Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 May 1.
Published in final edited form as: Sci Total Environ. 2016 Feb 18;551-552:622–630. doi: 10.1016/j.scitotenv.2016.02.018

Potential sources and racial disparities in the residential distribution of soil arsenic and lead among pregnant women

Harley T Davis a, C Marjorie Aelion b,*, Jihong Liu a, James B Burch a, Bo Cai a, Andrew B Lawson c, Suzanne McDermott a
PMCID: PMC4808624  NIHMSID: NIHMS761747  PMID: 26897405

Abstract

Exposure to arsenic (As) or lead (Pb) has been associated with adverse health outcomes, and high-risk populations can be disproportionately exposed to these metals in soils. The objectives of this study were: to examine if predicted soil As and Pb concentrations at maternal residences of South Carolina (SC) low-income mothers differed based on maternal race (non-Hispanic black versus white), to examine whether differences in predicted residential soil As and Pb concentrations among black and white mothers differed by socioeconomic status (SES), and to examine whether such disparities persisted after controlling for anthropogenic sources of these metals, including direction from, and distance to industrial facilities. Kriged soil As and Pb concentrations were estimated at maternal residences in 11 locations in SC, and models with maternal race and both individual and US Census block group level SES measures were examined. US Environmental Protection Agency Toxics Release Inventory (TRI) facility As and Pb releases categorized by distance and direction to block groups in which mothers resided were also identified, as were proxy measures for historic use of leaded gasoline (road density) and Pb-based paint (categories of median year home built by US Census block group). Consistent racial disparities were observed for predicted residential soil As and Pb concentrations, and the disparity was stronger for Pb than As (betas from adjusted models for black mothers were 0.12 and 2.2 for As and Pb, respectively, all p<0.006). Higher road density and older homes in block groups were more closely associated with higher predicted soil As and Pb concentrations than on-site releases of As and Pb categorized by facility location. The findings suggest that non-Hispanic black mothers in this study population had elevated residential As and Pb soil concentrations, after adjusting for SES, and that soil As and Pb concentrations were not associated with recent industrial releases.

Keywords: Neighborhood deprivation, racial disparity, residential soil metals, industrial facility distance/direction, maternal exposure, road density

Graphical Abstract

graphic file with name nihms761747f2.jpg

1. Introduction

Exposure to metals in soils has the potential to impact human health, and metals such as arsenic (As) and lead (Pb) are pervasive and long-lived in this environmental medium (Aelion et al., 2014; Nriagu and Pacyna, 1988). While As occurs naturally in the environment, elevated soil concentrations are often due to inputs from anthropogenic sources, such as mining, smelting, and other industrial activities (Hinwood et al., 2004; Luo et al., 2008). Arsenic can also leach into soils from chromated copper arsenate (CCA) treated wood (Mielke et al., 2010; Shalat et al., 2006), which was routinely used in residential areas until the early 2000s. Elevated concentrations of Pb in soils are most often the result of anthropogenic inputs, especially in residential locations. Soil Pb concentrations are strongly associated with the historic use of leaded gasoline (Datko-Williams et al., 2014; Kayhanian, 2012) and lead-based paints (Mielke et al., 2008; Mielke and Reagan, 1998), as well as industrial practices (Landsberger et al., 1999; Luo et al., 2009).

Arsenic or Pb exposure can elicit neurological (Ahamed et al., 2008; Bellinger, 2008; Llop et al., 2013; Mukherjee et al., 2005; Naujokas et al., 2013; Pabello and Bolivar, 2005), and cardiovascular impacts (Balakumar and Kaur, 2009; Kim et al., 2008; Moon et al., 2013; Poreba et al., 2011), and contribute to adverse reproductive outcomes (Ahamed et al., 2009; Ahmad et al., 2001; Ahmed et al., 2011; Jelliffe-Pawlowski et al., 2006; Mukherjee et al., 2005; Myers et al., 2010; Torres-Sanchez et al., 1999; Yang et al., 2003). Contaminated soils can become a component of household dust (Hinwood et al., 2004; Petrosyan et al., 2004), and the contribution of soil to house dust can range from a third to half (Calabrese and Stanek, 1992; StellaLevinson, 2008). Since the dust can then be inadvertently ingested or inhaled (Caussey et al., 2003), monitoring As and Pb soil contamination, especially in residential areas, can be important for preventing exposure and the associated negative health outcomes in these settings.

High-risk populations, such as those of racial/ethnic minorities and lower socioeconomic status (SES), have been found to be disproportionately exposed to As and Pb in soils (Aelion et al., 2012, 2013; Calderon et al., 2003; Calderon et al., 2004; Campanella and Mielke, 2008; Diawara et al., 2006; Mielke et al., 1999) and are potentially more susceptible to any associated health impacts. This may result from living in neighborhoods located on prior industrial sites, or in close proximity to industries and/or high volume roadways (McClintock, 2012; Pellow, 2000). This highlights the importance of examining neighborhood features when assessing high-risk populations and their potential exposure to environmental contaminants like As and Pb.

This study used environmental sampling to examine whether a racial disparity existed in predicted soil As and Pb concentrations at the residences of mothers during pregnancy who gave birth while enrolled in the South Carolina (SC) Medicaid program (a federally funded insurance program for low-income families) from 1996–2001 (Aelion et al., 2008, 2009a, 2009b, 2012, 2013, 2014; Davis et al., 2009, 2014; Liu et al., 2010; McDermott et al., 2011, 2014; Zhen et al., 2008, 2009). We controlled for individual SES (maternal education and receipt of federally-provided food assistance commonly referred to as food stamps) and neighborhood SES, measured by neighborhood deprivation, a composite measure of Census variables with higher values indicating more deprivation. It was hypothesized that non-Hispanic black mothers would have higher predicted soil As and Pb concentrations at their residence relative to non-Hispanic white mothers, after controlling for SES, and that SES measures would be positively associated with higher predicted soil As and Pb concentrations. It was also hypothesized that neighborhood deprivation would modify the association between predicted soil As and Pb concentrations and maternal race, because non-Hispanic black mothers living in neighborhoods with higher deprivation (as compared to non-Hispanic black mothers in neighborhoods with lower deprivation) would have higher predicted soil As and Pb concentrations in their neighborhoods.

Proximal and distal sources of As and Pb were also expected to impact concentrations in residential soils, and if racial/ethnic minorities lived in locations in closer proximity to more sources, then their residential soil metal concentrations were expected to be higher. Sources investigated included road density and median home age (proximal) at the neighborhood level, as well as direction and distances of maternal residences from SC industrial facilities releasing As and/or Pb on-site (distal), categorized by median average annual release of As or Pb. It was hypothesized that living in close proximity, and within the path of prevailing winds (e.g., in the southwest and northeast direction) of As- and Pb-emitting industrial facilities with cumulative on-site releases at or above the median for all TRI facilities in the state for the time period of interest would be associated with higher predicted soil As and Pb concentrations at maternal residences, and that the racial disparity in potential exposure to As and Pb in soils would no longer be apparent after accounting for these proximal and distal As and Pb sources in the analysis. Associations between soil metal concentrations and distance from industrial facilities have been reported (Aelion et al., 2009b; Bermudez et al., 2009; Douay et al., 2007). We chose to further examine the relation between soil metal concentrations and locations of industries by categorizing them by both the direction and distance to the location of potential exposure and cumulative on-site releases dichotomized at the median.

