CME AVAILABLE FOR THIS ARTICLE AT ACOEM.ORG
Association of Brominated Trihalomethane and Haloacetic
Acid Exposure With Fetal Growth and Preterm Delivery
in Massachusetts
Zorimar Rivera-Núñez, PhD and J. Michael Wright, ScD
Objectives: We examined the relationship between adverse reproductive outcomes and exposure to several disinfection by-products including haloacetic
acids (HAAs), trihalomethanes (THMs), and the brominated THMs sum
(THMBr). Methods: Second- and third-trimester exposure scores were estimated from quarterly town-level average THM and HAA concentration data
and were examined relative to mean birth weight (BWT), small for gestational
age (SGA), and preterm delivery (PTD). Results: We found an increased
risk of PTD for second-trimester chloroform, bromodichloromethane, and
all HAA exposure metrics (adjusted odds ratio range: 1.04 to 1.15), but detected no associations for SGA and third-trimester exposures. Mean BWT
deficits were observed across all HAA (26 to 33 g) and THMBr (11 to 23 g)
exposure categories. Conclusions: We detected consistent associations for
adjusted mean BWT and THMBr exposures; these data reinforce the need to
consider different disinfection by-product exposure metrics in epidemiological studies.
D
isinfectants react with organic and inorganic matter to form
a complex mixture of several hundred different disinfection
by-products (DBPs) in drinking water.1,2 Various factors influence
the formation of DBPs within a water distribution system (eg, water source, amount, timing, and type of disinfection, residence time,
and specific DBP precursors such as bromide and iodine levels, and
natural organic matter). Trihalomethanes (THMs) and haloacetic
acids (HAAs) are the most abundant classes of DBPs present in
treated drinking water.3 Along with chlorite and bromate, maximum contaminant levels have been established for public drinking
water systems in the United States for the sum of four THMs (ie,
THM4: chloroform [TCM], bromodichloromethane [BDCM], dibromochloromethane [DBCM], and bromoform [TBM]) and the
sum of five HAAs (ie, HAA5: monochloroacetic acid [MCAA],
dichloroacetic acid [DCAA], trichloroacetic acid [TCAA], monobromoacetic acid [MBAA], and dibromoacetic acid [DBAA]).4
Toxicological data from animal studies show various adverse
reproductive and developmental effects associated with DBP exposure, including fetotoxicity, reduced fetal growth, and reduced
survival.5 The evidence for associations between DBP exposure and
From the George Warren Brown School of Social Work and Institute of Public
Health (Dr Rivera-Núñez), Washington University, St Louis, Missouri; and
National Center for Environmental Assessment (Drs Rivera-Núñez and
Wright), Office of Research and Development, US Environmental Protection
Agency, Cincinnati, Ohio.
No external funds were used to support this research.
The views expressed in this article are those of the authors and do not necessarily
reflect the views or policies of the US Environmental Protection Agency.
Authors Rivera-Nunez and Wright have no relationships/conditions/circumstances
that present potential conflict of interest.
The JOEM editorial board and planners have no financial interest related to this
research.
Supplemental digital contents are available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions of
this article on the journal’s Web site (www.joem.org).
Address correspondence to: J. Michael Wright, ScD, 26 W. Martin Luther King
Dr. (MS-A110), Cincinnati, OH 45268 ([email protected]).
C 2013 by American College of Occupational and Environmental
Copyright Medicine
DOI: 10.1097/JOM.0b013e3182a4ffe4
Learning Objectives
r Become familiar with previous data on possible reproductive and developmental effects of exposure to disinfection
byproducts (DBPs) in drinking water.
r Summarize the new findings on specific DBPs and their assor
ciations with the outcomes of birthweight, preterm delivery,
and small size for gestational age.
Discuss the implications for understanding and future epidemiologic studies of the health effects of exposure to DBPs.
adverse reproductive outcomes reported in epidemiological studies has been more mixed. Fairly consistent associations between
THM exposure and small for gestational age (SGA) have been
identified by meta-analyses, although minimal evidence exists for
preterm delivery (PTD).6 More recent studies report mixed results for THM exposures and fetal growth, although several had
low THM levels7 or limited THM and brominated THMs exposure gradients,8–10 which could impact their ability to detect
associations. Although HAAs (eg, HAA5, DCAA, difluoroacetic
acid, and trifluoroacetic acid) have been previously identified as
teratogens,11,12 few epidemiological studies have examined the relationship between HAA exposure and fetal growth and prematurity. Hinckley et al13 observed an association between term low
birth weight (LBW) and third-trimester DBAA exposures 5 μg/L or
more (odds ratio [OR]: 1.49; 95% confidence interval [CI]: 1.09 to
2.04). They also reported an increased risk of SGA for DCAA and
TCAA exposures (see the Supplementary Digital Content Table 1,
http://links.lww.com/JOM/A132). Porter et al14 and Levallois et al10
found some evidence of an increased risk of SGA for secondand third-trimester exposures to HAA5 (see the Supplementary
Digital Content Table 1, http://links.lww.com/JOM/A132). Other
studies found no association between HAA exposure and fetal growth
or PTD.9,15–17 The lack of adequate exposure gradients is an even
larger issue for HAAs, especially when examining individual HAA
constituents, given their lower overall concentrations in most drinking water supplies.
Several studies in Massachusetts have reported associations
between different DBP exposure measures and SGA or LBW.15,18–20
Lewis et al19 reported an association between second-trimester THM
exposures 70 μg/L or more and an increased risk of term LBW
(OR: 1.50; 95% CI: 1.07 to 2.10) in 27 communities served by the
Massachusetts Water Resources Authority from 1999 to 2001. In
Wright et al,15 we previously reported an increased risk of SGA
(OR range: 1.06 to 1.13) in births to residents of 109 towns in Massachusetts from 1995 to 1998 when comparing the highest THM
tertile with the lowest (≤33 μg/L). Small increased risks for SGA
were detected for the upper TCM (OR range: 1.05 to 1.11) and
BDCM (OR range: 1.10 to 1.15) tertiles in this study, but limited
HAA data resulted in limited exposure contrasts and precluded adjustments for HAA exposure in the THM analyses. More recently, we
observed an increased risk of SGA for users of chlorinated surface
JOEM r Volume 55, Number 10, October 2013
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
1125
JOEM r Volume 55, Number 10, October 2013
Rivera-Núñez and Wright
Statistical Analysis
water and all chloraminated water (OR range: 1.04 to 1.16) relative
to untreated ground water.20
Recent studies of fetal growth and DBPs have advanced exposure assessment by examining specific water-use activities and
biomarker data.7,8,10 Nonetheless, it is unclear from epidemiological studies of DBPs and adverse birth outcomes which DBP mixtures
or surrogate markers are most relevant. Elucidating the potential impact of exposure to brominated THMs remains a key research need
because they are more potent toxicants than TCM.21,22 In this study,
previous analyses from 1995 to 199815 have been extended to 1996
to 2004 to examine associations between exposure to individual and
summary measures of THMs and HAAs with birth weight (BWT),
SGA, and PTD. This large data set allowed for an assessment of
low and unexposed reference groups and also provided sufficient
exposure gradients to examine the brominated THMs.
