Cohort characteristics. Participants included in analyses did not differ in major demographic characteristics from the remainder of the CCCEH cohort, except for lower prenatal exposure to environmental tobacco smoke (31% vs. 39%) and sensitization to indoor allergens at 24 months (8% vs. 17%); however, indoor allergen-specific sensitization did not differ at 36 or 60 months or when measures from any of the three ages (24, 36, 60 months) among available samples for specific IgE were combined and comparisons between the two groups were reanalyzed (). MBzP was detected in all except one urine sample, and concentrations ranged widely [geometric mean (GM) = 13.6, IQR: 5.7–31.1 ng/mL]. MBzP concentrations were higher among African American (GM = 18.3; IQR: 8.7–35.4 ng/mL) than Dominican mothers (GM = 11.7, IQR: 4.9–26.4 ng/mL) (p < 0.001).
Characteristics of mother–child pairs included and excluded from analysis.
Eczema characteristics. By child age 24 months, 30% of mothers had reported at least once that their doctor had ever said that their child had eczema, with a higher proportion among African-American (48%, n = 62/129) than Dominican (21%, n = 51/247) children. The proportion of Dominican mothers reporting eczema in their children was comparable among mothers completing questionnaires in English versus Spanish. There was no significant difference in the proportion of children with ever-eczema by child sex overall (32% female, 28% male, p = 0.37) or after stratifying by race/ethnicity (data not shown). The proportion of children with ever eczema generally increased with the child’s age, from 10% at 3 months to 28% at 60 months (). However, women were sometimes inconsistent in their report of their child’s eczema. Among those who had a positive report of ever-eczema (at any of the 12 questionnaire time points), 59% of subsequent questionnaires were also positive.
Figure 1 Relative risk estimates of child ever-eczema for an IQR increase in prenatal log MBzP urinary concentration (IQR is 1.7 log units, 5.7–31.1 ng/mL), adjusting for specific gravity, race/ethnicity, and sex from separate modified Poisson regression (more ...)
MBzP and eczema. Adjusted RRs of child ever-eczema for an IQR increase in prenatal MBzP urinary concentrations at each of 12 questionnaire time points were consistently positive with larger magnitudes at the earlier ages (). Using a reclassified definition of early onset eczema, an IQR increase in log MBzP was associated positively with any report of eczema by 24 months (RR = 1.52; 95% CI: 1.21, 1.91; p < 0.001, n = 113/376) after adjusting for specific gravity, sex, and race/ethnicity. The parameter estimate for child’s sex was not significant (RR = 0.82 for males vs. females; 95% CI: 0.60, 1.10). Adjusting for prenatal exposure to environmental tobacco smoke, or maternal attributes (age, education, marital status, self-report of asthma, log total IgE) changed the main effect of MBzP by < 2%, and these variables were not retained in the adjusted model. ORs from a three-outcome multinomial logistic regression model were comparable for ever-eczema among inconsistent reporters (who said no ever-eczema after having said yes ever-eczema) (OR = 1.98; 95% CI: 1.23, 3.17; n = 63) and consistent reporters (OR = 2.02; 95% CI: 1.19, 3.43; n = 50) compared with those who never reported eczema (n = 263). A second sensitivity analysis defined the outcome for consistent reporters to those with three or more consecutive positive reports among those with earlier first report of eczema (OR = 2.25; 95% CI: 1.22, 4.13 vs. never eczema, and OR = 1.90; 95% CI: 1.23, 2.95 for inconsistent or < 3 positive reports vs. never eczema). To determine whether the association between an IQR increase in prenatal log MBzP concentrations and eczema was greater among children with early onset (by 24 months) versus late-onset eczema, multinomial analyses were performed after reclassifying eczema as early (29%, n = 100), late (13%, n = 43), or none (58%, n = 196). MBzP concentration was associated with early onset eczema (adjusted OR = 1.91; 95% CI: 1.23, 2.97). There was no association between MBzP and late onset eczema (adjusted OR = 0.90; 95% CI: 0.51, 1.58) for an IQR change in log MBzP from the multinomial logistic regression analysis. Moreover, there was no association in a modified Poisson regression model between MBzP concentration and the report at 60 months of “itchy rash at any time in the past 12 months that was coming and going for at least 6 months” (RR = 1.23; 95% CI: 0.81, 1.88; n = 56/341).
MBzP and eczema by ethnicity.
Because previously we reported that eczema may vary by race/ethnicity (Donohue et al. 2008
), we sought to understand better whether the association between phthalates and eczema was modified by race/ethnicity. Maternal African-American race versus Dominican ethnicity was a significant predictor of report of eczema by 24 months (RR = 2.09; 95% CI: 1.53, 2.86; p
< 0.001) after adjusting for MBzP concentration, specific gravity, and sex. There was a consistently higher probability of eczema for African Americans across the range of urinary concentrations of MBzP. Both ethnic groups had similar slopes for the exposure–response curve in stratified analysis, indicating a lack of multiplicative interaction (). However, there was a nonsignificant greater than additive interaction between African American ethnicity and each IQR increase in MBzP urinary concentration (RR African American 2.47; RR IQR MBzP 1.60; joint RR 3.59; RERI 0.52; bootstrap 95% CI: –0.11, 1.31).
Figure 2 Association between prenatal urinary concentration of MBzP and report of eczema by 24 months from a multivariable logistic regression adjusted with mean specific gravity and sex. The y-axis is shown on the logit scale and labeled with the corresponding (more ...)
MBzP, eczema, and allergic sensitization. A total of 26% (n = 91/355) of children were classified as sensitized to indoor aeroallergens at 24, 36, or 60 months of age. Among children with eczema by 24 months, 29% (n = 31/106) were sensitized compared with 24% among children without eczema (n = 60/249) (p = 0.31). Children with eczema by 24 months on average had higher total IgE levels at 60 months of age than children without eczema, regardless of whether they were defined as nonsensitized (GM = 39 IU/mL, n = 67 vs. 24 IU/mL, n = 152; p = 0.016) or sensitized to indoor aeroallergens (GM = 207 IU/mL, n = 22 vs. 107 IU/mL, n = 50; p = 0.056).
There was no observed association between an IQR increase in prenatal log MBzP urinary concentration and sensitization to indoor aeroallergens (RR = 0.89; 95% CI: 0.68, 1.16, n = 355), adjusting for specific gravity, race/ethnicity, and sex. Similarly, there was no observed association between an IQR increase in log MBzP concentration and log total IgE at 60 months in adjusted analyses (β = –0.14; 95% CI: –0.41, 0.13, n = 306). We also examined whether the association between prenatal MBzP urinary concentration and eczema differed by sensitization to any of three indoor aeroallergens. There was no significant interaction between sensitization to indoor allergens and higher MBzP concentration in predicting eczema using the RERI to test for departures from additivity (RR sensitization 1.19; RR IQR MBzP 1.54; joint RR 1.87; RERI 0.14; bootstrap 95% CI: –0.34, 0.42). Retesting the RERI while examining interactions between any seroatopy (i.e., using total IgE as a continuous variable) and higher MBzP concentration in predicting eczema found a slightly greater than additive effect of MBzP and log total IgE at 60 months of age (RR log total IgE 1.18; RR IQR MBzP 1.46; joint RR 1.72; RERI: 0.07; 95% CI: 0.01, 0.46; n = 295).
Other phthalates and eczema. Although urinary concentrations of the metabolites of several other phthalates correlated moderately with MBzP, in adjusted models with both prenatal MBzP and mono-n-butyl phthalate (MnBP; a major metabolite of di-n-butyl phthalate and a minor metabolite of BBzP) or mono(2-ethyl-5-hydroxyhexyl) phthalate [MEHHP; a metabolite of di(2-ethylhexyl) phthalate (DEHP)] urinary concentrations, the estimate for MBzP was changed by < 10%. In these two-metabolite models, only MBzP was a significant predictor of report of eczema by 24 months (RR per log unit MnBP = 1.13; 95% CI: 0.89, 1.43; n = 376; RR per log unit MEHHP = 1.06; 95% CI: 0.75, 1.51; n = 376) adjusting for log MBzP, specific gravity, race/ethnicity, and sex.