2. Methods

2.1. Study Design and Population

This study utilized data sets from a retrospective cohort study initiated in 2006 that examined associations between maternal exposure to residential soil metal concentrations and both intellectual disability (ID) and developmental delay (DD) among children in a Medicaid population of mother-child pairs giving birth to singletons (Kim et al., 2009, 2010; Liu et al., 2010; McDermott et al., 2011, 2014; Zhen et al., 2008, 2009). Medicaid, a federal assistance program that is managed by individual states, offers insurance coverage to income qualifying, medically verified pregnant women throughout pregnancy, and up to 60 days postpartum (SC DHHS, 2013). Information on mothers and children who were enrolled in SC Medicaid during pregnancy from 1996–2001 was obtained from SC birth certificates, Medicaid billing records, and the SC Department of Social Services (SC DSS). United States (US) Census 2000 block group level data for SC were also utilized.

Eleven areas in SC were selected for sampling in previous studies based on prevalence of ID/DD in SC (McDermott et al., 2011, 2014). Nine of these areas had an ID/DD prevalence that was significantly higher than the statewide background prevalence (3.5%) for all Medicaid mothers (Zhen et al., 2008). The areas were identified using Bayesian local-likelihood cluster analysis of geocoded maternal residences by month of pregnancy (Zhen et al., 2008, 2009), ranged in size from 60–490 km2 (mean of 130 km2) and were sampled from 2006 to 2011. The exact geographic locations of the maternal addresses and sampling areas were undisclosed to protect participant confidentiality. The sampling areas were located throughout SC, covered 1,500 km2, and are representative of different land uses and categories (Davis et al., 2014); five were classified as urban, five as rural, and the largest sampling area was mixed urban/rural.

In each of the 11 areas, a regular 120-node grid was laid out and single grab surface soil samples of approximately 25 g were collected as close to each grid node as possible. The distances between sampling locations were approximately 0.5 km; for the largest sampling area, distances were 3 km. Soil sample locations were selected to maximize the probability of collecting undisturbed native soil with no visible contamination or previous development. Samples were analyzed for total As and Pb (in mg/kg dry weight) using inductively coupled plasma optical emission spectroscopy (ICP-OES) by an independent environmental laboratory. Both the soil sampling protocol and metal analysis procedures have been previously described in detail (Aelion et al., 2008, 2009a, 2009b, 2012, 2013, 2014; Davis et al., 2009, 2014). These measured soil metal concentrations were then used to spatially interpolate As and Pb soil metal concentrations at each mother’s geocoded address by month of pregnancy using the ordinary kriging method, which provides the best linear unbiased predictor. Kriging methodological details are described in detail in Zhen et al. (2008, 2009).

For the current study, mothers from the original study were excluded if data were missing related to maternal age, maternal race, baby birth weight, baby gender and gestational age (~1.3% of mothers). Mothers were also excluded if their infant had an improbable clinical estimate of gestation (<21 weeks, n=15) or birth weight (<500 g, n=39). The analysis was restricted to only non-Hispanic black and non-Hispanic white (henceforth referred to as black and white) mothers, as mothers of other races/ethnicities made up <1% of the sample population. The analysis was also restricted to mothers whose residence at month 6 of pregnancy was spatially linked with the US Census 2000 block group in which the residence was located, and to only the first birth of a mother (temporally) if she gave birth to more than one child during the study period (n=8,108). Additionally, maternal education was imputed for mothers missing these data (14%) using the maximum likelihood method (Allison, 2012).

2.2. Variables

Predicted soil As and Pb concentrations at the mother’s address at month 6 of pregnancy were used as the main outcome variables of interest for characterization of potential environmental exposure to these metals. Maternal race was the main independent variable, with white mothers set as the reference category. Additionally, both individual and neighborhood level measures of SES were included as covariates in statistical models. At the individual level, maternal education was dichotomized to less than a high school education and a high school diploma or above (reference category) and utilization of food stamps during pregnancy (from SC DSS records) was also examined, with mothers not receiving food stamps during pregnancy set as the reference category.

The neighborhood deprivation index (NDI) was derived from US Census 2000 block group data as a contextual measure of neighborhood SES (Messer et al., 2006b). A principal component analysis (PCA) was performed on 20 US Census 2000 block group level variables. After identifying 10 variables (proportion of the population with less than a high school education, proportion of population unemployed, proportion of the population identifying as non-Hispanic black, proportion of households renter-occupied, proportion of households crowded, proportion of female-headed households with dependent children, proportion of households in poverty, proportion of households with an income less than $30,000 USD per year, proportion of households on public assistance, and proportion of households with no car) with factor loadings greater than or equal to the median, the PCA was performed again and the values of these variables were weighted by the PCA communality estimates. The NDI was dichotomized at the median (4.8) for mothers in the study population, with values below the median (less deprivation) set as the reference category.

Percent of block group area covered by roads (henceforth referred to as road density) was calculated by estimating the percentage of each block group’s area covered by roads based on road length and average width (Aelion et al., 2012). This continuous measure was used as a proxy for historic use of leaded gasoline, and was modeled as a continuous variable. Road density was compared with 2006 and 2014 traffic density calculations by US Census 2000 block group, another proxy for estimating impact from leaded gasoline usage (Rowangould, 2003); Pearson correlation coefficients were 0.58 and 0.56 for 2006 and 2014, respectively (all p<0.0001). This suggests good agreement between our road density measure and the other proxy measure. The block group median year home built was used to categorized block groups into the following: median year home built pre-1950, built 1950–1977, and built post-1977 (Roberts et al., 2003). This measure was used to proxy the historic residential use of Pb-based paint.

To determine block group direction from, and distance to sources of industrial emissions, all United States (US) Environmental Protection Agency (EPA) Toxics Release Inventory (TRI) facilities with As (n=21) and/or Pb (n=192) emissions from 1996 to 2011 were identified (US EPA, 2013). We identified facilities reporting any on-site releases of either As or Pb including releases to both air and land; facilities reporting only off-site releases of these metals were not included in the analysis. Cumulative releases (in thousands of pounds) by facility were calculated for all years the facility reported releases in the time period of interest. Facility locations were then mapped in ArcMap Version 10.2 (ESRI, 2013). Since the actual location of the maternal residence was not known due to a confidentiality agreement of the original research, the block group centroid was utilized for direction and distance measures. For direction, the latitude and longitude coordinates (in decimal degrees) of each TRI facility location were subtracted from the coordinates of each block group centroid in which mothers resided. Then, depending on the signs of these values, the direction of the facility from the block group centroid was assigned to one of the following categories: northeast (NE), northwest (NW), southeast (SE), or southwest (SW). For this study, only the majority prevailing wind directions in SC (NE and SW) were used in the analyses (Supplementary Table 1). The straight-line distance (in km) from the facility location to the centroid of each mothers’ block group was also measured. Using these distances, all As- and Pb-emitting TRI facilities were categorized into four distance categories: ≤5 km, 6–10 km, 11–20 km, and >20 km from the residential block group centroid.

After determining the direction from and distance to each block group centroid from As-and Pb-emitting TRI facilities, mothers were categorized based on their block group location into 14 different direction, distance, or direction/distance combined categories for As- and Pb-emitting facilities separately. These variables were all dichotomized by the median cumulative release of either As and Pb for each direction, distance, and direction/distance category into either below the median (reference level) or at/above the median. Thus mothers were categorized as either below the median As (or Pb) for cumulative releases, or at/above the median, for each direction and/or distance category.

