Polychlorinated Biphenyl Exposure and Neuropsychological Status Among Older Residents of Upper Hudson River Communities

Edward F. Fitzgerald; Erin E. Belanger; Marta I. Gomez; Michael Cayo; Robert J. McCaffrey; Richard F. Seegal; Robert L. Jansing; Syni-an Hwang; Heraline E. Hicks


Environ Health Perspect. 2008;116(2):209-215. 

In This Article

Materials and Methods

Study Population, Recruitment, and Interview

Participants were randomly selected from several sources, including an on-line telephone directory search and a digital database from infoUSA (Omaha, NE). Excluding persons who were not 55–74 years of age, there were 1,281 candidates in Fort Edward and Hudson Falls and 1,423 in Glens Falls. Participants were then screened by telephone for eligibility.

Overall, 1,125 (88%) candidates in the study area and 1,228 (86%) in the comparison area were able to be contacted. Of these persons, 85% agreed to be screened; 482 (50%) were eligible in the study area, 419 (41%) in the comparison area. People who had not lived in their respective area for at least 25 years (27%) were excluded. Persons with a history of stroke, severe head injury, signs of Parkinson or Alzheimer disease, or other significant cognitive or motor impairment (17%) were also excluded, as persons with severe cognitive disabilities would have difficulty completing the neuropsychological tests and interview. To avoid confounding with occupational PCB exposures, we also excluded anyone who had worked for 1 year or more at the capacitor plants or any other job that entailed PCB exposure (21%). Persons from Glens Falls were frequency matched to the Hudson Falls and Fort Edward residents on both age and sex.

Structured in-person interviews were conducted from 2000 to 2002; questions included sociodemographic characteristics, height, weight, residential, occupational, and dietary history, cigarette smoking, alcohol consumption, health conditions, medication use, hobbies, and other characteristics and behaviors. Occupational histories were reviewed by two industrial hygienists for the probability of occupational exposure to lead, mercury, solvents, or pesticides.

Among those who were eligible and invited to participate, the response rates were 38% in the study area and 41% in the comparison area. There were no differences in participation according to age or sex. After the data collection was completed, 53 participants were excluded because they failed to have blood drawn, the laboratory was unable to analyze the serum for PCBs, or their interview indicated a reason for exclusion not previously reported.

Given their similarity in serum PCB concentrations, the study and comparison groups were combined. As a result, the total sample consisted of 253 persons 55–74 years of age. Sixty-seven men and 66 women were residents of Hudson Falls or Fort Edward, New York, and 60 men and 60 women were residents of Glens Falls (Figure 1). More detailed descriptions of the populations, recruitment, and interviews are given elsewhere (Fitzgerald et al. 2007).

Figure 1.

Map of the area of study: Hudson Falls, Fort Edward, and Glens Falls, New York.

Serum PCB analysis. Fasting serum samples were analyzed by dual capillary gas chromatography with microelectron capture detection for the 30 congeners that typically constitute more than 95% of the total PCB residue in human serum (Humphrey et al. 2000); their sum was calculated as total PCB. The method detection limit was 0.02 ppb per congener; nondetectable concentrations were assigned a value of one-half the detection limit. However, data are only presented for individual congeners if 50% or more of the samples had a detectable concentration. To provide for comparability with other studies, cholesterol and triglycerides were assayed enzymatically for 245 persons and their PCB concentrations expressed on a lipid basis (Phillips et al. 1989), although Schisterman et al. (2005) recently have argued that this method of lipid adjustment may introduce bias. In addition, nine dioxin-like PCB congeners were measured in the serum of about 90% of the study participants, and PCB toxic equivalent quantities (TEQs) were calculated (Van den Berg et al. 2006). Twenty-two organochlorine pesticides were also measured, but most were nondetectable. The only major exceptions were dichlorodiphenyl trichlorethane (DDT) and its metabolite p,p-dichlorodiphenyl dichloroethene (DDE). Because the concentrations of DDE were on average 20 times greater than those for DDT, both compounds were summed and evaluated jointly. More detail on the chemical analysis, including quality assurance/quality control procedures, is reported elsewhere (Fitzgerald et al. 2007).

Blood metals analysis. For blood lead analysis, whole blood was diluted 1:10 in a phosphate matrix modifier containing Triton X-100 and dilute nitric acid, and 12 μL was deposited into a graphite furnace atomic adsorption spectrometry instrument that incorporated longitudinal Zeeman background correction with a transversely heated graphite atomizer (Parsons and Slavin 1993). For the blood mercury analysis, an inductively coupled plasma–mass spectrometer was used (Palmer et al. 2006). The collection tubes were trace metal-free to avoid background contamination.

