Trihalomethanes in Drinking Water and Spontaneous Abortion

Abstract

Some studies indicate that chlorination by-products in drinking water may contribute slightly to breast cancer risk. This ecologic study describes the association between total trihalomethane levels in publicly supplied water and the incidence of female invasive breast cancer. We included 71 North Carolina water suppliers serving at least 10,000 customers in the summer of 1995 as the units of analysis. We estimated incidence rates using 6,462 cases who were either white or black and between 35 and 84 years old and were linked by zip codes to the water supplier. We treated ecologic measurements of age, income, education, urban status, and race as potential confounders. Total trihalomethane levels were not associated materially with breast cancer risk, adjusting for potential confounders. The rate ratio for 80.0 parts per billion (ppb) or more vs less than 40.0 ppb total trihalomethanes was 1.1 [95% confidence interval (CI) = 0.9-1.2]. When stratified by race, the observed association for the aforementioned total trihalomethane category was not very different in black women (rate ratio = 1.2; 95% CI = 0.8-1.8) than in white women (rate ratio = 1.1; 95% CI = 0.9-1.3). These ecologic data are compatible with trihalomethanes in drinking water being either unrelated or weakly related to breast cancer risk.

Influence of exposure assessment methods on risk estimates in an epidemiologic study of total trihalomethane exposure and spontaneous abortion

“…We expected the categorical ingestion results obtained using the unweighted utilitywide average and without the Zone A correction to be similar to the findings published in Waller et al (1998 ), as this method corresponded closely to our previous method. In fact, this method produced an AOR for high exposure (TTHM !75 g/l and intake !5 gl / day) of 1.9 (95% CI 1.1-3.2 ) that was very similar to our previously published AOR of 1.8.…”

“…Unlike our original analysis, if there were no TTHM measurements taken during the first trimester, we did not expand the time interval in order to capture a measurement and reduce missing data. Values also differed from those described in Waller et al ( 1998 ) because of additions and corrections to the sampling database and changes in utility assignment. Regardless, for 88% of subjects with nonmissing data, the current utility-wide estimate was within 5 g/l of the previous value.…”

“…Ingestion and Showering Metrics Individual daily cold tap water ingestion at 8 weeks' gestation was determined as in Waller et al (1998 ), using information from the telephone interview. We created a categorical ingestion exposure metric that combined first trimester trihalomethane concentration, dichotomized at 75 g /l, and cold tap water ingestion, dichotomized at 5 glasses / day.…”

“…Logistic regression models were created to adjust for multiple covariates. For comparison purposes, the models included the same covariates used in Waller et al (1998 ); these were gestational age at interview ( 8 vs. > 8 weeks ), maternal age at interview (!35 vs. < 35 years ), cigarette smoking (any vs. none ), history of pregnancy loss (!2 vs. < 2 prior SABs ), maternal race (black and Asian vs. white, Hispanic), and employment during pregnancy.…”

“…This paper describes a reanalysis of the data from Waller et al (1998 ), in which we explored how alternative methods of exposure assessment affect estimates of the association between TTHM exposure and risk of SAB. The new methods followed two basic approaches.…”