Statistical Analysis System software, Version 9.2 (SAS Institute, Cary, NC), was used for the statistical analyses. Births in
the lowest DBP exposure category served as the reference for comparison with the upper categories. Pearson correlation coefficients
were used to compare THM4, HAA5, and individual DBPs. Logistic regression was used to estimate ORs and 95% CIs for SGA
and PTD. Linear regression was used to estimate the change in
mean BWT relative to DBP exposure. Statistical significance was
based on α = 0.05 or less. The following potential confounders
were examined: maternal age, race/ethnicity, education, smoking
(ie, number of cigarettes per day), parity, adequacy of prenatal care,
prenatal source of payment (ie, private, government, and other), income (median household income), marital status, maternal medical
(eg, diabetes, chronic hypertension, lung disease, and renal disease)
and reproductive health factors (eg, hydramnios/oligohydramnios,
incompetent cervix, preeclampsia, previous PTD, and weight gain
during pregnancy), season, and THM4/HAA5 concentrations. Adequacy of prenatal care was determined from birth certificate data
by the Kotelchuck Index, which uses the time at the initiation of
prenatal care (ie, when prenatal care began) and the number of
prenatal visits from when prenatal care began until delivery.25 Income data were obtained from the 2000 Census (Geolytics, Inc, East
Brunswick, NJ) and was assigned using the census tract corresponding to maternal residential ZIP codes. Confounding factors were
selected on the basis of percentage change (>10%) in regression
coefficient estimates compared with univariate models. The following covariates were identified as confounders and adjusted for in
the multivariate BWT and PTD regression models: maternal age,
race/ethnicity, education, prenatal care source of payment, income,
and marital status. The SGA models were adjusted for the same covariates except maternal race. A quadratic polynomial was used for
maternal age, whereas categorical variables were included for maternal race/ethnicity, education, prenatal source of payment, and marital
status. Income was modeled as a continuous variable. We examined
effect-measure modification using stratified analyses and interaction
terms. We used the Brown and Forsythe test to examine homogeneity across effect-measure modifier groups within an exposure
group.26
MATERIALS AND METHODS
Study Population
We conducted a retrospective cohort study with a semiecologic study design.23 The Massachusetts Department of Public
Health (Boston, MA) supplied individual-level birth certificate data
on 712,394 live infants born from 1996 to 2004 in the Commonwealth of Massachusetts. After restricting the data set to singleton
births to Massachusetts residents with plausible values of gestational
age (ie, between 22 and 45 weeks) and BWT (ie, >200 g), 672,120
live births were available for analyses. Gestational age was derived
from clinical estimates according to birth certificates. Small for gestational age was defined as infants with a BWT below the tenth
percentile for their gestational age at birth specific to sex and maternal race (African American vs all other races combined) categories.
The mean BWT and SGA analyses were restricted to term births
from 37 to 45 gestational weeks. Preterm delivery was defined as
infants born at fewer than 37 gestational weeks. Infants of at least
37 gestational weeks and weighing at least 2500 g served as the
comparison group for the estimated ORs for PTD.
Exposure Data
The study population (n = 672,120 births) resided in 351
Massachusetts towns. Two hundred sixty-five towns (n = 619,984
births) were served by 276 public water systems (PWSs), while residents of 75 towns (n = 52,136 births) had no public water supplies
and relied on private well water. Births from 11 towns (n = 20,608
births) were excluded from the analyses because source water and
disinfection exposure measures could not be assigned to specific
residents due to one or more PWSs having multiple water sources
and/or different types of disinfection. Details of assignment of exposure measures by town, distribution of water source and disinfection,
and analytical protocols for water sample analyses have been previously described.20,24
RESULTS
There were 37,136 (5.7%) preterm births in this study population and 68,409 (11.1%) of the term births were classified as SGA.
Mean BWT reductions in infants from 171 to 270 g were noted for
previously reported maternal risk factors for BWT such as mothers
who were not married (vs married), were African American (vs other
races), were smokers (vs nonsmokers), had inadequate prenatal care
(vs adequate), and had used public (vs private) sources of prenatal
care payment (Table 1). Table 2 shows the distribution of secondand third-trimester DBP exposures.
Second- and third-trimester mean THM4 exposure concentrations were 37.5 and 38.1 μg/L, respectively, with 5.9% and 6.1%
of the maternal exposures exceeding 80 μg/L, respectively. The
mean concentrations for second- and third-trimester HAA5 exposure were 20.0 and 20.1 μg/L, respectively. The Pearson coefficient
between second- and third-trimester THM4 and HAA5 concentrations was 0.74 (see the Supplementary Digital Content Table 2,
http://links.lww.com/JOM/A132).
Exposure Assessment
We evaluated DBP exposure measures for THM4, TCM,
BDCM, DBCM, HAA5, TCAA, DCAA, and DBP9 (sum of
TCM, BDCM, DBCM, TBM, TCAA, DCAA, MBAA, MCAA,
and DBAA). Details of the exposure assessment methods can
be found in the Supplementary Digital Content Methods (http:
//links.lww.com/JOM/A132) and elsewhere.20 The THMBr metric
was composed of the sum of BDCM, DBCM, and TBM. We performed three sensitivity analyses that examined (1) a population
with no exposure to THM4 as the reference group; (2) imputation of
values for missing data; and (3) PWSs restricted to only those with
low spatial variability (ie, <20 μg/L change in DBP concentration
among samples collected during the same day [36 towns]). Details
of imputation procedures can be found in the Supplementary Digital
Content Methods (http://links.lww.com/JOM/A132).
1126
Birth Weight
Statistically significant reductions in mean BWT were observed for all individual THM and HAA third-trimester exposure
categories in the unadjusted models (Table 3). BWT deficits of 33
to 89 g were detected for all THM4, TCM, BDCM, and THMBr
exposure categories compared with the lowest category. THM4 exposures more than 10 μg/L were associated with BWT reductions of
C
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
JOEM r Volume 55, Number 10, October 2013
DBP Exposure in Relation to Adverse Reproductive Outcomes
TABLE 1. Maternal Characteristics of 651,512 Singleton Births in 340 Massachusetts Towns,
1995 to 2004
Total births‡
Infant sex
Male
Female
Marital status
Married
Unmarried
Previously married within 300 d of births
Number of previous birth
0
1
2 or 3
>3
Maternal weight gain during pregnancy, lb
<0
0–25
25–50
>50
Maternal age, yr
12–20
21–25
26–30
31–35
36–40
41–64
Maternal race
White
African American
Asian
American Indian
Other
Maternal education
Below high school graduate
High school graduate
Associate/certification
Bachelor’s
Graduate
Median household income, $
≤30,219
>30,219–38,738
>38,738–47,805
>47,805–61,042
>61,042
Prenatal care§
No prenatal care
Inadequate
Intermediate
Adequate
Adequate plus
Prenatal care source of payment
Public
Private
Other
C
Study Population, n (%)
BWT, g
SGA, %*
PTD, %†
651,512 (100.0)
3,411
10.5
5.7
333,467 (51.2)
318,045 (48.8)
3,469
3,349
10.2
10.7
6.0
5.3
471,348 (72.3)
178,770 (27.4)
1,388 (0.2)
3,458
3,287
3,354
8.9
14.5
13.1
5.1
7.3
6.5
291,045 (44.7)
220,016 (33.8)
124,169 (19.1)
15,443 (2.4)
3,349
3,462
3,465
3,390
13.0
8.2
8.4
10.2
6.4
4.7
5.3
8.4
6,297 (1.0)
250,873 (38.5)
365,174 (56.1)
23,914 (3.7)
3,223
3,307
3,473
3,631
14.8
13.3
8.7
5.7
11.6
7.8
4.2
3.4
62,937 (9.7)
106,021 (16.3)
179,992 (27.6)
200,719 (30.8)
88,296 (13.6)
13,547 (2.1)
3,249
3,345
3,424
3,465
3,455
3,407
16.0
12.9
10.1
8.5
8.9
10.6
7.4
5.8
5.3
5.2
5.9
7.4
484,815 (74.4)
48,793 (7.5)
37,758 (5.8)
1,229 (0.2)
78,262 (12.0)
3,459
3,245
3,246
3,369
3,297
9.3
10.4
17.0
13.0
14.4
5.1
9.2
5.5
6.7
7.2
91,458 (14.0)
249,727 (38.3)
65,701 (10.1)
163,788 (25.1)
80,686 (12.4)
3,271
3,386
3,445
3,483
3,470
15.4
11.4
9.1
7.9
8.3
7.4
6.0
5.6
4.8
4.6
129,195 (19.8)
128,804 (19.8)
128,949 (19.8)
129,137 (19.8)
128,758 (19.8)
3,311
3,383
3,430
3,456
3,473
13.9
11.6
10.1
8.8
7.9
7.1
6.0
5.2
5.2
4.9
3,302 (0.5)
55,298 (8.5)
52,386 (8.0)
309,297 (47.5)
231,229 (35.5)
3,246
3,312
3,480
3,500
3,302
12.2
13.8
10.2
9.3
11.2
13.5
6.9
2.9
1.7
11.3
181,387 (27.8)
463,836 (71.2)
6,232 (1.0)
3,302
3,454
3,348
14.1
9.0
12.2
7.0
5.2
8.4
(continued)
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
1127
JOEM r Volume 55, Number 10, October 2013
Rivera-Núñez and Wright
TABLE 1. (Continued)
Maternal smoking||
0
1–5
6–10
>10
Water source
Mixed
Surface
Ground
Disinfection treatment
Chlorination
Chloramination
Other
Untreated
Season
Winter
Spring
Summer
Fall
Study Population, n (%)
BWT, g
SGA, %*
PTD, %†
583,733 (89.6)
25,076 (3.8)
24,684 (3.8)
16,859 (2.6)
3,434
3,245
3,185
3,164
9.4
17.6
20.6
22.6
5.5
7.1
7.5
8.0
68,266 (10.5)
413,816 (63.5)
168,029 (25.8)
3,456
3,377
3,475
9.2
11.5
8.4
5.0
6.1
5.0
320,815 (49.2)
201,580 (30.9)
55,237 (8.5)
72,180 (11.1)
3,415
3,391
3,372
3,474
10.3
11.1
11.6
8.5
5.6
6.0
6.0
5.0
151,992 (23.3)
166,289 (25.5)
172,115 (26.4)
161,116 (24.7)
3,397
3,418
3,416
3,410
10.8
10.1
10.3
10.6
6.0
5.8
5.6
5.5
*SGA was defined as infants with a BWT below the tenth percentile for their gestational age at birth, specific to sex and maternal
race (African American vs all other races combined) categories.
