Arsenic in US correctional facility drinking water, 2006-2011

Abstract

Background:

Little is known about the quality of drinking water in US correctional facilities (e.g. detention centers, prisons, jails, etc.). Our objective was to determine if incarcerated persons are at risk for chronic, elevated arsenic exposure relative to the non-incarcerated US population, particularly in the Southwestern US where public water and groundwater arsenic concentrations are high compared to the rest of the US.

Methods:

We analyzed 230,158 arsenic monitoring records from 37,086 community water systems (CWSs) from the Environmental Protection Agency’s (EPA) Third Six Year Review of Contaminant Occurrence dataset (covering 2006-2011). We compared six-year average arsenic concentrations and the odds of exceeding the EPA’s 10 μg/L maximum contaminant level (MCL) for CWSs exclusively serving correctional facilities versus all other CWSs in the Southwestern US, where groundwater arsenic concentrations are high.

Results:

Average six-year water arsenic concentrations were higher for Southwestern correctional facility CWSs (6.41 μg/L, 95% CI 3.48, 9.34) compared to all other Southwestern CWSs (3.11 μg/L, 95% CI 2.97, 3.24) and to other CWSs across the rest of the US (1.39 μg/L, 95% CI 1.35, 1.42). In the Southwest, 26.1% (N=6) of correctional facility CWSs versus 5.8% (509) of other CWSs reported six-year arsenic averages exceeding 10 μg/L, corresponding to an odds ratio of 5.70 (95% confidence interval 2.24, 14.52). Correctional facility CWSs in the Southwest were also more likely to report six-year averages exceeding 5 μg/L (the MCL for New Jersey and New Hampshire, N=8, odds ratio 2.77, 95% CI 1.17, 6.54).

Discussion:

Persons incarcerated in the Southwestern US were at disproportionate risk of elevated drinking water arsenic exposure and related disease from 2006-2011. Strict enforcement of EPA regulations and additional technical and financial support for CWSs serving correctional facilities in the Southwest is necessary to protect the health and human rights of incarcerated persons.

INTRODUCTION

Tap water is likely the sole water source available to incarcerated populations, who lack access to alternative drinking water (e.g. bottled water, domestic wells) or point-of-use treatment devices in the event of compromised drinking water quality. Incarcerated populations are also at disproportionate risk of numerous adverse health outcomes including premature mortality (Patterson 2013). To our knowledge, no studies have systematically evaluated drinking water quality in US correctional facilities. Over 90,000 people are reliant on drinking water from community water systems (CWSs, public water systems that serve the same population year round) that exclusively serve correctional facilities located in the Southwestern US. Relative to the rest of the US, the Southwestern US has high concentrations of naturally occurring inorganic arsenic in domestic wells and in public water systems (Alfredo et al. 2014; Ayotte et al. 2017). Elevated arsenic concentrations in Southwestern groundwater reflect the higher likelihood of increased pH, dissolved solids, and redox differences which result from the local arid oxidizing environment (Ayotte et al. 2017). Although the majority of CWSs in the US rely on groundwater, contaminated surface water can also be a relevant source of arsenic in public drinking water systems (Hoover et al. 2017).

Inorganic arsenic is a potent carcinogen of the lung, bladder, and skin and is associated with numerous adverse health outcomes, including cardiovascular disease at low- to moderate-levels of exposure common in US populations (Farzan et al. 2013; International Agency for Research on Cancer 2004; Moon et al. 2018; National Research Council 2013; Grau-Perez et al. 2017; Kuo et al. 2017). Moreover, inorganic arsenic is a known immunotoxicant and is associated with compromised humoral and cell-mediated immune responses and increased risk of lower respiratory tract infections (Ahmed et al. 2014; Attreed et al. 2017; Islam et al. 2007; Nain and Smits 2012; Rahman et al. 2011). The prevalence of infectious diseases, including tuberculosis and viral hepatitis, are significantly higher in incarcerated populations where overcrowding, delays in medical treatment, and limited access to basic hygiene resources are common (Dolan et al. 2016; Hammett et al. 2002).

Although the US Environmental Protection Agency’s (EPA) goal maximum contaminant level (MCL) for arsenic is 0 μg/L, EPA set the current arsenic MCL at 10 μg/L given feasibility and treatment costs (United States Environmental Protection Agency 2000). Public water systems may switch or mix source water, or install arsenic removal treatment systems to reduce arsenic concentrations below the EPA’s MCL.

Our objective was to estimate drinking water arsenic exposure in US correctional facilities to determine the risk for chronic, elevated water arsenic exposure among the incarcerated compared to the non-incarcerated population in the Southwestern US. Using the most recent publicly available EPA data, we compared distributions of water arsenic concentrations for all other CWSs versus CWSs exclusively serving correctional facilities in the Southwestern US. For comparison, we also evaluated water arsenic concentrations in all other CWSs versus CWSs exclusively serving correctional facilities in the non-Southwestern US.

METHODS

US EPA Six Year Review data

We obtained drinking water arsenic monitoring records from the EPA’s Third Six-Year Review (SYR3) of Contaminant Occurrence dataset (United States Environmental Protection Agency 2016). The SYR3 contains monitoring records for public drinking water systems from 2006-2011 voluntarily sent in by states and primacy agencies to support EPA’s review of National Primary Drinking Water Regulations (United States Environmental Protection Agency 2016). SYR3 data represent 95 percent of all public drinking water systems and 92 percent of the total population served by public drinking water systems nationally. From an initial set of 297,354 arsenic monitoring records from 54,845 public water systems, we excluded records from 17,747 systems not categorized as CWSs (CWSs are public water systems that serve the same population year-round), records from 11 CWSs in American Samoa, and records from 1 CWS missing state information (final N= 230,158 records from 37,086 CWSs). We categorized CWSs as serving correctional facilities if system names included the terms “correction,” “detention,” “jail,” “juvenile,” “penitentiary,” or “prison;” two additional facilities for women were found via search for “women” (total N = 155; system names were also checked manually, e.g. “VALLEY STATE PRISON FOR WOMEN”). We categorized the Southwestern US as California, Nevada, Utah, Colorado, Arizona, New Mexico, and Texas, based on the United States Geological Survey groundwater arsenic estimates (Ayotte et al. 2017). EPA-approved analytical methods for arsenic compliance monitoring include inductively coupled plasma-mass spectrometry and atomic absorption.

Statistical analysis

For monitoring records reported as non-detects, we replaced values with the limit of detection divided by the square root of two when the detection limit was available, or with EPA’s minimum reporting limit for arsenic divided by the square root of two (0.35 μg/L) when the detection limit was not reported. We computed six-year average arsenic concentrations for each CWS (N= 37,086). Few CWSs reported records for both “raw” and “finished” (i.e. treated) water samples within the same year (N= 1,182 CWSs), likely reflecting that CWSs without elevated arsenic concentrations in source water are far less likely to utilize treatment systems to reduce arsenic concentrations and therefore do not report “finished” samples for arsenic monitoring. Within a given year, CWSs were assigned the overall water arsenic concentration regardless of whether monitoring samples were labeled as “raw” or “finished.” When the yearly average of arsenic in finished water samples was lower than in raw water samples by at least 1 μg/L (N=294 CWSs), we calculated the yearly average with finished water samples only (assuming only finished water was distributed). Of these 294 CWSs which was assigned at least one yearly arsenic average using finished samples only, 38 reported six year-average arsenic concentrations greater than 10 μg/L. Data from EPA’s fourth Six Year Review (covering years 2012-2018) is not yet available (anticipated release 2023).

We plotted the distribution of six-year arsenic concentration averages using raincloud plots and calculated the percentage of CWSs reporting six-year average arsenic concentrations exceeding the 10 μg/L MCL, separately for correctional facility CWSs vs. all other CWSs. Odds ratios comparing the odds of six-year arsenic averages exceeding the 10 μg/L MCL for correctional facility CWSs versus all other CWSs were determined from logistic regression models. Model 1 evaluated the crude odds ratio of arsenic average exceeding 10 μg/L for correctional facility CWSs versus all other CWSs. Model 2 was adjusted for the size of the population served (using standard EPA categories: <501 / 501 – 3,300 / 3,301 – 10,000 / 10,001 – 100,000 / > 100,000) and source water type (groundwater vs surface water, as reported in the Six Year Review monitoring records; CWSs which reported relying on groundwater under the influence of surface water were categorized as groundwater). At minimum, arsenic monitoring records reported in the SYR3 database should be aggregated to three-year averages to account for non-differential missingness of yearly records by source water type and whether a CWS had a previous MCL violation; this differential missingness results from EPA’s Standardized Monitoring Framework for arsenic (United States Environmental Protection Agency 2004). As a sensitivity analysis, we repeated our analysis using three-year arsenic averages (2006-2008 and 2009-2011), and calculated odds ratios of three-year arsenic averages exceeding the 10 μg/L MCL for correctional facility CWSs versus all other CWSs using generalized estimating equations with unstructured correlation matrices to account for multiple observations per system. We also evaluated odds ratios comparing the odds of six-year arsenic averages exceeding 5 μg/L (the arsenic MCL for New Jersey, recently passed but not yet implemented in New Hampshire) in correctional facility CWSs versus all other CWSs in the Southwest.

RESULTS

Of 8,750 total CWSs located in the Southwest, we identified 23 CWSs exclusively serving correctional facilities; these were located in California (N=15), Arizona (3), New Mexico (2), Nevada (2) and Texas (1). These 23 CWSs served 92,634 people. In the Southwest, six-year average water arsenic concentrations were higher for CWSs exclusively serving correctional facilities versus all other CWSs at the 10^th (1.41 versus 0.35 μg/L) through 95^th (11.34 versus 10.89 μg/L) quantiles (Table 1). Distribution plots for correctional facility CWSs in the Southwest indicate an even spread of water arsenic concentrations across the distribution; for all other CWSs in the Southwest, the distribution was bi-modal, with a larger proportion of systems with very low arsenic (Figure 1). CWSs exclusively serving correctional facilities reported significantly higher mean water arsenic concentrations (6.41 μg/L, 95% CI 3.48, 9.34) compared to all other CWSs in the Southwest (3.11 μg/L, 95% CI 2.97, 3.24) (Figure 1). The mean (95% CI) six-year arsenic concentration for the 15 correctional facilities in California was 7.62 μg/L (95% CI 3.43, 11.80); the mean six-year arsenic concentration for all other CWSs in California (N=2,720) was 3.85 μg/L (95% CI 3.58, 4.12). Correctional facility CWSs in the Southwestern US also reported a significantly higher proportion of systems with six-year arsenic averages exceeding the 10 μg/L MCL compared to all other CWSs (26.1% vs. 5.8%, p<0.001, Table 1), and a significantly higher odds of MCL exceedance in both Model 1 (crude, OR 5.70, 95% CI 2.24, 14.52) and Model 2 which adjusted for tap water source (ground versus surface water) and the size of population served by the CWS (OR 7.52, 95% CI 2.86, 19.78) (Table 1). The six CWSs exclusively serving correctional facilities in the Southwestern US who reported six-year average water arsenic concentrations exceeding 10 μg/L all reported groundwater as the source water type; these facilities are the Rio Consumnes Correctional Center in CA, the Kern Valley State Prison in CA, the CDCR High Desert State Prison in CA, the Chuckawalla Valley/Ironwood State Prison in CA, the Federal Correction Institution in AZ, and the Valley State Prison for Women in CA. Sensitivity analyses comparing the odds of six-year arsenic averages exceeding 5 μg/L comparing CWSs exclusively serving correctional facilities (N=8) versus all other CWSs in the Southwest (N=1,410) remained significant for Model 1 (OR 2.77, 95% CI 1.17, 6.54) and Model 2 (OR 2.97, 95% CI 1.24, 7.15).

Table 1.

Distribution of six-year average arsenic concentrations (μg/L) in community water systems (CWSs) comparing correctional facility CWSs to other CWSs, separately for the Southwestern and non-Southwestern US for the period of 2006-2011.

Region System Type	Southwest			Non-Southwest
	Correctional facilities	Other CWSs	p-value	Correctional facilities	Correctional facilities	Other CWSs	p-value
N	23	8,727		132	28,204
Total population served	92,634	78,103,370		223,030	183,293,294
Minimum	0.31	0.23		0.28	0.10
10%	1.41	0.35		0.35	0.35
25%	1.57	0.35		0.35	0.35
50%	3.10	1.41		0.35	0.35
75%	10.15a	3.25		0.95	0.95
90%	10.83a	7.40		3.15	3.71
95%	11.34a	10.89		5.38	6.20
Maximum	41.11	283.33		9.31	208.33
Geometric mean	3.49	1.41		0.63	0.65
Arithmetic mean	6.41	3.11		1.16	1.39
N (%) CWSs above MCL	6 (26.1%)	509 (5.8%)	p<0.001b	0 (0%)	458 (1.6%)	p=0.1b
Odds ratio (95% CI) of MCL exceedance c
Model 1d	5.70 (2.24, 14.52)	1 (reference)		1.50 (0.66, 3.40)	1 (reference)
Model 2e	7.52 (2.86, 19.78)	1 (reference)		1.63 (0.72, 3.71)	1 (reference)

^a10.15 is the value of the 17th ordered observation, 10.83 is the value for the 20th ordered observation, and 11.34 is the value for the 21st ordered observation.

^bp-value evaluated via logistic regression and chi-squared test;

^cOdds ratios from logistic regression models; models comparing three-year (2006-2008 vs 2009-2011) MCL exceedance for correctional facility vs. non-correctional facility systems yielded similar results (these models used generalized estimating equations with unstructured error structure to account for multiple records per system).

^dModel 1 is unadjusted.

^eModel 2 is adjusted for population-served size (standard EPA categories: <501 / 501 – 3,300 / 3,301 – 10,000 / 10,001 – 100,000 / > 100,000) and source water type (groundwater vs surface water; CWSs reliant on groundwater under the influence of surface water were categorized as groundwater sources). Values below the detection limit were replaced by the detection limit divided by the square root of two. MCL= maximum contaminant level (10 μg/L). R code for full replication (data management and statistical analysis) is publicly available on GitHub (https://github.com/annenigra/prison-arsenic).

Figure 1. Distribution of average arsenic concentrations (μ/L) in correctional facility community water systems (CWSs) versus all other CWSs separately for the Southwestern and non-Southwestern US for the period of 2006-2011.

Filled polygons represent density plots. Box plot upper, middle, and lower hinges correspond to the 25^th, 50^th, and 75^th percentiles, respectively. The six-year average arsenic concentration for each CWS is represented by a dot. For each subgroup, the arithmetic mean and 95% confidence interval (CI) is displayed below the boxplot. The 10 μg/L maximum contaminant level is indicated by the red dashed line. The x-axis is truncated at 54.60 μg/L. Code for the figure is adapted from Allen et al. (Allen et al. 2018)

In the non-Southwestern US, mean water arsenic concentrations did not differ between CWSs exclusively serving correctional facilities (1.16 μg/L, 95% CI 0.58, 1.74) (N=132) and all other CWSs (1.39 μg/L, 95% CI 1.35, 1.42) (N=28,204) (Figure 1). Distribution plots indicate a larger proportion of systems had very low arsenic concentrations both for correctional facility CWS and for all other CWSs (log-right skewed distribution) (Figure 1). None of the CWSs exclusively serving correctional facilities in the non-Southwestern US reported six year average water arsenic concentrations exceeding the 10 μg/L MCL, compared to 1.6% of all other CWSs in the non-Southwest (Table 1). Analyses considering three-year water arsenic averages (2006-2008 and 2009-2011) yielded similar results and interpretations.

DISCUSSION

To our knowledge, this is the first nationwide analysis of drinking water quality in US correctional facilities. In this analysis of compliance monitoring records for 37,086 CWSs across the US from 2006-2011 published by EPA, correctional facility CWSs in the Southwest reported average water arsenic concentrations twice as high as other CWSs located in the Southwest, and more than a quarter reported six-year averages exceeding the EPA’s regulatory MCL of 10 μg/L. Beyond arsenic, incarcerated persons may be at risk of elevated exposure to other drinking water contaminants that may display unique local and regional geographic patterns such as uranium, lead, and nitrate (Hoover et al. 2017). Studies evaluating drinking water quality and other environmental exposures in correctional facilities in the US, however, are lacking.

Arsenic is a naturally occurring element; concentrations in groundwater are influenced by aquifer geochemistry and geologic sources, including ion competition, pH, and reductive dissolution (Ayotte et al. 2017). The odds of six-year water arsenic averages exceeding 10 μg/L in correctional facility CWSs versus all other CWSs in the Southwest remained significant after adjustment for source water type and the size of the population served, indicating that elevated water arsenic concentrations in correctional facilities in the Southwest are not fully explained by differences in water source type (groundwater versus surface water utilization). Fifteen of the twenty-three CWSs exclusively serving correctional facilities in the Southwest were located in California. Our findings suggest that CWS arsenic concentrations in California are slightly higher than concentrations for the entire Southwest (mean 3.85 μg/L versus 3.11 μg/L), indicating that correctional facility locations may partially explain our findings of elevated water arsenic in Southwestern correctional facilities. CWSs reduce arsenic concentrations in drinking water by switching or mixing source water or installing arsenic removal treatment systems. Although CWSs are required by EPA to make yearly consumer confidence reports available to customers, whether incarcerated persons consuming but not purchasing water reliably have access to consumer confidence reports is not known. Regardless, incarcerated populations have limited or no access to alternative drinking water supplies or water treatment devices. Many correctional facilities across the US are served by CWSs which also serve other communities. The quality of drinking water in these correctional facilities may be somewhat protected by the presence of outside stakeholders (e.g. municipalities, community groups, other residents) who are served by the same CWS. We are aware of one correctional facility (Kern Valley State Prison in California) which has been subject to lawsuits filed by persons incarcerated at the facility, and installed an arsenic treatment system in 2012 after citation by the California Department of Health for violating the US EPA arsenic MCL (an additional citation was issued in 2018) (California Water Boards 2018; Sabalow et al. 2019).

Disparities and injustices in water quality may contribute to the excess burden of disease experienced by incarcerated and formerly incarcerated people. Mass incarceration is a substantial public health problem; approximately 2.2 million people, disproportionately black and low-income men, are incarcerated in the US (Wildeman and Wang 2017). Incarcerated populations are at elevated risk for several chronic diseases that are associated with chronic arsenic exposure, including hypertension, diabetes, and infectious diseases (Dolan et al. 2016; Grau-Perez et al. 2017; Kaeble and Cowhig 2016; Kinner et al. 2020; Moon et al. 2018; Wildeman and Wang 2017). The extent to which environmental exposures disproportionally affecting incarcerated populations contribute to increased disease rates or exacerbate the severity of disease in incarcerated populations is not yet known. Although mounting evidence supports adverse health outcomes are associated with chronic, low-level exposure to inorganic arsenic (the US EPA’s MCL goal is 0 μg/L for arsenic and New Hampshire and New Jersey have passed MCLs of 5 μg/L), correctional facility CWSs lack incentives to further reduce water arsenic concentrations that meet regulatory standards (Kuo et al. 2017; Grau-Perez et al. 2017; Moon et al. 2018). Additional state and federal technical and financial assistance is needed to reduce drinking water arsenic concentrations in correctional facilities across the US (including in facilities that are in compliance with the EPA MCL) given the susceptibility of incarcerated populations to adverse health outcomes and the lack of alternative water sources.

Persons incarcerated in the Southwestern US were at disproportionate risk of elevated drinking water arsenic exposure and related disease from 2006-2011. Immediate, aggressive enforcement of water standards for CWSs exclusively serving correctional facilities is critical to protect the health and human rights of all incarcerated persons, including adolescents, pregnant women, and the young children of incarcerated women.