Share this article on:

Exposure to Trihalomethanes in Drinking Water and Small-for-gestational-age Births

Summerhayes, Richard J.; Morgan, Geoffrey G.; Edwards, Howard P.; Lincoln, Douglas; Earnest, Arul; Rahman, Bayzidur; Beard, John R.

doi: 10.1097/EDE.0b013e31823b669b

Background: Trihalomethanes in drinking water have been associated with higher occurrence of small-for-gestational-age (SGA) births, although results have been inconsistent.

Method: We geocoded residential address for mother of live, singleton, term births to 33 water distribution systems in a large metropolitan area of New South Wales, Australia (314,982 births between 1998 and 2004) and classified births into <10th percentile and ≥10 percentile of weight for gestational age. Mean trihalomethane exposure was estimated by trimester and for the entire pregnancy based on monthly sampling in each of the 33 water distribution systems. We estimated the relative risk (RR) of SGA for exposure to trihalomethanes using log-binomial regression adjusting for confounding.

Results: SGA births increased with mother's third-trimester exposure to chloroform (RR=1.04 [95% confidence interval=1.02–1.06], across an interquartile range [IQR]=25 μg/L) and bromodichloromethane (1.02 [1.01–1.04], 5 μg/L). Larger associations were found for SGA less than third percentile. Smoking modified the effects of trihalomethane exposure, with generally larger associations in births to nonsmoking mother and weaker or protective associations in births to smoking mothers.

Conclusions: Mothers' exposures during pregnancy to total trihalomethane as well as to chloroform and bromodichloromethane were associated with SGA. These associations were modified by maternal smoking during pregnancy.

Supplemental Digital Content is available in the text.

From the aUniversity Center for Rural Health - North Coast, Medical School, University of Sydney, Lismore, New South Wales, Australia; bSchool of Health and Human Sciences, Southern Cross University, Lismore, New South Wales, Australia; cNorth Coast Area Health Service, New South Wales, Australia; dCentre for Quantitative Medicine, Duke-NUS Graduate Medical School, Singapore, Singapore; and eSchool of Public Health and Community Medicine, University of New South Wales, Sydney, New South Wales, Australia.

Submitted 19 October 2011; accepted 30 August 2011.

Supported by the Australian Research Council Linkage Grant (LP0348628) and Network for Spatially Integrated Social Science. The authors reported no other financial interests related to this research.

Institution where this work was performed: Northern Rivers University Department of Rural Health, Medical School, University of Sydney, Lismore, NSW Australia.

Supplemental digital content is available through direct URL citations in the HTML and PDF versions of this article (

Correspondence: Richard J. Summerhayes, University Centre for Rural Health - North Coast, 61 Uralba St, Lismore NSW 2480 Australia. E-mail:

Chlorine has been widely used to disinfect drinking water for nearly 100 years, leading to dramatic reductions in morbidity and mortality from water-borne diseases such as typhoid and cholera. Water disinfection is acknowledged as one of the great public health achievements.1 An unintended consequence of water disinfection is chemical reaction of chlorine with natural organic matter and bromide in water to produce a range of compounds called disinfection byproducts (DBPs).2 Trihalomethanes (THMs) were the first DBPs discovered, in 1974. Since then, more than 600 DBP species have been identified, with more than 50% of the total organic halogen content still unidentified.2 Only a small number of DBPs have been assessed by toxicologic studies and even fewer in epidemiologic studies.2

Several toxicologic and epidemiologic reviews have assessed the evidence for associations between DBPs and various health outcomes. There is some evidence for links with bladder cancer, neural tube defects, and small-for-gestational-age- (or intrauterine growth restriction) and stillbirths. Results have been more mixed for colorectal cancers, spontaneous abortion, low birth weight and term low birth weight, and cardiovascular, respiratory, and urinary tract defects. Studies on the associations of DBPs with other cancers, other birth defects, preterm birth, and neonatal death have generally been negative.3 8 A recent meta-analysis showed a small association between total THMs (THM4) and SGA while also identifying study limitations.9 Major limitations include the use of indirect estimates of DBP exposure; total trihalomethane as the only exposure metric; exposure assessment based on utility or municipality-wide average exposure; exposure data often available only on a quarterly basis; and important missing confounders such as maternal smoking, alcohol intake, socioeconomic status or baby's sex. With these limitations of past studies in mind, we investigated associations of SGA births with exposure to chloroform, bromodichloromethane (BDCM), dibromochloromethane (DBCM), and total THM (THM4) in drinking water in the Sydney/Illawarra metropolitan region of New South Wales, Australia.