†PTD was defined as infants born at less than 37 gestational weeks.
‡Singleton births, 22 to 45 gestational weeks and 200 g or more born to residents of Massachusetts.
§Adequacy of prenatal care was determined using the Kotelchuck Index.
||Number of cigarettes per day.
BWT, birth weight; PTD, preterm delivery; SGA, small for gestational age.
TABLE 2. Maternal Second- and Third-Trimester Exposures to DBPs (μg/L) Among Residents of
Massachusetts Public Water Systems, 1995 to 2004
DBPs
Second trimester
THM4
TCM
BDCM
THMBr
HAA5
DCAA
TCAA
DBP9
Third trimester
THM4
TCM
BDCM
THMBr
HAA5
DCAA
TCAA
DBP9
Mean
10%
20%
50%
75%
90%
Range
37.5
30.1
6.0
7.5
20.0
8.7
10.1
56.8
0
0
0
0
0
0
0
0
10.3
4.6
1.4
1.6
0
0
0
0
35.6
27.0
5.2
6.0
18.7
8.4
9.0
57.1
56.4
46.5
8.4
10.7
31.2
13.5
15.8
90.4
72.9
63.4
12.6
17.0
42.3
18.8
21.9
108.1
0–271.5
0–265.9
0–49.4
0–104.8
0–172.9
0–47.1
0–122.6
0–442.6
38.1
30.6
6.1
7.7
20.1
8.8
10.2
57.7
0
0
0
0
0
0
0
0
10.4
4.6
1.4
1.6
0
0
0
0
36.2
27.4
5.3
6.0
18.9
8.4
9.1
58.6
57.3
47.1
8.5
10.8
31.2
13.5
15.8
91.2
73.4
64.1
12.8
17.4
42.1
18.7
21.9
108.7
0–273.5
0–265.9
0–49.5
0–104.8
0–173.9
0–47.1
0–122.4
0–443.6
DBP, disinfection by-product; DBP9, sum of chloroform (TCM), bromodichloromethane (BDCM), dibromochloromethane (DBCM),
bromoform (TBM), monochloroacetic acid (MCAA), dichloroacetic acid (DCAA), trichloroacetic acid (TCAA), bromoacetic acid
(BMAA), and dibromoacetic acid (DBAA); HAA5, sum of MCAA, DCAA, TCAA, BMAA, and DBAA; THM4, sum of TCM, BDCM,
DBCM, and TBM; THMBr, sum of BDCM, DBCM, and TBM.
1128
C
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
JOEM r Volume 55, Number 10, October 2013
DBP Exposure in Relation to Adverse Reproductive Outcomes
TABLE 3. Associations Between Third-Trimester DBP Exposure and Change in Mean BWT and SGA Among Term Births
Term
DBP Metrics* Births (n)
THM4, μg/L
≤10
>10–30
>30–45
>45–63
>63
TCM, μg/L
≤5
>5–21
>21–36
>36–52
>52
BDCM, μg/L
≤1
>1–4
>4–6
>6–10
>10
THMBr, μg/L
≤2
>2–5
>5–8
>8–13
>13
HAA5, μg/L
0
>0–15
>15–24
>24–33
>33
DCAA, μg/L
0
>0–6
>6–10
>10–15
>15
TCAA, μg/L
0
>0–7
>7–11
>11–17
>17
DBP9, μg/L
0
>0−49
>49–73
>73–97
>97
Unadjusted
BWT
(95% CI), g
Adjusted†
BWT
(95% CI), g
95,077
96,742
94,850
95,403
95,088
REF
− 60 (−64 to −55)
− 72 (−77 to −68)
− 87 (−91 to −83)
− 83 (−87 to −78)
REF
− 9 (−15 to −2)
− 17 (−24 to −11)
− 23 (−30 to −17)
− 23 (−29 to −16)
98,901
95,182
94,097
95,009
93,912
REF
− 50 (−54 to −46)
− 73 (−77 to −69)
− 89 (−93 to −84)
− 83 (−87 to −79)
REF
− 1 (−7 to 5)
− 9 (−15 to −2)
− 13 (−19 to −7)
− 15 (−21 to −8)
97,520
94,994
94,309
94,439
94,793
REF
− 60 (−64 to −55)
− 63 (−68 to −59)
− 62 (−67 to −58)
− 49 (−53 to −45)
97,129
91,473
95,991
94,284
95,041
Adjusted‡
BWT
(95% CI), g
Unadjusted
SGA OR
(95% CI)
Adjusted§
SGA OR
(95% CI)
Adjusted‡
SGA OR
(95% CI)
REF
0 (−13 to 13)
− 2 (−15 to 11)
− 4 (−18 to 10)
− 1 (−15 to 13)
REF
1.28 (1.24 to 1.31)
1.32 (1.28 to 1.36)
1.42 (1.38 to 1.47)
1.39 (1.35 to 1.43)
REF
1.01 (0.96 to 1.05)
1.02 (0.97 to 1.07)
1.07 (1.02 to 1.12)
1.06 (1.01 to 1.11)
REF
0.96 (0.87 to 1.06)
1.00 (0.90 to 1.10)
1.04 (0.94 to 1.15)
1.01 (0.92 to 1.13)
REF
11 (2 to 21)
8 (−2 to −18)
4 (−6 to 14)
12 (1 to 22)
REF
1.27 (1.22 to 1.30)
1.33 (1.29 to 1.37)
1.43 (1.39 to 1.47)
1.41 (1.37 to 1.45)
REF
1.01 (0.96 to 1.05)
1.00 (0.95 to 1.04)
1.04 (1.00 to 1.10)
1.04 (0.99 to 1.09)
REF
0.91 (0.85 to 0.98)
0.96 (0.89 to 1.03)
0.99 (0.92 to 1.07)
0.96 (0.89 to 1.04)
REF
− 11 (−17 to −5)
− 14 (−21 to −8)
− 20 (−26 to −14)
− 16 (−22 to −10)
REF
0 (−14 to 13)
5 (−8 to 18)
− 2 (−15 to 12)
9 (−5 to 22)
REF
1.26 (1.22 to 1.30)
1.28 (1.24 to 1.32)
1.26 (1.22 to 1.30)
1.22 (1.18 to 1.26)
REF
1.04 (1.00 to 1.08)
1.08 (1.03 to 1.12)
1.09 (1.04 to 1.14)
1.09 (1.04 to 1.13)
REF
0.93 (0.85 to 1.02)
0.92 (0.84 to 1.00)
0.93 (0.84 to 1.01)
0.91 (0.83 to 1.00)
REF
− 56 (−60 to −52)
− 60 (−64 to −56)
− 52 (−56 to −48)
− 33 (−37 to −28)
REF
− 10 (−16 to −4)
− 17 (−23 to −11)
− 19 (−25 to −13)
− 13 (−19 to −7)
REF
− 19 (−35 to −2)
− 21 (−38 to −5)
− 23 (−39 to −7)
− 11 (−27 to 6)
REF
1.22 (1.18 to 1.26)
1.24 (1.21 to 1.28)
1.22 (1.18 to 1.26)
1.12 (1.08 to 1.15)
REF
1.00 (0.97 to 1.04)
1.06 (1.02 to 1.10)
1.08 (1.04 to 1.12)
1.05 (1.00 to 1.09)
REF
0.94 (0.84 to 1.06)
0.96 (0.86 to 1.08)
0.95 (0.85 to 1.07)
0.93 (0.83 to 1.05)
62,486
46,966
53,771
54,248
53,933
REF
− 62 (−68 to −56)
− 82 (−88 to −77)
− 97 (−103 to −92)
− 89 (−94 to −83)
REF
− 26 (−41 to −11)
− 28 (−43 to −12)
− 33 (−49 to −17)
− 26 (−42 to −10)
REF
− 28 (−43 to −13)
− 30 (−46 to −14)
− 36 (−52 to −19)
− 29 (−46 to −13)
REF
1.16 (1.11 to 1.21)
1.35 (1.30 to 1.40)
1.39 (1.33 to 1.44)
1.34 (1.29 to 1.39)
REF
1.05 (0.94 to 1.18)
1.09 (0.97 to 1.23)
1.08 (0.96 to 1.22)
1.04 (0.92 to 1.17)
REF
1.04 (0.93 to 1.17)
1.08 (0.96 to 1.21)
1.06 (0.94 to 1.20)
1.02 (0.90 to 1.15)
71,168
39,995
54,854
51,906
53,481
REF
− 50 (−56 to −44)
− 80 (−85 to −74)
− 89 (−94 to −83)
− 76 (−82 to −71)
REF
0 (−10 to 10)
− 4 (−14 to 6)
− 5 (−15 to 5)
1 (−9 to 11)
REF
1 (−9 to 11)
− 3 (−13 to 7)
− 4 (−14 to 6)
