Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2025 Jan 9;59(2):1232–1242. doi: 10.1021/acs.est.4c05437

Exploring Demographic Disparities in Private Well Water Testing in North Carolina

Wesley Hayes †, C Nathan Jones ‡, Khalid K Osman §, Lauren A Eaves ∥,⊥, Wilson Mize #, Jon Fowlkes #, Rebecca C Fry ∥,⊥, Kelsey J Pieper †,*
PMCID: PMC11755715  PMID: 39786966

Abstract

graphic file with name es4c05437_0005.jpg

The natural, built, and social environments shape drinking water quality supplied by private wells. However, the combined effects of these factors are not well understood. Using North Carolina as a case study, we (i) estimate the demographic characteristics of the private well population; (ii) evaluate representation in well testing records; and (iii) demonstrate how spatial scale influences knowledge of well-using household demographics and representation in testing. We leverage a statewide database of 117,960 well testing records collected over 20 years and a national model predicting well locations. An estimated 25% well-using households identify as Black, Indigenous, and Persons of Color (BIPOC) and 15% have incomes below the poverty threshold. While there is robust well sampling (an average of 4,269 wells tested annually), we observed that most testing records were from predominately White block groups (BGs). Well-using households that did not participate in state testing were 2.4 times more likely to be from predominately BIPOC BGs compared predominately White BGs. Due to the spatial heterogeneity of the well population, demographic differences in well populations were more evident using higher resolution data. Multifaceted testing approaches that couple government-driven efforts with localized studies that engage underrepresented communities are needed to facilitate evidence-based management.

Keywords: well water, demographics, environmental justice, well water testing, spatial scale

Short abstract

NC well population is demographically and geographically diverse, which is not well-represented in state testing records.

Introduction

Characteristics of and interactions between the natural, built, and social environments shape access to safe drinking water supplied by private wells. Well water contamination is driven by the natural (e.g., geologic setting, natural disasters) and built (e.g., well type, proximity to wastewater systems) environments.1−6 However, linkages between well water contamination and social structures within private well communities (i.e., social environment) continue to be a critical point of uncertainty. One method to quantify these social structures is to use demographic characteristics like race and ethnicity, income, and degree of urbanization. However, developing an understanding of who relies on private wells is challenging because there are no current national data on well locations. The 1990 US Census was the last comprehensive survey to collect household-level data on a home’s source of drinking water. Several national-level geospatial models have been developed to predict current trends in well water use by extrapolating the 1990 data to present day using characteristics such as housing density and proximity to roadways.7,8 These models estimate that 14–23 million households (37–53 million residents) rely on private wells for their drinking water supply. Studies estimate that the majority of well users are non-Hispanic White residents of varying socio-economic status.9

Researchers continuously report high rates of well water contamination, with studies documenting up to 58% of well water samples exceeding at least one health-based standard required for municipal systems.1,6 Quantifying disparities in waterborne exposures is challenging due to sampling biases, where participants in well water studies are most often older, college-educated, affluent, and White.5,6,10 While studies report that most well users do not regularly test their drinking water or install treatment that removes contaminants of health concern,1,5,6 environmental justice-oriented studies document disparities within the well population based on demographic characteristics. For example, high-income White well-using households were 10 times more likely to participate in testing and four times more likely to install treatment than low-income Black, Indigenous, and People of Color (BIPOC) well-using households.11 Additionally, BIPOC and low-income households have higher rates of well water contamination than high-income and White households.10−13 Private wells are also often viewed as rural drinking water systems because of the high visibility of well water resilience in rural communities and the lack of nearby municipal supply. However, private wells are prevalent in both urban and rural settings, with a larger quantity of well users in urban areas but a larger percentage of the population using wells in rural areas.14,15

In this study, we focus on North Carolina (NC) where approximately 1 in 4 residents rely on private wells for their drinking water source.7 Private well programs at local health departments provide testing opportunities for residents reliant on private wells.15,16 These programs have generated an extensive set of well testing records that have been used by researchers to explore well water contamination and adverse health outcomes in NC.17−24 Using race, ethnicity, and poverty rates, which represent a few key demographic characteristics,25−27 our study aims to explore the linkages between testing, contamination, and social structures in the NC well community. Here, we (i) estimate the demographic characteristics of the NC private well population; (ii) evaluate representation in well testing records generated through NC private well programs; and (iii) demonstrate how spatial scale influences our understanding of well-using household demographics and representation in testing.

Methods

Data Sets

Our study used the NCWELL database22 which curated 20 years of well testing records generated by the NC Department of Health and Human Services (DHHS). The 15A NC Administrative Code28 defines testing requirements for newly constructed wells, which are enforced in all counties through private well programs at local health departments. Through these programs, environmental health specialists collected well water samples typically at the wellhead or other outdoor spigot after flushing for 5+ minutes and submitted samples for analysis to the DHHS State Laboratory of Public Health (State Lab). As previously reported, these samples were representative of groundwater quality, not drinking water quality.17 Testing fees paid by residents varied based on the local health department, with inorganic analytes ranging from $25–$200 (median of $90).15 It is important to note that our study does not include testing records from private laboratories.

We used the NCWELL database to evaluate the quantity, distribution, and analytes tested. Between 1998 and 2019, there were 122,130 tests performed by the State Lab for up to 28 metals (e.g., lead, arsenic) or anions (e.g., nitrate). The NCWELL database contained 117,960 tests (97%), as tests without a valid sampling data or sampling location and non-water samples were removed. In this study, we focused on the number of well-using households that received a test, as one or more tests (e.g., inorganic panel, nitrate/nitrate) may have been performed at a single residence. Using the geocoded locations provided in the NCWELL database, we determined that 93,913 well-using households were within the database. The number of tests conducted and well-using households per year varied substantially (Figures S1 and S2; Table S1). Prior to July 1, 2008, well water sampling was conducted at the request of residents for contaminants of interest. Since, newly constructed wells have been required to be tested within 30 days of completion for microbial and inorganic contaminants per the 15A NC Administrative Code.28 While existing wells (i.e., systems older than 30 days) can still be tested at the request of residents, testing services for existing wells are infrequently and inconsistently offered within the state.15

We used well location predictions from Murray et al. (2021) to evaluate the number of well-using households predicted in each block and block group (BG; typically contains 250 and 550 housing units).7 This approach updated the 1990 Census water supply data and extrapolated well counts to 2010 populations using the changes in total number of housing units. This method has not been validated in NC but provides the most recent estimates available. The authors estimated that there are 1.6 million wells serving 3.6 million people in NC (Figure S2). Since well location predictions were developed based on the 2010 Census boundaries, we used demographic estimates from the 2010 Census and 2015 American Community Survey. To calculate BIPOC rates, we used the 2010 estimates for total population and total White alone non-Hispanic or Latino population at the block, BG, tract, and county levels. To calculate poverty rates, we used the 2015 estimate for total population and population living below the poverty limit at the BG level. In addition, we used the 2018 NC Department of Transportation Smoothed Urban Boundaries to classify urban and rural BGs. This boundary was generated using the US Census data urban-rural classification, which is based population and/or housing density.29 We classified BGs as urban and rural by identifying if the centroid of the BG was within outside urban areas and clusters (Section S1).

Demographic Characterization

We calculated the estimated demographics of households reliant on well water (“well population”), households reliant on well water that participated in testing through the local health department (“well testing population”), and all households in the state (“state population”).

We estimated demographics of the well and state populations using a weighted average approach (herein referred to as our “geospatial analysis”). We used the following equation:

graphic file with name es4c05437_m001.jpg

where W is the weighted average of the demographic estimate within the targeted spatial extent (e.g., county, state); n is the number of blocks or BGs within the targeted spatial extent; wi is the weight applied to the demographic estimate determined by the number of total households or well-using households within the block or BG from Murray et al. (2021); and Xi is the demographic estimate of the block or BG from the Census. We used block level estimates for BIPOC percentage and BG level estimates for percent poverty [Xi]. To calculate the demographics of the well population, we weighted these estimates by the number of well-using households in the block or BG [wi]. To calculate the demographics of households in the state population, we repeated this approach and weighted estimates by the total number of households in the block or BG [wi].

Our geospatial analysis provided demographics estimates using high resolution data, as we calculated weighted averages based on number of well-using households within the targeted spatial extent. However, this approach assumed that there was equal probability that households within the spatial aggregation unit would be supplied by well water. This assumption may introduce error, particularly in urban fringe areas, as researchers have documented that BIPOC and low-income households can be excluded from neighboring municipal services. For example, MacDonald Gibson et al. (2014) reported that roughly 10% of blocks with >75% White residents compared to 50% of blocks with >75% Black residents were reliant on well water in fringe areas.30

We used a demographic index (herein referred to as our “index analysis”) to consider the intersection of race, ethnicity, and poverty at the BG level. Specifically, we categorized BGs as having a lower or higher BIPOC percentage using a 50% threshold and a lower or higher rate of poverty using the 2015 NC poverty rate of 16.4% (Figure S3). Herein, we refer to BGs that had greater than 50% BIPOC population with a poverty rate below the statewide rate as “low-poverty BIPOC BGs” or those with a poverty rate above as “high-poverty BIPOC BGs”. In keeping, we refer to BGs that had less than 50% BIPOC population with a poverty rate below the statewide rate as “low-poverty White BGs” or those with a poverty rate above as “high-poverty White BGs”. We removed 2 BGs (containing 8 estimated well-using households) in the well population and 7 BGs (containing 8 well-using households that had a testing record) in the well testing population because these BGs did not have data on race/ethnicity and/or poverty rates.

To calculate the demographics of the well and well testing population, we paired our demographic index with the number of well-using households from Murray et al. (2021) and number of well-using households that had a test in the NCWELL database, respectively. All analyses were conducted at the BG level. This approach reduced demographic estimates to a binary variable (i.e., above or below a threshold), which lowered data resolution. However, this simplified approach enabled direct comparison across BGs and allowed for the combined consideration of race, ethnicity, and income. In addition, we delineated the well and well testing population into urban and rural populations using the NC Department of Transportation urban boundaries.

Statistical Analysis

We used the Test of Proportions to compare testing rates between various groups (e.g., urban/rural, demographic indices) and calculated odds ratios to determine the relationship between demographics and well-using households that had (i) tested through the local health department as documented in the NCWELL database and (ii) had measurable arsenic concentrations. We used data from our index analyses to determine the impact of BG-level demographics on testing rates and detection of arsenic. The reference group was any well-using household or tested well-using household that was not within the target category. We explored arsenic because of the high rates of measurable arsenic at the county and state level (Table S2). However, over the twenty-year period, the minimum reporting limit (MRL) for arsenic changed from 10 ppb (09/1998–10/1999) to 1 ppb (11/1998–01/2009) to 5 ppb (2/2009–5/2019). We used samples analyzed after February 2009 to explore arsenic using a MRL of 5 ppb. Between August and December 2015, 31 samples were analyzed using a MRL of 1 ppb, and we assigned samples that were 1–5 ppb to be <5 ppb. In total, 65,496 arsenic samples were included in our analyses. All analyses were done in R version 4.2.2 using the tidyverse package.

Exploring the Impact of Scale on Outcomes

The modifiable areal unit problem is a well-known challenge associated with aggregating spatial data. However, how the scale effects impacts well water knowledge is not well understood. We explored the impact of this effect on estimated and predicted outcomes using both the geospatial and index approaches. We quantified how modifying the aggregation unit changed both the distribution of estimated demographics and the relationship between demographics and testing and contamination. For our spatial aggregation units, we explored BGs (n = 4,997 of 6,155), tracts (n = 1,935 of 2,195), and counties (n = 100 of 100) that contained well-using households. First, we examined differences in BIPOC percentage weighted averages based on spatial aggregation unit in our geospatial approach. To do this, we repeated our analysis but changed our data inputs from BG level estimates to tract and county level estimates for BIPOC percentage [Xi] and weighted the estimates by the number of well-using households in the tract or county [wi]. We used the Kruskal–Wallis test to compare the median weighted-average BIPOC percentages across the different spatial scales. Second, we examined differences in relationships between demographic estimates and testing and arsenic contamination based on the scales of data aggregation in our index approach. To do this, we repeated our odds ratio analysis but subsetted our data by county. This resulted in 100 county-level odds ratios when exploring well testing records by indices and 14 county-level odds ratios when exploring measurable arsenic by BIPOC percentage. We removed counties that had no measurable arsenic results as we could not explore the intersection of race, ethnicity, and poverty rates because of insufficient sample size.

Results and Discussion

Demographics of the Well Population

We used well location predictions from Murray et al. (2021)7 and demographic data from the Census in our geospatial and index analyses. Results from this effort document that (i) an estimated 1 in 4 well-using households in NC identify as BIPOC (Table S3); (ii) the urban well population is more affluent than the rural well population but both are racially and ethnically diverse (Table 1; Figure 2); and (iii) demographic differences in well populations are more evident using higher resolution data (Figure 3).