Back to Top | Article Outline


We obtained birth data from 1997 to 2004 from the NSW Midwives Data Collection, a mandatory population-based surveillance system covering all births in New South Wales public and private hospitals and the small number of home births.10 Information for all live births and stillbirths of at least 20 weeks' gestation or birth weight of at least 400 g is recorded by either the attending midwife or medical practitioner. A validation study of 2% of the birth data in 1998 (n=1703) against other medical records demonstrated high agreement, with more than 70% of the data items having a kappa value of 0.75 or more. These items include gestational age, smoking during pregnancy, preeclampsia, and duration of pregnancy at first antenatal care visit.11

The birth data were linked to the state-wide Birth Defects Register. We excluded premature babies,12 babies with birth defects, nonsingleton births, and infants with gestational age <22 weeks or >43 weeks, as well as those with birth weight >5 standard deviations from the average for gestational age (considered biologically implausible). Where accurate self-reported date of the last menstrual period was not available, gestational age was estimated using a clinical estimate based on ultrasound or clinical examination. A detailed description of exclusion criteria, methods used to derive percentile birth weights stratified by sex and gestational week, and maternal and infant covariates has been published.13

We geocoded mother's residential address at the time of delivery14 and assigned an area-based measure of mothers socioeconomic status (the index of relative socioeconomic disadvantage15) at the Census Collection District level (approximately 80–200 households). The geocoded births were mapped to digital boundaries of the Sydney/Illawarra water distribution system; births falling outside the extent of the spatial layers were excluded. All mapping was done using ArcGIS v9 (2004, ESRI, Redlands CA).

Back to Top | Article Outline

Exposure Estimation

THM data were obtained for the Sydney/Illawarra water supply from 1998 to 2004; these data are described in detail elsewhere.16 The Sydney/Illawarra water utility has a three-level hierarchical structure with 14 delivery systems containing 33 distribution systems and 180 water supply zones. Monthly THM monitoring generally rotates through 3–6 sites in each distribution system on a 3–6-month cycle. During the study period, 5341 THM observations were available. Less than 1% of values were below the detection limit for chloroform and BDCM and less than 5% for DBCM; however, more than 85% of bromoform values were below the detection limit. As with some other studies, we set values flagged as below the detection limit to two-thirds of the detection limit value (detail in eAppendix A1,,18 Due to the large proportion of below-detection-limit values for bromoform, we assessed bromoform only as a component of THM4.

We assigned THM exposure at the distribution-system level, which are defined by water treatment plant and method of disinfection. The 33 distribution systems range in area from 13 to 349 km, with the number of births per system ranging from 62 to 34,008. We used distribution/mo data to calculate average THM and THM4 exposure per trimester and for the whole pregnancy for each birth to a mother whose usual place of residence was within the distribution system. We log-transformed THM concentrations due to their skewed distribution and back-transformed the values for regression modeling and reporting.

To calculate distribution-system-level exposures, the THM data were averaged within each zone to obtain a zone/mo THM concentration (68% of zone/mo values were missing). We then averaged across zones within a distribution system to obtain the distribution/mo THM concentration (13% of distribution/mo values were missing). We estimated missing distribution/mo THM data by replacing missing values with monthly values from the related delivery level of the hierarchy to obtain complete monthly THM data for each distribution system. We categorized disinfectant regimen within a zone using zone/mo monochloramine concentrations exceeding 0.3 mg/L to indicate chloraminated water (Sydney Water Corporation, 2007, written communication). Distribution systems that included zones receiving both chlorinated and chloraminated water were classified as mixed for that month. We used the intracluster correlation coefficient to assess the variability in monthly THM concentrations by year within and between the spatial units of Sydney/Illawarra water supply hierarchy and found greater variability between distribution systems than within.

We had limited data on 9 haloacetic acids and trichloroacetonitrile. A description of the data and correlations with THMs is provided in eAppendix A7 and A8 (

Back to Top | Article Outline

Statistical Analysis

We used a log-binomial model to calculate relative risks (RRs) of SGA<10th percentile of weight for gestational age births compared with appropriate-for-gestational-age (AGA) births (≥10th percentile) for exposure to THMs during each trimester and for the entire pregnancy, adjusting for a core group of covariates identified as potential confounding factors.

Sandwich-error estimates of the relative risks as recommended by Zou19 were calculated in SAS version 9.1.3 (SAS Institute, Cary, NC). In addition, the RELRISK8 SAS macro was used to check the accuracy of log Poisson estimates against log-binomial estimates; differences were negligible.20 We examined the shape of the exposure-response relationship between THM and SGA with an equivalent generalized additive model (GAM)21 including a smoothing spline (3 degrees of freedom) for THM exposure using S-Plus version 6.1 (Insightful Corp., Seattle, WA).

The association of potential confounding factors with SGA was first assessed individually in univariate models. All relevant factors (P < 0.25) were then combined in forced multivariable models to estimate adjusted relative risks. We determined the final list of confounding factors to be included in the multivariate model by assessing colinearity through changes in the effect estimates of the log-odds and their associated standard errors when other covariates were included. The final core model included baby's sex, year of birth, season of birth, duration of pregnancy at first antenatal care visit (≤12weeks or >12 weeks), maternal smoking anytime during pregnancy (yes/no), maternal age, indigenous mother, maternal country of birth (Australia, Asia, other), previous pregnancy, preexisting diabetes, preexisting hypertension, gestational diabetes, preeclampsia, and socioeconomic status (SES). We assessed potential mediation of effects of THM exposure on SGA by the major confounding factors of socio-economic status and smoking by including interactions terms in the model.

We assessed any potential geographical clustering in the residuals of the core multivariate model and found that the intracluster correlation coefficients at the distribution-system level were very small, indicating that more complex models were not required. Correlation between THMs and trimesters were assessed using the Spearman correlation coefficient.

We performed a number of sensitivity analyses to test robustness of the results, including the possible influence of disinfection type and potential threshold effects of the associations between THMs and SGA (eAppendices A,

We present results for SGA<10% (<10th percentile weight for gestational age) for comparison with previously published studies, and also for SGA<3% (<3rd percentile weight for gestational age) (summarized in eAppendix B,

Back to Top | Article Outline


There were 362,013 live singleton births during the study period in the Sydney/Illawarra Local Government Areas. The geocoded births were mapped to Sydney/Illawarra distribution systems; 6248 births (2%) outside the distribution systems were excluded. We also excluded 40,783 infants (11%) for the following reasons: birth defects, stillbirths, multiple births, missing birth weight or gestational age, gestational age <22 weeks or >43 weeks, birth weight >5 standard deviations away from the average for gestational age, births with missing covariate and exposure data, and preterm births before 37 weeks' gestation. This left 314,982 live, singleton, term births for analysis (eAppendix A2,

The characteristics of the study population are presented in Table 1. We found higher proportions of term SGA <10% births among teenage mothers (15.1%), indigenous mothers (15.9%), mothers born in Asia (15.0%), mothers who smoked during the third trimester (17.8%), mothers from lower socioeconomic areas (13.2%), first pregnancies (12.7%), and mothers who attended antenatal care classes after 12 weeks of pregnancy (11.5%). Births excluded due to missing covariate and exposure data were more likely to be SGA if the mothers were indigenous (34.0%), Asian (17.6%), smoked anytime during pregnancy (25.3%), or from lower socioeconomic areas (19.6%) (eAppendix A3,

Table 1

Table 1

Mothers living in areas supplied by chlorinated waters typically had higher THM exposures than women living in areas supplied by chloraminated water (86% of all births) or by mixed water. Table 2 summarizes the mean third-trimester and entire-pregnancy THM exposure for SGA and AGA births. eAppendix A4 ( provides THM exposures stratified by disinfection type. Mean THM exposures were marginally higher for mothers with SGA births than those with AGA births.

Table 2

Table 2

BDCM and chloroform were highly correlated; neither was correlated with DBCM (eAppendix A5, THMs were highly correlated across trimesters (eAppendix A6, Because the third trimester is generally regarded as the critical window for exposure, we have focused on the third-trimester results.