2.3. Statistical Analyses

Predicted soil As and Pb concentrations were modeled individually with each main predictor of interest in bivariate analyses using analysis of variance for categorical variables and simple linear regression for continuous variables. Collinearity of variables was examined using regression analysis with collinearity and variance inflation factors (VIF) options. A value of ≥10 was used as a cutoff to identify variables that may have posed a collinearity issue, as well as simple correlations between variables. No variables were identified as collinear.

To test all hypotheses, hierarchical linear modeling (HLM) was used to account for multiple maternal residences within a given block group. As the main outcomes were continuous, maximum likelihood estimation and Satterthwaite’s method for determination of degrees of freedom were used. Additionally, the slope of the intercept was allowed to vary for different block groups (random intercept). The crude model contained only the maternal race variable with either predicted soil As or Pb concentrations as the outcome. Model 1 further included maternal education and the food stamps variable, and Model 2 was additionally adjusted for NDI. Model 3 was additionally adjusted for home age category and road density. To investigate the potential modifying role of NDI, an interaction term between maternal race and NDI was added in Model 4. A backward selection process was used on Model 4 for each outcome to identify the best fit models for both outcomes. The variable with the highest p-value was removed from the model until all variables had p-values less than 0.05. However, variables were retained, regardless of p-value, if the change to the maternal race parameter estimate was >10% after removal.

To examine the impact of As- and Pb-emitting TRI locations categorized by distance and/or direction on predicted residential soil As and Pb concentrations, we first compared mean predicted soil As and Pb concentrations between study population mothers by dichotomized As (or Pb) cumulative releases for individual direction and/or distance categories. To investigate the final hypothesis, HLM was again used to model each metal outcome predicted by these release dichotomized direction and/or distance categories. Each direction/distance category was evaluated in a separate model, and models were additionally adjusted for parameters retained in As and Pb best fit models. SAS Version 9.4 (SAS Institute, 2010) was used for all statistical analyses and a p-value of 0.05 was used to determine statistical significance.

3. Results

The majority of the study population identified as black, had at least a high school diploma, and received food stamps during pregnancy (Table 1). Overall, mean predicted As and Pb soil concentrations were 4.9 mg/kg and 58.2 mg/kg, respectively (Table 1), and were highly correlated with each other in this study population (r = 0.71, p<0.0001; data not shown). While mean predicted soil As concentrations were higher for mothers who received food stamps during pregnancy (p<0.0001), mean predicted soil Pb concentrations were not different among mothers stratified by receipt of food stamps during pregnancy (Table 1). Mean predicted soil concentrations of both As and Pb also were elevated among mothers living in block groups at or above the median NDI value (p<0.0001; Table 1), which indicates more deprivation. Additionally, while 66% of black mothers lived in block groups with NDI values at or above the median, less than one-fourth (21%) of white mothers did (data not shown). Most mothers (67%) lived in block groups where the median year homes were built was between 1950 and 1977. Block groups with homes built pre-1950 had significantly higher As and Pb concentrations; for the three categories, mean predicted concentrations of both As and Pb doubled as homes got older (Table 1). Predicted soil As and Pb concentrations were also positively associated with higher road density within block groups (β = 0.66 for As and 7.2 for Pb, and both p-values <0.0001; data not shown).

Table 1.

Mean predicted soil arsenic (As) and lead (Pb) concentrations (standard deviation) for maternal and neighborhood (United States Census 2000 block group) variables and p-values for differences in As and Pb concentration.

No. mothers
(%)		 Soil As (mg/kg) Soil Pb (mg/kg)
Mean (SD) P-valuea Mean (SD) P-valuea
Study Population 8,108 (100) 4.9 (4.2) NAe 58.2 (57.2) NA
Maternal Race
Non-Hispanic black 5,252 (65) 5.6 (4.6) <0.0001 64.4 (60.7) <0.0001
Non-Hispanic white 2,856 (35) 3.5 (2.9) 46.8 (48.1)
Maternal education
High school or above 5,463 (67) 4.7 (4.2) <0.0001 56.9 (58.0) 0.003
Not high school graduate 2,645 (33) 5.2 (4.3) 60.9 (55.4)
Food stampsb
Food stamps 4,837 (60) 5.2 (4.5) <0.0001 58.7 (56.9) 0.33
No food stamps 3,271 (40) 4.4 (3.7) 57.4 (57.7)
NDIc
NDI < 4.8 4,046 (50) 3.5 (3.0) <0.0001 46.5 (51.8) <0.0001
NDI ≥ 4.8 4,062 (50) 6.3 (4.7) 69.8 (59.9)
Median year home builtd
Pre-1950 710 (8.8) 10 (5.4) <0.0001 128 (59.6) <0.0001
1950 – 1977 5,419 (67) 5.1 (3.8) 59.1 (54.5)
Post-1977 1,979 (24) 2.6 (2.7) 30.6 (38.4)
Open in a new tab
a

P-values for comparisons of mean As and Pb concentrations between categories via analysis of variance

b

Mother received food stamps during pregnancy

c

NDI: neighborhood deprivation index; standardized composite measure of 10 US Census 2000 block group variables (break in NDI at median)

d

Categories based on median year home built of the US Census 2000 block group in which a mother’s residence at month 6 of pregnancy was located

e

NA: not applicable

Parameter estimates for variables included in crude models, Models 1–3, and the best fit models, are shown in Table 2. For As, the racial disparity was significant in all models. The adjusted parameter estimate for predicted soil As concentrations at black mothers’ residences was 0.12–0.13 mg/kg higher than white mothers. Lower maternal education was consistently associated with higher predicted As concentrations at the mother’s residence. Receipt of food stamps during pregnancy was not a predictor of As in any model. Predicted soil As concentrations were 1.2 mg/kg higher for mothers in block groups with NDI values at or above the median (Model 2), but this parameter estimate did not remain significant after adjustment of year home built and road density. Mothers living in block groups with a median year home built before 1950 had predicted As concentrations 2 mg/kg higher than mothers where homes were built post-1977. Road density was also a significant predictor of predicted As concentrations in all models in which it was included. The interaction term between maternal race and NDI was not significant (β = 0.029, p = 0.8; data not shown), and the best fit model for predicted soil As concentrations contained maternal race, maternal education, median year home built, and road density.

Table 2.

Parameter estimates (standard error) and p-values for models of predicted soil arsenic (As) and lead (Pb) concentrations.