2 (−9 to 12)
REF
1.12 (1.07 to 1.16)
1.32 (1.27 to 1.37)
1.34 (1.29 to 1.40)
1.32 (1.27 to 1.37)
REF
0.98 (0.91 to 1.05)
1.02 (0.95 to 1.09)
1.00 (0.93 to 1.08)
0.99 (0.92 to 1.07)
REF
0.97 (0.85 to 1.11)
1.13 (1.00 to 1.28)
1.12 (0.98 to 1.27)
1.08 (0.94 to 1.23)
71,181
40,240
53,246
55,004
51,733
REF
− 55 (−60 to −49)
− 78 (−83 to −73)
− 80 (−85 to −74)
− 82 (−87 to −76)
REF
0 (−10 to 10)
− 1 (−10 to 9)
− 4 (−14 to 6)
0 (−11 to 10)
REF
0 (−10 to 10)
− 1 (−11 to 9)
− 5 (−15 to 6)
− 1 (−12 to 10)
REF
1.17 (1.12 to 1.22)
1.36 (1.31 to 1.41)
1.30 (1.26 to 1.35)
1.31 (1.26 to 1.36)
REF
1.01 (0.94 to 1.08)
1.04 (0.97 to 1.12)
1.01 (0.94 to 1.09)
0.99 (0.92 to 1.07)
REF
0.98 (0.87 to 1.12)
1.04 (0.92 to 1.18)
1.01 (0.89 to 1.16)
0.97 (0.85 to 1.11)
59,751
48,937
53,641
53,492
54,052
REF
− 73 (−78 to −67)
− 80 (−86 to −75)
− 97 (−103 to −92)
− 88 (−94 to −83)
REF
− 39 (−62 to −18)
− 42 (−64 to −19)
− 45 (−68 to −22)
− 39 (−62 to −16)
–
–
–
–
–
REF
1.20 (1.15 to 1.25)
1.32 (1.27 to 1.37)
1.39 (1.33 to 1.44)
1.33 (1.27 to 1.38)
REF
0.90 (0.76 to 1.06)
0.94 (0.79 to 1.11)
0.94 (0.79 to 1.11)
0.91 (0.77 to 1.08)
–
–
–
–
–
BWT, birth weight; CI, confidence interval; DBP, disinfection by-product; DBP9, sum of chloroform (TCM), bromodichloromethane (BDCM), dibromochloromethane
(DBCM), bromoform (TBM), monochloroacetic acid (MCAA), dichloroacetic acid (DCAA), trichloroacetic acid (TCAA), bromoacetic acid (BMAA), and dibromoacetic acid
(DBAA); HAA5, sum of MCAA, DCAA, TCAA, BMAA, and DBAA; OR, odds ratio; REF, reference; SGA, small for gestational age; THM4, sum of TCM, BDCM, DBCM,
and TBM; THMBr, sum of BDCM, DBCM, and TBM.
*Third-trimester exposures.
†Models adjusted for source, disinfection, maternal age, race, education, marital status, prenatal care source of payment, and income.
‡Models also adjusted for THM4 or HAA5 exposures.
§Models adjusted for source, disinfection, maternal age, education, marital status, prenatal care source of payment, and income.
9 to 23 g compared with exposures 10 μg/L or less after adjustment
for sociodemographic confounders, water source, and disinfection.
Although THM4 results were largely null after further adjustment
for HAA5, mean BWT reductions were still observed for THMBr
(11 to 23 g).
Statistically significant deficits in mean BWT (50 to 97
g) for all exposure categories in the HAA5, TCAA, and DCAA
unadjusted models were found relative to the reference group
C
(0 μg/L). Reductions in mean BWT of 28 to 36 g were detected for all
HAA5 exposure categories after adjustment for sociodemographic
confounders, water source, disinfection, and THM4 exposures. No
associations were detected for DCAA and TCAA in the multivariate
models. The unadjusted (73 to 97 g) and adjusted (39 to 45 g) models for the DBP9 metric (reference group = 0 μg/L) showed larger
mean BWT deficits than the THM4 and HAA5 results. Mean BWT
results for second-trimester exposures were similar to the results for
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
1129
JOEM r Volume 55, Number 10, October 2013
Rivera-Núñez and Wright
by inadequate exposure contrasts and very narrow exposure categories that can increase the potential for exposure misclassification.
We detected risks of SGA for elevated THM4, BDCM, and THMBr
exposures similar in magnitude to the small risks observed in our
previous study of 1995 to 1998 birth data (OR range: 1.08 to 1.13).15
After additional adjustment for HAA5, however, we did not detect
associations between SGA and any of the THM exposure metrics.
We also found limited evidence of an association between HAA exposures and SGA in the fully adjusted models presented here and in
our earlier study. The null results detected for HAA and SGA are
similar to that of Wright et al15 as well as Hoffman et al,16 although
Hoffman et al16 saw some evidence of an association between SGA
and total organic halide (TOX) (see the Supplementary Digital Content Table 1, http://links.lww.com/JOM/A132). Three other studies
have shown small increased risks of SGA for DCAA and TCAA
exposures10,13,14 (see the Supplementary Digital Content Table 1,
http://links.lww.com/JOM/A132). In a recent biomarker study designed to better characterize DBP exposure, Costet et al27 measured
TCAA levels in urine from pregnant women in France, in addition
to environmental THM and HAA concentrations. The authors reported an increased risk of SGA (OR: 1.8; 95% CI: 0.9 to 3.7) for
women with detectable levels of urinary TCAA. Although this study
dichotomized the TCAA exposure because of limitations of their
analytical protocol (ie, a high limit of detection), they showed some
evidence of an association (OR range: 1.2 to 1.4) between THMBr
(ie, DBCM and TBM) and a measure of SGA on the basis of environmental concentration data. It is unclear whether the improved
exposure assessment in the Costet et al27 study contributed to the
larger detected associations between SGA and DBP exposure, but
future research examining DBP biomarkers is warranted.