SGA was associated with an interquartile range (IQR) increase in third-trimester exposure to chloroform (RR=1.04, [95% CI=1.02–1.06], IQR=25 μg/L) and BDCM (1.02, [1.01–1.04], 5 μg/L), with similar associations in other trimesters and for the entire pregnancy (Table 3 and eAppendix A9, The results for SGA<3% were similar to those for SGA<10%, although the magnitude of the RRs were generally larger (eAppendix B, For SGA<3%, the results for the Hunter Region were similar to those for Sydney/Illawarra, but with wider confidence intervals. We found no association between THM exposure and SGA<10% births in the Hunter (eAppendix C6,

Table 3

Table 3

We categorized chloroform and BDCM into deciles to examine possible nonlinear associations and found a similar pattern of generally elevated associations in the first several deciles of exposure that increases in magnitude in the higher deciles. The third-trimester associations are shown in Table 4; cut points and full results are included in eAppendix A10 and A11 ( We categorized DBCM as a dichotomous variable due to the small range of DBCM exposure concentrations and found a weak association. The GAM model concentration response plots for chloroform and BDCM suggest a broadly linear response (eAppendix A12, In contrast, the GAM results for SGA<3% suggested a possible threshold for third-trimester associations with chloroform and BDCM (eAppendix B Figure B2-B3,

Table 4

Table 4

Back to Top | Article Outline

Sensitivity Analyses

Sensitivity analyses showed results similar to the primary analysis. We investigated the effects of stratifying our analysis by disinfection type and generally found the larger magnitude associations in chloraminated drinking water compared with chlorinated and mixed water (eAppendix A13, The effect estimates for SGA <3% in eAppendix B5 ( were similar, but in mixed-disinfectant systems, the magnitude of association was larger for third-trimester exposure to chloroform (RR=1.13 [CI=1.04–1.22] for SGA<3% compared with RR=1.07 [CI=1.03–1.12] for SGA<10%). We looked at the effect of excluding estimated distribution/mo data from the assessment of trimester and entire pregnancy exposure and found results similar to the primary analysis (eAppendix A14,

Mothers in areas of lower SES had exposure to higher concentrations (∼14%) of THMs (eAppendix A15, Smokers also had higher levels of THM exposure (∼10%) than nonsmokers (eAppendix A16, We found no evidence on an interaction between SES and THMs, but we did find an interaction between smoking and THMs. In a stratified analysis for smokers and nonsmokers, the association between SGA and third-trimester exposure to chloroform, BDCM, and DBCM increased for nonsmokers, compared with the nonstratified analysis, whereas there was a protective effect in smokers. The stratified-analysis results for SGA and third-trimester THM exposure are summarized in Table 3, with results in eAppendix A17,

We examined whether the associations of chloroform and of BDCM with SGA are independent of DBCM. The two-pollutant model suggests that the effects of chloroform on SGA are independent of the effects of DBCM, whereas the results for BDCM and DBCM are less clear (eAppendix A18a-b,

We examined the relationship between THM exposure and full-term birth weight (with gestational age included as a covariate in a linear regression model). There was a decrease in mean birth weight (−5 g [95% CI=−9 to −1]) with an IQR increase in chloroform, and an increase in mean birth weight of 4 g (2–5) with an IQR increase in DBCM (eAppendix A19,

We investigated the influence of the potential improvement in exposure assessment accuracy due to our use of mother's address data geocoded to distribution systems by comparing our primary results with those produced by an exposure assessment using mother's address mapped to 47 Local Government Areas, a large spatial unit often used for health studies in Australia. We found associations between SGA and THMs similar to the primary results, although with wider confidence intervals, the methods and results are detailed in eAppendix A20–A24 (

Back to Top | Article Outline


Our results suggest that maternal exposure to chloroform, BDCM, and THM4 during pregnancy is associated with reduced fetal weight. Although we found associations across all trimesters, the third-trimester association was generally the largest, consistent with this critical period of fetal growth. We found no clear evidence of potential threshold effects of THMs on SGA<10%, although there was some suggestion of a threshold for SGA<3%. Our investigation of potential threshold effects was limited by the lack of an unexposed population.

The associations between THM4 third-trimester exposure and SGA (RR=1.03) for a 25 μg/L increase (RR=1.07 and RR=1.10, respectively, for upper 2 deciles [73–86 μg/L and ≥86 μg/L], respectively compared with first decile [<32 μg/L]) are broadly consistent with previous studies.22 26 A recent meta-analysis by Grellier et al9 on 9 studies found a similar small association of third-trimester THM4 exposures with SGA (OR [per 10 μg/L]=1.01; odds ratio for THM4 concentrations above 80 μg/L vs. <80 μg/L=1.08).

The associations we found between chloroform and SGA are broadly consistent with 3 previous studies,23,27,28 whereas 3 other studies did not find an association.25,29,30 The associations we found between BDCM and SGA are similar to one previous study,23 whereas 5 other studies reported no association.25,27 30 No previous study has found associations between DBCM and SGA.22,27,28,30 However, the weak association we found with DBCM third-trimester dichotomous exposure (>5.9 μg/L) should be viewed with caution due to the narrow distribution of DBCM exposure in Sydney/Illawarra, and the protective association with DBCM exposure in our analysis of birth weight as a continuous measure.

Although we found stronger relative risks between THM and SGA<3% compared with SGA<10%, it is unclear if these differences are meaningful. The SGA<3% classification includes more pathologically growth-restricted births compared with the SGA<10% group.31,32 Risks for adverse fetal and perinatal birth outcomes increase as birth weight decreases.32 The use of the lower third-percentile birth weight has been recommended to define poor fetal growth in term infants31 and the 2.5th percentile was suggested in a review of DBP studies and birth outcomes.3 Although there is some epidemiologic evidence supporting the association of THMs with SGA, the potential biologic mechanisms for such an effect are not well defined.6

The associations between SGA and exposure to chloroform and BDCM were larger for nonsmokers than overall. Among smokers, these associations were protective. This observation may be related to the relatively large smoking effect on SGA (more than a doubling of risk in smokers compared with nonsmokers), making it difficult to tease out the potential of effects of subtle risk factors such as THMs on SGA, where the effects on SGA may be small. Also, only 12% of mothers in our study smoked at all during pregnancy. The complex relationship between socioeconomic status and smoking may not provide sufficient control of this important confounding factor. The proportion of Hunter mothers who smoked during pregnancy (22%) was nearly double the proportion in Sydney/Illawarra. Possible uncontrolled confounding by smoking may be related to some of the differences between the Hunter and Sydney/Illawarra SGA<10% results (eAppendix C,

The unexpected relationships of SES and smoking with THM concentrations reflect the SES status of suburbs serviced by chlorinated and chloraminated distribution systems within the Sydney/Illawarra region. Chlorinated systems generally have higher THM exposures than chloraminated systems (third-trimester average during the study period: chlorinated THM4=78 μg/L; chloraminated THM4=50 μg/L).

In New South Wales, women >16 years of age who smoke tend to drink less water and greater amounts of other liquids than nonsmokers, and they also tend to drink more bottled water.33 Potential unmeasured confounding from personal drinking habits, as well as the imprecise measure of smoking in our study, may have contributed to the negative association between THMs and SGA in smokers.

Our study comprised one of the largest study populations for analysis of THMs and SGA, and we were able to exclude premature babies and babies with birth defects. We worked closely with the water supply utilities to identify and account for variations in water source and disinfection regimes. Monthly THM data enabled us to account for temporal variations in disinfection byproducts related to season and operational changes in water management. Previous retrospective studies of SGA have generally been based on quarterly THM data, which do not capture shorter-term temporal variation. Our sensitivity analyses provided results similar to our primary analyses.

A potential limitation of this study was the lack of individual exposure data. THMs are volatile, and exposure can occur from activities such as showering and bathing. The THM exposure for a 10-minute shower may be equivalent to drinking 2 L of water.34 Although we attempted to reduce exposure misclassification by geocoding births to distribution systems within the Sydney/Illawarra utility, the influence of misclassification remains unclear. Sensitivity analysis indicated that exposure misclassification at the local-government-area level is nondifferential compared with exposure assessed at the distribution-system level. The increased width of confidence intervals with local-government-area estimates reinforces the importance of assessing exposure by smaller spatial units.

Our study had no information on whether the mother changed residence during pregnancy. A US study on pregnant women and birth defects estimated 15% of mothers changed address over short distances (mean, 5 km) during pregnancy, most often in the first trimester.35 We assumed that homes were connected to the public water supply (not a private well or rainwater tank) and that mothers did not filter or treat their water. Ninety-eight percent of Sydney's population is connected to the public water supply. A recent survey found that 87% of residents use the public water supply as their principal source of drinking water and that 65% do not filter or treat the water.36,37

Despite controlling for a number of important risk factors for SGA, our study lacked data on a number of potential confounders identified in other studies, such as number of cigarettes smoked a day, employment status, marital status, prepregnancy BMI, and caffeine intake.9,25

A large number of maternal, infant, and environmental factors are associated with SGA.12,31 These risks far outweigh the small associations between exposure to THMs and SGA found in Sydney/Illawarra.

