The aim of this research was to characterize the body burden and covariates for exposure to three neurotoxicants among childbearing-aged women living in the US from 1999 through 2004. The magnitude of exposure to multiple environmental chemicals is underscored by the observation reported here that 23% of childbearing-aged women had three and another 33% had two xenobiotic levels at or above the median, where one would expect 12.5% and 25%, respectively. These findings support the need for health outcomes research resulting from co-exposures to all three neurotoxicants.
Lead-PCBs were identified as a common binary combination among childbearing-aged women who had two xenobiotic levels at or above the median. In their study of 138 adolescent Akwesasne (Mohawk) girls, Denham et al. (2005)
found a statistically significant interaction between lead, four estrogenic PCB congeners (52, 70, 101/90, 187) and the delayed attainment of menses (P
<0.05); this relationship was nonlinear. Few studies have examined this binary combination (Boucher et al., 2012
; Stewart et al., 2006
). Mechanistic studies are needed to describe the joint toxic action of this particular binary chemical combination.
The odds of having two or more xenobiotic levels at or above the median rose exponentially with age. This study confirmed previously reported findings of a strong correlation between age with PCBs (Axelrad et al., 2009
), lead (Mushak, 1998
) and methyl mercury (Caldwell et al., 2009
). This study’s oldest cohort of women (aged 40–49) had a markedly higher risk [OR=29.81; 95% CI 7.66–115.99]. While five US studies have examined blood for lipid-adjusted levels of PCB congeners (118, 138, 153, 180) in older women (Laden et al., 2001
), data correlating xenobiotic levels with age by decade were not available for comparison. In this study, the women aged 40–49 were born between 1950 and 1963 when pollution levels were significantly higher than current levels. If historic emissions are a valid explanation, some women older than 49 may have equally high or higher xenobiotic levels. Their co-exposures and the potential relationship to neurodegenerative disease among this age cohort should be examined.
Consuming any fish in the prior 30 days was associated with having two or more xenobiotic levels at or above the median. Domestic and imported sea food and freshwater fish consumption are significant predictors of adult methyl mercury and PCB levels (Grandjean et al., 1992
; Gunderson, 1995
) and to a lesser extent, lead (Falco et al., 2006
). Since the half-lives of mercury in blood and PCBs in serum are approximately 70 days, the 30-day recall reflects xenobiotic levels more appropriately than 24-hour recall (Tran et al., 2004
). Individual cell size was too small to analyze individual consumption data on predatory species (i.e., shark, swordfish or mackerel) known to biomagnify methyl mercury and PCBs. Consuming tuna, salmon or haddock was significantly related (P
<0.05) to higher body burden. Relative risk quadrupled when these fish were consumed more than once per week ().
Odds of childbearing-aged women having xenobiotic levels at or above the median based on fish consumption (NHANES weighted data 1999–2004).
These findings support reducing environmental chemical exposures associated with fish consumption. Though the US Food and Drug Administration (2011)
recommended a tolerance level for methyl mercury of 1 and 2 μg/g for PCBs in edible fish entering interstate commerce, this recommendation is neither legally enforceable nor applicable to intrastate commerce, recreational, or subsistence fishing. A nutrition rating system for fish at points-of-sale would increase awareness among consumers. “Food labels should give clear guidance about their healthfulness and encourage healthier choices through simplicity, visual clarity, and the ability to convey meaning without written information” (Institute of-Medicine (IOM), 2011
), p. 1).
