Page 11: Guidelines for Canadian Drinking Water Quality: Guideline Technical Document – Trihalomethanes

9.0 Effects in humans

9.1 Cancer epidemiology

Epidemiological studies conducted prior to 1993 that explored associations between and adverse health outcomes often had limitations, particularly in the area of exposure measurement. In the case-control epidemiological studies conducted prior to 1993, associations were found between ingestion of chlorinated drinking water and the incidences of colon cancer for those aged 60 years or more (Cragle et al., 1985) and bladder cancer among non-smokers (Cantor et al., 1985, 1987). In the investigation by Cantor et al. (1985), which involved 1244 cases and 2500 control subjects who had never been exposed in high-risk occupations for bladder cancer and for which detailed information on geographic mobility, water source (non-chlorinated ground source or chlorinated surface source for 50% of their lifetime), and potential confounders was collected, there was a positive association between bladder cancer risk, level of tap water ingestion, and duration of exposure, predominantly among study subjects with long-term residence in communities with chlorinated surface water (NAS, 1987). Among non-smokers, there was an association between water intake and relative risk, and the odds ratio for those over 60 with more than median surface water intake compared with lifelong groundwater consumers was 2.3.

There has been an ongoing effort since 1993 to improve the design of these epidemiological studies in order to more clearly identify both the possible agents of concern in chlorinated drinking water and the associated adverse health effects. More recent analytical epidemiological investigations of bladder cancer have been conducted in Colorado (McGeehin et al., 1993), Ontario (King and Marrett, 1996), and Iowa (Cantor et al., 1996). Data reported thus far from a study in Iowa indicate that risk of bladder cancer is not associated with estimates of past exposure to chlorination by-products, except among men who had ever smoked, for whom bladder cancer risk increased with duration of exposure after control for cigarette smoking. No increased relative risk of bladder cancer was associated with exposure to chlorinated municipal surface water supplies, to chloroform, or to other THM species in a cohort of women, but the follow-up period of 8 years was very short, resulting in few cases for study. In Ontario, King and Marrett (1996) found an increased bladder cancer risk with increasing duration of exposure and THM levels. The association was statistically significant and of higher magnitude only after 35 or more years of exposure. The authors use a concept of THM-years to express the cumulative exposure to THM, which incorporates both levels of exposure to THMs and the period of exposure and is measured in µg/L-years. The bladder cancer incidence was about 40% higher among persons exposed to greater than 1956 µg/L-years of THMs in water compared with those exposed to less than 584 µg/L-years. Although it is not possible to conclude on the basis of available data that this association is causal, observation of associations in well-conducted studies where exposures were greatest cannot be easily dismissed. In addition, it is not possible to attribute these excesses to chloroform per se, although it is generally the DBP present at highest concentration in drinking water (IPCS, 2000). In 2002, an expert panel convened by Health Canada to identify critical endpoints for assessment of health risks related to THMs in drinking water also agreed that THMs are used in epidemiological studies as a surrogate for exposure to CBDPs more generally, and the complexity of CDBP mixtures in drinking water makes the assignment of causation to any single component or class of components extremely difficult (Health Canada, 2003a).

In 2002, Health Canada commissioned a review of the non-bladder cancer epidemiology of THMs in drinking water (SENES Consultants Ltd., 2002). The studies reviewed focussed on colon, rectal, pancreatic, kidney, brain, and haematological/lymphoreticular cancer sites. There were only a few studies with significant odds ratios for colon, rectal, brain and pancreatic cancer; studies were not significant for kidney and the blood-related cancers.

For colon cancer, there were two studies showing a statistically increased risk of colon cancer with exposure to chlorinated drinking water. King et al. (2000a) showed a significant association only for the male cohort, whereas Doyle et al. (1997) showed one only in females, as only females were considered. The results of the King et al. (2000a) study suggest that there may be different risk factor profiles for the different sexes insofar as there was no significant risk for females. However, the Iowa cohort (Doyle et al., 1997) indicates that this may not be the case.