Table 1. Demographic Characterization Using our Index Analysis for the Well and Well Testing Populations and Urban and Rural Well Populations at the Block Group Leveld.

    BIPOC &
 White &
  Total well population High-poverty Low-poverty High-poverty Low-poverty
Statewide
Well populationa
# of well-using households 1,604,750 145,119 88,788 460,978 909,865
Distribution of well population - 9.0% 5.5% 28.7% 56.7%
# of BG with well-using households 4,997 801 309 1,391 2,496
Distribution of BGs with well-using households - 16.0% 6.2% 27.8% 49.9%
Well testing populationb
# of well-using households that tested 93,877 4,853 1,694 29,561 57,769
Distribution of well testing population - 5.2% 1.8% 31.5% 61.5%
% of well-using households that tested 5.8% 3.3% 1.9% 6.4% 6.3%
# of BG with well-using households that tested 4,648 691 255 1,360 2,342
% of BGs with well-using households that tested 93.0% 86.3% 82.5% 97.8% 93.8%
Arsenic testingc
# of well-using households that tested for arsenic 57,602 2,876 872 18,818 35,036
# of well-using households that had >5 ppb arsenic 1,805 46 11 452 1,296
% well-using household that had >5 ppb arsenic 3.1% 1.6% 1.3% 2.4% 3.7%
# of BG with well-using households that tested for arsenic 3,950 488 193 1,200 2,069
% of BGs with well-using households that tested for arsenic 79.0% 60.9% 62.5% 86.3% 82.9%
Urban areas
Urban well populationa
# of well-using households in urban areas 604,962 42,425 67,021 101,625 393,891
Distribution of urban well population - 7.0% 11.1% 16.8% 65.1%
# of BG with well-using households in urban areasc 2,640 463 239 547 1,391
Distribution of BGs with well-using households in urban areas - 17.5% 9.1% 20.7% 52.7%
Urban well testing populationb
# of urban well-using households that tested 23,988 1,041 806 4,946 17,195
Distribution of urban well testing population - 4.3% 3.4% 20.6% 71.7%
% of urban well-using households that tested 4.0% 2.5% 1.2% 4.9% 4.4%
# of urban BG with well-using households that testedc 2,354 374 184 530 1,266
% of urban BGs with well-using households that tested 89.2% 80.8% 77.0% 96.9% 91.0%
Rural areas
Rural well populationa
# of well-using households in rural areas 999,788 102,694 21,767 359,353 515,974
Distribution of rural well population - 10.3% 2.2% 35.9% 51.6%
# of BG with well-using households in rural areasc 2,357 338 70 844 1,105
Distribution of BGs with well-using households in rural areas - 14.3% 3.0% 35.8% 46.9%
Rural well testing populationb
# of rural well-using households that tested 69,889 3,812 888 24,615 40,574
Distribution of rural well testing population - 5.5% 1.3% 35.2% 58.1%
% of rural well-using households that tested 7.0% 3.7% 4.1% 6.8% 7.9%
# of rural BG with well-using households that testedc 2,294 317 71 830 1,076
% of rural BGs with well-using households that tested 97.3% 93.8% 100% 98.3% 97.4%
Open in a new tab
a

Data on number of well-using households was retrieved by Murray et al. (2021).7 Urban and rural boundaries were delineated using the NC DOT urban boundaries.29 We removed the 8 well-using households with missing Census data.

b

Data on number of well-using households that tested was from NCWELLs.22 Removed the 8 well-using households that tested with missing Census data.

c

There were 681 well-using households with a testing record in BGs with no predicted wells based using the method described in Murray et al. (2021).7

d

Testing since February 2009 in NCWELLs. Removed the 5 well-using households that tested for arsenic with missing Census data.

Figure 2.

Figure 2

Open in a new tab

Estimated demographic characteristics of the (A) well and well testing populations and (B) urban and rural well population at the block group level using our index analysis approach. Block groups were categorized as predominately BIPOC or White using a 50% threshold and as low- or high-poverty rate using the 2015 statewide rate of 16.4% as a threshold.

Figure 3.

Figure 3

Open in a new tab

Distribution of weighted average BIPOC percentage at target spatial aggregation unit.

The Well Population is Demographically Diverse with Substantial Differences Between Urban and Rural Populations

From our geospatial analysis, we estimated that 25.3% of well-using households identify as BIPOC and 14.9% have incomes below the poverty threshold (Table S3). We estimated that 32.6% of households in the state population identify as BIPOC and 17.6% have incomes below the poverty threshold, which is in keeping with prior studies suggesting that the well population is predominately White with varying socio-economic status.9 Among well-using BIPOC households, we estimate that 58% identify as Black or African American alone and 26% as Hispanic or Latino (Table S4). We observed that more than a third of the population (38% of well-using households) were in located in urban areas. We observed that the demographic characteristics of the well population varied substantially across the state (Figures 1 and S4 and S5; Table S5), with estimated county-level BIPOC percentages of 4.5% to 76% (median of 23%) and percent poverty of 7.6% to 29% (median of 17%). There was additional variability within counties (Table S6). For example, the estimated BG-level BIPOC percentage in Alamance County ranged from 1.6%–96% (average of 33%) and percent poverty ranged from 0% to 74% (average of 20%).

Figure 1.

Figure 1

Open in a new tab

Demographic estimates for the well-using population in North Carolina using our geospatial and index approaches. County-level weighted averages for (A) BIPOC percentage and (B) poverty percentage for the well-using population. Weighted averages were calculated using block BIPOC and block group poverty rates. (C) Composite demographic estimates for each block group with a private well. Block groups were categorized as predominately BIPOC or White using a 50% threshold and as low- or high-poverty rate using the 2015 statewide rate of 16.4% as a threshold.

When exploring the intersection of race, ethnicity, and income among the well population using our index approach, we observed that 85% of well-using households were in predominately White BGs (Table 1; Figure 2A), of which 34% were in high-poverty BGs and 66% in low-poverty BGs. This was consistent with our geospatial analysis findings, but composite demographic estimates provided additional insights. Among well-using households within predominately BIPOC BGs (14.5%), 62% of well-using households were in high-poverty BGs and 38% were in low-poverty BGs. We found further demographic differences between well-using households in urban and rural areas (Figure 2B), as 24% of urban well-using households were estimated to be in high-poverty BGs compared to 46% of rural well-using households (Test of Proportions, p < 0.001). We observed approximately the same proportion of well-using households in high-poverty BIPOC BGs in both urban and rural settings (7% of total in urban vs 10% of total in rural), but urban settings had more well-using households in low-poverty BIPOC BGs (11% vs 2.2%; p < 0.001).

