The Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS) is a longitudinal birth cohort study of predominantly Mexican-American families in California’s Salinas Valley. Detailed methods for CHAMACOS are published elsewhere (Eskenazi et al. 2004
). Eligible pregnant women (≥ 18 years old, < 20 weeks gestation, Spanish- or English-speaking, qualifying for low-income health insurance, and planning to deliver at the public hospital) were recruited between October 1999 and October 2000 from community clinics. The cohort included 601 women, 526 of whom delivered live-born singletons.
Women were interviewed twice during pregnancy (at ~ 13 and 26 weeks gestation), after delivery, and when children were 6 months old, and 1, 2, 3.5, 5, and 7 years old. Mothers completed the Peabody Picture Vocabulary Test (PPVT) or Test de Vocabulario en Imágenes Peabody (TVIP) of verbal intelligence (Dunn and Dunn 1981
) at the 6-month visit and the Center for Epidemiologic Studies Depression Scale (CES-D) (Radloff 1977
) at the 1-year visit. Age-appropriate versions of the HOME (Home Observation for Measurement of the Environment) survey were completed at most postdelivery visits (Caldwell and Bradley 1984
). Birth weight and gestational duration were abstracted from medical records.
Neurobehavioral assessments were performed by bilingual psychometricians, and children were assessed in their dominant language. A total of 310 children were assessed at 5 years (mean = 60.0 ± 2.6 months) and 323 at 7 years (85.2 ± 2.9 months). The present analysis excludes four children with autism, Down syndrome, cerebral palsy/hydrocephalus, or deafness and 63 children who lacked PBDE measurements.
Compared with children in the cohort who were not followed, children included in the present analyses were more likely to be female and born full term, with mothers who were older, breastfed longer, and were less likely to smoke or drink during pregnancy (data not shown). They did not differ according to other sociodemographic characteristics or by their maternal prenatal PBDE levels [median = 24.9 ng/g lipid; interquartile range (IQR) 14.0–42.1] for those followed versus 23.8 ng/g lipid (IQR = 14.9–41.3) for those not followed].
Mothers provided written informed consent at both visits, and children provided verbal assent at 7 years of age. Study activities were approved by the University of California at Berkeley (UC) Committee for the Protection of Human Subjects. A technical assistance agreement was established between the Division of Laboratory Sciences at the National Center for Environmental Health, Centers for Disease Control and Prevention (CDC), and UC Berkeley.
At the 5-year visit, mothers completed the Child Behavior Checklist (CBCL)/1.5–5 (CBCL) (Achenbach and Rescorla 2000
). We analyzed two subscales as continuous raw scores: the Attention Problems scale and the DSM-IV–oriented Attention Deficit/Hyperactivity Disorder (ADHD) Problems scale. We also analyzed a “borderline clinical range” (≥ 93rd percentile in the standardization sample) indicator variable for each scale (Achenbach and Rescorla 2000
). In addition, children were assessed on the Conners’ Kiddie Continuous Performance Test (K-CPT) (Conners and Staff 2001
), a 7-min computerized vigilance task that assesses reaction time, accuracy, and impulse control. We determined continuous T
-scores (standardized to a nonclinical population) for errors of commission, errors of omission, and hit reaction time (Conners and Staff 2001
). We also examined the continuous ADHD Confidence Index score, which indicates the probability that children are correctly classified as having clinical ADHD, and a binary variable indicating a Confidence Index score ≥ 70th percentile.
At child’s age 7 years, mothers and teachers completed the Conners’ ADHD/DSM-IV Scales (CADS) (Conners 2001
) and the Behavior Assessment System for Children, 2nd edition (BASC) (Reynolds and Kamphaus 2004
). CADS data from four subscales (Conners ADHD index score, and DSM-IV–based Inattentive, Hyperactive/Impulsive, and Total ADHD scores) were analyzed both as continuous, standardized scores (T
-scores; mean ± SD = 50 ± 10) and as a binary variable indicating scores in the “Moderately” or “Markedly Atypical” range (T
-score ≥ 65) (Conners 2001
). BASC data from Hyperactivity and Attention Problems subscales were analyzed as standardized T
-scores and as a binary “at-risk” or “clinically significant” variable (T
-score ≥ 60) (Reynolds and Kamphaus 2004
At ages 5 and 7 years, children’s gross motor skills were assessed using select subscales of the McCarthy Scales of Children’s Abilities (McCarthy 1972
). Their fine motor dexterity was assessed with a pegboard test (Wide Range Assessment of Visual Motor Ability; WRAVMA) (Adams and Sheslow 1995
) (age-standardized mean = 100 ± 15) and with a finger-tapping task [at 5 years: Behavioral Assessment and Research System (BARS) (Rohlman et al. 2003
); and at 7 years: Reitan Neuropsychology Laboratory (Tucson, AZ)]. We standardized McCarthy gross motor and finger tap scores within our study population (z
-scores, mean = 0 ± 1).
At 5 years of age, children completed tests of receptive verbal intelligence in both English and Spanish using the PPVT and TVIP, respectively (Dunn and Dunn 1981
). We analyzed children’s continuous standardized scores (mean = 100 ± 15) in their language of best performance. We assessed children’s performance intelligence (PIQ) with the Wechsler Preschool and Primary Scale of Intelligence, 3rd edition (WPPSI-III) (mean = 100 ± 15).
