Participants and procedures. Children in this binational investigation participated in two affiliated studies, one located in California [Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS)] and the other in Mexico (Proyecto Mariposa). Written consent for maternal and child participation was obtained from all mothers. Assent was obtained from the CHAMACOS 7-year-olds. Human subjects protocols for both studies were approved by institutional review boards at University of California, Berkeley, and for Proyecto Mariposa, at the National Institute of Public Health in Mexico.
Detailed methods about the CHAMACOS study are described elsewhere (Eskenazi et al. 2003
). Briefly, between 1999 and 2000, pregnant women were enrolled from prenatal clinics serving a low-income, Spanish-speaking farm worker population in the Salinas Valley in Monterey County, California. Women were eligible if they were ≤ 20 weeks gestation at enrollment, were ≥ 18 years of age, spoke Spanish or English, qualified to receive poverty-based government health insurance, and planned to deliver at the local county hospital. Of the 601 participants enrolled, 526 were followed to delivery of a live-born singleton infant. A total of 346 women provided a prenatal blood sample of sufficient volume for analysis. Structured interviews collecting demographic and other information were conducted in English or Spanish by bilingual, bicultural study staff during pregnancy and at various ages during childhood. Blood was collected when the children were 7 years of age (n =
339) and occurred between March 2007 and November 2008. For the present analysis, we excluded twins (n =
6), children lacking 7-year PBDE measurements (n =
59), and children whose mothers were not born in Mexico or of Mexican ancestry (n =
8). In addition, two children were missing 7-year lipid measurements. The final sample size was 264. Seven-year-old children with missing chemical measurements were not appreciably different in sociodemographic characteristics, including the length of time their mothers had lived in the United States, from those included in these analyses. There were 160 mother–child pairs with serums analyzed for PBDEs and organochlorine (OC) pesticides. We found no statistical difference in demographic characteristics between these 160 mother–child pairs and those without maternal blood (n =
104) except that the latter mothers had lived in the United States longer. Additionally, there were no differences in child PBDE or DDT/DDE levels between the two groups.
Detailed methods for Proyecto Mariposa have been described previously (Rosas et al. 2009
). We recruited a low-income population in Mexico that was demographically similar to the CHAMACOS population. In both populations, the children had access to health care and were receiving government benefits for nutrition. In Mexico, between May and June 2006, we recruited a convenience sample of women and their 5-year-old children from government-run community health clinics serving families enrolled in the social welfare program Oportunidades. We enrolled participants from high-migration communities in the states of Michoacán, Guanajuato, and Jalisco, the states from which most of the CHAMACOS women originated.
Women and their children were eligible to participate if their child was approximately 5 years of age, the mother spoke fluent Spanish and was at least 18 years of age when her child was born, and the mother and child were currently receiving Oportunidades benefits, had lived exclusively in Mexico, and had not been to the United States for longer than 1 month. Of 317 families in the Proyecto Mariposa sample, 283 provided birth and demographic information and child blood samples of sufficient volume to perform chemical measurements. In Proyecto Mariposa, women were interviewed in Spanish by trained interviewers using a structured questionnaire similar to that used for CHAMACOS, although information about early childhood and pregnancy was obtained retrospectively for Proyecto Mariposa.
Blood collection and chemical analysis.
Maternal blood samples were collected from the CHAMACOS women around the 26th week of gestation. Child blood samples were collected from the CHAMACOS and Proyecto Mariposa cohorts at the time of the interview and were immediately processed, with the sera stored at –80°C until shipment on dry ice to the U.S. Centers for Disease Control and Prevention (CDC; Atlanta, GA), where they were analyzed. PBDEs and OC pesticides were measured in serum using gas chromatography/isotope dilution high-resolution mass spectrometry (Sjödin et al. 2004
). Samples were analyzed for 10 tri- to hepta-brominated congeners, BDEs 17, 28, 47, 66, 85, 99, 100, 153, 154, and 183 and for OC pesticides including DDT and DDE. Total lipids were determined based on the measurement of triglycerides and total cholesterol in serum using standard enzymatic methods (Roche Chemicals, Indianapolis, IN, USA) (Phillips et al. 1989
Data analysis. Statistical analyses were conducted using Stata for Windows, version 10.1 (StataCorp LP, College Station, TX, USA).
We compared demographic characteristics for the two populations using t
-tests for continuous and chi-square tests for categorical characteristics. Breast-feeding duration was censored at age 24 months. We calculated z
-scores for children’s body mass index (BMI) (kilograms per square meter) using sex-specific BMI-for-age percentile data issued in 2000 by the CDC (National Center for Health Statistics 2005
). CHAMACOS mothers’ length of time living in the United States at pregnancy was examined both as a continuous and as a categorical variable (≤ 1 year, 2–5 years, 6–10 years, or 11 years to lifetime).
Concentrations of PBDEs and OC pesticides were expressed on a blood lipid basis. PBDE congeners with < 65% detection frequency in both populations (BDEs 17, 66, and 183) were not included in these analyses. Values below the limit of detection (LOD) for which a signal was detected were coded with the concentration obtained; when no signal was detected, values were coded as the lowest concentration obtained for that congener divided by the square root of two (Hornung and Reed 1990
). To compare PBDE levels between groups, we examined individual congeners and also calculated a direct sum of the seven congeners with at least 65% detected in either of the two populations (BDEs 28, 47, 85, 99, 100, 153, 154).
We transformed the individual and summed PBDE and OC pesticide concentrations to the log10 scale to approximate a normal distribution. We used t-tests to compare GMs and Pearson correlations to examine the correlations among PBDE congeners. We also employed Pearson correlations and GMs to examine relationships between serum concentrations for the CHAMACOS mother–child pairs.
We calculated for both the CHAMACOS and Proyecto Mariposa cohorts the unadjusted beta coefficients for the association between exposure concentrations (PBDEs, DDT, and DDE) and child’s breast-feeding duration.
We determined the unadjusted and adjusted beta coefficients and GMs and 95% confidence intervals (CIs) for children’s PBDEs (BDEs 47, 99, 100, and 153), DDT, and DDE concentrations by the mothers’ duration (years) living in the United States at the time of pregnancy. The mothers’ time living in the United States was examined as both a continuous variable (left-censored at 18 years) and as a categorical variable. Proyecto Mariposa mothers lived in Mexico for their lifetime, that is, 0 years in the United States. The multivariable models were adjusted for the child’s BMI z-score, breast-feeding duration, and maternal parity (parous or not). p-Values ≤ 0.05 were considered statistically significant.