Previous studies have reported little evidence of an association between PTD and either THM or HAA.6,9 Small increased risks
(OR range: 1.09 to 1.13) were detected for the upper three TCM exposure categories after adjustment for HAA5 and other confounders
in this study. Although they were largely not statistically significant,
we saw consistently elevated ORs in the fully adjusted models for
HAA5 (OR range: 1.09 to 1.15), DCAA (OR range: 1.04 to 1.10),
and TCAA (OR range: 1.07 to 1.14) exposure categories compared
with the reference group (0 μg/L). In addition, we observed elevated
ORs for PTD among the upper four DBP9 exposure categories (OR
range: 1.20 to 1.29) compared with the reference group (0 μg/L).
Recently, Horton et al9 examined a group of births from a site with
moderate levels of chlorine-containing DBPs and another group of
births from a site with moderate levels of bromine-containing DBPs.
In the brominated site, they found some evidence of an association
between PTD and TOX (OR range: 1.09 to 1.99) using TOX less
than 111 μg/L as the reference group. They also found a statistically
significant association between TOX and very PTD (ie, <32 weeks
of gestation; OR range: 2.43 to 4.17) that was stronger in magnitude
among residents from the brominated site. Lack of data precluded
examination of TOX exposures in this study, but we saw no association between very PTD and either THM or HAA exposures (data
not shown).
Overall, we saw larger associations for the THMBr and some
of the HAA metrics, although the available HAA data did not allow
for sufficient examination of the brominated species. Because THMs
and HAAs often comprise less than half of the TOX sum,3 TOX may
be a good surrogate of halogenated DBP mixtures beyond summary
THM and HAA measures typically examined in epidemiological
studies.28 More specific exposure measures of halogenated species,
such as TOCl, TOBr, and TOI, and nonhalogenated species should
also be examined given the differences in toxicity noted in previous studies.22 Joeong et al29 recently identified more than 90 DBPs
in 11 locations throughout Europe, including haloacids, halophenols, haloamides, halonitromethanes, haloketones, haloaldehydes,
and haloalkenes. They reported positive correlations between the
third-trimester exposures (see the Supplementary Digital Content
Table 3, http://links.lww.com/JOM/A132).
Small for Gestational Age
An increased risk of SGA was observed for THM4, HAA5,
and all individual THM and HAA exposure metrics in unadjusted
models (Table 3). After adjustment for sociodemographic confounders, water source, and disinfection, small but statistically significant associations (OR range: 1.04 to 1.09) were detected for the
upper THM4, BDCM, and THMBr categories compared with the
lowest categories. Null associations were detected after further adjustment for HAA exposures. Unadjusted models showed increased
risk of SGA for all HAA5 and DBP9 exposure categories (OR
range: 1.16 to 1.39). The adjusted results, however, were largely null
for HAA5, DCAA, TCAA, and DBP9 exposure categories. Overall, SGA results for second-trimester exposures were similar to the
results for third-trimester exposures (see the Supplementary Digital
Content Table 3, http://links.lww.com/JOM/A132).
Preterm Delivery
Compared with the reference categories, statistically significant associations were detected between PTD and all secondtrimester THM4 and TCM exposure categories (OR range: 1.16
to 1.30) in the unadjusted models (Table 4). Smaller associations
were detected for BDCM (OR range: 1.06 to 1.21) and THMBr
(OR range: 1.13 to 1.16) but only in the lower two exposure categories. Results for all THM exposure categories were largely null
after adjustment for sociodemographic covariates, water source, disinfection, and HAA5. The most consistent results across exposure
categories were noted for TCM (OR range: 1.09 to 1.13) and THMBr
(OR range: 1.07 to 1.10). Unadjusted models showed an increased
risk of PTD for all HAA5, TCAA, and DCAA exposure categories
(OR range: 1.15 to 1.36) relative to the lowest category (0 μg/L)
(Table 4). Statistically significant associations were found for TCAA
(OR range: 1.12 to 1.14) exposure categories after adjustment for sociodemographics and other confounders. We also detected increased
ORs for PTD after adjustment for sociodemographics, water source,
and disinfection in the DBP9 model (OR range: 1.20 to 1.29). PTD
results for first-trimester exposures were similar to results for secondtrimester exposures (see the Supplementary Digital Content Table 4,
http://links.lww.com/JOM/A132).
Effect-Measure Modification
We examined maternal age, race, education, prenatal care
source of payment, disinfection treatment, and income as potential
effect-measure modifiers. In the THM4 unadjusted models, infants
from African American mothers had the lowest mean BWT deficits
(range: 11 to 29 g) after whites (range: 21 to 46 g), other races (range:
37 to 55 g), and Asians (range: 42 to 60 g) (data not shown). Infants
from Native American mothers had increases in mean BWT (range:
41 to 103 g), but this was based on a small sample size. Infants from
mothers reporting public prenatal care source of payment had slightly
larger mean BWT deficits (range: 39 to 73 g) than mothers reporting
private prenatal care source of payment (range: 36 to 59 g) (data
not shown). The effect of THM4 on mean BWT, however, varied by
income only in the unadjusted THM4 models. For example, infants
born to women in the lowest income quintile had much larger mean
BWT deficits (105 g) than infants from women in the highest income
quintile (19 g) (see the Supplementary Digital Content Table 5,
http://links.lww.com/JOM/A132). We did not find evidence of
effect-measure modification by any of the other examined covariates.
DISCUSSION
SGA has been the most consistent adverse reproductive outcome associated with DBP exposures observed in previous epidemiological studies.6 Nevertheless, many of these studies were limited
1130
C
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
JOEM r Volume 55, Number 10, October 2013
DBP Exposure in Relation to Adverse Reproductive Outcomes
TABLE 4. Associations Between Second-Trimester DBP Exposure and PTD Among All Births
DBP Metrics*
THM4, μg/L
≤10
>10–29
>29–44
>44–62
>62
TCM, μg/L
≤5
>5–21
>21–35
>35–52
>52
BDCM, μg/L
≤1
>1–4
>4–6
>6–10
>10
THMBr, μg/L
≤2
>2–5
>5–7
>7–12
>12
HAA5, μg/L
0
>0–14
>14–21
>21–34
>34
DCAA, μg/L
0
>0–7
>7–11
>11–15
>15
TCAA, μg/L
0
>0–7
>7–10
>10–15
>15
DBP9, μg/L
0
>0–49
>49–72
>72–96
>96
All Births (n)
Unadjusted PTD OR (95% CI)
Adjusted† PTD OR (95% CI)
Adjusted‡ PTD OR (95% CI)
100,854
100,924
101,000
100,997
100,628
REF
1.21 (1.16 to 1.26)
1.23 (1.18 to 1.28)
1.26 (1.22 to 1.31)
1.22 (1.17 to 1.27)
REF
1.01 (0.96 to 1.08)
1.02 (0.96 to 1.08)
1.02 (0.96 to 1.09)
0.98 (0.92 to 1.04)
REF
1.04 (0.92 to 1.17)
1.06 (0.94 to 1.20)
1.07 (0.94 to 1.21)
1.04 (0.91 to 1.19)
103,472
100,402
100,478
99,938
100,003
REF
1.16 (1.12 to 1.21)
1.30 (1.26 to 1.35)
1.29 (1.24 to 1.34)
1.24 (1.19 to 1.29)
REF
1.00 (0.94 to 1.06)
1.08 (1.02 to 1.14)
1.06 (0.99 to 1.12)
1.00 (0.94 to 1.07)
REF
1.10 (1.00 to 1.21)
1.13 (1.03 to 1.24)
1.13 (1.03 to 1.25)
1.09 (0.99 to 1.20)
103,206
100,425
99,623
100,047
99,924
REF
1.17 (1.12 to 1.21)
1.21 (1.17 to 1.26)
1.06 (1.02 to 1.11)
1.08 (1.04 to 1.12)
REF
0.96 (0.91 to 1.01)
0.99 (0.94 to 1.04)
0.90 (0.86 to 0.95)
0.93 (0.88 to 0.98)
REF
1.05 (0.93 to 1.18)
1.15 (1.02 to 1.29)
1.05 (0.93 to 1.18)
1.09 (0.97 to 1.23)
103,705
96,787
100,111
100,307
100,107
REF
1.16 (1.12 to 1.21)
1.13 (1.09 to 1.17)
1.02 (0.98 to 1.06)
1.03 (0.99 to 1.07)
REF
0.97 (0.92 to 1.01)
0.96 (0.91 to 1.01)
0.89 (0.85 to 0.94)
0.92 (0.88 to 0.97)
REF
1.08 (0.93 to 1.25)
1.10 (0.95 to 1.28)
1.07 (0.92 to 1.24)
1.09 (0.94 to 1.27)
64,958
45,198
55,039
54,825
55,004
REF
1.21 (1.15 to 1.28)
1.32 (1.26 to 1.39)
1.32 (1.25 to 1.38)
1.35 (1.28 to 1.42)
REF
1.16 (0.99 to 1.35)
1.17 (1.00 to 1.36)
1.10 (0.94 to 1.29)
1.14 (0.97 to 1.34)
REF
1.15 (0.98 to 1.35)
1.15 (0.98 to 1.36)
1.09 (0.93 to 1.29)
1.13 (0.95 to 1.33)
73,251
39,119
56,016
52,258
54,578
REF
1.18 (1.12 to 1.25)
1.30 (1.24 to 1.37)
1.36 (1.30 to 1.43)
1.28 (1.22 to 1.35)
REF
1.10 (1.00 to 1.21)
1.10 (1.01 to 1.21)
1.11 (1.01 to 1.22)
1.05 (0.96 to 1.22)
REF
1.10 (1.00 to 1.21)
1.10 (1.00 to 1.21)
1.10 (1.00 to 1.21)
1.04 (0.94 to 1.15)
73,381
38,772
54,473
53,967
54,629
REF
1.15 (1.09 to 1.22)
1.33 (1.27 to 1.39)
1.27 (1.21 to 1.33)
1.36 (1.30 to 1.43)
REF
1.07 (0.97 to 1.18)
1.12 (1.02 to 1.23)
1.08 (0.98 to 1.19)
1.14 (1.03 to 1.25)
REF
1.07 (0.97 to 1.17)
1.12 (1.02 to 1.23)
1.08 (0.98 to 1.19)
1.14 (1.03 to 1.26)
62,714
46,613
55,734
54,051
54,293
REF
1.23 (1.16 to 1.29)
1.30 (1.23 to 1.36)
1.40 (1.33 to 1.47)
1.30 (1.24 to 1.37)
REF
1.23 (0.99 to 1.53)
1.24 (0.99 to 1.54)
1.29 (1.03 to 1.61)
1.20 (0.96 to 1.50)
–
–
–
–
–
CI, confidence interval; DBP, disinfection by-product; DBP9, sum of chloroform (TCM), bromodichloromethane (BDCM), dibromochloromethane
(DBCM), bromoform (TBM), monochloroacetic acid (MCAA), dichloroacetic acid (DCAA), trichloroacetic acid (TCAA), bromoacetic acid (BMAA), and
dibromoacetic acid (DBAA); HAA5, sum of MCAA, DCAA, TCAA, BMAA, and DBAA; OR, odds ratio; REF, reference; PTD, preterm delivery; THM4, sum
of TCM, BDCM, DBCM, and TBM; THMBr, sum of BDCM, DBCM, and TBM.
*Second-trimester exposures.
†Models adjusted for source, disinfection, maternal age, race, education, marital status, prenatal care source of payment, and income.
‡Models also adjusted for THM4 or HAA5 exposures.
C
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
1131
JOEM r Volume 55, Number 10, October 2013
Rivera-Núñez and Wright
sis of effect-measure modification by income is limited by the potential misclassification of the group-level income data (ie, Census
data). We did not observe effect-measure modification for maternal race/ethnicity or prenatal care source of payment in the mean
BWT model for either THM or HAA exposures. Although we did
not see much evidence of effect-measure modification by socioeconomic status determinants, potential interactions between low socioeconomic status markers and DBP exposures should be further
examined.
Our large sample size provided ample statistical power to
detect associations that were small in magnitude and to adjust for
numerous confounding factors. Maternal medical (eg, diabetes and
chronic hypertension) and reproductive health factors (eg, incompetent cervix and preeclampsia) were strong predictors of mean
BWT but were not identified as confounders in our analyses. The
strongest individual confounder was residential source of water,
which changed the mean BWT by 80%. Other variables identified
as confounders in the THM4 model included maternal race, prenatal
care source of payment, income, marital status, HAA5 exposure,
maternal age, maternal education, and disinfection type. These same
sociodemographic covariates and THM4 exposure were also identified as confounders in the HAA5 model. As expected, gestational
age was a strong predictor of mean BWT. Because bias may result if
gestational age is an intermediate in the causal pathway of DBPs and
BWT,36 we did not adjust our models for gestational age here or in
our previous studies.15 In our sensitivity analysis, however, we found
comparable results adjusting for gestational age (data not shown).
With the exception of THMBr exposure models, we found that the
THM4 results (mean BWT and SGA) were confounded by HAA5. In
contrast, HAA5 results were largely similar after THM4 adjustment.
Because of limited DBP exposure data, most studies have not developed multipollutant models to examine potential confounding by
other DBP exposures. A particular strength of this study is our ability to adjust for other DBP exposures, water source, and disinfection
as confounders. Although we adjusted the multivariate models by
these exposure metrics, confounding from other unmeasured DBPs
or correlated environmental exposures may still be a concern here
and in previous studies.
A key limitation of our study and most of the previous research cited herein is the potential for exposure misclassification
because of individual variability in water-use practices and differences in temporal and spatial DBP formation. Our analyses are limited mostly to town quarterly average data to estimate individual
exposures. A strength of this study included sensitivity analyses
in which we imputed data for some systems with missing quarterly THM4 data (78,925 births). These findings were similar to
the main results (see the Supplementary Digital Content Table 6,
http://links.lww.com/JOM/A132); however, most of the imputed values were obtained from systems with limited data or low DBP levels.
Although spatial variability is difficult to assess given the limited
number of samples collected at different locations in most water distribution systems, it could have an impact on epidemiological studies
if the use of average aggregated values does not adequately characterize exposure to residents, relying on certain water systems.37 We did
not find any BWT deficits when we restricted our THM4 analysis
to 36 systems with low spatial variability (see the Supplementary
Digital Content Table 6, http://links.lww.com/JOM/A132). The
THMBr model results (8 to 22 g deficits) were similar to the main results including all systems, whereas the BDCM results were stronger
(17 to 34 g deficits) (data not shown). Although not statistically significant, we observed larger associations for THM4 (OR range: 1.60
to 1.92) and TCM exposures (OR range: 1.42 to 1.72) with PTD in the
sensitivity analysis than the main results. THMBr results (OR range:
1.03 to 1.17), however, were comparable to the main results. This
sensitivity analysis restricted to systems with low spatial variability
was based on at least a 20-μg/L change in THM4 concentration
number of DBPs detected, higher genotoxicity and cytotoxicity index, and reported levels of SGA in epidemiological cohorts. The
coherence reported between analytical chemistry findings, in vitro
toxicology results, and epidemiologic results in that study lends some
support to epidemiologic findings of an association between adverse
reproductive effects and exposure to DBPs. Nevertheless, additional
characterization of the chemical mixture found in different study
populations is needed to further elucidate previously reported epidemiological associations.
Previous studies have not detected associations between HAA
exposures and mean BWT.15,16 The considerably smaller sample size
and limited concentration gradient may have prevented the detection
of associations between mean BWT and third-trimester HAA5 exposures (reference group 4 to 30 μg/L) in our previous study of
1995 to 1998 birth data.15 Our current results showed statistically
significant associations between mean BWT and all HAA exposure
categories (28 to 36 g deficits). Similar to most other studies, we
did not have enough data to evaluate the brominated HAAs. Nevertheless, the null associations between mean BWT and chlorinated
HAAs suggest that the association found between HAA5 and mean
BWT may be largely attributable to the brominated proportion of the
HAA mixture. A recent study examining a DBP biomarker reported
even stronger mean BWT deficits (82 to 160 g) for the upper two
urinary TCAA quartiles.30 The authors did not collect environmental
THM and HAA data; therefore, comparison between urinary TCAA
and environmental DBP concentrations, and examination of BWT
changes was not possible.
Compared with our previous study, we saw slightly higher
mean BWT deficits ranging from 13 to 28 g for third-trimester
THM4 exposure categories more than 10 μg/L after adjusting for
sociodemographic factors alone (data not shown). We did not, however, observe any mean BWT reductions associated with THM4,
TCM, and BDCM exposures after additional adjustment for water
source, disinfection, and HAA exposures. We detected mean BWT
reductions from 11 to 23 g across all THMBr exposure categories
in the fully adjusted model. These data suggest that the brominated
proportion of the THM and HAA mixture may be a more relevant exposure surrogate to help characterize the association between mean
BWT and DBP exposures. The BWT and PTD results also indicate
confounding by other DBP exposure metrics that are important to
consider in epidemiological studies.
Determining the influence of different measures of socioeconomic status on adverse reproductive outcomes is a challenge
in epidemiological studies because it is not always clear whether
determinants of socioeconomic status are confounders or effectmeasure modifiers.31 For example, low income has been associated
with race/ethnicity, reduced access to health care, lower education,
inadequate housing, and poor nutrition in studies of women of reproductive age.32,33 Residual confounding by income may occur in this
study because of variability in income levels within a census tract,
because we used census tract-level data to adjust for income in the
multivariate models. Additional control for socioeconomic status,
however, was accomplished through adjustment of other individuallevel socioeconomic status predictors (ie, maternal race, education,
and prenatal care source of payment). Although residual confounding
from factors associated with LBW and PTD, such as nutritional status, drug use, or psychological factors (eg, stress and depression),34
is a possibility, we are not aware of any evidence that these variables
are associated with DBP exposures.
Previous research has found some limited evidence of effectmeasure modification by race/ethnicity and prenatal care source of
payment in studies of THMs and LBW and PTD.19,35 We detected
some variation in the mean BWT and THM4 relationship by income
quintiles in the unadjusted THM4 model (see the Supplementary
Digital Content Table 5, http://links.lww.com/JOM/A132), but this
was not evident in the multivariate regression models. Our analy1132
C
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
JOEM r Volume 55, Number 10, October 2013
between locations for samples collected on the same date. We recognize that the full extent of spatial variability may not be captured
by existing DBP sample locations in some PWSs and that other approaches for quantifying the degree of spatial variability should be
considered for different DBP metrics.38 Nevertheless, our THMBr
results seem robust to the impact of spatial variability in aggregated
exposure measures.
Given that inter- and intra-individual variability in water-use
practices can substantially modify DBP exposures on the basis
of aggregated exposure estimates, exposure misclassification bias
may be present in epidemiological studies relying on these types of
surrogate measures.39–41 In addition, attenuation of effect estimates
may occur if the reference group includes a combination of moderately and highly exposed individuals as observed in previous DBP
studies.13,18,19,35,42,43 Another strength of this study was the ability to
examine low or unexposed reference groups for our categorical analyses given the large exposure contrasts for most DBP metrics. As part
of our sensitivity analyses, we examined an unexposed population
(THM4 = 0 μg/L) as the reference group (see the Supplementary
Digital Content Table 6, http://links.lww.com/JOM/A132). We saw
larger mean BWT deficits for THM4 (33 to 43 g) and THMBr
(28 to 40 g) in the fully adjusted model using this unexposed
reference group than for the main THM4 (0 to 4 g) and THMBr
(11 to 23 g) results on the basis of the 10 μg/L or less reference
group. Similar to this sensitivity analysis, our DBP9 model included
an unexposed reference group where we saw the strongest mean
BWT deficits (39 to 45 g). Associations for PTD for this sensitivity
analysis were similar in magnitude for THM4 (OR range: 1.27 to
1.31), TCM (OR range: 1.08 to 1.20), and THMBr (OR range: 1.28
to 1.33), and the SGA results were similar to the THM4 main results
(OR range: 0.92 to 0.97). Exposure misclassification may be greater
for the volatile THMs given the larger role of dermal and inhalation
exposures resulting from noningestion water-use activities.44,45
Because an unexposed population as a reference group uses drinking
water systems without DBPs present, exposure misclassification
from interindividual variability in water-use practices would be
minimized in this group regardless of their frequency and duration.
Another potential source of exposure misclassification is residential mobility during pregnancy. Mobility during pregnancy can
impact the ability to link DBP exposures to the critical exposure window being examined. For example, third-trimester exposures may be
less subject to misclassification because women in some populations
have been shown to be less likely to move late in pregnancy.46 If this
occurred in our study population, we would expect less exposure
misclassification from mobility in the BWT or SGA analyses and
third-trimester exposure metrics.
Recent studies have helped advances in DBP exposure assessment methods by interpolating missing data,47 developing dose
estimates on the basis of individual water-use data,7,10 and through
the use of exposure biomarkers.27,30,48 Nonetheless, given the potential number of DBPs present in drinking water, the delineation
of which surrogate DBP mixture to target remains elusive. More
studies are needed to address this surrogacy issue by quantifying
associations between individual DBPs or specific DBP mixtures (eg,
DBP9 or a TOX exposure metric not attributable to THMs and HAAs
as reported by Horton et al9 ). The primary analysis and sensitivity
analyses of our data showed consistent BWT deficits for THMBr exposures. In our analysis of the correlations between THM4 and the
individual THMs, THMBr had the lowest correlation (r = 0.55) with
THM4 compared with other THM metrics (TCM, r = 0.97; BDCM,
r = 0.68; HAA5, r = 0.74; DCAA, r = 0.69; TCAA, r = 0.72). The
association between THMBr and fetal growth, and the lack of association between chlorinated HAAs and fetal growth, suggests that the
risk of adverse birth outcomes may be due to brominated THMs or
other nonchlorinated DBPs for which no exposure data were available. Our study also provides evidence that THM4 may not be an
C
DBP Exposure in Relation to Adverse Reproductive Outcomes
adequate surrogate of brominated exposures or for the key DBP
mixture related to fetal growth retardation. Thus, more research is
warranted on the potential impact of brominated compounds (THMs,
HAAs, and other DBP metrics) on fetal growth retardation in studies
with sufficient exposure contrasts and unexposed reference populations designed to minimize attenuation of effect estimates and allow
for better characterization of exposure–response relationships.
REFERENCES
1. Boorman GA, Dellarco V, Dunnick JK, et al. Drinking water disinfection byproducts: review and approach to toxicity evaluation. Environ Health
Perspect. 1999;107(suppl 1):207–217.
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. Krasner SW, McGuire MJ, Jacangelo JG, Patania NL, Reagan KM, Aieta EM.
The occurrence of disinfection by-products in United States drinking water.
J Am Water Works Assoc. 1989;81:41–53.
4. US EPA. National primary drinking water regulations: stage 2 disinfectants
and disinfection by-products rule. Fed Reg. 2006;71:388–493.
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. Grellier J, Bennett J, Patelarou E, et al. Exposure to disinfection by-products,
fetal growth, and prematurity: A systematic review and meta-analysis. Epidemiol. 2010;21:300−313.
7. Grazuleviciene R, Nieuwenhuijsen MJ, Vencloviene J, et al. Individual exposures to drinking water trihalomethanes, low birth weight and small for
gestational age risk: a prospective Kaunas cohort study. Environ Health Perspect. 2011;10:32.
8. Patelarou E, Kargaki S, Stephanou EG, et al. Exposure to brominated trihalomethanes in drinking water and reproductive outcomes. J Occup Environ
Med. 2011;68:438–445.