Neuropsychological assessment. The neuropsychological tests were selected because they provide sensitive, clinically relevant measures of nervous system functions that are altered by aging and by damage to central dopamine processes. Not only have these tests been shown to be altered in studies of individuals occupationally exposed to PCBs (Singer 1988; Troster et al. 1991), but they also have been used in PCB research involving nonoccupationally exposed groups (Schantz et al. 1999, 2001), thereby facilitating a comparison of results between studies. In addition, olfactory function was assessed, as a decreased sense of smell is among the first signs of idiopathic diseases involving the depletion of brain dopamine (Doty et al. 1992). The New Adult Reading Test-Revised (NART-R) was used to assess intelligence (Nelson and O'Connell 1978), given that intellectual ability is a key determinant of performance on cognitive tests. The Test of Memory Malingering (TOMM) was used to discriminate between participants who were putting forth an adequate level of effort and those who were not (Tombaugh 1996).

Memory and learning was assessed by two different tests: the California Verbal Learning Test (CVLT) (Delis et al. 2000) and the Weschler Memory Scale (WMS) Form I-Russell's Revision (Russell 1975). The CVLT test uses semantic associations as a strategy to learning words in a 16-item list. The WMS was used to assess immediate and delayed recall of verbal and visual material.

Executive functioning pertains to abstract reasoning, concept formation, and other higher-order cognitive abilities. It was assessed with four different tests: the Trail Making Test-Parts A and B, the Stroop Color–Word Test (SCWT), and the Wisconsin Card Sorting Test (WCST). The Trail Making Test-Parts A and B is a subtest of the Halstead-Reitan Battery (Reitan and Wolfson 1993) and assesses visual scanning and attention. The SCWT consists of three parts and assesses the participant's ability to shift a perceptual set (Treverry et al. 1994). The WCST was used to assess concept formation and set shifting strategies.

The visual spatial domain pertains to the ability to interact with the environment visually and spatially. Two tests were used: the Digit Symbol Substitution Test (DSST), and the Block Design subtest (BDT) from the Wechsler Adult Intelligence Test-Revised (Wechsler 1981). The DSST was used to assess passive associative learning, visual speed, attention, and intense effort. The BDT was used to measure visuospatial organization.

Three tests were employed to assess motor function: the Static Motor Steadiness Test (SMST), the Grooved Pegboard Test (GPT), and the Finger Oscillation Test (FOT). The SMST assessed for tremorlike movements in the participants (Lezak et al. 2004). The GPT (Klove 1963) evaluated complex visuomotor coordination and visual–spatial orientation. The FOT is a subtest of the Halstead-Reitan Neuropsychological Battery and was used as a measure of motor speed and coordination (Reitan and Wolfson 1993).

Simple reaction time was measured after the participant was told to respond to the appearance of a visual stimulus after an auditory warning.

Depression and anxiety were assessed using the Beck Depression Inventory (BDI) and the State-Trait Anxiety Inventory (STAI) respectively. The BDI is a 21-item self-rated scale that measures the graded severity of certain symptoms of depression (Beck et al. 1961). The STAI was used to measure both state anxiety and trait anxiety. The instrument consisted of two scales of 20 items each (Speilberger et al. 1970); one measured state anxiety and the other measured trait.

Finally, olfactory function was measured using the Smell Identification Test (SIT), a 40-item scratch and sniff multiple-choice test (Doty 1983) that is widely used to assess olfactory functioning both clinically and experimentally.

Statistical analyses. Multiple linear regression analysis was used to test for association between PCB exposure and neuropsychological test scores after controlling for significant background variables that could potentially confound such associations. Following the method of Schantz et al. (1999, 2001), potential confounding variables were first screened for each neuropsychological test in a bivariate analysis. Candidate variables included age, sex, education, income, NART-R score, BMI, marital status, cigarette smoking, alcohol consumption, health conditions, medication use, employment status, physical activity level, hours of sleep, blood lead and mercury concentrations, serum DDT + DDE level, and occupational or hobby exposure to lead, mercury, solvents, or pesticides. Regarding smoking, participants reported the number of cigarettes smoked over the past year and the past 10 years, which in turn was then converted to packs smoked (nonsmokers were assigned zero). In addition to being analyzed as outcome variables, depression and anxiety were also considered as potential confounders in the analysis of the cognitive and motor tests, as they may affect performance in those domains. Variables that were significant in the bivariate analysis at p < 0.20 and considered biologically plausible by our consulting psychologist (R.J.M.) were regressed on the neuropsychological test scores using stepwise procedures to add (p < 0.10) or remove (p > 0.10) the variables one at a time. New regression models were then created with the addition of serum PCB concentrations to estimate their associations with the neuropsychological test scores after adjustment for potential confounders.

Serum PCB and DDE concentrations were expressed on a lipid basis and log transformed; blood lead and mercury concentrations were log transformed. When the distribution of scores on a given neuropsychological test was skewed, the scores were also log transformed. For a few tests, the log transformations failed to achieve normality, so the scores were dichotomized at the median and logistic regression performed. For some neuropsychological tests, serum PCB levels were divided into quartiles and a test for linear trend conducted, coding the categories as 1 through 4. The data were also stratified by age, sex, and income to assess effect modification.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: