places with large emission sources, we identified high-exposure events (HEEs; TVOC > 500 ppb) using geocoded TVOC time-course data overlaid on Google Earth maps. Across the 45 participant-days, the MEs ranked from highest to lowest median TVOC were: Home-In (165 ppb), Other-In (86 ppb), In-Vehicle (52 ppb), and Out (46 ppb). For the two participants living in single-family houses with attached garages, the median exposures for Home-In were substantially higher (209, 416 ppb) than the three participant homes without attached garages: one living in a single-family house (129 ppb), and two living in apartments (38, 60 ppb). The daily average Home-In exposures exceeded the estimated Leadership in Energy and Environmental Design (LEED) building guideline of 108 ppb for 60% of the participant-days. We identified 94 HEEs across all participant-days, and 67% of the corresponding peak levels exceeded 1000 ppb. The MEs ranked from the highest to the lowest number of HEEs were: Home-In (60), Other-In (13), In-Vehicle (12), and Out (9). For Other-In and Out, most HEEs occurred indoors at fast food restaurants and retail stores, and outdoors in parking lots, respectively. For Home-In HEEs, the median TVOC emission and removal rates were 5.4 g h^-1 and 1.1 h^-1, respectively. Our study demonstrates the ability to determine individual sensor-based time-resolved TVOC exposures in different MEs, in support of identifying potential sources and exposure factors that can inform exposure mitigation strategies.

Environmental Health Surveillance System for a Population Using Advanced Exposure Assessment

Human exposure to air pollution is a major public health concern. Environmental policymakers have been implementing various strategies to reduce exposure, including the 10th-day-no-driving system. To assess exposure of an entire population of a community in a highly polluted area, pollutant concentrations in microenvironments and population time–activity patterns are required. To date, population exposure to air pollutants has been assessed using air monitoring data from fixed atmospheric monitoring stations, atmospheric dispersion modeling, or spatial interpolation techniques for pollutant concentrations. This is coupled with census data, administrative registers, and data on the patterns of the time-based activities at the individual scale. Recent technologies such as sensors, the Internet of Things (IoT), communications technology, and artificial intelligence enable the accurate evaluation of air pollution exposure for a population in an environmental health context. In this study, the latest trends in published papers on the assessment of population exposure to air pollution were reviewed. Subsequently, this study proposes a methodology that will enable policymakers to develop an environmental health surveillance system that evaluates the distribution of air pollution exposure for a population within a target area and establish countermeasures based on advanced exposure assessment.

Necessity of personal sampling for exposure assessment on specific constituents of PM2.5: Results of a panel study in Shanghai, China

Many epidemiological studies have evaluated the health risks of ambient fine particulate matter (PM_2.5). However, few studies have investigated the potential exposure misclassification caused by using ambient PM_2.5 concentrations as proxy for individual exposure to PM_2.5 in regions with high-level of air pollution. This study aimed to compare the differences between personal and ambient PM_2.5 constituent concentrations, and to predict the personal exposure of sixteen PM_2.5 constituents. We collected 141 72-h personal exposure filter samples from a panel of 36 healthy non-smoking college students in Shanghai, China. We then used the liner mixed effects models to predict personal constituent-specific exposure using ambient observations and several possible influencing factors including time-activity patterns, temporal variables, and meteorological conditions. The final model of each component was further evaluated by determination coefficient (R²) and root mean square error (RMSE) from leave-one-out-cross-validation (LOOCV). We observed ambient concentrations were higher than personal concentrations for all PM_2.5 components except for Mn, Fe, Ca, and V. Especially, ambient NH₄⁺, As, and NO₃^- concentrations were 3.65, 5.65 and 7.33-fold higher than their corresponding personal concentrations, respectively. The ambient level was the strongest predictor of their corresponding personal PM_2.5 components with the highest marginal R² (R_M²: 0.081 ~ 0.901), meteorological conditions (R_M²: 0.000 ~ 0.357), time-activity pattern (R_M²: 0.000 ~ 0.083) and temporal indicators (R_M²: 0.031 ~ 0.562) were also important predictors. Our final models predicted at least 50% of the variance of all personal PM_2.5 constituents and even over 90% for K, Pb, and SO₄^2-. LOOCV analysis showed that R² and RMSE ranged from 0.251 to 0.907 and 0.000 to 0.092 μg/m³, respectively. Our results showed that ambient concentration of most PM_2.5 constituents along with time-activity patterns, temporal variables, and meteorological conditions, could adequately predict personal exposure concentration. Prediction models of individual PM_2.5 constituent may help to improve the accuracy of exposure measurement in future epidemiological studies.

Effects of ambient ozone exposure on circulating extracellular vehicle microRNA levels in coronary artery disease patients

Exposure to ambient air pollutants such as ozone (O₃) and particulate matter (PM) is associated with increased cardiovascular morbidity and rate of mortality, but the underlying biological mechanisms have yet to be described. Emerging evidence shows that extracellular vehicle (EV) microRNAs (miRNAs) may facilitate cell-to-cell and organ-to-organ communications and play a role in the air pollution-induced cardiovascular effects. This study aims to explore the association between air pollutant exposure and miRNA changes related to cardiovascular diseases. Using a panel study design, 14 participants with coronary artery diseases were enrolled in this study. Each participant had up to 10 clinical visits and their plasma samples were collected and measured for expression of miRNA-21 (miR-21), miR-126, miR-146, miR-150, and miR-155. Mixed effects models adjusted for temperature, humidity, and season were used to examine the association between miRNA levels and exposure to 8-hr O₃ or 24-hr PM_2.5 up to 4 days prior. Results demonstrated that miR-150 expression was positively associated with O₃ exposure at 1-4 days lag and 5day moving average while miR-155 expression tracked with O₃ exposure at lag 0. No significant association was found between miRNA expression and ambient PM_2.5 at any time point. β-blocker and diabetic medication usage significantly modified the correlation between O₃ exposure and miR-150 expression where the link was more prominent among non-users. In conclusion, evidence indicated an association between exposure to ambient O₃ and circulating levels of EV miR-150 and miR-155 was observed. These findings pointed to a future research direction involving miRNA-mediated mechanisms of O₃-induced cardiovascular effects.

Fine-Scale Modeling of Individual Exposures to Ambient PM2.5, EC, NOx, CO for the Coronary Artery Disease and Environmental Exposure (CADEE) Study

Air pollution epidemiological studies often use outdoor concentrations from central-site monitors as exposure surrogates, which can induce measurement error. The goal of this study was to improve exposure assessments of ambient fine particulate matter (PM2.5), elemental carbon (EC), nitrogen oxides (NOx), and carbon monoxide (CO) for a repeated measurements study with 15 individuals with coronary artery disease in central North Carolina called the Coronary Artery Disease and Environmental Exposure (CADEE) Study. We developed a fine-scale exposure modeling approach to determine five tiers of individual-level exposure metrics for PM2.5, EC, NOx, CO using outdoor concentrations, on-road vehicle emissions, weather, home building characteristics, time-locations, and time-activities. We linked an urban-scale air quality model, residential air exchange rate model, building infiltration model, global positioning system (GPS)-based microenvironment model, and accelerometer-based inhaled ventilation model to determine residential outdoor concentrations (Cout_home, Tier 1), residential indoor concentrations (Cin_home, Tier 2), personal outdoor concentrations (Cout_personal, Tier 3), exposures (E, Tier 4), and inhaled doses (D, Tier 5). We applied the fine-scale exposure model to determine daily 24-h average PM2.5, EC, NOx, CO exposure metrics (Tiers 1-5) for 720 participant-days across the 25 months of CADEE. Daily modeled metrics showed considerable temporal and home-to-home variability of Cout_home and Cin_home (Tiers 1-2) and person-to-person variability of Cout_personal, E, and D (Tiers 3-5). Our study demonstrates the ability to apply an urban-scale air quality model with an individual-level exposure model to determine multiple tiers of exposure metrics for an epidemiological study, in support of improving health risk assessments.

Development of TracMyAir Smartphone Application for Modeling Exposures to Ambient PM2.5 and Ozone

Air pollution epidemiology studies of ambient fine particulate matter (PM_2.5) and ozone (O₃) often use outdoor concentrations as exposure surrogates. Failure to account for the variability of the indoor infiltration of ambient PM_2.5 and O₃, and time indoors, can induce exposure errors. We developed an exposure model called TracMyAir, which is an iPhone application ("app") that determines seven tiers of individual-level exposure metrics in real-time for ambient PM_2.5 and O₃ using outdoor concentrations, weather, home building characteristics, time-locations, and time-activities. We linked a mechanistic air exchange rate (AER) model, a mass-balance PM_2.5 and O₃ building infiltration model, and an inhaled ventilation model to determine outdoor concentrations (Tier 1), residential AER (Tier 2), infiltration factors (Tier 3), indoor concentrations (Tier 4), personal exposure factors (Tier 5), personal exposures (Tier 6), and inhaled doses (Tier 7). Using the application in central North Carolina, we demonstrated its ability to automatically obtain real-time input data from the nearest air monitors and weather stations, and predict the exposure metrics. A sensitivity analysis showed that the modeled exposure metrics can vary substantially with changes in seasonal indoor-outdoor temperature differences, daily home operating conditions (i.e., opening windows and operating air cleaners), and time spent outdoors. The capability of TracMyAir could help reduce uncertainty of ambient PM_2.5 and O₃ exposure metrics used in epidemiology studies.