Updated weight of evidence for an association between adverse reproductive and developmental effects and exposure to disinfection by-products
Introduction
Chlorination of public water supplies was introduced in Britain and the United States in the first decade of the 20th century. This public health advance virtually eliminated water borne diseases such as cholera, typhoid, and dysentery in developed countries. However, concomitant with drinking water disinfection is the formation of disinfection by-products (DBP), the result of reaction of chemical disinfectants with naturally occurring organic matter (i.e., vegetable and plant material) primarily in surface water. This paper focuses on the by-products of chlorination, the most widely practiced method of water disinfection. The first articles investigating the potential for adverse reproductive and developmental effects of DBP exposure in animals appeared in 1974; the first investigation of these effects in humans appeared in 1989.
The 1996 amendments to the Safe Drinking Water Act directed the U.S. Environmental Protection Agency (EPA) to develop rules balancing the risks between microbial pathogens and DBP. The Stage 1 Disinfectants and Disinfection By-products Rule was proposed in 1998 and the Stage 2 Rule in 2003. Among the most prevalent DBP in chlorinated water are the four trihalomethanes (THM), chloroform, bromoform, chlorodibromomethane (CDBM), and bromodichloromethane (BDCM). Together, these four compounds comprise total THM (TTHM). Other DBP include haloacetic acids (HAA) and haloacetonitriles (HAN). Altogether, EPA has identified nearly 600 DBP (U.S. EPA, 2002).
In 2001, Graves et al. published a critical review of the toxicological and epidemiological evidence to date. This paper is an extension of that effort and updates the weight of evidence analysis. The goal of this paper is to view the totality of the toxicological and epidemiological evidence to judge the overall weight of evidence concerning DBP exposure and reproductive and developmental effects.
Like outcomes are grouped and include the following effects: low birth weight, very low birth weight, preterm delivery, intrauterine growth retardation/small for gestational age, term low birth weight, small body length, and head circumference, congenital anomalies/birth defects (total and by type) including central nervous system anomalies, spontaneous abortion/miscarriage (i.e., early pregnancy loss), stillbirth/fetal death (i.e., late pregnancy loss), and neonatal death.
The tables include all statistical tests concerning an outcome in studies published since Graves et al. (2001). In this paper, “significance” refers to statistical significance at the standard level (i.e., significance at the P = 0.05 level). In the case of odds ratios1 (OR) or relative risks2 (RR) presented with confidence intervals, “statistical significance” means the lower limit of the 95th percentile confidence interval is ⩾1.0. OR and RR are followed by 95th percentile confidence intervals (CI) in parentheses unless otherwise noted. In addition, OR and RR are adjusted values unless otherwise noted. Of particular relevance was our frequent observation of attempt in individual studies to express the belief in the presence of a genuine association between exposure and reproductive outcomes by portraying numerical differences that were not statistically significant as if they were demonstrated associations. We have concluded that such statements do not constitute “positive” findings.
Table 1 lists recent toxicology literature (since Graves et al., 2001) by type of effect from those occurring before or early in pregnancy, through gestation, to after delivery. The toxicological effects are discussed as introductions to similar effects observed in humans. Findings are placed into perspective by considering the doses and circumstances at which effects are manifest, by examining the degree of concordance among the findings, and by considering differences with known or anticipated human exposures.
Table 2 lists all of the epidemiological studies of reproductive and developmental effects and DBP exposure. Study details are presented only for those investigations published since Graves et al. (2001). In addition, this paper summarizes several review articles.
Table 1, Table 2 together with the first tables in Graves et al. (2001) presents the totality of toxicological and epidemiological evidence concerning DBP exposure and reproductive and developmental effects.
As pointed out in Graves et al. (2001), the primary difficulty with the epidemiologic studies has been and continues to be the assessment of exposure. In all epidemiological studies but two, exposure assessment consists of indirect measures of exposure, based on water type (e.g., surface water versus ground water), method of water treatment (e.g., chlorination versus no chlorination), or on routine monitoring of municipal water supplies matched to maternal residence. In a recent article, Dodds et al. (2004) analyzed TTHM, BDCM, and chloroform concentrations in tap water from individual cases and controls. This individual ascertainment of exposure by integration of THM levels from inhalation, ingestion and from absorption exposures at home and at work is a welcome advance in the epidemiological study of DBP and fertility.
Weight of evidence consists of a critical examination of all scientific observations (positive as well as negative). Weight of evidence was judged by applying various characteristics to data bases. The principal elements include the quality of studies, reproducibility and replication of findings, consistency of findings across studies, and concordance among diverse findings (Tardiff and Rodricks, 1987). This paradigm is not the simple counting of studies with either positive or negative findings.
Section snippets
Recent reviews
The appearance of review articles is indicative of interest in the topic, and comprehensive review articles dealing with reproductive and developmental effects and DBP exposure are briefly described in the following paragraphs.
Reif et al. (2000) is a short report on work done for Health Canada that includes an extensive list of citations. In each of three areas—fetal growth, fetal viability, and fetal malformations—the authors explicitly state that the epidemiological evidence is inconsistent.
