Bibliographic record and links to related information available from the Library of Congress catalog
Note: Electronic data is machine generated. May be incomplete or contain other coding.
INTRODUCTION The interaction between chlorine and organic material in drinking water sources produces a wide range of chemical disinfection by-products (DBPs) of potential health concern, including trihalomethanes (THMs), haloacetic acids (HAAs), and other halogenated and non-halogenated compounds. Cancer has been the greatest concern in regard to these DBPs, but over the past 15 years, reproductive health has come to be of increasing ig interest to the research community and to the public as well. A range of reproductive health outcomes has been studied in relation to exposure to DBPs, but the greatest attention is now focused on pregnancy loss. This possibility was raised most strongly in a study conducted in Northern California by Waller et al. (1998). That research team found that women who drank five or more glasses of cold tap water per day containing at least 75 gg/L of total THMs had an odds ratio of 1.8 (95% CI = 1.1-310) compared to all other women, and women who consumed five or more glasses per day of cold tap water containing at least 18 pg/L bromodichloromethane (CHBrC12) had an odds ratio of 3.0 (95% CI = 1.4-6.6) after making statistical adjustments for other THMs. That study added markedly to the plausibility of an effect of DBPs on pregnancy loss but fell far short of providing definitive support. RESEARCH OBJECTIVES Our study was planned and conducted primarily to address the hypothesis that exposure to DBPs causes pregnancy loss. We made several key methodologic improvements over previous research, most critically in regard to the assessment of DBP exposure, the range of DBP exposure evaluated, and the quality of assessment of pregnancy outcome. We sought to address the following study questions: 1. Is living in an area served by water with elevated levels of DBPs associated with increased risk of pregnancy loss compared to living in areas with lower DBP levels? 2. Are elevated levels of chlorinated DBPs. brominated DBPs. or any individual DBP species among THMs or HAAs associated with increased risk of pregnancy loss? 3. Are women who receive water with elevated levels of DBPs and ingest greater amounts of that water at increased risk of pregnancy loss relative to those women who drink lesser amounts of water from the same source? 4. Do the patterns of association between DBPs and pregnancy loss differ for losses that occur less than 12 weeks after the last menstrual period (LMP) versus losses that occur later than 12 weeks after the LMP, or with different stages of development prior to loss as assessed by ultrasound? 5. Is there an association between exposure to DBPs and reduced fetal growth, as measured by small-for-gestational-age births or preterm birth? 6. What is the contribution of tap water THMs to blood THM levels? APPROACH Three study sites were selected to reflect a wide range of DBP concentrations and speciation typical of those found across the US. One site had moderate levels of chlorinated DBPs (Site 1), one had very low levels of all DBPs (Site 2), and one had moderate levels of brominated DBPs and lower levels of the species containing chlorine only (Site 3). For the sites with moderte DBP levels, we sought locations that used chloramines as a terminal disinfectant. With chioramination, there is little additional DBP formation in the distribution system, resulting in minimal spatial variation within the water supply service area. In addition to variation in exposure resulting from the different water sources, patterns of water use for drinking, bathing, and showering were ascertained in detail because those behaviors affect personal exposure to DBPs. We sought to recruit women in each of the three areas who were planning a pregnancy or who were pregnant at less than 12 weeks' gestation. The study was marketed through prenatal care providers and through multiple forms of community advertising at drug stores, doctor's offices, and other settings. We successfully recruited and completed data collection for a total of 3132 women, 252 recruited prior to conception that became pregnant and enrolled in the study, and 2514 recruited early in pregnancy. Potentially eligible women were screened by telephone and, if eligible, were contacted for a telephone interview within 2 weeks of recruitment (<16 weeks' gestation) to assess behaviors and other factors that might influence the health of their pregnancy. Participants were scheduled for an ultrasound assessment between 6 and 7 weeks' gestation and no later than 14 weeks to accurately determine the dates of their pregnancy and to determine whether it was progressing normally. Participants then had a follow-up interview at 20-25 weeks' gestation to complete data collection and assess the continuation of their pregnancy. Pregnancy losses were identified by self-report, with medical records sought for confirmation, and live birth outcomes (birth weight, duration of gestation) were identified through vital records, medical records, or self-report. Recruitment occurred in Site I from December 2000 through February 2004, in Site 2 from June 2002 through March 2004, and in Site 3 from September 2002 through April 2004. Tap water was sampled at frequent intervals (weekly or every other week) at the three study sites and analyzed for the four currently regulated THMs (THM4), the nine bromine- and chlorine-containing HAAs (HAA9). and total organic halide (TOX). Based on the measurements, DBP concentrations were then assigned to each weekly interval of pregnancy. More intensive sampling programs were implemented at Sites I and 3 during intermittent periods in which free chlorine was used in the distribution system. Because sites were chosen that either used year- round chloramination except for those intermittent periods or had low overall DBP levels, there was little within-system variation and therefore all women in a given study area were assigned the same exposure score for any given week. Based on previous epiderniologic and toxicological studies, we focused most intensively on THM4. CHBrC12, and HAA9. Overall median THM4 levels were 60.7, 3.6, and 57.8 gg/L at sites 1, 2, and 3, respectively. Overall median HAA9 concentrations were 41.5, 3.3, and 44.7 ig/L, respectively. We examined several indices of exposure: concentration of DBPs in tap water, ingested amount of DBPs, DBP exposure from showering and bathing alone, and integrated DBP exposure combining ingestion with showering and bathing. These exposure estimates were linked to specific time intervals in the pregnancy: 4 weeks before the LMP through 3 weeks after LMP (periconceptional), weeks 4-8 after LMP (early gestation), and weeks 9-20 after LMP (later gestation). In the interview, women were asked about changes in water use during pregnancy and these reported changes were taken into account in the calculation of the exposure indices. DBP exposure was calculated directly from the measured weekly or bi-weekly concentrations by generating a mean of the measured values during the pregnancy window. Ingested amount was based on these measured concentrations but modified as a function of the reportd. amount of water ingested during the time window. Laboratory experiments were conducted to derive estimates of changes in individual DBP concentrations due to heating water for the preparation of hot beverages and using either tap or pitcher filters prior to cold water ingestion. These estimates were combined with self-reported information on the individual's daily ingestion of cold and hot water, as well as their ingestion of filtered and unfiltered water. Showering and bathing exposure was calculated combining information on tap water DBP concentration, reported frequency and duration of those activities, and estimated absorption of DBPss as a result of these activities. Finally, integrated exposure combined those pathways of exposure into a summary estimate of total exposure to DBPs. In the study of pregnancy loss, we examined the various indices of DBP exposure in relation to the probability of the pregnancy surviving or being lost through the period of gestation. We adjusted statistically for other known and suspected risk factors for pregnancy loss that were found to be predictive in our data, including maternal age, race/ethnicity, education, marital status, history of prior pregnancy loss, and alcohol consumption. CONCLUSIONS 1. Comparing our results to the primary findings of the Northern California study, we did not find the same associations relating THM4 concentration combined with water consumption to the risk of pregnancy loss. Waller et al. (1998) found that women who drank 5 or more glasses per day of tap water with >75pg/L THMs had twice the risk of pregnancy loss compared to other women (measured statistically as an odds ratio in which he odds of pregnancy loss among women with higher DBP exposure is compared to the odds of loss among women with lower DBP exposure). In contrast, we found women who drank 5 or more glasses per day of tap water with >75ug/L THMs had the same risk (an odds ratio of 1.0) compared to all other women, indicating no association between exposure and pregnancy loss. Waller et all (1998) also comparing the upper quartiles of CHBrC12 and DBCM (CHBr2CI) exposure to the lower three quartiles and found odds ratios of 2.0 and 1.3; in our study, the same comparisons yielded odds ratios of 1.6 and 1.7, providing some indication that these specific DBPs may be associated with an increased risk of pregnancy loss.. 2. In examining THM4 using the multiple indices described above, results were generally not supportive of an association with pregnancy loss with the possible exception of an increased risk for losses at >12 weeks' gestation. The array of results for CHBrCl2, HAA9, and the other groups of DBPs considered provided sporadic support for elevated risk that varied across pregnancy time window, exposure index, and agent. CHBrC12 results were marginally stronger than those for THM4, and the results for TOX had the most consistently association with pregnancy loss, both for tap water concentration and ingested amount. 