Maternal
racea 		Maternal
educationb 		Food
stampsc 		NDId Year home
builte 		Road
density
Crudef As 0.13 (0.04)
<0.0001		 NAj NA NA NA NA
Model 1g
As		 0.12 (0.04)
0.003		 0.085 (0.03)
0.01		 0.058 (0.03)
0.09		 NA NA NA
Model 2h
As		 0.12 (0.04)
0.006		 0.084 (0.03)
0.01		 0.057 (0.03)
0.09		 1.2 (0.28)
<0.0001		 NA NA
Model 3i
As		 0.12 (0.04)
0.006		 0.085 (0.03)
0.01		 0.058 (0.03)
0.09		 0.30 (0.29)
0.31		 1.7 (0.66)
0.009		 0.34 (0.05)
<0.0001
Best Fit
Model As		 0.13 (0.04)
0.001		 0.089 (0.03)
0.007		 NA NA 1.8 (0.66)
0.007		 0.36 (0.05)
<0.0001
Crudef Pb 2.2 (0.60)
0.0003		 NAj NA NA NA NA
Model 1g
Pb		 2.3 (0.62)
0.0002		 0.51 (0.48)
0.29		 −0.29 (0.50)
0.57		 NA NA NA
Model 2h
Pb		 2.2 (0.62)
0.0003		 0.51 (0.48)
0.29		 −0.29 (0.50)
0.56		 9.4 (4.0)
0.02		 NA NA
Model 3i
Pb		 2.2 (0.62)
0.0004		 0.50 (0.48)
0.30		 −0.29 (0.50)
0.56		 −4.6 (4.2)
0.28		 24.9 (9.5)
0.009		 4.6 (0.70)
<0.0001
Best Fit
Model Pb		 2.1 (0.60)
0.0006		 NA NA NA 24.3 (9.5)
0.01		 4.4 (0.68)
<0.0001
Open in a new tab
a

Non-Hispanic white mothers are the referents

b

Mothers with less than a high school education are the referents

c

Mothers who did not receive food stamps during pregnancy are the referents

d

NDI: neighborhood deprivation index; categorized to below and at or above the median (4.8); block groups below the median are the referents

e

Only parameter for comparison between referent category (median year home built post-1977) and pre-1950 home category shown.

f

Contains maternal race variable only

g

Additionally adjusted for maternal education and food stamps

h

Addtionally adjusted for block group NDI

i

Additionally adjusted for block group year home built category and road density

j

NA: not applicable

For predicted soil Pb concentration, maternal race parameter estimates were significant in all models, and the racial disparity was stronger than that for As (Table 2). For the study population, parameter estimates for predicted soil Pb concentrations at black mothers’ residences were 2.2–2.3 mg/kg higher than white mothers. Neither maternal education nor receipt of food stamps predicted soil Pb concentrations. As with As, the NDI also was associated with predicted Pb soil concentrations in Model 2 but did not predict soil Pb after adjustment for median year home built category and road density, both of which were significant (Table 2). The interaction term between NDI and maternal race was not significant (β = −1.2, p = 0.3) for predicted soil Pb concentrations (data not shown). For Pb, the best fit model contained maternal race, median year home built category, and road density.

We examined crude models to determine if predicted soil As and Pb concentrations were higher for mothers living in block groups in the direction of prevailing winds from, and in closer proximity to, TRI facilities dichotomized by median cumulative releases (Figure 1A–B). Mean predicted soil As concentrations were significantly higher for mothers residing in block groups in the SW direction from As-emitting TRI facilities that had cumulative releases at or above the median, compared to mothers that did not (Figure 1A); the reverse held true for the NE direction. Predicted soil As concentrations were significantly lower for mothers in close proximity (≤5 and 6–10 km) to As-emitting TRI facilities with cumulative releases at/above the median, regardless of direction (Figure 1A). Mean predicted soil Pb concentrations were significantly lower for mothers residing in block groups in the SW and NE directions from Pb-emitting TRI facilities with cumulative releases at or above the median, compared to mothers who did not. Mean predicted soil Pb concentrations were significantly higher for mothers living within 6–20 km of Pb-emitting TRI facilities with cumulative releases at or above the median (Figure 1B). As with As, mean predicted soil Pb concentrations were significantly lower for mothers residing in block group located within 5 km of Pb-emitting TRI facilities (Figure 1B). Overall, there was not an apparent pattern in differences between mean predicted As and Pb soil concentrations when mother’s residences were categorized by direction and/or distance and dichotomized to below or at/above the median.

Figure 1.

Figure 1

Open in a new tab

A–B. Comparisons of mean predicted soil A) arsenic (As) and B) lead (Pb) concentrations at maternal residences (with 95% confidence interval error bars) categorized by block group centroid direction and/or distance from As- or Pb-emitting Toxics Release Inventory (TRI) facilities and stratified by median cumulative As or Pb releases.

Parameter estimates from mixed models including cumulative releases from As- and Pb-emitting TRI facilities dichotomized at the median and categorized by direction from and/or distance to block groups are shown in Table 3. For As, all significant parameter estimates were negative, suggesting lower predicted residential soil As concentrations were associated with on-site cumulative releases of As that were at or above the median in the direction of prevailing winds, and in close proximity (≤5 km) to As-emitting TRI facilities after adjusting for maternal SES, road density, and median year home built category (Table 3). This was also generally true of parameter estimates for Pb facility release categories, though significant estimates were positive for the 6–10 and 11–20 km distance categories (Table 3). Maternal race parameter estimates remained significant in all As and Pb models (Table 3), and they were similar in magnitude to estimates from crude and adjusted models (Table 2). Thus, there was little additional variation explained due to inclusion of stratified cumulative releases of As and Pb categorized by direction from and/or distance to As- or Pb-emitting TRI facilities on predicted soil As and Pb concentration in mixed models.

Table 3.

Parameter estimates (standard errors) for arsenic (As) and lead (Pb) emitting TRI facilities categorized by amount of on-site releases (below or at/above the median) and direction, distance, or direction/distance combined (with respect to US Census 2000 block group from facility) and maternal race in As and Pb best fit models.

As modelsa Pb modelsb
Distance and/or
direction		 Maternal race Distance and/or
direction		 Maternal
race
Direction only
SW 0.32 (0.25) 0.13 (0.04)** −18.5 (3.6)** 2.1 (0.6)**
NE −0.96 (0.24)** 0.13 (0.04)** −31.9 (3.5)** 2.2 (0.6)**
Distance only
≤5 km −2.4 (0.71)** 0.14 (0.04)** −12.4 (3.8)** 2.1 (0.6)**
6–10 km −0.71 (0.49) 0.14 (0.04)** 21.2 (3.8)** 2.1 (0.6)**
11–20 km 0.01 (0.31) 0.13 (0.04)** 10.5 (3.6)** 2.1 (0.6)**
>20 km 0.36 (0.28) 0.14 (0.04)** −8.6 (3.6)** 2.0 (0.6)**
Direction/distance combined
SW, ≤5 km −0.6 (2.0) 0.13 (0.04)** −11.3 (4.4)** 2.1 (0.6)**
SW, 6–10 km 0.15 (0.88) 0.13 (0.04)** 4.3 (3.7) 2.1 (0.6)**
SW, 11–20 km 0.33 (0.41) 0.13 (0.04)** 23.5 (3.7)** 2.1 (0.6)**
SW, >20 km 0.09 (0.25) 0.13 (0.04)** −9.6 (3.6)** 2.1 (0.6)**
NE, ≤5 km −3.7 (1.0)** 0.14 (0.04)** −20.1 (9.4)** 2.1 (0.6)**
NE, 6–10 km −1.5 (0.71)** 0.14 (0.04)** −6.2 (4.0) 2.1 (0.6)**
NE, 11–20 km −0.45 (0.62) 0.13 (0.04)** 2.2 (3.7) 2.1 (0.6)**
NE, >20 km −0.89 (0.24)** 0.13 (0.04)** −26.6 (3.7)** 2.1 (0.6)**
Open in a new tab
a

As models additionally adjusted for maternal education, block group year home built category, and block group road density; parameter estimate are for mothers living within that direction, distance, or direction/distance combined category of TRI facilities with on-site releases at or above the median for As

b

Pb models additionally adjusted for block group year home built category and road density; parameter estimate are for mothers living within that direction, distance, or direction/distance combined category of TRI facilities with on-site releases at or above the median for Pb