Results from the studies involving rectal cancer were inconclusive. Of the studies examined, the only study showing significance was a population-based case-control study by Hildesheim et al. (1998). Hildesheim et al. (1998) and Doyle et al. (1997) both used the Iowa population and cancer registry for their studies. Their methodologies differed, in that Hildesheim et al. (1998) used a case-control design, examining rectal and colon cancers for both men and women, while Doyle et al. (1997) used a cohort design, examining only women in the population, prospectively, for colon and rectal cancers. Doyle et al. (1997) found an association only for colon cancer, while Hildesheim et al. (1998) found one for rectal cancer.

The only recent study involving the association between brain cancer and exposure to THMs indicated that such an association exists (Cantor et al., 1999). This study involved the same Iowa-based cohort used by Hildesheim et al. (1998) and Doyle et al. (1997).

In summary, even though recent studies suggest that some association exists between colon, rectal, and brain cancer and exposure to DBPs in drinking water, the data presented in the studies are not sufficient to reliably confirm a dose-response or causal relationship (SENES Consultants Ltd., 2002).

The only study that found any significant relationship between treated water and pancreatic cancer was an ecological study by Koivusalo et al. (1995) involving 56 communities in 1950 in Finland. The inherent limitations and uncertainties associated with ecological studies make it difficult to acknowledge the outcome of this study and raise concerns about confidence in the results.

Several studies have attempted to estimate exposures to THMs or chloroform and the other THM species, but the studies did not consider exposures to other DBPs or other water contaminants, which may differ between surface water and groundwater sources. Because inadequate attention has been paid to assessing exposure to water contaminants in epidemiological studies, it is not possible to properly evaluate the increased relative risks that have been reported. Specific risks may be due to other DBPs, mixtures of by-products, or other water contaminants, or they may be due to other factors for which chlorinated drinking water or THMs may serve as a surrogate (WHO, 1998; IPCS, 2000).

9.2 Reproductive epidemiology

Epidemiological studies have raised concerns regarding the potential effects of exposure to DBPs in drinking water and reproductive and developmental outcomes, supported in part by the findings that some DBPs cause reproductive and developmental toxicity in laboratory animals, albeit at doses much higher than those encountered by humans. In 1997, both Health Canada and the U.S. EPA held scientific panel workshops that concluded that the evidence at the time was insufficient to establish a causal relationship between chlorinated water or THMs and adverse pregnancy outcomes (Mills et al., 1998; IPCS, 2000).

Reif et al. (2000) conducted a critical review of the most recent epidemiological evidence. This review examined studies that used either 1) qualitative exposure assessment, which examined associations between source of water supply or method of disinfection and risk of adverse reproductive outcome or 2) quantitative exposure assessment, relying predominantly on reported concentrations of THMs in drinking water supplies. Reif et al.'s (2000) conclusions were as follows:

  1. Effects on fetal growth: The epidemiological evidence for an association between THMs and effects on fetal growth is inconsistent. Weak but statistically significant associations (odds ratios: 1.2-2.6) with birth weight, low birth weight, and intrauterine growth retardation were described in epidemiological studies at concentrations of ≥61 µg THMs/L (Gallagher et al., 1998), >80 µg THMs/L (Bove et al., 1992), and >100 µg THMs/L (Bove et al., 1995). Increases in risk for intrauterine growth retardation were also reported at concentrations of chloroform and BDCM ≥10 µg/L, although the latter was not statistically significant (Kramer et al., 1992). Conversely, two studies (Savitz et al., 1995; Dodds et al., 1999) were unable to demonstrate a statistically significant association with any of these related outcomes. Among these studies, all adjusted for an indicator of socioeconomic status and for race, or restricted the analysis to caucasians. Smoking was controlled for in all but one (Bove et al., 1995) study. The two largest studies, each with good statistical power, reached different conclusions despite relative similarity in exposure assessment and other methods (Bove et al., 1995; Dodds et al., 1999).