The Spatial Scale of Analysis Influenced the Observed Outcomes

We observed the impact of scaling effect when modifying the resolution of our input data for our geospatial analysis–the median weighted average BIPOC percentage of the well population increased from 19.6% using BG-level data to 30.2% using county-level data (Figure 3; Kruskal–Wallis, p < 0.05). This 10.6% difference in median values represents an estimated 170,104 BIPOC well-using households. While this is a well-known challenge associated with aggregating spatial data, discrepancies using such well data stem from the spatial heterogeneity of the well population (Figures 1 and S6). Discriminatory housing policies and practices such as redlining and gentrification can greatly influence the demographic structure of a community, resulting in populations that are not spatially uniform in their composition and/or access to water infrastructure.14,31−33 For example, 98% of BGs in the western region were estimated to be predominantly White with varying poverty rates whereas 56% of BGs in the eastern region had high-poverty rates with varying racial and ethnic backgrounds (Table S7). While smaller scales may accurately represent the well population, researchers often use state data sets such as the NCWELL database to evaluate county- to regional-level trends in well water quality. To address the scaling effect, researchers advocate analyzing trends across multiple spatial scales as disparities can be evident at some scales but not others–local issues may not be evident at larger scales while data constraints can limit analysis at small scales.34−37

The challenges associated with the spatial scale of analysis were further compounded by potential error in the well location prediction model used in this study.7 This model extrapolated 1990 Census data to 2010 populations and we speculate that well location prediction errors (and resulting errors in our population estimates) are most likely highest in transitional areas (i.e., peri-urban setting) where homes in block and BGs are supplied by both municipal service and well water. This is an important nuance for the well population because there are documented cases of municipal underbounding in several southern and western states, which primarily impacts Black, Hispanic/Latino, and Indigenous communities.38,39 This has also been documented in NC in communities such as Irongate Drive and Rogers Road, where BIPOC and low-income households have been excluded from neighboring municipal water systems.14,30,40 Identifying underbounded communities can be challenging when using large aggregation unit or low spatial resolution data and relying on modeled well locations, underscoring the importance of collecting and using higher resolution data to provide a more nuanced understanding of the well population.

Overall, our results highlight that the well population is diverse both demographically and geographically. Nonrepresentative sampling and data sets can greatly impact our ability to empower, engage, and equip all people reliant on private wells, especially those in underrepresented communities.41 Developing an evidence-based understanding of the demographic characteristics of the well population is critical first step for developing effective outreach and engagement strategies that promote accessible, available, and affordable strategies for all well users.

Demographics of the Well Testing Population

We used the well testing records from the NCWELL database,22 well location predictions from Murray et al. (2021),7 and demographics data from the Census to explore rates of testing and contamination using our index analysis approach. Results from this effort document that (i) DHHS testing records have been instrumental in advancing well water knowledge (Figure S1); (ii) most testing records were from predominately White BGs (Table 1); and (iii) data aggregation decisions can impact outcomes observed (Figure 4).

Figure 4.

Figure 4

Open in a new tab

Odds ratio of well-using households being from block groups based on not having a well testing record in the NCWELL database (A) across the state using our demographic indices and (B) within each county by block group race/ethnicity and (C) having measurable arsenic (>5 ppb) across the state and within each county arsenic by block group race/ethnicity. Only 71 counties had testing records in both predominately BIPOC and White block groups and 15 counties had testing records with detectable arsenic in both predominately BIPOC and White block groups. Note: odd ratios above 10 or below 0.1 are not shown (ntesting = 1 and narsenic = 1, see Tables S8 and S9). Reference groups were all tests not included in the demographic category selected.

There is Extensive Well Water Sampling in NC Because of State Well Regulations

Since state-mandated sampling requirements were promulgated in 2008, an average of 6,984 wells have been tested annually for 13–28 inorganic analytes. Prior to this, only 3,414 wells were tested annually for three inorganic analytes (Figure S1 and Table S1).22 The extensive influx of well water data has greatly enhanced public health surveillance of well water contamination and adverse health outcomes in NC. For example, researchers have harnessed and analyzed DHHS well testing records, which has led to novel insights on the occurrence of well water contaminants (e.g., arsenic, hexavalent chromium); adverse health outcomes for infants and children (e.g., birth defects, hearing loss) and the general population (e.g., emergency room visits); and hurricane-induced water contamination and exposures.17−24 The spatial distribution of well testing records is also noteworthy, with 93% of BGs in the state having at least one private wells that has been tested by the State Lab. Further, tens of thousands of well testing records are available for both rural and urban areas (69,889 rural wells and 23,988 urban wells have been tested). Testing coverage is also extensive with 97% of rural BGs and 89% of urban BGs with wells have at least one well testing record. This is important as well water hazards in urban and rural settings are different (e.g., differences in housing densities, contamination risks, impact from natural disasters),3,42,43 and require different stewardship approaches. Although many advocate for such state-level regulatory testing requirements, many states have been slow to adopt this type of legislation.44 Thus, NC is an example of the benefit of regulatory-driven sampling requirements to enhance environmental health surveillance.

Most of the Testing Records in the NCWELL Database Were from Predominately White BGs

When exploring the demographic characteristics of well-using households with testing records in the NCWELL database, we found that 93% of well-using households that tested were in predominately White BGs, which was a significant shift compared to the underlying population of 85% (Test of Proportion, p < 0.001; Table 1). In keeping, well-using households that did not have a test within the NCWELL database were 2.4 times more likely to be from predominately BIPOC BGs compared predominately White BGs (Figure 4A). Income was not as strong a driver, as well-using households that did not have a test were only 1.1 times more likely to from high-poverty BGs compared to low-poverty BGs. When considering relationships by race, ethnicity, and poverty rates, we observed similar trends–well-using households that did not have tests within the NCWELL database were 3.3 and 1.9 times more likely to be from low- and high-poverty BIPOC BGs compared to all other BGs.1,6 Our team has documented similar disparities during community science campaigns–low-income BIPOC households were less likely to have participated in testing compared to high-income White households.11 However, we also observed that high-income BIPOC households were more likely to have participated in testing than low-income BIPOC households, which was not consistent with the findings in this study. The differences in magnitude and discrepancy in income-related findings are likely attributed to differences in the populations included in the studies (e.g., only four BIPOC communities were sampled during community science campaigns compared to state-level testing) and scale of analysis (BG- vs household-level results) as well as modeling errors associated with predicted well location (i.e., overpredicting the occurrence of private wells in urban areas). However, both these studies highlight that there are disparities in access to well water testing that need to be addressed.