Our results suggest associations between mother's exposure to THMs and SGA births. These associations are very small compared with the risks for SGA observed with smoking or low socioeconomic class. Furthermore, the associations may not be directly related to THM4 or its components (including chloroform or BDCM) but rather to some unmeasured DBPs correlated with these THMs. The US Environmental Protection Agency ranks drinking water risks as a high priority because almost everyone drinks public water or uses it for bathing. Potential toxic substances can have important public health effects due to the large exposed populations, even if the risk is small.38 Still, the health risks associated with DBPs remain uncertain, whereas the effects of disinfection in reducing pathogens are well known. Reduction of disinfection byproducts may be desirable, but it should never compromise effective disinfection.39

Back to Top | Article Outline


We acknowledge the following organizations and individuals for their work on this manuscript: NSW Health for data; Mark Angles, Peter Cox, Vicky Whiffin, David Holland and Adam Lovell from Water Services Association of Australia (formerly SWC) for expert advice on Sydney Water data and spatial boundaries; David Holland, Sydney Water Corporation, for advice on disinfection changes and data extraction; Bruce Cole and Pam O'Donoghue from the Hunter Water Corporation for expert advice on the Hunter Water data, water source blending, and spatial boundaries; and Paul Byleveld from NSW Health. We also thank Nel Glass and Stephen Kermode from Southern Cross University, David Muscatello, Therese Dunn, and Paul Houlder for their valuable contributions to this manuscript. This work is part of a PhD thesis by Richard Summerhayes.

