Point-of-use chlorine residuals and disinfection byproduct occurrences in rural households served by public water utilities in Appalachian Virginia

More than 300,000 people in the rural Appalachian region of the United States (US) do not have consistent access to in-home safe drinking water (Krometis et al. 2017; Mueller & Gasteyer 2021). A 2021 analysis of US Census American Community Survey data and EPA Enforcement and Compliance History Online data on a national scale identified Appalachia as one of the regions with the highest rates of incomplete plumbing and water violations (Mueller & Gasteyer 2021). Our understanding of which Appalachian communities and populations have the highest risks of exposure to contaminated drinking water is limited by a lack of data at the household point-of-use (Krometis et al. 2017; Darling et al. 2023). Of the ∼345 community water systems (CWSs) in Central Appalachia, ∼20% (67 systems) serve populations of less than 500 people, and many CWS in Central Appalachia have histories of US Safe Drinking Water Act (SDWA) violations (Bagi 2002; Hughes et al. 2005; Mueller & Gasteyer 2021). Consumers of water from some systems in Central Appalachia face frequent water outages, with the most serious violator CWSs declaring boil-water advisories for years (Marcillo & Krometis 2019).

The primary goal of any CWS under the US SDWA is to ensure the microbial safety of the supplied water, which requires successful implementation and management of several physical and chemical processes, such as sedimentation, filtration, and disinfection (Hughes et al. 2005; Weinmeyer et al. 2017). Most water systems maintain microbial water quality via the addition of relatively low concentrations of chlorine to the treated water prior to distribution. This residual disinfectant helps maintain the safety of the potable water during pipeline transport from the CWS to the consumer point-of-use. However, the use of disinfectants such as chlorine is not without risk, as the chemical reaction between chlorine and organic compounds still present in finished drinking water, or added via leaks, can generate disinfection byproducts (DBPs) (Charrois et al. 2004; Mazhar et al. 2020). Exposure to DBPs in drinking water has been associated with various negative health impacts, including an increased risk of cancer, reproductive and developmental problems, and potential damage to the liver, kidneys, and central nervous system (Krasner 2009; Grellier et al. 2015; Villanueva et al. 2015).

Considering the associated health risks, selected DBPs, like trihalomethanes (THMs) and haloacetic acids (HAAs), are regulated by various countries to protect public health. In the United States, the maximum contaminant level (MCL) under the SDWA for total THMs (which include chloroform, bromodichloromethane, dibromochloromethane, and bromoform) is 80 μg/L (EPA 2010; Villanueva et al. 2023). The US also regulates bromate, chlorite, and five HAAs (EPA 2010). The MCL for total THMs in the European Union is very similar to that in the US (i.e., 100 μg/L), despite the fact that many nations, including China, New Zealand, and Turkey, have much more stringent THM regulations, which an equivalent MCL set at 1 μg/L (Villanueva et al. 2023). THMs make up a significant portion of all DBPs, both regulated and unregulated, in chlorinated drinking water (Grellier et al. 2015; Villanueva et al. 2023).

The aim of drinking water guidelines is to strike a balance between the immediate, well-defined risks posed by pathogenic microorganisms and the potential long-term hazards associated with exposure to DBPs (Charrois et al. 2004). It is often much more challenging for small water systems, such as those in Appalachia, to address water safety risks in drinking water compared to larger water systems (Charrois et al. 2004; Pons et al. 2015). Small systems often lack access to advanced water treatment technologies and skilled water operators, while struggling with comparatively high operating costs for fewer customers (Kot et al. 2011; Pons et al. 2015). Moreover, the combination of prolonged water age due to complex distribution systems and resource constraints in many lower-income rural settings makes it challenging for many smaller CWS to control DBPs given a primary focus on meeting SDWA guidelines for microbial contaminants. A recent analysis of data from the EPA Safe Drinking Water Information System indicated that CWSs in rural areas are more susceptible to exceeding the MCL for total THM in comparison to larger water systems (Strosnider et al. 2017). The research revealed that 6.6% of CWSs in rural areas exceeded the MCL for total THM, while only 2.1% of large central metropolitan water systems were found to have committed similar violations (Strosnider et al. 2017).

Previous researchers identified the formation of DBPs in the drinking water of small communities in the US via point-of-use testing (Binkley et al. 2015). However, comprehensive data regarding DBP exposure levels among rural residents in the Appalachian region has been lacking, except for a recent study in Martin County, Kentucky, USA, focusing on DBPs in the rural public drinking water system (Unrine et al. 2024). A comprehensive systematic review and meta-analysis of existing research spanning two decades (from 2000 to 2019), along with a recent manual search of relevant literature, also revealed a lack of studies addressing drinking water DBPs in rural Appalachia (Darling et al. 2023). While earlier research has explored DBPs in private well waters (Aelion & Conte 2004; Elliott et al. 2018), drinking water treatment plants (Wang et al. 2017; Sayess et al. 2021), and municipal water systems (Lewis et al. 2021; Manley et al. 2022) in rural regions of the US, none of these studies examined the potential of DBP formation in small CWSs in rural Appalachia, where a unique geography and socioeconomic history have resulted in long-standing recalcitrant water inequities (Cook et al. 2015; Mueller & Gasteyer 2021). The purpose of this study was therefore to characterize the concentration of THMs, a measure of DBPs, in household-level drinking water samples for households with utility-supplied water in a predominantly rural region of southwest Virginia and assess associations with pH, free chlorine, temperature, and metal ions that may affect the formation of THMs. This work was designed as an in-depth examination of a limited number of homes, and included measures of multiple potential drivers of THMs within multiple household samples (e.g., point-of-use and source) in order to guide future broader efforts in the region.

This study was conducted with consenting households (HHs) in Wise and Lee County, VA, with support from the Appalachian Service Project (ASP), a non-profit organization that provides home repairs and renovations for low-income HHs across Central Appalachia. Water samples were collected from households with utility-supplied water and analyzed in July 2022 by researchers from Virginia Tech, East Tennessee State University (ETSU), and The University of Virginia's College at Wise.

Water quality was examined in each household by collecting two or three sets of samples: (1) tap water (first flush); (2) source water (tap after a 5-min flush); and then if the homeowner indicated these were their primary drinking water source: (3) either bottled water or roadside spring water. Acid-washed and autoclaved sampling bottles were used for collecting water samples, and all the bottles were pre-labeled with household-linked code numbers. Tap/drinking water samples were collected from the household's drinking water source as if a typical household member were about to drink water at that moment, i.e., turning on the faucet, using a bottled water dispenser, or opening a new single-use bottle of water. Source water samples (i.e., 5-min flushed-tap water samples) were collected after sterilizing the faucet using a 70% isopropyl alcohol solution. Immediately after collecting water samples from each household, duplicate measurements were taken for physicochemical parameters such as pH, temperature, and conductivity using Thermo Scientific™ Elite PCTS Pocket Testers (Waltham, MA, USA). Sampling bottles were transported in coolers with ice packs to ensure a steady transport temperature of 4 °C for additional analyses in laboratories at one of the three universities.

