National Health and Nutrition Examination Survey

Polychlorinated biphenyls (PCBs) are a class of chlorinated aromatic hydrocarbon chemicals that once were used as heat-exchanger, transformer, and hydraulic fluids, and as additives to paints, oils, joint caulking, and floor tiles. The different types of PCBs are known as congeners, which are compounds that are distinguished by the number of chlorine atoms and their location on the biphenyl structure. Peak production occurred in the early 1970s, and production was banned in the U.S. after 1979. More than 1.5 million pounds of PCBs were manufactured in the U.S. prior to 1977. Concern about these chemicals continues because of their persistence in the environment and accumulation in wildlife and the animal food chain. PCBs enter the food chain by various routes, including migration into food from external sources, contamination of animal feeds, and accumulation in the fatty tissues of animals. PCBs are found a higher concentrations in fatty foods, (e.g., dairy products and fish), and food is the main source of exposure for the general population. PCBs can be transferred from mother to infant via breast milk. Exposure to these chemicals nearly always occurs as mixtures rather than individual PCB congeners.

Study participants aged 3 years and older from NHANES 2001 - 2002 with stored serum available to prepare pooled samples.

Twenty nine non-dioxin-like polychlorinated biphenyls were measured in serum by gas chromatography/isotope-dilution high-resolution mass spectrometry (GC/ID-HRMS). The analytical method for PCBs is described in Sjodin et al., 2004 and Barr et al., 2003. Reported results met the Division of Laboratory Sciences’ quality control and quality assurance performance criteria for accuracy and precision (similar to specifications outlined by Westgard, 1981).

Specimens were processed, stored, and shipped to DLS, NCEH, CDC (Atlanta, Georgia). The analytical approach used, including data processing is described in detail in Sjodin, et al (2004). Reported results met the Division of Laboratory Sciences’ quality control and quality assurance performance criteria for accuracy and precision (similar to specifications outlined by Westgard, 1981).

The NHANES quality assurance and quality control (QA/QC) protocols meet the 1988 Clinical Laboratory Improvement Act mandates. Detailed QA/QC instructions are discussed in the NHANES Laboratory/Medical Technologists Procedures Manual (LPM).

Read the General Documentation on Laboratory Data file for detailed QA/QC protocols. The analytical methods are described in the Description of the Laboratory Methodology section.

The sampling scheme for NHANES 2001–2002, a complex multistage area probability design, included 11,039 persons. Data were collected through household interviews and through standardized physical examinations conducted in mobile examination centers. On the basis of self-reported data, a composite race/ethnicity variable helped define three major racial/ethnic groups: non-Hispanic blacks, non-Hispanic whites, and Mexican Americans. Informed consent was obtained from all study participants. Serum samples used to prepare pools for analysis were selected from those obtained by venipuncture from 2150 participants, a random one-third subsample of people 12 years of age and older from NHANES 2001-2002 and representative of the U.S. general population for this age range. A total of 1,832 individual serum samples were used to prepare the 54 pooled serum samples analyzed for this subset of NHANES 2001-2002.

After collection of samples, serum specimens were divided into aliquots and were stored cold (2–4°C) or frozen until they were shipped on dry ice to CDC’s National Center for Environmental Health (NCEH). Serum samples were stored frozen after receipt at NCEH at –70°C until analyzed. The 2,150 individual serum samples available were categorized in 24 demographic groups (Table 1), each representing a combination of race/ethnicity, gender, and age (12–19 years, 20–39 years, 40–59 years, and 60 years and older). The number of samples available, the number of these samples that were usable, and the number of pools formed in each demographic group are presented in Table 1. On the basis of the number of individual serum samples per demographic group, multiple pools were available for 14 demographic groups, and one pool was available for 10 demographic groups (Table 1). To ensure that no individual sample overly influenced the pooled results, all serum samples included in any one pool (25.5 mL each) were of equal volume (i.e., each individual sample contributed 750 μL). Most pools included 34 individual serum samples that were randomly selected. Only 31 individual serum samples were available, however, for the pool representing non-Hispanic black men 60 years of age or older. In addition, one of the two pools representing Mexican American men between 20 and 39 years of age consisted of only 33 individual specimens. 

In addition, residual serum from 3–11 year old NHANES 2001–2002 participants, previously analyzed for cotinine, a marker of environmental tobacco smoke, was used to create pools representing 3-5 year old and 6-11 year old participants. The 1049 individual samples available were categorized in 12 demographic groups, each representing a combination of race/ethnicity, sex, and age (3–5 years and 6–11 years). A total of 937 randomly selected individual samples were used to prepare 24 pools (two per demographic group). The number of samples available, the number of these samples that were usable, and the number of pools formed in each demographic group are presented in Table 1. To ensure that no individual sample overly influenced the pooled results, all samples included in any one pool were of equal volume (i.e., 0.5 mL). Eleven of the individual serum samples from non-Hispanic black females 3–5 years of age contributed <0.5 mL. Each pool included 21 (3–5 year old) or 57 (6–11 year old) individual samples randomly selected (Table 1).

Table 1. Number of individual serum samples available, number of usable samples, and number of pools formed from NHANES 2001–2002 participants per demographic group.

Guidelines for Working with Pooled-Samples from NHANES 2001-2002
Because the samples constituting the pools originated from NHANES 2001-2002, which was designed to be representative of the non-institutionalized US population, the pooled results should provide good coverage of the US population. We cannot be assured, however, that estimates based on the pooled samples are unbiased for at least two reasons: 1) the sampling weight information was lost when the samples were pooled because each pool consists of equal volumes of each sample in the pool; and 2) the measured value for a pooled sample (consisting of individual samples that tend to be log-normally distributed) is comparable to an arithmetic average of log-normal results and thus represents a biased estimate of the central tendency (geometric mean or median) of the samples making up the pool. Also, because samples were pooled across the design cells of the original NHANES sampling design, it is not possible to obtain unbiased estimates of standard errors. These bias and standard error issues are being addressed in later NHANES surveys involving pooled samples. 

Pooled-Sample weights
The sampling weight information was lost when the samples were pooled because each pool consists of equal volumes of each sample in the pool. 

Variance estimation
The sampling scheme for NHANES 2001–2002 is a complex multistage, probability sampling design that selects participants who are representative of the civilian, non-institutionalized U.S. population. Over-sampling of certain population subgroups is done to increase the reliability and precision of health status indicator estimates for those groups. For PCBs, instead of using the full NHANES sample, a random one-third subsample of NHANES participants was used. Curtin et al. (2012) provide documentation of the construction of sampling weights for this one-third subsample. Also, because of the use of stratified multistage selection, incorporation of the sampling design is needed to calculate sampling variances (NCHS, 1994). These variances cannot be calculated for NHANES 2001-2002 pooled sample estimates, however, because samples were pooled across the design cells of the original NHANES 2001-2002 sampling design.

Links to NHANES Data Files
Although this pooled-sample data file cannot be directly linked to other NHANES 2001-2002 data files, it does contain demographic information (RIAGENDR, RIDRAETN, and RIDAGGRP) for each pooled-sample.

Variables in the data file
Each variable in the data file is given in Table 2 with its definition. The definition used for the numeric constant for categorical variables is given in the column: Definition of Categorical Variables.

The NHANES abbreviation used for each analyte measured with the IUPAC name is given in Table 3. The code “XYZ” in the variable names “ADJUSTED_RESULT_XYZ” and “RESULT_COMMENT_XYZ” as defined in Table 2 corresponds to the Value of the “NHANES code” in Table 3.