Selecting an Appropriate Spatial Aggregation Unit is Essential for Accurately Capturing Demographic Relationships

Relationships between demographics and participation in well testing observed in our study varied based on the scale of data aggregation. Aggregating data at the county-level (i.e., calculating odds ratios for the 100 counties), we observed that well-using households that did not have a test were from 7.4 times less likely to 14.7 times more likely to be from predominately BIPOC BGs compared to predominately White BGs (Figure 4B). While the statewide odds ratio we observed was consistent with prior studies highlighting that BIPOC well-using households are 2.3 times less likely to have tested compared to White well-using households,11 this statistic was not always representative of trends in specific counties–the magnitude of this relationship was dampened for some counties and an entirely opposite trend for others.

We continued to explore the impact of spatial aggregation units by examining relationships between demographics and arsenic in well water. We observed that well-using households with measurable arsenic were 2.2 times more likely to be from predominately White BGs than from BIPOC BGs. When repeating our analysis at the county-level (n = 15 counties), we observed that well-using households with measurable arsenic were 2.9 to 41.9 times more likely to be from predominately BIPOC BGs than White BGs (Figure 4C). Arsenic is an ideal example to illustrate the impact of aggregation scale as the highest concentrations and most frequent testing of arsenic in NC are in counties within the Carolina Slate Belt (e.g., Anson, Cabarrus, Davidson, Montgomery, Randolph, Stanly, and Union). To illustrate, 1,985 samples were submitted in Union County (3.4% of all arsenic samples submitted), and 25.4% of samples had measurable arsenic (Table S2). Our results underscore that knowledge and consideration of the underlying characteristics of the community (e.g., local geology in this instance) is imperative to identify and address environmental justice concerns. Scholars are calling for hyperlocal assessments of environmental challenges to identify environmental justice concerns and achieve just and equitable outcomes.45−47

Implications

Regulatory-Driven Sampling Efforts Can Enhance Well Water Surveillance but Have Important Limitations

State-level sampling requirements can generate extensive data on the well community at larger scales, which greatly enhance public health surveillance efforts as evident in NC. However, databases associated with new construction and real estate transfer may have limitations due to underlying housing disparities.48,49 In the NCWELL database, well testing records were primarily associated with newly constructed systems. Thus, our results were most likely shaped by preexisting underlying housing disparities −88% of new home buyers in the US are White.49 Although homeownership rates have increased over the past decade, rates have only increased by 0.4% among Black residents compared to 29% among White residents.48 While real estate-driven well testing mechanisms are powerful in generating extensive data,50,51 understanding the population engaged is essential for ensuring equitable representation and subsequent distribution of resources. Lessons learned from other states highlight the importance of required testing compliance. The Oregon Domestic Well Testing Act has a voluntary compliance approach, which has resulted in variability in the reporting of well testing records with participation declining since 1996.51 In contrast, the New Jersey Private Well Testing Act has mandated testing, and within 4.5 years, the state database contained testing records for 13% of their well population.50 Iowa has demonstrated how removing financial barriers can also enhance participations. Through their Grants to Counties Water Well Program, the state provides free testing to all residents reliant on a private well. Similar to NC, this effort has produced an extensive testing database (e.g., 135,789 nitrate samples have been collected over 20 years).52 However, researchers exploring Ontario’s free well water testing program continue to highlight infrequent participation in testing due to other barriers such as convenience.53,54

Despite NC having one of the largest well water testing databases, this testing mechanism is not satisfying the state well water testing recommendations. The 93,913 private wells that have tested through the State Lab represent roughly 6% of the predicted 1.6 million wells in North Carolina (Table 1). While the NCWELL database represents testing conducted through the private well programs at local health departments, there are several private laboratories. However, costs are a well-known barrier for participating in well water testing,55 and private laboratories are generally more expensive than the State Lab. For all private wells to be tested according to DHHS’s recommendations, every well would be tested yearly coliform bacteria and biyearly inorganics, which would result in an average of 2.4 million samples annually. Well testing fees vary based on counties, but the median coliform testing fees are $50 and inorganic fees are $90.15 When assuming all private wells comply with DHHS’s recommendation, well water testing would generate an average of $152 million in analytic fees annually. Insufficient testing rates are attributed to well-known barriers such as costs, perception, and convenience,55 but the capacity of the testing infrastructure (e.g., capacity of laboratories and health departments) is also a major barrier.

Multifaceted Testing Approaches That Couple Government-Driven Sampling Efforts with Localized Studies are Needed

An evidence-based understanding of the well community is necessary to develop effective outreach and engagement strategies that promote accessible, available, and affordable strategies for all well users. Knowledge of how geology and system construction influence water quality and treatment needs is well-established,1−5,11 but understanding of social drivers shaping the ability of these communities to address natural and anthropogenic contaminants must be expanded. Regulatory-driven sampling efforts largely focus on sampling newly constructed wells and/or during real estate translations, which may not engage the full well population (e.g., inclusion of rental properties, wells constructed prior to testing regulations, wells in data sparse location).44,50,51 Even when testing is available through health departments, some do not allow samples to be collected from construction standards (e.g., no grouting, inadequate wellhead).15 Community science sampling efforts often use convenience sampling approaches to generate large data sets, but there are known testing barriers such as lack of knowledge, risk perception, and costs that shape participation.53 Thus, optimized sampling approaches are needed. Challenges associated with spatial heterogeneity discussed in this study are not unique to private wells. For example, the US EPA provides standardized methods for representative sampling of chemical, physical, and biological conditions to address spatial heterogeneity of streams in streams.57,58 A similar method for well water sampling would establish a more just and equitable baseline for informing policy decisions and provide well users with greater understanding and confidence in the safety of their drinking water.

Acknowledgments

The research was supported by the NASA Applied Sciences Program – Water Resources (Grant No. 80NSSC22K0921); the NASA Rapid Response and Novel Research in Earth Science (Grant No. 80NSSC21K1168); the USEPA “Untapping the Crowd: Consumer Detection and Control of Lead in Drinking Water” (Grant No. 8399375); the UNC Superfund Research Program (Grant No. P42-ES031007); and the National Academies of Sciences, Engineering, and Medicine’s Gulf Research Program Early-Career Research Fellowship.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.4c05437.