**

Denotes significant estimate (p<0.05)

4. Discussion

A consistent racial disparity in predicted residential soil As and Pb concentrations was observed, which remained significant after controlling for maternal and neighborhood demographics, as well as proximal and distal sources of these metals. This suggests that black mothers in the study population were more likely to live at locations with higher soil concentrations of both As and Pb, which may mean they could be at increased risk for exposure. Predicted soil concentrations of both As and Pb were generally low in these sampling areas, and differences between black and white mothers were less than 2.5 mg/kg for Pb and less than 0.15 mg/kg for As in adjusted models. Even so, the predicted concentrations reported for residential locations in the current study are similar to those reported in studies of background As and Pb concentrations in both SC and the US. Canova (1999) reported ranges of 0–210 mg/kg and 0–200 mg/kg for As and Pb, respectively. Shacklette and Boerngen (1984), however, reported much lower ranges (0–4.1 mg/kg for As and ≤10 mg/kg for Pb).

Other studies have also documented racial disparities in soil Pb concentrations (Campanella and Mielke, 2008; Diawara et al., 2006; Mielke et al., 1999), though mean soil concentrations of Pb reported in these studies were much higher than in the sampling areas examined in our study (5 and 58 mg/kg for As and Pb, respectively). This is most likely due to more anthropogenic sources of Pb in those locations. For example, Mielke et al. (1999) reported median soil Pb concentrations of 120 mg/kg in New Orleans, LA, and the US Census 1990 tract population percentage of blacks was ~60% in tracts categorized as high metal (soil Pb concentrations ≥316 mg/kg), compared to just 36% in tracts categorized as low metal. Using the same 11 sampling areas utilized in the current study, Davis et al. (2014) reported significant associations between soil Pb concentrations measured on a uniform grid (not kriged to residential location) and US Census 2000 block group population percentages of non-Hispanic blacks. An aggregate analysis at the US Census 2000 block group level also identified a significant association between percentage of black study population mothers and mean block group Pb concentrations using a subset of four (three urban and one mixed rural/urban) of the 11 sampling areas (Aelion et al., 2012). The current study identified a racial disparity within the study population of Medicaid mothers rather than relying on Census population data, and corroborates the findings at the individual level, which strengthen the findings in this and the previous two studies.

There are fewer studies reporting racial disparities in the potential for exposure to As in soils than for Pb. Aelion et al. (2012), using a subset of the original study sampling areas, found that block groups with higher percentages of black mothers had higher mean soil As concentrations relative to whites in two of four areas examined. In contrast, Davis et al. (2014) reported that soil As concentrations were not associated with the US Census 2000 block group percentages of non-Hispanic blacks. In the current study, we observed a small but significant disparity in soil As concentrations for study population mothers. Soil As concentrations (average of 12.6 mg/kg) measured by Diawara et al. (2006) were higher than what was reported for predicted As soil concentrations in the current study, and they reported that blocks with higher soil As concentrations had higher block population percentages of Hispanics.

The racial disparity in predicted soil As and Pb concentrations at maternal residences was not modified by block group NDI in this study. This may have been due to overlap between these two variables, as the NDI calculation included the block group proportion of the population identifying as non-Hispanic black. Gee and Payne-Sturges (2004) reported on how neighborhood resources, community stressors, and structural factors (all of which are related to NDI) are driven by racial residential segregation (i.e., higher concentrations of individuals of the same race/ethnicity). The lack of a significant interaction also may be related to the fact that this study’s population was limited to mothers giving birth while enrolled in Medicaid (of low income by definition) and more homogeneous in SES than populations examined in other studies that have included NDI. For example, the study population in Messer et al. (2006b) included over 200,000 mothers from Pennsylvania, Maryland, Michigan, and North Carolina, and Messer et al. (2006a) examined over 13,000 mothers from North Carolina. Ma et al. (2014) included mothers from all over SC, and the variability in NDI for all block groups in SC was higher than that of the subset of block groups examined in the current study (data not shown). Additionally, our study population was based on prevalence of ID and DD in study areas, suggesting population homogeneity on risk factors associated with these outcomes.

It was also of note that the best fit models for As and Pb were different, even though these metals were highly correlated. Predicted soil Pb concentrations were associated with road density (a proxy for historic leaded gasoline use) and median year home built category (a proxy for Pb-based paint use) in the study areas, whereas predicted soil As concentrations were associated with maternal SES measures as well as road density and year home built. Given that As in soils may originate from both natural and anthropogenic sources (as compared to Pb, which is primarily anthropogenic) in these sampling areas (Aelion et al., 2013; Davis et al, 2009), this association may be related to the SES homogeneity in the study population. Further investigation of the association between soil As and Pb concentrations and other individual level SES measures may help elucidate the findings of the current study related to maternal education and predicted soil As concentrations.

In models examining cumulative releases from As- and Pb-emitting facilities stratified at the median and categorized by direction, distance, and combined direction/distance to block groups, the racial disparity in predicted soil As and Pb concentrations persisted in both As and Pb models and most significant direction, distance, and combined direction/distance parameter estimates were negative. Additionally, mean predicted soil As and Pb concentrations were not consistently higher for mothers living in the prevailing wind directions, or in close proximity to As- and/or Pb-emitting TRI facilities with cumulative releases at or above the median. These findings suggest that not only were residential predicted soil As and Pb concentrations not positively associated with higher releases from TRI facilities located in close proximity and in the direction of prevailing winds, but also that these distal releases did not impact the magnitude of the racial disparity for either the predicted soil As or Pb concentration outcomes. While distance was associated with higher concentrations of chromium in Aelion et al. (2009b) using a subset of the sampling areas from the current study, it appears that on-site releases of As and Pb at nearby TRI facilities in the direction of prevailing winds did not significantly impact predicted soil As and Pb concentrations. Prevailing wind and distance are proxies for air-borne distribution of contaminants; additionally, it is possible that soil deposition may occur at much smaller geographic distances than those used in this study. In addition, it was not possible to differentiate between on-site releases to land and air after 2000 using the TRI data. Further studies on industrial contribution to soil metal concentrations may reveal significant differences using more comprehensive data.

Both road density and median year home built category were predictors of As and Pb in all models, and these parameter estimates were generally greater than those for direction, distance, or direction/distance combined release estimates. While releases from industrial facilities are ongoing, and use of leaded gasoline and paints has been phased out in the US, our findings show that the long-term historic use of leaded gasoline and paints within residential areas of SC has the potential to currently impact human health. Mielke et al. (2011) reported that estimated releases of Pb to soils in New Orleans, LA would be 5–10 times higher for leaded gasoline than Pb-based paint. However, both of these proxy measures are interrelated and generally coincide with low income and high minority populations. Regardless of the environmental source of As or Pb, black mothers in this study population, on average, lived in areas with higher soil As and Pb concentrations. These findings could be specific to the locations where samples were collected, but may reflect environmental inequality of residences of black mothers in SC.