While heavy alcohol consumption and binge drinking appeared to increase the odds of having two or more xenobiotic levels at or above the median, moderate alcohol consumption tended to decrease these odds slightly but not significantly when compared to non- and seldom drinkers. While alcohol consumption has been associated with reduced cardiovascular disease risk (Brien et al., 2011
), no evidence exists of a similar effect on xenobiotic body burden. Alcohol potentiation of prenatal methyl mercury- and lead-related toxicities has been demonstrated in animal studies (Gupta and Gill, 2000
; Maia et al., 2009
). Gender-based alcohol studies show greater severity of alcohol-related neurological damage among women than men (Mancinelli et al., 2009
). In this study, all 16–19 year olds were categorized as non- and seldom drinkers because NHANES restricted these data. Fryar et al. (2009)
estimated as many as 18.5% females aged 16–17 are heavy alcohol consumers or binge drinkers. Comparatively, in this study, 19% of this age cohort had serum cotinine levels >10 μg/dL indicating they were active smokers. Misclassification may have underestimated the true prevalence of alcohol consumption among the youngest cohort of women and contributed to the overall instability in this relationship between alcohol and body burden (Shrader-Frechette, 2008
A history of breastfeeding at least one child for one month or more was inversely correlated with increased body burden [OR=0.56; 95% CI 0.33–0.94]. Conversely, current breastfeeding tended to increase the odds of these women having two or more xenobiotics at or above the median, however this relationship was not significant [OR=1.97; 95% CI 0.56–6.89]. All three xenobiotics have been measured in breast milk (Agency for Toxic Substances and Disease Registry (ATSDR), 2004
; Gundacker et al., 2002
). These findings suggest breastfeeding increases chemical exposures for infants and children while reducing total maternal body burden with a potentially lasting effect.
As stated previously, environmental chemicals addressed in this study are known to transfer from maternal blood through the placenta to the fetus (Needham et al., 2011
). Woodruff et al. (2011)
found xenobiotic levels in pregnant women were higher than non-pregnant women when levels were adjusted for covariates (i.e., age, race-ethnicity, education, marital status, parity, body mass index and smoking). Multiple chemical exposures among pregnant women should be described more fully and compared to non-pregnant women.
As an estimate of risk, race-ethnicity was not statistically significant in this study, but there were differences observed. Health disparities among racial and ethnic minorities are well known (Morello-Frosch and Shenassa, 2006
; Payne-Sturges and Gee, 2006
). The odds of minority women having two or more of these xenobiotics at or above the median were higher than for non-Hispanic whites (). This study used data weighted to the US population dominated by non-Hispanic whites (73%). Since race and ethnicity are social and not biological constructs, this “bioethnic conscription” may act as an indirect surrogate for socioeconomic disadvantage (Montoya, 2007
). However, neither three socioeconomic indicators (food security, time in longest employment and marital status) nor any of the categorical income variables factored into the model (P
>0.20). In the absence of bias and real effect, the effect of race-ethnicity may be a random variation. Examining these data for each racial-ethnic group would allow for a more detailed comparison.
The findings of this study should be used to inform healthcare practitioners and environmental health professionals of the wide-spread prevalence of childbearing-aged women’s exposure to lead, mercury and PCBs. Emphasis should be placed on bioaccumulation, maternal exposures and intergenerational transfers during gestation and lactation. Longitudinal prospective studies should focus on the long-term health impacts of bioaccumulation from multiple environmental chemical exposures. Prospective studies spanning more than two generations should examine transgenerational consequences of these exposures.
4.1. Study limitations
The goodness-of-fit for the logistic regression model without interactions was fair (R2
=0.25). A coefficient of determination less than 0.40 is not uncommon with cross-sectional studies (Lehmann, 1975
; Murray, 2005
). To improve this metric, interaction among independent variables could be more fully described within the model. Adding to the dataset (i.e., NHANES 2005–2010) would sustain adequate cell counts required for sequential nested model operations; 33% of two-way interactions were strongly significant (P
<0.001). Comparing this study’s best-fit logistic regression model to similar models for each individual chemical as well as models for binary chemical combinations could lead to a better understanding of exposure covariates. Aside from data adequacy, the body burden does not identify sources of exposure. Bioaccumulation and intergenerational transfers complicate this identification. Overall, there is a limit to understanding these complex relationships using cross-sectional studies.
This study examined three chemicals—only a fraction of all chemicals detected in the environment and in humans. No inference should be made with regard to exposures to other chemicals. Only associations could be made about the relationships between dependent and independent variables since all data were collected at a single point in time. While these findings can be generalized to the population of childbearing-aged women who lived in the United States 1999–2004, no inferences should be made about exposures among other populations inside or outside the United States, nor should the results be extrapolated in terms of exposure risk for any given individual.
In conclusion, these findings are among the first description of body burden and risk factors for multiple chemical exposures among US childbearing-aged women. This study further supports increasing age, any fish consumption, and heavy alcohol consumption as significant risk factors for body burden. Prior history of breastfeeding lowered the body burden. Limited evidence was found of increased risk of exposure for minority status independent of other risk factors.