3. Live birth outcomes were addressed in the form of preterm birth (<37 weeks completed gestation), small-for-gestational-age (SGA) births (<10 0h percentile of weight for gestational age), and a continuous measure of birth weight among term births. Pregnancy windows were defined by the trimester of pregnancy (weeks 0-12, 13-26. 27+). Preterm birth showed a modest but consistent tendency to be rarer among women with higher exposure (i.e., an inverse association). CHBrC1 was unrelated to risk of preterm birth and HAA9 showed a weak inverse association as did chloroform, another of the THM4 species. 4. Analysis of THM4 exposure in the third trimester and SGA births generated evidence of a positive association based on the dichotomy of 80 pg/L (OR = 2.1, 95% C = I 1 -3.8). Restriction to sites I and 3 enhanced the association. BDCM was also related to increased risk of SGA births, whereas HAA9 and chloroform were not. 5, Term birth weight was largely unrelated to DBP exposures, except for somewhat lower birth weights among women who were served by water with >80 [giL of THM4 in the second and third trimesters. 6. Blood THM levels varied in the expected direction by season in Site 1 (higher in summer than in winter), and generally showed some contrast across the three sites generally, but to a much lesser extent than would have been expected based on tap water concentrations of DBPs. IMPLICATIONS OF FINDINGS Policy recommendations do not follow directly from the findings of this (or any single) epidemiologic study, but we can comment on the overall nature of how these findings may shift priorities. Relative to the earlier study in Northern California (Waller et al., 1998), we found less support for an adverse effect of DBPs on pregnancy loss. logically leading to a somewhat lower level of concern with that possibility than was present prior to the conduct of our study. The failure to provide strong evidence in support of the hypothesized associations is worth noting as well, in that the methodological refinements in our epidemiologic study should have generated more persuasive evidence of adverse effects if such effects are indeed associated with DBP exposures. Nonetheless, there were sporadic indications of increased risk of pregnancy loss and fetal growth restriction associated with higher exposure to selected DBPs that may warrant further consideration. FUTURE RESEARCH The logical next steps in the evolution of research on DBPs and reproductive health outcomes are not obvious. We offer the following suggestions for consideration: 1. Our results do not provide encouragement for undertaking additional, large-scale research of a similar nature in different geographic locations. To the extent that researchers can identify settings in which exposure levels differ appreciably over season within communities, there are some clear logistical and scientific advantages, namely that the study base can be more circumscribed, saving substantial time and money required to manage multiple field sites. In addition, the demographic and recruitment differences across communities in studies such as ours make it very difficult to isolate any direct impact of DBPs in accounting for varying health patterns across communities. 2 Exposure assessment remains a key limiting factor in our study and in all studies that have preceded it. While we constructed indices that were intended to move well beyond tap water concentrations to reflect actual exposure, in the process of doing so, the limitations in making such inferences with accuracy (or even knowing the accuracy) were ,quite obvious. Work to develop and validate stronger approaches to characterizing exposure in epidemiologic studies is a high priority for new studies to improve upon ours. In particular, the high prevalence of use of bottled water and point of use devices calls for more extensive research into the impact on human exposure at a population level. 3. Enrollment prior to conception is very challenging, and incurs substantial sacrifices in terms of numbers of participants and their social and demographic profile, but provides a marked increment in the quality of information on the course of their pregnancy. To the extent that the logistical challenges can be overcome, there are notable advantages to enrollment before conception as compared to enrollment in very early pregnancy.