It is acknowledged that this research has several limitations related to study location and population. The areas of study were all in SC but varied in land use category and included a range of urban and rural sites; thus they are generalizable to many areas. The study population was chosen based on prevalence of ID/DD in a Medicaid population, which is representative of lower income individuals in developed countries and limits the generalizability of these results to the general population of mothers in the US or high-income countries similar to the US. Future studies should be conducted to examine potential racial disparities in exposure to metals in soils using a larger and more SES-diverse population of mothers and children. Also, any health behaviors or outcomes related to ID/DD prevalence may be higher in this population and should be considered with respect to the outcome investigated. For example, under the same restriction/exclusion criteria, the study population was more likely to be black, not have at least a high school diploma, and have received food stamps during pregnancy as compared to all Medicaid mothers in SC (data not shown).

The study also was limited by the availability of historical environmental data related to industrial releases. Given the observed importance of historical releases of metals (e.g., leaded gasoline) to current soil concentrations, examining locations of not only current TRI facilities, but also historical facilities that may have emitted As and/or Pb by direction and/or distance would be beneficial. There are also inherent limitations with both proxy measures used (road density and median year home built category). However, we compared road density to traffic density, and the categories of block group median year home built have been implemented in other studies. Another uncertainty is that predicted soil As and Pb concentrations at maternal residences were used rather than measured soil concentrations at maternal residences; however, kriging is well-documented as an acceptable method for estimation of soil concentrations for a variety of contaminants, and estimated concentrations are generated from actual concentrations measured at known locations. Simplified methods were also used for determining direction and distance from facilities, as well as for categorizing As and Pb releases from TRI facilities. Given the limitations associated with the confidentiality agreement (not knowing maternal residence locations), and the complexity of air-borne deposition of metals, more refined analyses, including smaller distances and distance categories, angular directions or a different categorization method, may be needed to effectively characterize the contribution of TRI emissions to As or Pb concentrations in residential soils. Finally, unequal sample sizes within direction/distance categories may have impacted our findings for As, given the low number of As-releasing facilities identified in the state.

5. Conclusions

Significantly higher predicted concentrations of soil As and Pb at maternal residences were observed among black mothers as compared to whites in this low-income study population. This disparity persisted after controlling for individual and neighborhood level demographics, as well as possible industrial sources of these metals. The disparity was larger for Pb than for As, though both were small in magnitude. Block group median year home built and road density were associated with elevated residential soil concentrations of both As and Pb in all adjusted models. These were more strongly associated with As and Pb in soils than releases from TRI facilities categorized by direction and/or distance, suggesting limited contribution of air-borne metals from these facilities. The results suggest the importance of both individual and neighborhood level characteristics as predictors of these metals in residential soils, and the need for further investigations of environmental justice issues that may impact health outcomes in susceptible populations.

Supplementary Material

Highlights.

  1. Small but persistent racial disparity in distribution of soil As and Pb concentrations.

  2. Neighborhood deprivation and SES did not modify racial disparity in distribution of soil As and Pb.

  3. Older homes and road density were strongly associated with soil As and Pb.

  4. Industries categorized by distance and direction were not associated with soil As and Pb.

Acknowledgments

Funding for this research was provided by a National Institute of Environmental Health Sciences (NIEHS), National Institutes of Health (NIH) grant (2R01ES012895-04) to McDermott, Aelion, and Lawson, and by a University of South Carolina Support to Promote Advancement of Research and Creativity (SPARC) Graduate Research Grant. We thank B. Bey, J. Davis, M. Engle, S. Jayasinghe, and F. Nemeth for their help with soil collection.

Footnotes

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

References

  1. Aelion CM, Davis HT, McDermott S, Lawson AB. Metal concentrations in rural topsoil in South Carolina: potential for human health impact. Science of the Total Environment. 2008;402:149–156. doi: 10.1016/j.scitotenv.2008.04.043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Aelion CM, Davis HT, Liu Y, Lawson AB, McDermott S. Validation of Bayesian kriging of arsenic, chromium, lead and mercury in surface soils based on internode sampling. Environmental Science and Technology. 2009a;43:4432–4438. doi: 10.1021/es803322w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Aelion CM, Davis HT, McDermott S, Lawson AB. Soil metal concentrations and toxicity: associations with distances to industrial facilities and implications for human health. Science of the Total Environment. 2009b;407:2216–2223. doi: 10.1016/j.scitotenv.2008.11.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Aelion CM, Davis HT, Lawson AB, Cai B, McDermott S. Associations of estimated residential soil arsenic and lead concentrations and community-level environmental measures with mother-child health conditions in South Carolina. Health & Place. 2012;18:774–781. doi: 10.1016/j.healthplace.2012.04.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Aelion CM, Davis HT, Lawson AB, Cai B, McDermott S. Associations between soil lead concentrations and populations by race/ethnicity and income-to-poverty ratio in urban and rural areas. Environmental Geochemistry and Health. 2013;35:1–12. doi: 10.1007/s10653-012-9472-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Aelion CM, Davis HT, Lawson AB, Cai B, McDermott S. Temporal and spatial variation in residential soil metal concentrations: implications for exposure assessments. Environmental Pollution. 2014;185:365–368. doi: 10.1016/j.envpol.2013.10.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Ahamed M, Fareed M, Kumar A, Siddiqui WA, Siddiqui MKJ. Oxidative stress and neurological disorders in relation to blood lead levels in children. Redox Report. 2008;13:117–122. doi: 10.1179/135100008X259213. [DOI] [PubMed] [Google Scholar]
  8. Ahamed M, Mehrotra PK, Kumar P, Siddiqui MKJ. Placental lead-induced oxidative stress and preterm delivery. Environmental Toxicology and Pharmacology. 2009;27:70–74. doi: 10.1016/j.etap.2008.08.013. [DOI] [PubMed] [Google Scholar]
  9. Ahmad SA, Salim Ullah Sayed MH, Barua S, et al. Arsenic in drinking water and pregnancy outcomes. Environmental Health Perspectives. 2001;109:629–631. doi: 10.1289/ehp.01109629. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Ahmed S, Khoda SM, Rekha RS, et al. Arsenic-associated oxidative stress, inflammation, and immune disruption in human placenta and cord blood. Environmental Health Perspectives. 2011;119:258–264. doi: 10.1289/ehp.1002086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Allison PD. Handling missing data by maximum likelihood. SAS Global Forum. 2012;2012 Paper 312–2012. [Google Scholar]
  12. Balakumar P, Kaur J. Arsenic exposure and cardiovascular disorders: an overview. Cardiovascular Toxicology. 2009;9:169–176. doi: 10.1007/s12012-009-9050-6. [DOI] [PubMed] [Google Scholar]
  13. Bellinger DC. Neurological and behavioral consequences of childhood lead exposure. PLOS Medicine. 2008;5:e115. doi: 10.1371/journal.pmed.0050115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Bermudez GMA, Moreno M, Invernizzi R, Pla R, Pignata ML. Heavy metal pollution in topsoils near a cement plant: the role of organic matter and distance to the source to predict total and HCl-extracted heavy metal concentrations. Chemosphere. 2009;78:375–381. doi: 10.1016/j.chemosphere.2009.11.012. [DOI] [PubMed] [Google Scholar]
  15. Calabrese EJ, Stanek EJ. What proportion of household dust is derived from outdoor soil? Journal of Soil Contamination. 1992;1:253–263. [Google Scholar]
  16. Calderon J, Ortiz-Perez D, Yanez L, Diaz-Barriga F. Human exposure to metals. Pathways of exposure, biomarkers of effect, and host factors. Ecotoxicology and Environmental Safety. 2003;56:93–103. doi: 10.1016/s0147-6513(03)00053-8. [DOI] [PubMed] [Google Scholar]
  17. Calderon RL, Abernathy CO, Thomas DJ. Consequences of acute and chronic exposure to arsenic in children. Pediatric Annals. 2004;33:461–466. doi: 10.3928/0090-4481-20040701-11. [DOI] [PubMed] [Google Scholar]
  18. Campanella R, Mielke HW. Human geography of New Orleans’ high-lead geochemical setting. Environmental Geochemistry and Health. 2008;30:531–540. doi: 10.1007/s10653-008-9190-9. [DOI] [PubMed] [Google Scholar]
  19. Canova JL. Elements in South Carolina inferred background soil and stream sediment samples. South Carolina Geology. 1999;41:11–25. [Google Scholar]
  20. Caussey D, Gochfeld M, Gurzau E, Neagu C, Ruedel H. Lessons from case studies of metals: investigating exposure, bioavailability, and risk. Ecotoxicology and Environmental Safety. 2003;56:45–51. doi: 10.1016/s0147-6513(03)00049-6. [DOI] [PubMed] [Google Scholar]
  21. Datko-Williams L, Wilkie A, Richmond-Bryant J. Analysis of U.S. soil lead (Pb) studies from 1970–2012. Science of the Total Environment. 2014;468–469:854–863. doi: 10.1016/j.scitotenv.2013.08.089. [DOI] [PubMed] [Google Scholar]
  22. Davis HT, Aelion CM, McDermott S, Lawson AB. Identifying natural and anthropogenic sources of metals in urban and rural soils using GIS-based data, PCA, and spatial interpolation. Environmental Pollution. 2009;157:2378–2386. doi: 10.1016/j.envpol.2009.03.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Davis HT, Aelion CM, Lawson AB, Cai B, McDermott S. Associations between land cover categories and soil concentrations of arsenic, lead and barium, and population race/ethnicity and socioeconomic status. Science of the Total Environment. 2014;490:1051–1056. doi: 10.1016/j.scitotenv.2014.05.076. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Diawara MM, Litt JS, Unis D, et al. Arsenic, cadmium, lead, and mercury in surface soils, Pueblo, Colorado: implications for population health risk. Environmental Geochemistry and Health. 2006;28:297–315. doi: 10.1007/s10653-005-9000-6. [DOI] [PubMed] [Google Scholar]
  25. Douay F, Roussel H, Fourrier H, Heyman C, Chateau G. Investigation of heavy metal concentrations on urban soils, dust and vegetables nearby a former smelter site in Mortagne du Nord, Northern France. Journal of Soils and Sediments. 2007;3:143–146. [Google Scholar]
  26. Environmental Systems Resource Institute (ESRI) ArcMap Version 10.2.1. California: ESRI, Redlands; 2013. [Google Scholar]
  27. Gee GC, Payne-Sturges DC. Environmental health disparities: a framework integrating psychosocial and environmental concepts. Environmental Health Perspectives. 2004;112:1645–1653. doi: 10.1289/ehp.7074. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Hinwood AL, Sim MR, Jolley D, et al. Exposure to inorganic arsenic in soil increases urinary inorganic concentrations in residents living in old mining areas. Environmental Geochemistry and Health. 2004;26:27–36. doi: 10.1023/b:egah.0000020897.15564.93. [DOI] [PubMed] [Google Scholar]
  29. Jelliffe-Pawlowski LL, Miles SQ, Courtney JG, Materna B, Charlton V. Effect of magnitude and timing of maternal pregnancy blood lead (Pb) levels on birth outcomes. Journal of Perinatology. 2006;26:154–162. doi: 10.1038/sj.jp.7211453. [DOI] [PubMed] [Google Scholar]
  30. Kayhanian M. Trend and concentrations of legacy lead (Pb) in highway runoff. Environmental Pollution. 2012;160:169–177. doi: 10.1016/j.envpol.2011.09.009. [DOI] [PubMed] [Google Scholar]
  31. Kim J-I, Lawson AB, McDermott S, Aelion CM. Variable selection for spatial random field predictors under a Bayesian mixed hierarchical spatial model. Spatial and Spatio-temporal Epidemiology. 2009;1:95–102. doi: 10.1016/j.sste.2009.07.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Kim J-I, Lawson AB, McDermott S, Aelion CM. Bayesian spatial modeling of disease risk in relation to multivariate environmental risk fields. Statistic in Medicine. 2010;29:142–157. doi: 10.1002/sim.3777. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Kim K-R, Lee S-W, Paik NW, Choi K. Low-level exposure among South Korean lead workers, and estimates of associated risk of cardiovascular disease. Journal of Occupational and Environmental Hygiene. 2008;5:399–416. doi: 10.1080/15459620802081587. [DOI] [PubMed] [Google Scholar]
  34. Landsberger S, Iskander F, Basunia S, Barnes D, Kaminski M. Lead and copper contamination of soil from industrial activities and firing ranges. Biological Trace Element Research. 1999;71–72:387–396. doi: 10.1007/BF02784226. [DOI] [PubMed] [Google Scholar]
  35. Liu Y, McDermott S, Lawson AB, Aelion CM. The relationship between mental retardation and developmental delays in children and the levels of arsenic, mercury, and lead in soil samples taken near their mother’s residence during pregnancy. International Journal of Hygiene and Environmental Health. 2010;213:116–123. doi: 10.1016/j.ijheh.2009.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Llop S, Lopez-Espinosa M-J, Rebagliato M, Ballester F. Gender differences in the neurotoxicity of metals in children. Toxicology. 2013;311:3–12. doi: 10.1016/j.tox.2013.04.015. [DOI] [PubMed] [Google Scholar]
  37. Luo W, Lu Y, Wang G, Shi Y, Wang T, Giesy JP. Distribution and availability of arsenic in soils from the industrialized urban area of Beijing, China. Chemosphere. 2008;72:797–802. doi: 10.1016/j.chemosphere.2008.03.003. [DOI] [PubMed] [Google Scholar]
  38. Luo W, Lu Y, Tong X, et al. Distribution of copper, cadmium, and lead in soils from former industrialized urban areas of Beijing, China. Bulletin of Environmental Contamination and Toxicology. 2009;82:378–383. doi: 10.1007/s00128-008-9627-y. [DOI] [PubMed] [Google Scholar]
  39. Ma X, Fleischer NL, Liu J, Hardin JW, Zhao G, Liese AD. Neighborhood deprivation and preterm birth: an application of propensity score matching. Annals of Epidemiology. 2014;25:120–125. doi: 10.1016/j.annepidem.2014.11.021. [DOI] [PubMed] [Google Scholar]
  40. McClintock N. Assessing soil lead contamination at multiple scales in Oakland, California: implications for urban agriculture and environmental justice. Applied Geography. 2012;35:460–473. [Google Scholar]
  41. McDermott S, Wu J, Cai B, Lawson AB, Aelion CM. Probability of intellectual disability is associated with soil concentrations of arsenic and lead. Chemosphere. 2011;84:31–38. doi: 10.1016/j.chemosphere.2011.02.088. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. McDermott S, Bao W, Tong X, Cai B, Lawson AB, Aelion CM. Are different soil metals near homes of pregnant women associated with mild and severe intellectual diability in children? Developmental Medicine & Child Neurology. 2014;56:888–897. doi: 10.1111/dmcn.12442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Messer LC, Kaufman JS, Dole N, Savitz DA, Laraia BA. Neighborhood crime, deprivation, and preterm birth. Annals of Epidemiology. 2006a;16:455–462. doi: 10.1016/j.annepidem.2005.08.006. [DOI] [PubMed] [Google Scholar]
  44. Messer LC, Laraia BA, Kaufman JS, et al. The development of a standardized neighborhood deprivation index. Journal of Urban Health. 2006b;83:1041–1062. doi: 10.1007/s11524-006-9094-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Mielke HW, Reagan PL. Soil is an important pathway of human lead exposure. Environmental Health Perspectives. 1998;106:217–229. doi: 10.1289/ehp.98106s1217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Mielke HW, Gonzales CR, Smith MK, Mielke PW. The urban environment and children’s health: soils as an integrator of lead, zinc, and cadmium in New Orleans, Louisiana, U.S.A. Environmental Research Section A. 1999;81:117–129. doi: 10.1006/enrs.1999.3966. [DOI] [PubMed] [Google Scholar]
  47. Mielke HW, Gonzales C, Powell E, Mielke PW. Urban soil-lead (Pb) footprint: retrospective comparison of public and private properties in New Orleans. Environmental Geochemistry and Health. 2008;30:231–242. doi: 10.1007/s10653-007-9111-3. [DOI] [PubMed] [Google Scholar]
  48. Mielke HW, Gonzales CR, Cahn E, Brumfield J, Powell ET, Mielke PW. Soil arsenic surveys of New Orleans: localized hazards in children’s play areas. Environmental Geochemistry and Health. 2010;32:431–440. doi: 10.1007/s10653-010-9286-x. [DOI] [PubMed] [Google Scholar]
  49. Mielke HW, Gonzales CR, Mielke PW. The continuing impact of lead dust on children’s blood lead: comparison of public and private properties in New Orleans. Environmental Research. 2011;111:1164–1172. doi: 10.1016/j.envres.2011.06.010. [DOI] [PubMed] [Google Scholar]
  50. Moon KA, Gullar E, Umans JG, et al. Association between exposure to low to moderate arsenic levels and incident cardiovascular disease. Annals of Internal Medicine. 2013;159:649–659. doi: 10.7326/0003-4819-159-10-201311190-00719. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Mukherjee SC, Saha KC, Pati S, et al. Murshidabad-one of nine groundwater arsenic-affected districts of West Bengal, India. Part II: dermatological, neurological, and obstetric findings. Clinical Toxicology. 2005;43:835–848. doi: 10.1080/15563650500357495. [DOI] [PubMed] [Google Scholar]
  52. Myers SL, Lobdell DT, Liu Z, et al. Maternal drinking water arsenic exposure and perinatal outcomes in Inner Mongolia, China. Journal of Epidemiology and Community Health. 2010;64:325–329. doi: 10.1136/jech.2008.084392. [DOI] [PubMed] [Google Scholar]
  53. Naujokas MF, Anderson B, Ahsan H, et al. The broad scope of health effects from chronic arsenic exposure: update on a worldwide public health problem. Environmental Health Perspectives. 2013;121:295–302. doi: 10.1289/ehp.1205875. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Nriagu JO, Pacyna JM. Quantitative assessment of worldwide contamination of air, water and soil by trace metals. Nature. 1988;333:134–139. doi: 10.1038/333134a0. [DOI] [PubMed] [Google Scholar]
  55. Pabello NG, Bolivar VJ. Young brains on lead: adult neurological consequences? Toxicological Sciences. 2005;86:211–213. doi: 10.1093/toxsci/kfi221. [DOI] [PubMed] [Google Scholar]
  56. Pellow DN. Environmental inequality formation: toward a theory of environmental injustice. American Behavioral Scientist. 2000;43:581–601. [Google Scholar]
  57. Petrosyan V, Orlova A, Dunlap CE, Babayan E, Farfel M, von Braun M. Lead in residential soil and dust in a mining and smelting district in northern Armenia: a pilot study. Environmental Research. 2004;94:297–308. doi: 10.1016/S0013-9351(03)00113-0. [DOI] [PubMed] [Google Scholar]
  58. Poreba R, Gac P, Poreba M, Andrzejak R. Environmental and occupational exposure to lead as a potential risk factor for cardiovascular disease. Environmental Toxicology and Pharmacology. 2011;31:267–277. doi: 10.1016/j.etap.2010.12.002. [DOI] [PubMed] [Google Scholar]
  59. Roberts JR, Hulsey TC, Curtis GB, Reigart JR. Using geographic information systems to assess risk for elevated blood lead levels in children. Public Health Reports. 2003;118:221–229. doi: 10.1093/phr/118.3.221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Rowangould GM. A census of the US near-roadway population: public health and environmental justice considerations. Transportation Research Part D. 2003;25:59–67. [Google Scholar]
  61. SAS Software Version 9.4. Copyright 2010. Cary, NC: SAS Institute; [Google Scholar]
  62. Shacklette HT, Boerngen JG. Element concentrations in soils and other surficial materials of the conterminous United States. US Geological Survey Professional Paper 1270. 1984 [Google Scholar]
  63. Shalat SL, Solo-Gabriele HM, Fleming LE, et al. A pilot study of children’s exposure to CCA-treated wood from playground equipment. Science of the Total Environment. 2006;367:80–88. doi: 10.1016/j.scitotenv.2006.01.002. [DOI] [PubMed] [Google Scholar]
  64. South Carolina Department of Health and Human Services (SC DHHS) [Accessed 04/01/2013];Medicaid Optional Coverage for Women and Infants. Available at: https://www.scdhhs.gov/eligibility-groups/optional-coverage-women-and-infants-ocwi.
  65. StellaLevinson HR. Method validation for the determination of soil constituent in household dust. Thesis, State University of New York. Ann Arbor: ProQuest/UMI; 2008. (Publication No. 1454157.) [Google Scholar]
  66. Torres-Sanchez LE, Berkowitz G, Lopez-Carrillo L, Torres-Arreola L, Rios C, Lopez-Cervantes M. Intrauterine lead exposure and preterm birth. Environmental Research Section A. 1999;81:297–301. doi: 10.1006/enrs.1999.3984. [DOI] [PubMed] [Google Scholar]
  67. United States Environmental Protection Agency (US EPA) [Accessed 04/01/2013];Toxics Release Inventory (TRI) Explorer. Available at: http://iaspub.epa.gov/triexplorer/tri_release.chemical.
  68. Yang C-Y, Chang C-C, Tsai S-S, Chuang H-Y, Ho C-K, Wu T-N. Arsenic in drinking water and adverse pregnancy outcome in an arseniasis-endemic area in northeaster Taiwan. Environmental Research. 2003;91:29–34. doi: 10.1016/s0013-9351(02)00015-4. [DOI] [PubMed] [Google Scholar]
  69. Zhen H, Lawson AB, McDermott S, Pande Lamichhane A, Aelion CM. A spatial analysis of mental retardation of unknown cause and maternal residence during pregnancy. Geospatial Health. 2008;2:173–182. doi: 10.4081/gh.2008.241. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Zhen H, McDermott S, Lawson AB, Aelion CM. Are clusters of mental retardation correlated with clusters of developmental delay? Geospatial Health. 2009;14:17–26. doi: 10.4081/gh.2009.207. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS761747-supplement.docx (12.3KB, docx)

RESOURCES