Association Between Serum Perfluorooctanoic Acid (PFOA) and Thyroid Disease
Association Between Serum Perfluorooctanoic Acid (PFOA) and Thyroid Disease
Data were from three independent cross-sectional waves of NHANES: 1999–2000, 2003–2004, and 2005–2006. NHANES surveys assess the health and diet of the noninstitutionalized civilian population of the United States and are administered by the National Center for Health Statistics (NCHS). The study protocol for NHANES was approved by the NCHS Institutional Review Board.
Solid-phase extraction coupled to high-performance liquid chromatography/turbo ion spray ionization/tandem mass spectrometry with isotope-labeled internal standards was used for the detection of PFOA and PFOS, with a limit of detection of 0.2 ng/mL (Kuklenyik et al. 2005). The laboratory methods and comprehensive quality control system were consistent in each NHANES wave, and documentation for each wave is available (NCHS 2009; NHANES 2006, 2007).
Serum polyfluorinated chemicals (PFCs) were measured in a one-third representative random subset of persons ≥ 12 years of age in each NHANES wave. Data from individuals < 20 years of age were excluded, because questions relating to disease prevalence were asked only for adults.
In all NHANES waves, adult respondents were asked about physician-diagnosed diseases. Associations were examined between PFOA and PFOS concentrations and thyroid disease outcomes. Individuals were asked whether they had ever been told by a doctor or health professional that they had a thyroid problem (in the 1999–2000 survey the questions related to goiter and other thyroid conditions) and whether they still had the condition. We further defined thyroid disease by considering those people who said they currently had thyroid disease and were taking any thyroid-related medication, including levothyroxine, liothyronine, "thyroid desiccated," and "thyroid drugs unspecified" for hypothyroidism and propylthiouracil and methimazole for hyperthyroidism. No details were available on specific thyroid disease diagnosis, and the PFC samples did not overlap with the thyroid hormone measurement subsamples in NHANES.
To assess disease specificity, associations were examined between PFOA and the other NHANES disease categories elicited: ischemic heart disease (combining any diagnoses of coronary heart disease, angina, and/or heart attack), diabetes, arthritis, current asthma, chronic obstructive pulmonary disease (COPD; bronchitis or emphysema), and current liver disease.
NHANES uses a complex cluster sample design with some demographic groups (including less-privileged socioeconomic groups and Mexican Americans) oversampled to ensure adequate representation. Prevalence estimates and models were therefore survey-weighted using the NHANES primary sampling unit, strata, and population weights, unless otherwise stated.
Multivariate logistic regression modeling was used to estimate odds ratios (ORs) of thyroid disease outcomes by quartile of PFOA and PFOS concentrations, and associations of other physician-diagnosed diseases. Because thyroid disease prevalence is markedly higher in women, we used sex-specific models. Because the distribution of PFC concentrations is skewed (with most people having relatively low exposures and with considerably more variance at the higher exposure end), all available data were pooled, and PFOA and PFOS concentrations were divided into population-weighted quartiles. Using the Hsieh method (Hsieh et al. 1998), our estimated power to detect an association of OR ≥ 1.8 with current treated disease comparing the top PFOA quartile (Q4) with bottom quartile (Q1) is 67% in women. Combining the lowest two quartiles (Q1 and Q2) into a larger control group provides 80% power. The corresponding minimum detectable effect size in men is OR > 2.9. Assumptions for the power calculations include a significance level of 5% and a multiple correlation coefficient of 0.2 relating PFOA exposure to potential confounders.
Models were adjusted for the following potential confounding factors: year of NHANES study; age; sex; race/ethnicity, from self-description and categorized into Mexican American, other Hispanic, non-Hispanic white, non-Hispanic black, and other race (including multiracial); education, categorized into less than high school, high school diploma (including GED), more than high school, and unknown education; smoking (from self-reported status asked for those ≥ 20 years of age), categorized into never smoked, former smoker, smoking some days, smoking every day, and unknown smoking status; body mass index (BMI; weight in kilograms divided by the square of measured height in meters), categorized into underweight (BMI < 18.5), recommended weight (BMI = 18.5–24.9), overweight (BMI = 25.0–29.9), obese (BMI = 30.0–34.9), and unknown BMI; and alcohol consumption (in adults ≥ 20 years of age, based on responses to the question "In the past 12 months, on those days that you drank alcoholic beverages, on the average day, how many drinks did you have?"), categorized into 0, 1, 2, 3, 4, and ≥ 5 drinks per day, and unknown alcohol consumption. Regression analyses were conducted using STATA/SE (version 10.1; StataCorp LP, College Station, TX, USA).