9. Horton BJ, Luben TJ, Herring AH, et al. The effect of water disinfection
by-products on pregnancy outcomes in two southeastern US communities. J
Occup Environ Med. 2011;53:1172–1178.
10. Levallois P, Gingras S, Marcouz S, et al. Maternal exposure to drinking-water
chlorination by-products and small-for-gestational-age neonates. Epidemiol.
2012;23:267–276.
11. Hunter E, Rogers E, Schmid J, Richard A. Comparative effects of haloacetic
acids in whole embryo culture. Teratology. 1996;54:57–64.
12. Narotsky MG, Best D, McDonald A, Godin E, Hunter E, Simmons J. Pregnancy loss and eye malformations in offspring of F344 rats following gestational exposure to mixtures of regulated trihalomethanes and haloacetic acids.
Reprod Toxicol. 2011;31:59–65.
13. Hinckley AF, Bachand AM, Reif JS. Late pregnancy exposures to disinfection by-product and growth-related birth outcomes. Environ Health Perspect.
2005;113:1808–1813.
14. Porter CK, Putman 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.
15. 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.
16. Hoffman CS, Mendola P, Savitz DA, et al. Drinking water disinfection byproduct exposure and fetal growth. Epidemiol. 2008;19:729–737.
17. Hoffman CS, Mendola P, Savitz DA, et al. Drinking water disinfection byproduct exposure and duration of gestation. Epidemiol. 2008;19:738–746.
18. Wright JM, Schwartz J, Dockery DW. Effect of trihalomethane exposure on
fetal development. J Occup Environ Med. 2003;60:173–180.
19. Lewis C, Suffet IH, Hoggatt K, Ritz B. Estimated effects of disinfection byproducts on birth weight in a population served by a single water utility. Am
J Public Health. 2006;163:38–47.
20. Wright JM, Rivera-Núñez Z. Effect of water disinfection type on adverse fetal
outcomes. J Am Water Works Assoc. 2011;103:67–75.
21. Colman J, Rice GE, Wright JM, et al. Identification of developmentally toxic
drinking water disinfection byproducts and evaluation of data relevant to mode
of action. Tox App Pharmacol. 2011;254:100–126.
22. Plewa MJ, Wagner ED, Muellner MG, Hsu KM, Richardson SD. Comparative mammalian cell toxicity of N-DBPs and C-DBPs. In:Karanfil T, Krasner
SW, Westerhoff P, Xie Y, eds. ACS Symposium Series, Vol 995: Disinfection
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
1133
JOEM r Volume 55, Number 10, October 2013
Rivera-Núñez and Wright
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
By-products in Drinking Water: Occurrence, Formation, Health Effects. Washington, DC: American Chemical Society; 2008:36–50.
Künzli N, Tager IB. The semi-individual study in air pollution epidemiology:
a valid design as compared to ecologic studies. Environ Health Perspect.
1997;105:1078–1083.
Parvez S, Rivera-Núñez Z, Meyer A, Wright JM. Temporal variability in
trihalomethane and haloacetic acid concentrations in Massachusetts public
drinking water systems. Environ Res. 2011:499–509.
Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care
Index and proposed Adequacy of Prenatal Care Utilization Index. Am J Public
Health. 1994;84:1414–1420.
Brown MB, Forsythe AB. Robust tests for equality of variances. J Am Stat
Assoc. 1974;69:364–367.
Costet N, Garlantézec R, Monfort C, et al. Environmental and urinary markers
of prenatal exposure to drinking water disinfection by-products, fetal growth,
and duration of gestation in the PELAGIE birth cohort (Brittany, France,
2002–2006). Am J Epidemiol. 2012;175:263–375.
Bull RJ, Rice GE, Teuschler LK. Determinants of whether or not mixtures of
disinfection by-products are similar. J Toxicol Environ Health A. 2009;72:737–
460.
Joeong C, Wagner E, Siebert V, et al. Occurrence and toxicity of disinfection
byproducts in European drinking waters in relation with HIWATE epidemiology study. Environ Sci Technol. 2012;46:12120–12128.
Zhou W-S, Xy L, Xie S-H, et al. Decreased birth weight in relation to maternal
urinary trichloroacetic acid levels. Sci Total Environ. 2012;416:105–110.
Savitz DA, Kaufman JS, Dole N, Siega-Riz AM, Thorpe JM Jr, Kaczor DT.
Poverty, education, race, and pregnancy outcome. Ethn Dis. 2004;14:322–
329.
Janevic T, Stein CR, Savitz DA, Kaufman JS, Mason SM, Herring AH.
Neighborhood deprivation and adverse birth outcomes among diverse ethnic
group. Ann Epidemiol. 2010;20:445–451.
Forssén UM, Wright JM, Herring AH, Savitz DV, Nieuwenhuijsen MJ, Murphy PA. Variability and predictors of changes in water use during pregnancy.
J Expo Sci Environ Epidemiol. 2009;19:593–602.
Kramer MS, Seguin L, Lydon J, Goulet L. Socio-economic disparities in pregnancy outcome: why do the poor fare so poorly? Paediatr Perinat Epidemiol.
2000;14:194–210.
Lewis C, Suffet IH, Hoggatt K, Ritz B. Estimated effect of disinfection byproducts on preterm birth in a population served by a single water utility.
Environ Health Perspect. 2007;115:290–295.
1134
36. Wilcox AJ, Weinberg CR, Basso O. On the pitfalls of adjusting for gestational
age at birth. Am J Epidemiol. 2011;174:1062–1068.
37. Waller K, Swan SH, Windham GC, Fenster L. Influence of exposure assessment methods on risk estimates in an epidemiologic study of total
trihalomethane exposure and spontaneous abortion. J Exp Anal Environ
Epidemiol. 2001;11:522–531.
38. Hinckley AF, Bachand AM, Nuckols JR, Reif JS. Identifying public water
facilities with low spatial vaiability of disinfection by-products for epidemiological investigations. J Occup Environ Med. 2005b;62:494–499.
39. Wright JM, Murphy PA, Nieuwenhuijsen MJ, Savitz DA. The impact of water
consumption, point-of-use filtration and exposure categorization on exposure
misclassification of ingested drinking water contaminants. Sci Total Environ.
2006;366:65–73.
40. Weinberg HS, Pereira VR, Singer PC, Savitz DA. Considerations for improving the accuracy of exposure to disinfection by-products by ingestion in
epidemiologic studies. Sci Total Environ. 2006;354:35–42.
41. Krasner SW, Wright JM. The effect of boiling water on disinfection byproduct exposure. Water Res. 2005;39:855–864.
42. Dodds L, King W, Woolcott C, Pole J. Trihalomethanes in public water
supplies and adverse birth outcomes. Epidemiol. 1999;10:233–237.
43. Savitz DA, Andrews KW, Pastore LM. Drinking water and pregnancy outcome
in central North Carolina: source, amount, and trihalomethane levels. Environ
Health Perspect. 1995;103:592–596.
44. Nuckols JR, Ashley DL, Lyu C, Gordon SM, Hinckley AF, Singer P. Influence of tap water quality and household water-use activities on indoor
air and internal dose levels of trihalomethanes. Environ Health Perspect.
2005;113:863−870.
45. Lynberg M, Nuckols JR, Langlois P, et al. Assessing exposure to disinfection
by-products in women of reproductive age living in Corpus Christi, Texas,
and Cobb County, Georgia: descriptive results and methods. Environ Health
Perspect. 2001;109:597−604.
46. Schulman J, Selvin S, Shaw G, Malcoe L. Exposure misclassification due
to residential mobility during pregnancy in epidemiologic investigations of
congenital malformations. Arch Environ Health. 1993;48:114–119.
47. Richter W, Hart TF, Luben T, Freud T, Nuckols JR. Evaluation of two methods
of interpolating quarterly trihalomethanes levels between sampling dates.
J Expo Sci Environ Epidemiol. 2009;19:405–413.
48. Rivera-Núñez Z, Wright JM, Blount BC, et al. Comparison of trihalomethanes
in tap water and blood: a case study in the United States. Environ Health
Perspect. 2012;120:661–667.
C
2013 American College of Occupational and Environmental Medicine
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.