    In a hospital-based study in Italy, Kanitz et al. (1996) reported lower mean birth weights among mothers older than 30 years of age consuming chlorinated water. Kallen and Robert (2000) also reported an effect of chlorine-treated systems on somatic parameters of body length and head circumference, as well as an association with low birth weight and preterm delivery. However, Jaakkola et al. (1999) reported no association between chlorinated water use and measures of fetal growth or prematurity. Yang et al. (2000) found no evidence of an association between low birth weight and chlorination in Taiwan, but municipalities using chlorination had a significantly higher rate of preterm delivery.
  2. Effects on fetal viability: Epidemiological evidence is inconsistent in associating DBPs with an increased risk of spontaneous abortion and stillbirth. Although these endpoints were grouped together in the Reif et al. (2000) report, their mechanisms of induction may differ. Increased rates of spontaneous abortion were reported in a cohort study by Waller et al. (1998) in California with heavy consumption of water (five or more glasses of cold tap water per day) containing ≥75 µg/L of THMs. When specific THMs were considered, only heavy consumption of water containing BDCM (≥18 µg/L) was associated with a risk of miscarriage (IPCS, 2000). An increased risk of spontaneous abortion associated with DBP formation is supported by findings from Aschengrau et al. (1989), who reported a doubling in risk for the consumption of surface water, compared with groundwater and mixed water systems. Savitz et al. (1995) found a statistically significant relationship with increasing concentration of THMs and with the highest sextile of exposure, but there was no relationship with ingested dose or with water source.

    An increased risk of stillbirth was reported for Nova Scotia women exposed to water containing more than 100 µg THMs/L (Dodds et al., 1999). In further analyses of these data, King et al. (2000b) found dose-dependent increases in adjusted risk for stillbirth with exposure to THMs, chloroform, and BDCM. Exposure to BDCM at levels ≥20 µg/L was associated with a doubling in risk. In New Jersey, Bove et al. (1992, 1995) found little evidence for an association with THMs at 80 µg/L, but did report a weak association between stillbirth and consumption of drinking water from surface water systems. Aschengrau et al. (1993) found an association between stillbirth and the use of a chlorinated versus chloraminated surface water supply.
  3. Effects on risk for fetal malformations: Relatively strong associations of several types of congenital anomalies with THMs were described by Bove et al. (1992, 1995). The highest risks were found for central nervous system, oral cleft, and major cardiac defects at THM concentrations above 80 or 100 µg/L. Other studies of neural tube defects (Dodds et al., 1999; Klotz and Pyrch, 1999) and cardiac anomalies (Shaw et al., 1991; Dodds et al., 1999) found lower risks or no evidence of an association with THMs. The literature to date presents an inconsistent pattern of association with congenital anomalies collectively and a lack of consistency with specific anomalies across the relatively few studies that have explored these outcomes.

A 2005 study published by the Awwa Research Foundation was conducted to address the hypothesis that exposure to CDBPs causes pregnancy loss (Savitz et al., 2005). To reflect a range of CDBP concentrations and speciation typical of those found across the United States, three study sites were selected which contained either very low levels of all CDBPs, moderate levels of all CDBPs, or combined low levels of chlorinated species and moderate levels of brominated species. Women in each of these areas who were planning a pregnancy or who were pregnant at less than 12 weeks gestation were recruited into the study. In contrast to the Waller et al. (1998) study, the Savitz et al. (2005) study found women who drank 5 or more glasses per day of tap water with >75 µg/L THMs had the same risk of pregnancy loss (an odds ratio of 1.0) as all other women, indicating no association between exposure and pregnancy loss. They did find some indication that BDCM and DBCM were associated with an increased risk of pregnancy loss. However, the results were generally not supportive of an association between pregnancy loss and exposure to CDBPs.

Page details

Date modified: