Phthalate Exposure and Allergy in the US Population
Phthalate Exposure and Allergy in the US Population
We used publicly available data from NHANES 2005–2006 [Centers for Disease Control and Prevention (CDC 2012)] to evaluate the association of phthalates and allergy. NHANES 2005–2006 collected detailed data on allergic symptoms and sensitization, so both questionnaire and biochemical measures of allergy are available for all NHANES participants > 1 year of age (n = 8,338). At the time of recruitment, all study participants provided informed consent. All data were anonymized prior to becoming publicly available. Urinary phthalate concentrations were measured in a random sample of participants ≥ 6 years of age (n = 2,548). Our analysis is limited to the 2,325 individuals who had complete information on allergy, urinary phthalate concentrations, and model covariates.
We assessed both self-reported current allergic symptoms and allergic sensitization as measured by specific IgE (sIgE). Information on current allergic symptoms was obtained from self-administered questionnaires completed at the NHANES clinic visit. Subjects < 16 years of age were interviewed with a proxy respondent, usually a parent, responsible for completing the interview. The questionnaire asked about six allergic conditions (asthma, wheeze, hay fever, allergy, itchy rash, and rhinitis) in the past year.
Serum samples were analyzed for allergen-specific IgEs using the Pharmacia Diagnostics ImmunoCAP 1000 System (Kalamazoo, MI, USA). A total of 19 allergen-specific IgEs (Dermatophagoides farinae, Dermatophagoides pteronyssinus, cat, dog, cockroach, Alternaria alternata, peanut, egg white, cow's milk, ragweed, rye grass, bermuda grass, oak, birch, shrimp, Aspergillus fumigatus, Russian thistle, mouse, and rat) were assessed. Individuals who tested positive (≥ 0.35 kU/L) to at least one allergen were considered allergen sensitized (sIgE positive). Information on sensitization to specific allergens from NHANES 2005–2006 has been published elsewhere (Salo et al. 2011).
Fifteen phthalate metabolites were measured in spot urine samples using high performance liquid chromatography–electrospray ionization–tandem mass spectrometry (HPLC-ESI-MS/MS) at the National Center for Environmental Health laboratory (CDC, Atlanta, GA, USA) (CDC 2009). Four of these analytes were primary [mono-(2-ethyl)-hexyl phthalate (MEHP)] or secondary metabolites [mono-2-ethyl-5-carboxypentyl phthalate (MECPP), mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), and mono(2-ethyl-5-oxohexyl) phthalate (MEOHP)] of DEHP. We summed the concentrations of all four metabolites to create a summary DEHP variable (ΣDEHP) for analysis; individual DEHP metabolites were not analyzed because of their common sources and the resulting high correlation among these metabolites (77–98%). We analyzed all chemicals that were detected in ≥ 25% of the population; for values below the detection limit, we assigned a value of the limit of detection (LOD) divided by the square root of 2 (Hornung and Reed 1990).
Information on covariates was obtained either via questionnaire (e.g., demographic characteristics, smoking status) or via measurement [e.g., body mass index (BMI)]. Urinary creatinine levels were measured using the Jaffe rate reaction with a CX3 analyzer (Beckman Instruments, Brea, CA, USA).
We used logistic regression models adjusted for study design using sampling weights to estimate associations of urinary phthalates with measures of allergic sensitization and allergic symptoms. Urinary phthalate concentrations were log10-transformed because of nonnormality of the distribution. Models were adjusted for age (continuous), race/ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, other), gender, creatinine (log10 transformed, continuous), BMI (categories), and cotinine (categories). Cotinine was classified as < LOD (0.015 ng/mL), low exposure (< 10 ng/mL) and high exposure (≥ 10 ng/mL). For adults, BMI was calculated as body weight in kilograms divided by height in meters squared and categorized as underweight or normal (< 25), overweight (25 to 30), or obese (≥ 30). For children, BMI was classified as the age percentile underweight or normal (< 85th percentile), overweight (85–95th percentile), obese (≥ 95th percentile) (Krebs et al. 2007). Similar modeling strategies have been employed for previous analyses of these outcomes in the NHANES 2005–2006 data (Salo et al. 2011). We also evaluated poverty income ratio [PIR, three categories: low (≤ 1.3), middle (1.3–3.5), and high (> 3.5) income] as a potential confounder because previous analyses have shown an association between socioeconomic status and phthalate concentrations in women (Kobrosly et al. 2012) and allergic sensitization in children in NHANES 2005–2006 (Visness et al. 2009). Adjustment for PIR did not substantially alter our odds ratio (OR) estimates. Therefore, to maximize the observations included in our models, we did not include PIR as a covariate (100 missing observations). Data for children (6–17 years of age) and adults were analyzed separately because both the covariate structure and the outcome prevalence differed between adults and children. Because phthalate concentrations and allergic sensitization rates differ by race/ethnicity, we assessed potential interaction by race/ethnicity in four categories by including three two-way interaction terms in our models and performed a likelihood ratio test (3 degrees of freedom) comparing the fit of models with and without the interaction terms to assess whether statistical interaction was present. In addition, we explored whether findings for monobenzyl phthalate (MBzP) and allergic symptoms were related to allergic sensitization by expanding our logistic regression models to four-level polytomous models (allergic sensitization + symptom, symptom without sensitization, sensitization without symptom, and no symptom + no sensitization) for each of the four symptoms (asthma, wheeze, rhinitis, hay fever). To test whether the ORs differed across the four strata, we used a contrast statement; a p-value for difference was the result of this contrast test. All statistical modeling was done using survey procedures in SAS, version 9.3 (SAS Institute Inc., Cary, NC, USA). A p-value ≤ 0.05 was considered statistically significant.
Methods
We used publicly available data from NHANES 2005–2006 [Centers for Disease Control and Prevention (CDC 2012)] to evaluate the association of phthalates and allergy. NHANES 2005–2006 collected detailed data on allergic symptoms and sensitization, so both questionnaire and biochemical measures of allergy are available for all NHANES participants > 1 year of age (n = 8,338). At the time of recruitment, all study participants provided informed consent. All data were anonymized prior to becoming publicly available. Urinary phthalate concentrations were measured in a random sample of participants ≥ 6 years of age (n = 2,548). Our analysis is limited to the 2,325 individuals who had complete information on allergy, urinary phthalate concentrations, and model covariates.
We assessed both self-reported current allergic symptoms and allergic sensitization as measured by specific IgE (sIgE). Information on current allergic symptoms was obtained from self-administered questionnaires completed at the NHANES clinic visit. Subjects < 16 years of age were interviewed with a proxy respondent, usually a parent, responsible for completing the interview. The questionnaire asked about six allergic conditions (asthma, wheeze, hay fever, allergy, itchy rash, and rhinitis) in the past year.
Serum samples were analyzed for allergen-specific IgEs using the Pharmacia Diagnostics ImmunoCAP 1000 System (Kalamazoo, MI, USA). A total of 19 allergen-specific IgEs (Dermatophagoides farinae, Dermatophagoides pteronyssinus, cat, dog, cockroach, Alternaria alternata, peanut, egg white, cow's milk, ragweed, rye grass, bermuda grass, oak, birch, shrimp, Aspergillus fumigatus, Russian thistle, mouse, and rat) were assessed. Individuals who tested positive (≥ 0.35 kU/L) to at least one allergen were considered allergen sensitized (sIgE positive). Information on sensitization to specific allergens from NHANES 2005–2006 has been published elsewhere (Salo et al. 2011).
Fifteen phthalate metabolites were measured in spot urine samples using high performance liquid chromatography–electrospray ionization–tandem mass spectrometry (HPLC-ESI-MS/MS) at the National Center for Environmental Health laboratory (CDC, Atlanta, GA, USA) (CDC 2009). Four of these analytes were primary [mono-(2-ethyl)-hexyl phthalate (MEHP)] or secondary metabolites [mono-2-ethyl-5-carboxypentyl phthalate (MECPP), mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), and mono(2-ethyl-5-oxohexyl) phthalate (MEOHP)] of DEHP. We summed the concentrations of all four metabolites to create a summary DEHP variable (ΣDEHP) for analysis; individual DEHP metabolites were not analyzed because of their common sources and the resulting high correlation among these metabolites (77–98%). We analyzed all chemicals that were detected in ≥ 25% of the population; for values below the detection limit, we assigned a value of the limit of detection (LOD) divided by the square root of 2 (Hornung and Reed 1990).
Information on covariates was obtained either via questionnaire (e.g., demographic characteristics, smoking status) or via measurement [e.g., body mass index (BMI)]. Urinary creatinine levels were measured using the Jaffe rate reaction with a CX3 analyzer (Beckman Instruments, Brea, CA, USA).
We used logistic regression models adjusted for study design using sampling weights to estimate associations of urinary phthalates with measures of allergic sensitization and allergic symptoms. Urinary phthalate concentrations were log10-transformed because of nonnormality of the distribution. Models were adjusted for age (continuous), race/ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, other), gender, creatinine (log10 transformed, continuous), BMI (categories), and cotinine (categories). Cotinine was classified as < LOD (0.015 ng/mL), low exposure (< 10 ng/mL) and high exposure (≥ 10 ng/mL). For adults, BMI was calculated as body weight in kilograms divided by height in meters squared and categorized as underweight or normal (< 25), overweight (25 to 30), or obese (≥ 30). For children, BMI was classified as the age percentile underweight or normal (< 85th percentile), overweight (85–95th percentile), obese (≥ 95th percentile) (Krebs et al. 2007). Similar modeling strategies have been employed for previous analyses of these outcomes in the NHANES 2005–2006 data (Salo et al. 2011). We also evaluated poverty income ratio [PIR, three categories: low (≤ 1.3), middle (1.3–3.5), and high (> 3.5) income] as a potential confounder because previous analyses have shown an association between socioeconomic status and phthalate concentrations in women (Kobrosly et al. 2012) and allergic sensitization in children in NHANES 2005–2006 (Visness et al. 2009). Adjustment for PIR did not substantially alter our odds ratio (OR) estimates. Therefore, to maximize the observations included in our models, we did not include PIR as a covariate (100 missing observations). Data for children (6–17 years of age) and adults were analyzed separately because both the covariate structure and the outcome prevalence differed between adults and children. Because phthalate concentrations and allergic sensitization rates differ by race/ethnicity, we assessed potential interaction by race/ethnicity in four categories by including three two-way interaction terms in our models and performed a likelihood ratio test (3 degrees of freedom) comparing the fit of models with and without the interaction terms to assess whether statistical interaction was present. In addition, we explored whether findings for monobenzyl phthalate (MBzP) and allergic symptoms were related to allergic sensitization by expanding our logistic regression models to four-level polytomous models (allergic sensitization + symptom, symptom without sensitization, sensitization without symptom, and no symptom + no sensitization) for each of the four symptoms (asthma, wheeze, rhinitis, hay fever). To test whether the ORs differed across the four strata, we used a contrast statement; a p-value for difference was the result of this contrast test. All statistical modeling was done using survey procedures in SAS, version 9.3 (SAS Institute Inc., Cary, NC, USA). A p-value ≤ 0.05 was considered statistically significant.