At age 7 years, children were assessed on four subdomains of the Wechsler Intelligence Scale for Children–Fourth Edition (WISC-IV) (Wechsler 2003
): Verbal Comprehension, Perceptual Reasoning, Working Memory, and Processing Speed. A Full-Scale IQ was also calculated (mean = 100 ± 15 for the Full-Scale IQ and all components).
Other questions. Mothers were also asked “Has a doctor, nurse, psychologist or teacher ever told you that your child might have 1) attention problems? or 2) learning problems?” Teachers were asked “Do you have any specific concerns about this student (in terms of) 1) emotional problems, 2) behavioral problems, or 3) learning problems?”
PBDE exposure assessment.
Blood samples were collected by venipuncture from mothers during pregnancy (mean = 26.7 ± 2.6 weeks gestation, n
= 219) or at delivery (n
= 60), and from children at the 7-year visit (n
= 272). PBDE serum levels in women with data at both time points were very strongly correlated (Pearson r
≥ 0.98, p
< 0.001)]. Samples were immediately processed and stored at –80o
C until shipment on dry ice to the CDC (Atlanta, GA). Samples were analyzed at CDC for 10 congeners (BDEs 17, 28, 47, 66, 85, 99, 100, 153, 154, and 183) using gas chromatography isotope dilution high-resolution mass spectrometry (Sjödin et al. 2004
). PBDE concentrations are expressed on a serum lipid basis (nanograms per gram lipids). Total serum lipid concentrations were determined based on the measurement of triglycerides and total cholesterol using standard enzymatic methods (Roche Chemicals, Indianapolis, IN) (Phillips et al. 1989
). The limits of detection (LODs) for BDE-47 ranged from 0.3 to 2.6 ng/g lipids for maternal samples, and 0.4 to 0.8 ng/g lipids for child samples. For all other congeners, LODs ranged between 0.2 and 0.7 ng/g lipids for maternal and 0.3 and 5.6 ng/g lipids for child samples, respectively. Quality control samples (blanks and spikes) were included in each run.
We used the sum of BDEs 47, 99, 100, and 153 congeners as our primary exposure measure. Values < LOD were assigned the machine-read value if a signal was detected. If not, all concentration levels < LOD were imputed at random based on a log-normal probability distribution using maximum likelihood estimation (Lubin et al. 2004
We assessed maternal exposure to organophosphate (OP) insecticides as measured by dialkyl phosphate (DAP) metabolites in maternal urine (at 13 and 26 weeks gestation), using an isotope dilution gas chromatography-tandem mass spectrometry method (Bradman et al. 2005
; Bravo et al. 2002
); lead in maternal prenatal and cord blood samples, using graphite furnace atomic absorption spectrophotometry; polychlorinated biphenyls (PCBs) in maternal serum using high-resolution gas chromatography/high-resolution mass spectrometry with isotope dilution quantification (Barr et al. 2003
); and maternal thyroid stimulating hormone (TSH; using immunochemiluminometric assay) and free thyroxine (T4
; using direct equilibrium dialysis followed by radioimmunoassay) (Bayer ADVIA Centaur system; Siemens Healthcare Diagnostics, Deerfield, IL) at 26 weeks gestation (Chevrier et al. 2010
; Nelson and Tomei 1988
Data analysis. PBDE levels were expressed on the log10 scale. To determine the shape of the dose–response function, we ran generalized additive models using cubic splines. If nonlinearity was detected (p < 0.10), additional models were run with categorized PBDE concentrations (quartiles). We re-ran all final models with PBDE concentrations expressed on a serum basis (picograms per gram serum) with total serum lipids as a covariate. We also ran models with the sum of all 10 PBDE congeners; individually for each of the four primary congeners (47, 99, 100, and 153); and excluding outliers (defined as being ≥ 3.5 SD away from the mean for log10 PBDEs or the outcome).
Variables were identified as potential confounders based on their relationship to neurodevelopment. We examined the following [see Supplemental Material, Table S3 for categories (http://dx.doi.org/10.1289/ehp.1205597
)]: maternal age, education, years in the United States, marital status, work outside the home, use of alcohol and tobacco during pregnancy, depression (CES-D), parity, and PPVT or TVIP score; housing density, household poverty, pregnancy exposure to environmental tobacco smoke, number of children in the home, father’s presence in the home, and HOME score at 6 months and 7 years; preschool and out-of-home child care attendance; psychometrician, location, and language of assessment; and child sex, birth weight, preterm delivery status, and handedness (motor outcomes only). Missing values (< 10%) for covariates were imputed by randomly selecting a value from the dataset.
We built separate models for attention, cognition, and motor outcomes, and used the same model for all outcomes within a category. In addition to child’s sex and months of age (continuous), final models included all covariates that changed the coefficient for the main exposure and any outcome within the group by > 10%. The covariates maintained in the models are listed in the footnote of the respective tables.
For sensitivity analyses, we adjusted for birth weight, gestational age at birth, maternal thyroid hormone (TSH and free T4
), DAPs, lead, and PCBs in separate models (Chevrier et al. 2010
; Harley et al. 2011
). We evaluated effect modification by child sex. In addition, we included maternal and child PBDE levels in the same models, although doing so reduced the sample size (n
Main effects were considered statistically significant with p < 0.05 based on two-tailed tests, and interactions were considered significant if p < 0.10. All analyses were conducted with STATA version 10.1 (StataCorp, College Station, TX).