Methods
Study Population
Data were from three independent cross-sectional waves of NHANES: 1999–2000, 2003–2004, and 2005–2006. NHANES surveys assess the health and diet of the noninstitutionalized civilian population of the United States and are administered by the National Center for Health Statistics (NCHS). The study protocol for NHANES was approved by the NCHS Institutional Review Board.
Assessment of PFOA/PFOS Concentrations
Solid-phase extraction coupled to high-performance liquid chromatography/turbo ion spray ionization/tandem mass spectrometry with isotope-labeled internal standards was used for the detection of PFOA and PFOS, with a limit of detection of 0.2 ng/mL (Kuklenyik et al. 2005). The laboratory methods and comprehensive quality control system were consistent in each NHANES wave, and documentation for each wave is available (NCHS 2009; NHANES 2006, 2007).
Serum polyfluorinated chemicals (PFCs) were measured in a one-third representative random subset of persons ≥ 12 years of age in each NHANES wave. Data from individuals < 20 years of age were excluded, because questions relating to disease prevalence were asked only for adults.
Disease Outcomes
In all NHANES waves, adult respondents were asked about physician-diagnosed diseases. Associations were examined between PFOA and PFOS concentrations and thyroid disease outcomes. Individuals were asked whether they had ever been told by a doctor or health professional that they had a thyroid problem (in the 1999–2000 survey the questions related to goiter and other thyroid conditions) and whether they still had the condition. We further defined thyroid disease by considering those people who said they currently had thyroid disease and were taking any thyroid-related medication, including levothyroxine, liothyronine, "thyroid desiccated," and "thyroid drugs unspecified" for hypothyroidism and propylthiouracil and methimazole for hyperthyroidism. No details were available on specific thyroid disease diagnosis, and the PFC samples did not overlap with the thyroid hormone measurement subsamples in NHANES.
To assess disease specificity, associations were examined between PFOA and the other NHANES disease categories elicited: ischemic heart disease (combining any diagnoses of coronary heart disease, angina, and/or heart attack), diabetes, arthritis, current asthma, chronic obstructive pulmonary disease (COPD; bronchitis or emphysema), and current liver disease.
Statistical Analysis
NHANES uses a complex cluster sample design with some demographic groups (including less-privileged socioeconomic groups and Mexican Americans) oversampled to ensure adequate representation. Prevalence estimates and models were therefore survey-weighted using the NHANES primary sampling unit, strata, and population weights, unless otherwise stated.
Multivariate logistic regression modeling was used to estimate odds ratios (ORs) of thyroid disease outcomes by quartile of PFOA and PFOS concentrations, and associations of other physician-diagnosed diseases. Because thyroid disease prevalence is markedly higher in women, we used sex-specific models. Because the distribution of PFC concentrations is skewed (with most people having relatively low exposures and with considerably more variance at the higher exposure end), all available data were pooled, and PFOA and PFOS concentrations were divided into population-weighted quartiles. Using the Hsieh method (Hsieh et al. 1998), our estimated power to detect an association of OR ≥ 1.8 with current treated disease comparing the top PFOA quartile (Q4) with bottom quartile (Q1) is 67% in women. Combining the lowest two quartiles (Q1 and Q2) into a larger control group provides 80% power. The corresponding minimum detectable effect size in men is OR > 2.9. Assumptions for the power calculations include a significance level of 5% and a multiple correlation coefficient of 0.2 relating PFOA exposure to potential confounders.
Models were adjusted for the following potential confounding factors: year of NHANES study; age; sex; race/ethnicity, from self-description and categorized into Mexican American, other Hispanic, non-Hispanic white, non-Hispanic black, and other race (including multiracial); education, categorized into less than high school, high school diploma (including GED), more than high school, and unknown education; smoking (from self-reported status asked for those ≥ 20 years of age), categorized into never smoked, former smoker, smoking some days, smoking every day, and unknown smoking status; body mass index (BMI; weight in kilograms divided by the square of measured height in meters), categorized into underweight (BMI < 18.5), recommended weight (BMI = 18.5–24.9), overweight (BMI = 25.0–29.9), obese (BMI = 30.0–34.9), and unknown BMI; and alcohol consumption (in adults ≥ 20 years of age, based on responses to the question "In the past 12 months, on those days that you drank alcoholic beverages, on the average day, how many drinks did you have?"), categorized into 0, 1, 2, 3, 4, and ≥ 5 drinks per day, and unknown alcohol consumption. Regression analyses were conducted using STATA/SE (version 10.1; StataCorp LP, College Station, TX, USA).