  • Additional data and figures demographic, testing, and water quality data and sampling information, and data and figures regarding odds ratios (PDF)

The authors declare no competing financial interest.

Supplementary Material

es4c05437_si_001.pdf (627.5KB, pdf)

References

  1. Swistock B. R.; Clemens S.; Sharpe W. E.; Rummel S. Water Quality and Management of Private Drinking Water Wells in Pennsylvania. J. Environ. Health 2013, 75 (6), 60–67. [PubMed] [Google Scholar]
  2. Pieper K. J.; Krometis L.-A. H.; Benham B. L.; Gallagher D. L. Simultaneous Influence of Geology and System Design on Drinking Water Quality in Private Systems. J. Environ. Health 2016, 79 (2), 1–9. [Google Scholar]
  3. Pieper K. J.; Jones C. N.; Rhoads W. J.; Rome M.; Gholson D. M.; Katner A.; Boellstor D. E.; Beighley R. E. Microbial Contamination of Drinking Water Supplied by Private Wells after Hurricane Harvey. Environ. Sci. Technol. 2021, 55 (12), 8382–8392. 10.1021/acs.est.0c07869. [DOI] [PubMed] [Google Scholar]
  4. Murphy H. M.; Mcginnis S.; Blunt R.; Stokdyk J.; Wu J.; Cagle A.; Denno D. M.; Spencer S.; Firnstahl A.; Borchardt M. A. Septic Systems and Rainfall Influence Human Fecal Marker and Indicator Organism Occurrence in Private Wells in Southeastern Pennsylvania. Environ. Sci. Technol. 2020, 54 (6), 3159–3168. 10.1021/acs.est.9b05405. [DOI] [PubMed] [Google Scholar]
  5. Murray R. T.; Rosenberg Goldstein R. E.; Maring E. F.; Pee D. G.; Aspinwall K.; Wilson S. M.; Sapkota A. R. Prevalence of Microbiological and Chemical Contaminants in Private Drinking Water Wells in Maryland, USA. Int. J. Environ. Res. Public Health 2018, 15 (8), 1686. 10.3390/ijerph15081686. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Pieper K. J.; Krometis L. A. H.; Gallagher D. L.; Benham B. L.; Edwards M. Incidence of Waterborne Lead in Private Drinking Water Systems in Virginia. J. Water Health 2015, 13 (3), 897–908. 10.2166/wh.2015.275. [DOI] [PubMed] [Google Scholar]
  7. Murray A.; Hall A.; Weaver J.; Kremer F. Methods for Estimating Locations of Housing Units Served by Private Domestic Wells in the United States Applied to 2010. J. Am. Water Resour. Assoc. 2021, 57 (5), 828–843. 10.1111/1752-1688.12937. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Johnson T. D.; Belitz K. Domestic Well Locations and Populations Served in the Contiguous U.S.: 1990. Sci. Total Environ. 2017, 607–608, 658–668. 10.1016/j.scitotenv.2017.07.018. [DOI] [PubMed] [Google Scholar]
  9. Hernandez A.; Pierce G. The Geography and Socioeconomic Characteristics of U.S. Households Reliant on Private Wells and Septic Systems. J. Am. Water Resour Assoc. 2023, 59, 1397. 10.1111/1752-1688.13135. [DOI] [Google Scholar]
  10. Flanagan S. V.; Spayd S. E.; Procopio N. A.; Marvinney R. G.; Smith A. E.; Chillrud S. N.; Braman S.; Zheng Y. Arsenic in Private Well Water Part 3 of 3: Socioeconomic Vulnerability to Exposure in Maine and New Jersey. Sci. Total Environ. 2016, 562, 1019–1030. 10.1016/j.scitotenv.2016.03.217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. George A.; Gray K.; Wait K.; Gallagher D.; Edwards M.; Currie J.; Hogan J.; Kwasikpui A. W.; Pieper K. J. Drinking Water Disparities in North Carolina Communities Served by Private Wells. Environ. Justice 2023, 10.1089/env.2022.0100. [DOI] [Google Scholar]
  12. Pace C.; Balazs C.; Bangia K.; Depsky N.; Renteria A.; Morello-Frosch R.; Cushing L. J. Inequities in Drinking Water Quality Among Domestic Well Communities and Community Water Systems, California, 2011–2019. Am. J. Public Health 2022, 112 (1), 88–97. 10.2105/AJPH.2021.306561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Smith T.; Krometis L. A. H.; Hagedorn C.; Lawrence A. H.; Benham B.; Ling E.; Ziegler P.; Marmagas S. W. Associations between Fecal Indicator Bacteria Prevalence and Demographic Data in Private Water Supplies in Virginia. J. Water Health 2014, 12 (4), 824–834. 10.2166/wh.2014.026. [DOI] [PubMed] [Google Scholar]
  14. Brown J.; Acey C. S.; Anthonj C.; Barrington D. J.; Beal C. D.; Capone D.; Cumming O.; Fedinick K. P.; Gibson J. M.; Hicks B.; Kozubik M.; Lakatosova N.; Linden K. G.; Love N. G.; Mattos K. J.; Murphy H. M.; Winkler I. T. The Effects of Racism, Social Exclusion, and Discrimination on Achieving Universal Safe Water and Sanitation in High-Income Countries. Lancet Glob. Health 2023, 11 (4), e606–e614 10.1016/S2214-109X(23)00006-2. [DOI] [PubMed] [Google Scholar]
  15. Wait K.; Katner A.; Gallagher D.; Edwards M.; Mize W.; Jackson C. L. P.; Pieper K. J. Disparities in Well Water Outreach and Assistance Offered by Local Health Departments: A North Carolina Case Study. Sci. Total Environ. 2020, 747, 141173. 10.1016/j.scitotenv.2020.141173. [DOI] [PubMed] [Google Scholar]
  16. Gibson J. M. D.; Pieper K. J. Strategies to Improve Private-Well Water Quality: A North Carolina Perspective. Environ. Health Perspect 2017, 125 (7), 076001. 10.1289/EHP890. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Wilson L.; Hayes W.; Jones C. N.; Eaves L. A.; Wait K. D.; George A.; Freeman B.; Mize W.; Fowlkes J.; Currie J.; et al. Lead Occurrence in North Carolina Well Water: Importance of Sampling Representation and Collection Techniques. Environ. Res. Lett. 2024, 19 (4), 044020. 10.1088/1748-9326/ad2b2c. [DOI] [Google Scholar]
