Effects of plumbing systems on human exposure to disinfection byproducts in water: a case study

Disinfection byproducts (DBPs) in municipal water have been a concern due to their possible association with cancer and non-cancer risks to humans (USEPA 2006, 2014; Richardson et al. 2007, 2008; Health Canada 2008). During disinfection, reactions between natural organic matter (NOM) and disinfectant (typically chlorine, chloramines, chlorine dioxide, ozone) form various DBPs, including trihalomethanes (THMs), haloacetic acids (HAAs), haloacetonitriles, haloketones and other known and unknown byproducts (USEPA 2006; Richardson et al. 2007, 2008; Health Canada 2008). In addition to the regulated THMs (e.g. chloroform, CHCl₃; bromodichloromethane, BDCM; dibromochloromethane, DBCM; and bromoform, CHBr₃), emerging THMs including iodo-THMs have been reported in drinking water (Krasner et al. 2006). The iodo-THMs are more toxic than the regulated THMs in drinking water (Woo et al. 2002) while iodo-THMs are not regulated (USEPA 2006; Health Canada 2008) and not much is known about their occurrences in drinking water. Occurrences of iodo-THMs in drinking water are not regularly monitored in many countries.

Exposure to DBPs can occur through ingestion of drinking water and inhalation and dermal contact during regular indoor activities (e.g. showering, bathing and cooking). Ingestion, inhalation and dermal contact are believed to be the major exposure pathways for DBPs (USEPA 2006; Chowdhury & Champagne 2009; Chowdhury 2013). Early studies have reported that inhalation exposure of volatile organic compounds emitted from supply waters could pose a threat to human health (Andelman 1985; McKone 1987). Past studies have also reported risks to humans from exposure to THMs through inhalation and dermal contact during showering and bathing (Lee et al. 2004; Nuckols et al. 2005; Xu & Weisel 2005; Savitz et al. 2006; Semerjian & Dennis 2007). In most of these studies (Nuckols et al. 2005; Xu & Weisel 2005; Uyak 2006; Savitz et al. 2006), exposure assessment was performed based on the levels of these compounds measured in water from distribution systems (WDS).

Consumers typically use water from taps in the house, while the regulatory/monitoring agencies measure DBPs within the WDS (e.g. Drinking Water Surveillance Program in Ontario, US Environmental Protection Agency). Depending on the size of plumbing system, water may stay in plumbing pipes (PP) for a considerable amount of time before it reaches the tap in the house (Dion-Fortier et al. 2009). This stagnation may be even longer during the period of off-peak hours (e.g. midnight to morning). Stagnation of water within PP allows additional reactions between residual NOM and free residual chlorine (FRC), which can increase DBPs. Such increased DBPs may pose an increased risk to human health. In addition, previous studies estimating exposure and risk during showering have assumed that THMs in warm water were equivalent to those of cold water. Warm water (35–45 °C) is typically used for showering, while warming of water during showering can increase THMs formation (Weisel & Chen 1994; Xu et al. 2002; Al-Omari et al. 2004; Chowdhury & Champagne 2009). Recent research also demonstrated that DBPs could increase between WDS and consumer taps due to stagnation within PP and hot water tanks (HWT) (Wu et al. 2001; Weinberg et al. 2006; Dion-Fortier et al. 2009). Use of water after a long period of stagnation in PP and/or heating in HWT may increase DBPs exposure and risks. A recent study addressed possible increase of THMs and health risks associated with warming of water during showering (Chowdhury 2013). However, changes in DBPs due to stagnation of water in PP and/or heating in HWT were not explained in this study.

Human exposure assessments are associated with uncertainty from different sources, including water ingestion rate, human body weight, exposure frequency, life expectancy, water temperature, shower stall volume, water flow rate, shower duration and air exchange rate (Xu & Weisel 2005; Chowdhury & Champagne 2009). Obtaining precise data for these parameters is difficult. Characterization of uncertainty in exposure assessment may provide flexibility in parameter values and a better understanding of human exposure and risks from DBPs. In this study, occurrences of THMs and HAAs in WDS, PP and HWT were investigated for two houses in Dhahran, Saudi Arabia. Possible effects of water stagnation in PP and HWT were investigated. The seasonal variability of DBPs was assessed. THMs and HAAs in WDS (entry points at the houses), PP and HWT were incorporated for human exposure and risk assessment. For each parameter, statistical distributions were developed to address uncertainty.

Occurrences of DBPs in municipal water are affected by disinfectant, type and concentration of NOM, pH, bromide ion, water temperature, contact time and seasonal variability (Singer 1994; Chowdhury & Champagne 2008). Concentrations of THMs and HAAs in WDS, PP and HWT were investigated for two houses in Dhahran, Saudi Arabia. Water samples were collected from January 2012 to December 2012 through 12 sampling programs. At each sampling location, seven types of samples were collected at three different times of the day. The samples were collected in the order of: (i) water entry points of the houses (WDS) after the last use of water in late evening (S1); (ii) cold water samples in early morning before the first use, i.e. first flush to represent PP (S2); (iii) HWT samples in the morning (S3); (iv) WDS samples in the morning (S4); (v) cold water samples from PP in afternoon before water use (S5); (vi) HWT samples in afternoon (S6); and (vii) WDS samples in afternoon (S7). For samples of HWT, a temperature of approximately 55 °C was maintained to prevent risk of burn injuries. The sample collections were performed at the same time of the day for each type of sample (e.g. S1, S2) throughout the experimental period.

In this study, THMs, HAAs, FRC, total chlorine (TC), UV absorbance (UV₂₅₄), total organic carbon (TOC), dissolved organic carbon (DOC), temperature, pH, turbidity and conductivity were measured for each sample. Samples for measuring THMs and HAAs were taken in 100 mL glass vials containing a dechlorinating agent of 100 mg/L (ammonium chloride) and the samples for pH, turbidity, UV₂₅₄, TOC, DOC and conductivity were collected in 125 mL plastic bottles. The samples were transported to the laboratory in a cooler (4 °C). Temperature and pH were measured in situ. THMs were measured by gas chromatography equipped with mass spectroscopy detection (Varian chromatograph, model 3900 equipped with quadrupole mass spectrometer). The analysis was conducted according to the USEPA method 551.1 (USEPA 1995a). The analysis of HAAs was conducted according to USEPA method 552.2 (USEPA 1995b) using gas chromatography with electron capture detector (Perkin Elmer Chromatograph, model AutoSystem XL). Further details on the extraction of THMs and derivatization of HAAs can be found in USEPA (1995a, 1995b). The FRC and TC were measured by HACH spectrophotometer (HACH DR 3900 model) following HACH methods 8021 and 8167, respectively. Turbidity was measured with a turbidimeter (HACH model 2100N). TOC and DOC were measured with a Shimadzu TOC analyzer (Model: TOC-L-CSN) following standard method 5310B (APHA, AWWA & WEF 1995). UV₂₅₄ was measured using a spectrophotometer (Genesys 10 UV VIS model) at 254 nm with a 10 mm optical path quartz cell. Prior to measuring DOC and UV₂₅₄, samples were filtered through 0.45 μm membrane filters.

The molecular weight (MW) and octanol-water partition coefficient (K_ow) of THMs are in the ranges of 119.4–252.8 g/mole and 93–128 g/mole, respectively. The MW and K_ow of HAAs are 94.5–296.7 g/mole and 1.7–21.4 g/mole, respectively. The technical report on the assessment of non-occupational exposure to nonionizing chemicals noted that chemicals with K_ow between 0.1 and 10⁵ and MW less than 700 g/mole might be absorbed through human skin (ECETOC 1994). It is plausible that THMs and HAAs may be absorbed through human skin during showering and/or swimming (ECETOC 1994), which may pose a risk to human health.

The USEPA demonstrates that early-life exposure has a greater contribution to cancer appearing later in life. The USEPA suggested the following age-dependent adjustment factors (ADAF) to represent such effects (USEPA 2005):

1. For exposure before 2 years of age (i.e. spanning a 2-year time interval from the first day of birth up until a child's second birthday), a 10-fold adjustment.

2. For exposures between 2 and <16 years of age (i.e. spanning a 14-year time interval from a child's second birthday up until their sixteenth birthday), a three-fold adjustment.

3. For exposures after turning 16 years of age, no adjustment.

It is important to emphasize that these adjustments are combined with corresponding age-specific estimates of exposure to assess cancer risk. This is a departure from the way cancer risks have historically been based upon the premise that risk is proportional to the daily average of lifetime dose. The USEPA (2005) showed an example of lifetime exposure for a carcinogenic chemical with slope factor of 2.0 per mg/kg-day and lifetime average dose of 0.0001 mg/kg-day. Without considering the ADAF, risk was estimated to be 0.0002. With ADAF, the risk was estimated to be 0.00033, which is 1.63 times the risk without the ADAF. In this study, the estimated cancer risks were multiplied by 1.63 to better protect human health. However, upon availability of precise exposure data for these age groups (e.g. 0–2, 2–16 and 16 + years), cancer risks can be predicted for each age group and then normalized over the lifetime.

Time of sampling did not affect the concentrations of THMs and HAAs. As such, THMs and HAAs in WDS (S1, S4 and S7) were combined to obtain a single exposure scenario. The inclusion of data from different times represents overall diurnal variability. Similarly, THMs and HAAs in PP (S2 and S5) and HWT (S3 and S6) were combined to obtain independent exposure scenarios for PP and HWT. These data were characterized through statistical distributions (Table 3). Using these distributions, 5,000 random values were generated through statistical software (Minitab). The statistical distributions for the other input parameters were obtained from Table 1. The CDI for 13 DBPs (four THMs and nine HAAs) are shown in Table 4. The highest CDI was noted for CHCl₃ followed by DCAA, TCAA and BDCM. In most cases, CDI from PP and HWT were higher than from WDS (Table 4). It is to be noted that the CDI represents ingestion, inhalation and dermal routes of exposure for THMs, and ingestion and dermal routes for HAAs, because HAAs are not partitioned significantly from water to air during showering.

On average, CDI of CHCl₃ for PP and HWT were 1.8 and 2.3 times, respectively, higher than the CDI for WDS. The CDI of DCAA and TCAA for PP and HWT were also higher than the CDI for WDS. For DCAA, ratios of CDI for PP to WDS, HWT to WDS, and HWT to PP were 1.5, 2.4 and 1.6, respectively (Table 4). For TCAA, ratios of CDI for PP to WDS, HWT to WDS and HWT to PP were 1.2, 0.8 and 0.7, respectively, indicating that the concentrations of TCAA might have been decreased in the HWT. Past studies have reported that the increase in water temperature could transform TCAA into CHCl₃ and CO₂ (Wu et al. 2001). This transformation phenomenon was observed when water temperature exceeded 40 °C (Dion-Fortier et al. 2009). Overall, for the 13 DBPs, CDI for PP and HWT were 0.6–1.8 and 0.5–2.3 times, respectively, higher than the CDI for WDS. The CDI for DBCM, CHBr₃, monobromoacetic acid (MBAA), DBAA and BCAA were much lower in comparison to the other DBPs (Table 4). It is to be noted that CDI of HAAs were through the ingestion and dermal routes and the intake through inhalation was assumed negligible due mainly to their non-volatile nature (Chowdhury et al. 2014). Further, coefficients for dermal permeation were not available for three HAAs (chlorodibromoacetic acid, CDBAA; bromodichloroacetic acid, BDCAA; and tribromoacetic acid, TBAA). As such, CDI through dermal contact was predicted for the remaining six HAAs. The coefficients of variation (CV), defined as the ratio of standard deviation to mean, were calculated for CDI. For CDI of THMs, CV varied in the ranges of 0.32–0.47, 0.33–0.52 and 0.30–0.45 for the WDS, PP and HWT, respectively. For HAAs, CV was in the ranges of 0.27–0.8, 0.44–0.68 and 0.36–0.52 for the WDS, PP and HWT, respectively.

Three regulated THMs (BDCM, DBCM and CHBr₃) and two HAAs (DCAA and TCAA) were used to predict human health risks. In predicting cancer risks, it is to be noted that the route-specific CDI need to be multiplied by the route-specific SF. However, route-specific SF were not available for all routes. For comparative purposes, the SF through ingestion route from USEPA were used. Upon availability of route-specific SF, these values can be updated in future.

The cancer risks and hazard indices for WDS, PP and HWT sourced water are shown in Table 5. The cancer risks from PP and HWT were 1.46 (0.40–4.3) and 1.68 (0.35–5.1) times the cancer risks from WDS, while the cancer risks from the HWT water were 1.27 (0.46–3.97) times the cancer risks from the PP water. The HI for PP and HWT were 1.73 (0.63–4.37) and 2.37 (0.53–5.12) times the HI for the WDS water. Cancer risks from HAAs were 3.0–3.5 times the cancer risks of THMs. Average cancer risk from THMs and HAAs in WDS were 2.12 × 10⁻⁰⁶ and 6.38 × 10⁻⁰⁶, respectively. HAAs in PP and HWT also had higher cancer risks than the corresponding THMs (Table 5). The higher cancer risks from HAAs may be partially explained by the higher concentrations of DCAA and TCAA than the three THMs (BDCM, DBCM and CHBr₃) considered for cancer risk assessment. Although CHCl₃ concentrations (Table 3) were much higher than DCAA and TCAA, CHCl₃ was not included in cancer risk assessment. However, in case of non-cancer risks, all four THMs were included, and the HI of THMs were higher than the HI from HAAs (e.g. DCAA and TCAA). The ratios of HI from HAAs to THMs were in the range of 0.78–0.93.

Various forms of chlorine or chloramines are used to protect drinking water throughout the WDS. Despite the efforts to remove NOM from drinking water, it is not entirely free from NOM. During the stagnation of water in PP and HWT, reactions between FRC and NOM are likely to form additional DBPs. In this study, water stagnation in PP and HWT has shown to increase the concentrations of DBPs in tap water. Such an increase may be attributed to extended reaction periods, temperature-driven higher reaction rates and/or decomposition of some DBPs. Past studies have reported decomposition of TCAA into CHCl₃ and CO₂ when temperature exceeded 40 °C (Wu et al. 2001; Dion-Fortier et al. 2009). This study observed seasonal variability of THMs and HAAs in WDS, PP and HWT. The higher temperature in summer increased THMs significantly while HAAs were variable. Past studies reported that some HAAs could serve as a source of nutrient for microbiological regrowth in WDS (McRae et al. 2004; Tung & Xie 2009). McRae et al. (2004) reported that monochloroacetic acid (MCAA) culture degraded MCAA and MBAA, while TCAA culture degraded TCAA and MCAA. Past studies have documented lower concentrations of HAAs at the extremities of a large WDS where bacterial activities were much higher, which might have degraded HAAs (Baribeau et al. 2004).

In multi-storey residential or office buildings, this can be an issue as water can be stagnant for hours to several days. Such stagnation can form additional DBPs and the FRC can be depleted, which can compromise microbiological water quality. There is a need to assess water quality from the taps in high-rise buildings. Stagnation of water is possible at the dead zone of a large WDS, which can form additional DBPs in this zone. Due to extended stay, FRC can be exhausted and microbiological quality of water can be compromised in the dead zones. Availability of a chlorine-boosting station may resolve this problem. Further, few regulatory agencies (e.g. USEPA, Health Canada) require DBPs concentrations on quarterly samples. The sampling program should be designed to represent the dead zone of a large WDS. In this study, DBPs were below the regulatory limits of the World Health Organization (WHO 2011) while local regulations are not available. The TOC and DOC were relatively higher while FRC were lower, which can be an issue with respect to water quality at the consumption points. Additional treatment may reduce TOC and DOC. Future study should perform comprehensive investigation through incorporating more sampling points from the dead zones of a WDS and the taps of high-rise buildings to protect water quality. A further challenge related to sampling points is to decide whether samples should be collected from the WDS or from the tap in the house. The populations are generally exposed to tap water. Assessment of human exposure based on tap water concentration may better protect human health.

This study analyzed THMs and HAAs in the WDS, PP and HWT, and predicted cancer risks to humans. Cancer risks from DBPs in the PP and HWT can be higher than the cancer risks from WDS. The study thus demonstrates that the use of DBPs data from the WDS is not adequate to represent the real exposure and risks of DBPs. The study has some limitations. The toxicological information for THMs and HAAs provided by the USEPA was developed using a set of transformations and extrapolations from animal bioassay data. Determination of slope factors was subjected to further assumptions, including a linear relationship between dose and response at low doses and 95-percentile upper value, while the true relationship may be different. Hence, the predicted cancer risks represent 95 percentile upper values. In addition to PP and HWT, indoor handling of drinking water such as aeration, freezing and filtration may also affect risks of DBPs. Further, DBPs are the mixture of many known and unknown compounds. Populations are generally exposed to the DBPs mixture, while many DBPs are yet to be identified. However, approaches to characterize cancer risks from DBPs mixture are not fully established yet. Despite several limitations, this study sheds light on the implications of plumbing systems for human exposure and risks from DBPs in drinking water, which may better protect human health in future.

The author would like to acknowledge the support provided by King Abdulaziz City for Science and Technology (KACST) through the Science & Technology Unit at King Fahd University of Petroleum & Minerals (KFUPM) for funding this work through project No. 12-WAT2402-04 as part of the National Science, Technology and Innovation Plan.