Back to Top | Article Outline


1. Centers for Disease Control and Prevention. Ten great public health achievement—United States, 1900–1999. Morb Mortal Wkly Rep. 1999;48:241–243.
2. Richardson SD, Plewa MJ, Wagner ED, Schoeny R, DeMarini DM. Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection by-products in drinking water: a review and roadmap for research. Mutat Res. 2007;636:178–242.
3. Reif RS, Hatch MC, Bracken M, Holmes LB, Schwetz BA, Singer PC. Reproductive and developmental effects of disinfection by-products in drinking water. Environ Health Perspect. 1996;104:1056–1061.
4. Nieuwenhuijsen MJ, Toledano MB, Eaton NE, Fawell J, Elliott P. Chlorination disinfectant byproducts in water and their association with adverse reproductive outcomes: a review. Occup Environ Med. 2000;57:73–85.
5. Tardiff RG, Carson ML, Ginevan ME. Updated weight of evidence for an association between adverse reproductive and developmental effects and exposure to disinfection by-products. Regul Toxicol Pharmacol. 2006;45:185–205.
6. Nieuwenhuijsen MJ, Grellier J, Smith R. The epidemiology and possible mechanisms of disinfection by-products in drinking water. Philos Transact A Math Phys Eng Sci. 2009;367:4043–4076.
7. Bove F, Shim Y, Zeitz P. Drinking water contaminants and adverse pregnancy outcomes: a review. Environ Health Perspect. 2002;110(suppl 1):61–74.
8. Graves CG, Matanoski GM, Tardiff RG. Weight of evidence of an association between adverse reproductive and developmental effects and exposure to disinfectant by-products: a critical review. Regul Toxicol Pharmacol. 2001;34:103–124.
9. Grellier J, Bennett J, Patelarou E. Exposure to disinfection by-products and adverse birth outcomes related to fetal growth and prematurity—a systematic review and meta-analysis. Epidemiology. 2010;21:300–313.
10. NSW Health. Midwives data collection (Mdc) reporting and submission requirements. Sydney, Australia: NSW Health Department; 2005.
11. Taylor L, Pym M, Bajuk B, Sutton L, Travis S, Banks C. Part 8: validation study NSW midwives data collection 1998. N S W Public Health Bull Suppl. 2000;11:97–99.
12. Savitz DA, Ananth CV, Berkowitz GS, Lapinski R. Concordance among measures of pregnancy outcome based on fetal size and duration of gestation. Am J Epidemiol. 2000;151:627–633.
13. Beard JR, Lincoln D, Donoghue D. Socioeconomic and maternal determinants of small-for-gestational age births: patterns of increasing disparity. Acta Obstet Gynecol Scand. 2009;88:575–583.
14. Summerhayes R, Holder P, Beard J. Automated geocoding of routinely collected health data in New South Wales. N S W Public Health Bull. 2006;17:33–38.
15. Australian Bureau of Statistics. Information paper: census of population and housing—socio-economic indexes for areas, Australia, 2001. Vol 2039.0. Canberra, Australia: Australian Bureau of Statistics; 2001.
16. Summerhayes RJ, Morgan GG, Lincoln D, et al.. Spatio-temporal variation in trihalomethanes in New South Wales. Water Research. 2011;45:5715–5726.
17. Whitaker H, Nieuwenhuijsen MJ, Best N, Fawell J, Gowers A, Elliot P. Description of trihalomethane levels in three UK water suppliers. J Expo Anal Environ Epidemiol. 2003;13:17–23.
18. Hongliang J, Liyan L, Yeqing S. Monitoring and modeling endosulfan in Chinese surface soil. Environ Sci Technol. 2010;44:9279–9284.
19. Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159:702–706.
20. Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol. 2005;162:199–200.
21. Hastie T, Tibshirani R. Generalized Additive Models. London: Chapman and Hall; 1990.
22. Infante-Rivard C. Drinking water contaminants, gene polymorphisms, and fetal growth. Environ Health Perspect. 2004;112:1213–1216.
23. Wright JM, Schwartz J, Dockery DW. The effect of disinfection by-products and mutagenic activity on birth weight and gestational duration. Environ Health Perspect. 2004;112:920–925.
24. Wright JM, Schwartz J, Dockery DW. Effect of trihalomethane exposure on fetal development. Occup Environ Med. 2003;60:173–180.
25. Hoffman CS, Mendola P, Savitz D. Drinking water disinfection by-product exposure and fetal growth. Epidemiology. 2008;19:729–737.
26. Bove F, Fulcomer M, Klotz JB, Esmart J, Dufficy EM, Savrin JE. Public drinking water contamination and birth outcomes. Am J Epidemiol. 1995;141:850–862.
27. Porter CK, Putnam SD, Hunting KL, Riddle MR. The effect of trihalomethane and haloacetic acid exposure on fetal growth in a Maryland county. Am J Epidemiol. 2005;162:334–344.
28. Kramer MD, Lynch CF, Isacson P, Hanson JW. The association of waterbourne chloroform with intrauterine growth retardation. Epidemiology. 1992;3:407–413.
29. Yang C-Y, Xiao Z-P, Ho S-C, Wu T-N, Tsai S-S. Association between trihalomethane concentrations in drinking water and adverse pregnancy outcome in Taiwan. Environ Res. 2007;104:390–395.
30. Hinckley A, Bachand A, Reif J. Late pregnancy exposures to disinfection by-products and growth-related birth outcomes. Environ Health Perspect. 2005;113:1808.
31. Lee PA, Chernausek SD, Hokken-Koelega AC, Czernichow P; International Small for Gestational Age Advisory Board. International Small for Gestational Age Advisory Board consensus development conference statement: management of short children born small for gestational age, April 24–October 1, 2001. Pediatrics. 2003;111:1253–1261.
32. Xu H, Simonet F, Luo ZC. Optimal birth weight percentile cut-offs in defining small- or large-for-gestational-age. Acta Paediatr. 2010;99:550–555.
33. NSW Health, Population Health Survey. Available at: Accessed August 24, 2011.
34. Weisel CP, Jo W-K. Ingestion, inhalation, and dermal exposures to chloroform and trichloroethane from tap water. Environ Health Perspect. 1996;104:48–51.
35. Schulman J, Selvin S, Shaw GM, Malcoe LH. Exposure misclassification due to residential mobility during pregnancy in epidemiologic investigations of congenital malformations. Arch Environ Health. 1993;48:114–119.
36. Australian Bureau of Statistics. Domestic water use—New South Wales. Vol Catalogue 4616.1. Canberra, Australia: Australian Bureau of Statistics; 2002.
37. NSW Health. NSW adult health survey 2002. N S W Public Health Bull Suppl. 2002;14(suppl 4).
38. US Environmental Protection Agency. Technical and Economic Capacity of States and Public Water Systems to Implement Drinking Water Regulations. Report to Congress. 1993. EPA 810-R-93-001. 125p.
39. NHMRC. Australian drinking water guidelines 6. Canberra, Australia: National Health and Medical Research Council, Australian Government; 2004.
Copyright © 2012 Wolters Kluwer Health, Inc. All rights reserved.