  18. Johnson C. D.; Nandi A.; Joyner T. A.; Luffman I. Iron and Manganese in Groundwater: Using Kriging and GIS to Locate High Concentrations in Buncombe County, North Carolina. Groundwater 2018, 56 (1), 87–95. 10.1111/gwat.12560. [DOI] [PubMed] [Google Scholar]
  19. Defelice N. B.; Johnston J. E.; Gibson J. M. Reducing Emergency Department Visits for Acute Gastrointestinal Illnesses in North Carolina (USA) by Extending Community Water Service. Environ. Health Perspect 2016, 124 (10), 1583–1591. 10.1289/EHP160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hochard J.; Li Y.; Abashidze N. Associations of Hurricane Exposure and Forecasting with Impaired Birth Outcomes. Nat. Commun. 2022, 13 (1), 6746. 10.1038/s41467-022-33865-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Sanders A. P.; Desrosiers T. A.; Warren J. L.; Herring A. H.; Enright D.; Olshan A. F.; Meyer R. E.; Fry R. C. Association between Arsenic, Cadmium, Manganese, and Lead Levels in Private Wells and Birth Defects Prevalence in North Carolina: A Semi-Ecologic Study. BMC Public Health 2014, 14 (1), 955. 10.1186/1471-2458-14-955. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Eaves L. A.; Keil A. P.; Rager J. E.; George A.; Fry R. C. Analysis of the Novel NCWELL Database Highlights Two Decades of Co-Occurrence of Toxic Metals in North Carolina Private Well Water: Public Health and Environmental Justice Implications. Sci. Total Environ. 2022, 812, 151479. 10.1016/j.scitotenv.2021.151479. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Naylor E.; Kelley T.; Easter L.; Iverson G. Evaluation of Nitrate Concentrations and Potential Sources of Nitrate in Private Water Supply Wells in North Carolina. J. Environ. Health Eng. 2018, 80 (9), 16–23. [Google Scholar]
  24. Langley R.; Kao Y.; Mort S.; Bateman A.; Simpson B.; Reich B. Adverse Neurodevelopmental Effects and Hearing Loss in Children Associated with Manganese in Well Water, North Carolina, USA. J. Environ. Occup. Sci. 2015, 4 (2), 62. 10.5455/jeos.20150403060427. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Di Fonzo D. F.; Fabri A.; Pasetto R. Distributive Justice in Environmental Health Hazards from Industrial Contamination: A Systematic Review of National and near-National Assessments of Social Inequalities. Soc. Sci. Med. 2022, 297, 114834. 10.1016/j.socscimed.2022.114834. [DOI] [PubMed] [Google Scholar]
  26. Switzer D.; Teodoro M. P. The Color of Drinking Water: Class, Race, Ethnicity, and Safe Drinking Water Act Compliance. J. Am. Water Works Assoc. 2017, 109 (9), 40–45. 10.5942/jawwa.2017.109.0128. [DOI] [Google Scholar]
  27. Troesken W.Water, Race, and Disease; The MIT Press, 2004. 10.7551/mitpress/7170.001.0001. [DOI] [Google Scholar]
  28. North Carolina Administrative Code NC DEQ: Groundwater Standards https://deq.nc.gov/about/divisions/water-resources/water-planning/classification-standards/groundwater-standards-0 (accessed 2024 November 19).
  29. Smoothed Urban Boundaries North Carolina Department of Transportation, 2023. https://www.nconemap.gov/datasets/NCDOT:ncdot-smoothed-urban-boundaries/about (accessed 2024 November 19).
  30. MacDonald Gibson J.; DeFelice N.; Sebastian D.; Leker H. Racial Disparities in Access to Community Water Supply Service in Wake County, North Carolina. Am. J. Public Health 2014, 104 (12), e45–e45 10.2105/AJPH.2014.10412e45.1.24228652 [DOI] [Google Scholar]
  31. Sterling C. W.; Krometis L.-A.; Pingel T.; Winling L. D. Connections Between Present-Day Water Access and Historical Redlining. Environ. Justice 2023, 10.1089/env.2022.0115. [DOI] [Google Scholar]
  32. Estien C. O.; Wilkinson C. E.; Morello-Frosch R.; Schell C. J. Historical Redlining Is Associated with Disparities in Environmental Quality across California. Environ. Sci. Technol. Lett. 2024, 11 (2), 54–59. 10.1021/acs.estlett.3c00870. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Krings A.; Schusler T. M. Equity in Sustainable Development: Community Responses to Environmental Gentrification. Int. J. Soc. Welf. 2020, 29 (4), 321–334. 10.1111/ijsw.12425. [DOI] [Google Scholar]
  34. Weigand M.; Wurm M.; Dech S.; Taubenböck H. Remote Sensing in Environmental Justice Research - a Review. ISPRS Int. J. Geo-Inform. 2019, 8 (1), 20. 10.3390/ijgi8010020. [DOI] [Google Scholar]
  35. Baden B.; Noonan D.; Turaga R. M. Scales of Justice: Is There a Geographic Bias in Environmental Equity Analysis?. J. Environ. Planni. Manage. 2007, 50 (2), 163–185. 10.1080/09640560601156433. [DOI] [Google Scholar]
  36. Cutter S.; Holm D.; Clark L. The Role of Geographic Scale in Monitoring Environmental Justice. Risk Anal. 1996, 16 (4), 517–526. 10.1111/j.1539-6924.1996.tb01097.x. [DOI] [Google Scholar]
  37. Fisher J. B.; Kelly M.; Romm J. Scales of Environmental Justice: Combining GIS and Spatial Analysis for Air Toxics in West Oakland, California. Health Place 2006, 12 (4), 701–714. 10.1016/j.healthplace.2005.09.005. [DOI] [PubMed] [Google Scholar]
  38. Wells E. C.; Vidmar A. M.; Webb W. A.; Ferguson A. C.; Verbyla M. E.; De Los Reyes F. L.; Zhang Q.; Mihelcic J. R. Meeting the Water and Sanitation Challenges of Underbounded Communities in the U.S. Environ. Sci. Technol. 2022, 56 (16), 11180–11188. 10.1021/acs.est.2c03076. [DOI] [PubMed] [Google Scholar]
  39. Aiken C. S. Race as a Factor in Municipal Underbounding. Annals Assoc. Am. Geogr. 1987, 77 (4), 564–579. 10.1111/j.1467-8306.1987.tb00181.x. [DOI] [Google Scholar]
  40. Heaney C. D.; Wing S.; Campbell R. L.; Caldwell D.; Hopkins B.; Richardson D.; Yeatts K. Relation between Malodor, Ambient Hydrogen Sulfide, and Health in a Community Bordering a Landfill. Environ. Res. 2011, 111 (6), 847–852. 10.1016/j.envres.2011.05.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Virginia Department of Health Private Well Water Information: Commonwealth of Virginia, 2023. https://www.vdh.virginia.gov/environmental-health/onsite-sewage-water-services-updated/organizations/private-well-water-information/ (accessed 2024 November 19).
  42. Stillo F.; Gibson J. M. Exposure to Contaminated Drinking Water and Health Disparities in North Carolina. Public Health 2017, 107, 180–185. 10.2105/AJPH.2016.303482. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Hochard J.; Abashidze N.; Bawa R.; Etheridge R.; Li Y.; Peralta A.; Sims C.; Vogel T. Air Temperature Spikes Increase Bacteria Presence in Drinking Water Wells Downstream of Hog Lagoons. Sci. Total Environ. 2023, 867, 161426. 10.1016/j.scitotenv.2023.161426. [DOI] [PubMed] [Google Scholar]
  44. Bowen K.; Krishna T.; Backer L.; Hodgins K.; Waller L. A.; Gribble M. O. State-Level Policies Concerning Private Wells in the United States. Water Policy 2019, 21 (2), 428–435. 10.2166/wp.2019.205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Valencia A.; Serre M.; Arunachalam S. A Hyperlocal Hybrid Data Fusion Near-Road PM2.5 and NO2 Annual Risk and Environmental Justice Assessment across the United States. PLoS One 2023, 18 (6), e0286406 10.1371/journal.pone.0286406. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Zhang Y.; Smith S. J.; Bell M.; Mueller A.; Eckelman M.; Wylie S.; Sweet E. L.; Chen P.; Niemeier D. A. Pollution Inequality 50 Years after the Clean Air Act: The Need for Hyperlocal Data and Action. Environ. Res. Lett. 2021, 16 (7), 071001. 10.1088/1748-9326/ac09b1. [DOI] [Google Scholar]
  47. Harrison T. R.; Clark A.; Clement A.; Lombard J.; Maranto G.; Parrish A.; Purkis S.; Reamer M.; Collins O.; Lewis C.; Cruz M.; Solache A. Advancing a Hyperlocal Approach to Community Engagement in Climate Adaptation: Results from a South Florida Pilot Study in Two Communities. PLOS Climate 2022, 1 (6), e0000041 10.1371/journal.pclm.0000041. [DOI] [Google Scholar]
  48. Annie Baumann-Mitchell Black-White Home Ownership Gap in North Carolina: North Carolina Housing Finance Agency, 2021. https://www.nchfa.com/black-white-home-ownership-gap-north-carolina. (accessed 2024 November 19).
  49. Lauren Cozzi. More Americans Own Their Homes, but Black-White Homeownership Rate Gap is Biggest in a Decade, NAR Report Finds; National Association of Realtors, National Assocation of Relators, 2023. https://www.nar.realtor/newsroom/more-americans-own-their-homes-but-black-white-homeownership-rate-gap-is-biggest-in-a-decade-nar. (accessed 2024 November 19).
  50. Flanagan S. V.; Spayd S. E.; Procopio N. A.; Chillrud S. N.; Braman S.; Zheng Y. Arsenic in Private Well Water Part 1 of 3: Impact of the New Jersey Private Well Testing Act on Household Testing and Mitigation Behavior. Sci. Total Environ. 2016, 562, 999–1009. 10.1016/j.scitotenv.2016.03.196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Spanier A. J.; Hoppe B. O.; Harding A. K.; Staab J.; Counter M. Private Well Testing in Oregon from Real Estate Transactions: An Innovative Approach toward a State-Based Surveillance System. Public Health Rep. 2011, 126 (1), 107–115. 10.1177/003335491112600115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Lade G. E.; Comito J.; Benning J.; Kling C.; Keiser D. Improving Private Well Testing Programs: Experimental Evidence from Iowa. Environ. Sci. Technol. 2024, 58, 14596. 10.1021/acs.est.4c02835. [DOI] [PubMed] [Google Scholar]
  53. Latchmore T.; Hynds P.; Brown R. S.; Schuster-Wallace C.; Dickson-Anderson S.; McDermott K.; Majury A. Analysis of a Large Spatiotemporal Groundwater Quality Dataset, Ontario 2010–2017: Informing Human Health Risk Assessment and Testing Guidance for Private Drinking Water Wells. Sci. Total Environ. 2020, 738, 140382. 10.1016/j.scitotenv.2020.140382. [DOI] [PubMed] [Google Scholar]
  54. Latchmore T.; Lavallee S.; Hynds P. D.; Brown R. S.; Majury A. Integrating Consumer Risk Perception and Awareness with Simulation-Based Quantitative Microbial Risk Assessment Using a Coupled Systems Framework: A Case Study of Private Groundwater Users in Ontario. J. Environ. Manage. 2023, 331, 117112. 10.1016/j.jenvman.2022.117112. [DOI] [PubMed] [Google Scholar]
  55. Morris L.; Wilson S.; Kelly W. Methods of Conducting Effective Outreach to Private Well Owners - A Literature Review and Model Approach. J. Water Health 2016, 14 (2), 167–182. 10.2166/wh.2015.081. [DOI] [PubMed] [Google Scholar]
  56. Paulsen S. G.; Mayio A.; Peck D. V.; Stoddard J. L.; Tarquinio E.; Holdsworth S. M.; Van Sickle J.; Yuan L. L.; Hawkins C. P.; Herlihy A. T.; Kaufmann P. R.; Barbour M. T.; Larsen D. P.; Olsen A. R. Condition of Stream Ecosystems in the US: An Overview of the First National Assessment. J. North Am. Benthol. Soc. 2008, 27 (4), 812–821. 10.1899/08-098.1. [DOI] [Google Scholar]
  57. Olsen A. R.; Kincaid T. M.; Kentula M. E.; Weber M. H. Survey Design to Assess Condition of Wetlands in the United States. Environ. Monit. Assess. 2019, 191 (1), 268–280. 10.1007/s10661-019-7322-6. [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

es4c05437_si_001.pdf (627.5KB, pdf)

Articles from Environmental Science & Technology are provided here courtesy of American Chemical Society

RESOURCES