Recent articles
In the Stage 2 rule making for disinfection by-products, EPA relied on the body of toxicological and epidemiological evidence, much of which was also included in the weight of evidence analysis by Graves et al. (2001). Exceptions are those studies published since that analysis was prepared. The subsequent studies and their contributions to the data base on DBP and reproductive and developmental effects are summarized below in chronological order. Ten new toxicological papers are listed in Table
Low birth weight
In Graves et al. (2001), retarded fetal development (i.e., reduced fetal weight or length) was suggested in the toxicological literature for chloroform and several HAA and HAN, while no effect on fetal weight was seen for bromoform, BDCM, CDBM, or chloral hydrate. In new toxicological studies, this effect was suggested for DBA (Christian et al., 2002 b) but not for BDCM (Christian et al., 2001). When birth weight was found to be lower in treated versus control animals, it was usually associated
Very low birth weight
Very low birth weight (VLBW) was defined as birth weight less than 1500 g in two previous papers and two new articles (Table 3). Like in three previous studies, the latest considering this outcome (Källén and Robert, 2000) observed no significant results. As in the case of low birth weight, the Toledano et al. observations in one of three areas studied are explained equally by the association of VLBW and SES. The weight of evidence now based on four studies is that no evidence exists of an
Preterm delivery
In new toxicology data, Christian et al. (2002a) saw no effect on length of gestation.
Källén and Robert (2000) found borderline significant results when considering preterm to be either <37 or <32 weeks gestation (Table 4). For <37 weeks gestation, Källén and Robert found an OR of 1.09 (CI 1.01–1.17) when comparing chlorination with sodium hypochlorite to no chlorination.
Jaakkola et al. (2001) found a borderline negative association when comparing chlorination/high color to no chlorination/low
Intrauterine growth retardation, small for gestational age, and term low birth weight
These outcomes are suggested by the same toxicological studies considered previously under LBW. The toxicological findings are inconsistent, and tend to be associated with substantial maternal toxicity derived from exposure to systemically toxic doses.
Intrauterine growth retardation (IUGR) or small for gestational age (SGA), previously dealt with by three authors, was considered by nine additional studies (Table 5). Jaakkola et al., 2001, Wright et al., 2003, Wright et al., 2004, Aggazzotti et
Small body length and small head circumference
A new study, Källén and Robert (2000), also included these outcomes (Table 6). They saw no association when considering low BMI (<12 kg/m3), but interestingly they reported an association for high BMI (>16 kg/m3) (OR 1.27, CI 1.19–1.37). Källén and Robert (2000) observed a significant association when considering small body length (<43 cm) (OR 1.97, CI 1.30–2.97). The authors also observed significant results for small head circumference (<31 cm) (OR 1.46, CI 1.07–1.98).
Källén and Robert (2000)
Congenital anomalies/birth defects
The toxicological literature generally discusses teratogenic effects by type. Recently, Christian et al. (2001) demonstrated that BDCM produced several ossification sites at a maternally toxic dose (900 ppm) in rats and rabbits. The BDCM NOAEL for developmental toxicity reported for rats was 45 mg/kg day and for rabbits was 55 mg/kg day. Christian et al. (2002 b) reported that DBA produced no congenital anomalies at the highest dose tested; however, other forms of reproductive toxicity were
Central nervous system anomalies
Previously two authors had considered central nervous system (CNS) anomalies as a group, and four publications included neural tube defects (NTD) or spina bifida, a specific type of NTD. Four new articles on DBP exposure included specific types of CNS anomalies as outcomes (Table 7).
Other individual congenital anomalies
The only new toxicological data dealing with congenital anomalies was work by Christian et al. (2001) on BDCM and skeletal defects in two species. They observed no effects at 450 ppm in a 15 day study despite the presence of mild maternal toxicity.
New epidemiological data for individual anomalies are listed in Table 8. Various types of anomalies were considered in both previous and recent articles.
Spontaneous abortion or miscarriage
In a far more comprehensive study than Waller et al., 1998, Savitz et al., 2005 reported no association between TTHM and SAB even at TTHM concentrations greater than 75 ppb, in contrast to findings reported by Waller et al.; and much lower odds ratios than those of Waller et al., yet not statistically significant associations between SAB and BDCM (Table 9). Savitz et al. also reported lower ORs than Waller et al. for association between SAB and either BDCM or DBCM; however, neither OR was
Stillbirth or fetal death
The constellation of data from laboratory animals provides no support for an association between stillbirth and DBP exposure. In gavage studies of rats, BDCM administered at 75 mg/kg day induced full liter resorptions; however, the means of administration was probably responsible for the observed effects, as discussed in Graves et al., 2001. Early stage fetal mortality, marked by fetal resorptions, was rarely reported and only at very high doses that produced primary toxicity leading to
Neonatal death
One reproductive toxicity study (Christian et al., 2002a) and another of developmental toxicity (Christian et al., 2001) reported no change of pup survival for rats exposed to BDCM in tap water versus controls.