Samples were tested for free chlorine using a Hach DR900 portable colorimeter and DBPs (i.e., THMs) using a Hach DR1900 spectrometer (Hach Company, Loveland, CO, USA). Specifically, THMs in this study were assessed using the colorimetric HACH THM Plus™ Method (Method 10132), as has been used in other recent studies as well (Ali et al. 2019). This method has a range of 10–600 ppb, with a sensitivity of 19 ppb under ideal conditions (HACH n.d.). This method measures 11 trihalogenated DBP species, consisting of 4 THM compounds (i.e., chloroform, bromodichloromethane, dibromochloromethane, and bromoform) regulated by the EPA SDWA, in addition to 7 other DBPs. It reports the total sum of these 11 compounds as chloroform (Ali et al. 2019). Free chlorine was analyzed within 8 h after sample collection. Samples collected for DBPs analysis were stored via refrigeration (≤ 5°C) prior to analysis for no more than 14 days, in keeping with standard methods. Water samples for cation analysis underwent acid digestion with 2% trace metal grade nitric acid in bottle prior to analysis for heavy metals and elements, including arsenic, cadmium, chromium, copper, iron, lead, manganese, and silver, using a Thermo Electron iCAP-RQ ICP-MS at Virginia Tech (Standard Methods 3030D, 3125B).

We uploaded pre-specified study protocols to the Open Science Framework before starting data collection (Cohen 2022) and study protocols were formally approved by the Virginia Tech Institutional Review Board (VT-IRB #22-293; ETSU Reliance Agreement). Since the EPA MCLs were established to balance health risk data, treatment methods, and associated cost considerations, our analysis involved examining our findings in relation to concentrations exceeding both full and half EPA MCLs, as well as secondary maximum contaminant levels (SMCLs), as in previous studies (Cohen et al. 2022). Generalized linear models were used with a logit link and a binomial distribution to identify factors that influence THM formation. The binary outcome variable was structured to be 1 if the THM concentration was higher than the EPA MCL of 80 μg/L, and zero otherwise; this analysis was repeated using 40 μg/L (half the recommendation) as the outcome as well. To report model findings as odds ratios (ORs), coefficients were exponentiated, along with corresponding 95% confidence intervals (CIs). To improve visualization, Log10 transformations were applied to selected water quality parameter comparisons. Before transformation, all observations were increased by one and this adjustment addressed observations with zero values and those between zero and one. All the statistical analyses (standard p < 0.05 threshold for significance) and modeling were conducted (by M.R.) using R (v4.3.2), and then replicated (by A.C.) using Stata (Stata/MP v16.1, StataCorp, College Station, TX, USA).

Excessive chlorine is often identified as a primary driver for the formation of DBPs, including THMs. The tap water samples collected from the households in this study did not surpass the EPA's recommended levels for free chlorine residuals (average of 0.40 mg/L; maximum of 1.28 mg/L). Free chlorine residuals in source water samples were similarly compliant (average of 0.73 mg/L; maximum of 2.03 mg/L) (see Table 1). The average residual chlorine concentrations observed in the utility-supplied water samples in this study conform to the guidance documents (Bartram et al. 2007; European Guidelines Working Group 2017; Singh et al. 2020) and expert recommendations (Rasheduzzaman et al. 2023), which advocate for a minimum of ≥0.2 mg/L of free chlorine in the plumbing system to ensure the protection against microbial contamination. However, these levels of free chlorine concentration do have the potential to spur for DBP formation, as reported in previous studies (Rodriguez et al. 2003; Lee et al. 2013). In a previous study on the presence of DBPs in tap water samples where the total THMs concentration ranged from 3.9 to 53.5 μg/L, the mean free chlorine residual concentration was only 0.40 mg/L (Lee et al. 2013).

While only 11% (n = 3) of source water samples were found to have concentrations that exceeded the EPA MCL for THMs, one-third (n = 9) of samples exceeded half of the US EPA MCL for THMs, as illustrated in Table 2. Samples that exceeded half of the EPA MCL for THMs exhibited higher mean pH, conductivity, and chloride concentrations, though no apparent differences were observed for arsenic, copper, aluminum, and iron. Previous studies also reported positive correlations between pH concentration and THM formation (Liang & Singer 2003; Saidan et al. 2016; Sriboonnak et al. 2021). In a previous study, Chowdhury et al. (2010) have reported that increased temperature and chlorine concentration led to an increase in the formation of THMs. Contrary to findings in prior studies (Chowdhury et al. 2010; Masoud et al. 2019), samples below half of the EPA MCL for THMs in this study reported higher average temperatures and chlorine concentrations.

Previous studies of DBP concentrations in point-of-use samples from homes served by major drinking water utilities have reported similar results, i.e., a majority of the water samples were in accordance with established standards (Rodriguez et al. 2003; Bujar et al. 2013). Given the potential health effects, such as bladder cancer, cardiac anomalies, and miscarriages associated with exposure to low levels of DBPs below the EPA MCL, it is important to recognize however that water from households that are in compliance are not necessarily without risk (Siddique et al. 2015; Pang et al. 2022). A prior study investigating the health risks associated with exposure to THMs in urban drinking water systems revealed that, in certain areas, the total cancer risk surpassed the unacceptable risk zone (1 × 10⁻⁴), even when the mean concentration of total THMs remained below the EPA MCL (Siddique et al. 2015). Previous research has also reported that households with elevated THM levels were more inclined to use bottled water for drinking (Font-Ribera et al. 2017), which could explain the increased use of bottled water in the majority of the HHs in our study area (Cohen et al. 2022).

Water samples in this study were obtained from households served by eight different utilities in Lee and Wise counties. Among these eight utilities, source water collected from homes served by three had THM concentrations that exceeded the EPA MCL for THM (see Table 3). Relationships between utilities and THM concentrations above the EPA MCL were not statistically significant, though it is worth noting that Utility H had the highest mean free chlorine concentration and did not exceed the EPA MCL for THMs in its samples (see Table 3), an association at odds with findings from other studies (Roth & Cornwell 2018). Likewise, Utility F, which had samples exceeding the MCL for THMs, also exhibited the highest mean conductivity among all the utilities (see Table 3), a trend consistent with findings from other studies (Unrine et al. 2024). The variations in THM concentrations among all these utilities could be attributed to differences in source water (Unrine et al. 2024), climate, topography, and geology, as natural organic matter (NOM), a precursor to DBPs, is influenced by these factors (Tak & Vellanki 2018).

Regression model results indicate that higher levels of free chlorine, pH, and conductivity were positively associated with the formation of THM levels above the EPA MCL (Table 4). However, none of these associations were statistically significant, and temperature was found to have a non-significant negative association (OR = 0.84, p = 0.49) which contrasts with findings from a recent study that explored DBP formation in a rural Appalachian county (Unrine et al. 2024). In their study, Unrine et al. (2024) reported a positive relationship between conductivity and temperature with THMs, while they observed a negative correlation between free chlorine and THMs. For models where THM concentration was higher than the half EPA MCL, positive associations were only found for pH and conductivity in all the models, but none of the associations were statistically significant. Free chlorine concentration and temperature have a non-significant negative association with THM formation above half the EPA MCL, which contradicts findings from previous studies (Chowdhury et al. 2010; Kumari & Gupta 2022). Nevertheless, previous researchers have also reported that a decrease in residual free chlorine results in an increase in THM concentration (Ye et al. 2009; Unrine et al. 2024) which supports the findings in this study.

When examining utility tap water samples, our final model indicated that higher levels of temperature, free chlorine, and conductivity were positively associated with the formation of THM levels above the EPA MCL (Table 5). However, none of these associations were statistically significant. Previous studies also found similar associations between THM concentration, temperature, and chlorine concentration (Chowdhury et al. 2010; Roth & Cornwell 2018). Similar results were also obtained in the final model, with THM concentrations exceeding half of the EPA MCL (see Table 5).

Logistic regression models were also conducted to assess the impact of selected chemical elements (As, Cu, Al, Cl, and Fe) on the formation of THMs above the EPA MCL of utility source water. Non-significant positive associations between THM formation above the MCL and the elements (As, Al, and Cl) were observed. Ye et al. (2009) also observed positive relationships between THM formation and aluminum (Al) ions, as well as a negative relationship with arsenic (As) ions. In our study, we observed non-significant negative associations between THM formation and elements (Cu and Fe). While some researchers have reported similar findings to ours regarding Cu and Fe (Liu et al. 2012), others have documented increased THM formation that appears to be the result of elevated Cu ion concentrations (Sharma et al. 2017).

One of the primary limitations of this study is the relatively limited number of participating households, which in turn limits generalizability. Moreover, the limited sample size could have influenced the lack of significant relationships for factors affecting THM formation. This is highlighted by a recent study investigating DBP formation in a rural Appalachian county, which identified significant relationships using drinking water samples from 97 individual homes (Unrine et al. 2024) versus the 27 households sampled in our study. Nonetheless, prior research studies concentrating on the nexus of drinking water and health in rural Appalachia and its neighboring regions have often depended on relatively small sample sizes (Patton et al. 2020, 2024; Mulhern et al. 2021; Cohen et al. 2022). The test method used in this study (THM Plus Method – Method 10132) for measuring the concentration of THMs reported the results as only chloroform concentration, which might potentially represent an overestimation of the total THMs present in the samples. Although this test method reports the sum of 11 DBP species as chloroform, the US EPA SDWA guideline values for THMs are specifically for a group of four chemicals, namely chloroform, bromodichloromethane, dibromochloromethane, and bromoform. However, earlier studies have indicated that chloroform is the predominant THMs compound in drinking water utilities, often constituting 90–95% of the total THM values (Rodriguez et al. 2003; Kumari & Gupta 2022). While seasonal variations significantly influence THM concentration (Rodriguez et al. 2003; Siddique et al. 2015; Ratpukdi et al. 2019; Unrine et al. 2024), in this study, samples were collected during the summer month of July, when household THM levels are expected to be elevated. For example, Rodriguez et al. reported that the average THM levels measured during the summer were frequently 2.5–5 times higher than the average levels measured in the winter (Rodriguez et al. 2003). Additionally, this study did not measure total organic carbon (TOC) values, which could serve as an indicator for DBPs in drinking water (Wallace et al. 2002; Tak & Vellanki 2018). However, previous researchers analyzing data from 73 water systems found no apparent correlation between TOC and THM concentration (Consonery et al. 2004). In our study, it is important to acknowledge that while regulatory monitoring typically involves quarterly samplings (EPA 2020), we only collected samples at a single time point. Therefore, the MCL violations identified in our research may not precisely reflect reported violations as per EPA standards. Given these limitations and our study design and sample size, we did not attempt to assess health outcomes potentially associated with DBP exposures, but believe this would be a worthwhile direction for future research studies focused on the Central Appalachia region.

Overall, the majority of samples collected in this study complied with EPA SDWA THM standards, though a minority of households have THM concentrations above the MCL. Even in households where THM levels are assessed to be in compliance with EPA MCLs, prolonged exposures to DBPs may still contribute to adverse health outcomes, including conditions such as cancer, liver disease, and kidney disease. Of the eight utilities examined, only three were associated with source water samples that surpassed the EPA MCL for THM; however, inter-utility differences in THM concentrations were non-significant. Under the SDWA, smaller systems are not required to sample and test for DBPs as frequently as larger systems. Given the relatively high numbers of smaller CWS in the study region, additional research is needed to better understand the extent to which our results may or may not be generalizable to other counties in Central Appalachia, as well as potential drivers of DBP formation in smaller rural systems.

We thank the Appalachia Service Project for their support and assistance with this study, and in particular, Melisa Winburn, Nicole Intagliata, Samantha Lansinger, Annalee Posey, and Caroline Nowak. We also extend thanks to Amanda Darling, Joshua MacRae, Blaine Pennala, Sarah Price, Breanna Lytton, Jeffrey Parks, and Ada Sloop.

Funding support for this study was provided by Virginia Tech, the University of Virginia, and East Tennessee State University. The funders had no role in study design or the decision to publish.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.