In the epidemiology literature, neonatal death was defined as death within one week of birth. One former (discussed in Graves et al., 2001) and one additional article (Table 10) considered this outcome. Källén and Robert (2000) found no significant results for this outcome. This
Conclusions
Conclusions remain very close to those in Graves et al. (2001) with the following alterations. Chromosomal abnormalities moved from no evidence of an association to inconsistent evidence. Urinary-tract defects moved from suggestive of a positive associations to mixed evidence of an association. Hydrocephaly and anencephaly were added to the list of outcomes. The weight of evidence for still birth/fetal death is presently changed to evidence of no association based in large measure on the recent
Discussion
The updated toxicity weight of evidence found little indication of previously unreported reproductive or developmental toxicity. The relevant data indicate that NOAELs and LOAELs are much higher than known levels of human exposure, and only limited data explore modes of action for reproductive toxicity. In a few instances, mild adverse reactions were reported in fetuses of dams treated at doses that produced maternal toxicity; these particular effects are generally considered secondary to
Acknowledgments
Partial financial support was supplied by the Chlorine Chemistry Council. The authors thank Dr. Carol Graves for her contributions to this paper.
References (51)
- et al.
Weight of evidence for an association between adverse reproductive and developmental effects and exposure to disinfection by-products: a critical review
Regul. Toxicol. Pharmacol.
(2001) - et al.
Drinking water chlorination and delivery outcome-A registry-based study in Sweden
Reprod. Toxicol.
(2000) - et al.
Chlorination by-products (CBPs) in drinking water and adverse pregnancy outcomes in Italy
J. Water Health
(2004) - et al.
Quality of community drinking water and the occurrence of late adverse pregnancy outcomes
Arch. Environ. Health
(1993) - et al.
Serum hormone characterization and exogeneous hormone rescue of bromodichloromethane-induced pregnancy loss in the F344 rat
Toxicol. Sci.
(2004) - et al.
Chronic exposure to dibromoacetic acid, a water disinfection byproduct, diminishes primordial follicle populations in the rabbit
Toxicol. Sci.
(2004) - et al.
Public drinking water contamination and birth outcomes
Am. J. Epidemiol.
(1995) - et al.
Drinking water contaminants and adverse pregnancy outcomes: A review
Environ. Health Perspect.
(2002) - et al.
Bromodichloromethane inhibits human placental trophoblast differentiation
Toxicol. Sci.
(2004) - et al.
Oral (drinking water) developmental toxicity studies of bromodichloromethane (BDCM) in rats and rabbits
Int. J. Toxicol.
(2001)
Oral (drinking water) two-generation reproductive toxicity study of bromodichloromethane (BDCM) in rats
Int. J. Toxicol.
Oral (drinking water) two-generation reproductive toxicity study of dibromoacetic acid (DBA) in rats
Int. J. Toxicol.
The design, evaluation, and interpretation of developmental toxicity tests
Trihalomethanes in public water supplies and adverse birth outcomes
Epidemiology
Relation between trihalomethane compounds and birth defects
Occup. Environ. Med.
Trihalomethanes in public water supplies and risk of stillbirth
Epidemiology
Trihalomethane levels in home tap water and semen quality
Epidemiology
Late pregnancy exposures to disinfection by-products and growth-related birth outcomes
Environ. Health Perspect.
Risk of specific birth defects in relation to chlorination and the amount of natural organic matter in the water supply
Am. J. Epidemiol.
Water chlorination and birth defects: a systemic review and meta-analysis
Arch. Environ. Health
Drinking water contaminants, gene polymorphisms, and fetal growth
Environ. Health Perspect.
Foetal growth and duration of gestation relative to water chlorination
Occup. Environ. Med.
Haloacid induced alterations in fertility and the sperm biomarker SP22 in the rat are additive: validation of an ELISA
Toxicol. Sci.
Relation between stillbirth and specific chlorination by-products in public water supplies
Environ. Health Perspect.
Exposure assessment in epidemiologic studies of adverse pregnancy outcomes and disinfection byproducts
J. Expo. Anal. Environ. Epidemiol.
Cited by (112)
Toxicity and underlying lipidomic alterations generated by a mixture of water disinfection byproducts in human lung cells
2024, Science of the Total EnvironmentAmbient Environment and the Epidemiology of Preterm Birth
2024, Clinics in PerinatologyThe antiestrogen-like activity and reproductive toxicity of 2,6-DCBQ on female zebrafish upon sub-chronic exposure
2022, Journal of Environmental Sciences (China)Halogen substitution reactions of halobenzenes during water disinfection
2022, ChemosphereCitation Excerpt :Nevertheless, about 50% of the total organic chlorine in DBPs belongs to unknown so far compounds (Krasner et al., 2006; Richardson and Temes, 2018). The spectrum of the diseases related to DBPs is rather wide (Richardson et al., 2007; Villanueva et al., 2015; Tardiff et al., 2006; Nieuwenhuijsen et al., 2009), while bladder and colorectal cancer are the dominant ones (Rahman et al., 2010; Villanueva et al., 2004). Humic matter is not the only organics present in water.