This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).

Recent evidence suggests that endocrine disrupting chemicals (EDCs) may cause perturbations in endogenous hormonal regulation that predispose to weight gain. Using data from NHANES (1999–2002), we investigated the association between body mass index (BMI), waist circumference (WC) and selected persistent organic pollutants (POPs) via multiple linear regressions. Consistent interaction was found between gender, ln oxychlordane and ln p,p’ DDT. Also, we found an association between WC and ln oxychlordane and ln hpcdd in subjects with detectable levels of POPs, whereas an association between WC and ln p,p’ DDT was observed in all subjects. Furthermore, ln Ocdd showed an increase with higher WC and BMI, whereas, ln trans-nonachlor decreased with higher BMI. Hence, BMI and WC are associated with POPs levels, making the chemicals plausible contributors to the obesity epidemic.

The prevalence of obesity has risen substantially in both adults and children, and is recognized as a serious public health problem [^{2}, which is the most commonly used proxy index of obesity, indicators of regional fat distribution such as waist circumference (WC) have been linked to cardiovascular disease [

While much has been written about the reasons behind the global obesity epidemic, industrial toxicants found in the environment, the food system, and humans have just begun to receive significant attention. Recent evidence suggests that environmental contaminants known or presumed to disrupt endocrine systems, appropriately termed endocrine disrupting chemicals (EDCs), may play a role in the growing problem of obesity [

Recently, a dose-response relation was observed between serum concentrations of POPs and metabolic syndrome [

We used data from the publicly available nationally representative cross-sectional surveys of the U.S. non-institutionalized civilian population conducted by the US National Center for Health and Statistics (NCHS), (

Anthropometric measures including height, weight, and WC were obtained via standardized protocols as described elsewhere (

Measures of polychlorinated dibenzo-

We selected five POPs (present in 80% of the NHANES population): 1,2,3,4,6,7,8-heptachlorodibenzo-

The primary analysis used three regression models to test all available participants for associations with overall adiposity as assessed by BMI and WC. Two models regressed BMI and WC on gender, ethnicity, age, age squared, and POPs. The third model regressed WC on the previous predictors while controlling for BMI. We will refer to these models as additive models. Each of these models was extended by including two-way interactions between gender and the POPs. Secondary/sensitivity analyses employed all of the same models as the primary analyses but evaluated only participants with detectable levels of POPs. All parameter interpretations will be done on using standard deviation units based on standardized regression parameters; _{j} = β̂_{j}_{j}_{y}

All analyses, including descriptive statistics, were conducted using SAS-Callable SUDDAN 9.0.1, which estimates standard errors using the sampling weights, strata, and primary sampling units (PSU) from NHANES taking into account for the complex sampling procedures used. For details on the sampling procedures used, visit the NHANES website at

Descriptive statistics for gender, ethnicity, age, BMI, and WC are displayed in

Results of the additive regression models for all participants are presented in

For the additive WC model, there are significant associations for gender (p < 0.00005), age (p < 0.00005), age squared (p < 0.00005), and ln Ocdd (p = 0.0052). The model accounted for 16% of the variance in WC. WC increases by 1.76 centimeters with each SD unit increase in ln Ocdd. The joint test p-value for the additive WC model is 0.0008.

For testing the additive WC model controlling for BMI, there was a significant association for gender (p < 0.00005) and ln hpcdd (p = 0.0091) with the model accounting for 88% of the WC variance (

For the ln hpcdd interaction, male BMIs increase by 0.413 and female BMIs decrease by 1.71 for every SD unit increase in ln hpcdd (

The interaction WC model accounted for 18% of the WC variance and showed significant associations for ln Ocdd (p = 0.0120) and ln DDT (p = 0.0010) (

Lastly, the interaction WC model controlling for BMI accounted for 88% of the WC variance and showed significant associations for ln hpcdd (p = 0.0060) with a significant gender by ln DDT interaction (p = 0.0056) (

Results of the additive regression models for participants with detectable POPs are presented in

The additive BMI model accounted for 13% of the BMI variance and showed significant associations for ethnicity (p = 0.0450), age (p < 0.00005), age squared (p < 0.00005), ln trans-nonachlor (p = 0.0483), and ln DDT (p = 0.0340) (

For the additive WC model, significant associations, and in the same direction, for gender (p < 0.00005), age (p < 0.00005), and age squared (p < 0.00005) appeared (

For the additive WC model controlling for BMI, the model accounted for 87% of the WC variance (

The interaction WC model accounted for 23% of the variance and showed significant associations for gender (p = 0.0112), age (p < 0.00005), age squared (p < 0.00005), ln Ocdd (p = 0.0123), and ln DDT (p = 0.0184) (

Finally, the interaction WC model controlling for BMI accounted for 88% of the WC variance with significant main associations for ethnicity(p = 0.0299), age (p = 0.0105), BMI (p < 0.00005), BMI squared (p = 0.0211), and ln hpcdd (p = 0.0214) (

Consistent findings for the additive models with all participant and participants with detectable POPs were as follows: The additive BMI models showed consistent associations for ethnicity, age, age squared, and ln DDT. For the additive WC model, the consistent associations were for gender, age, and ln Ocdd. Lastly, the additive WC model controlling for BMI show consistent finding for gender, age, BMI, and BMI squared.

Consistent findings for the interaction models for all participants and participants with detectable POPs were as follows: The interaction BMI model showed consistent associations for age, age squared, and ln Ocdd. The interaction WC model showed consistent associations for age, ln Ocdd, and ln DDT. In addition, it had consistent gender by ln oxychlordane and ln DDT interactions. Lastly, the interaction WC model controlling for BMI had consistent associations for age, BMI, BMI squared, and ln hpcdd.

Lastly, the additive and interaction models were estimated with all participants ages 19 and up (results not shown but available upon request). Results of these analyses were consistent with the results of the equivalent models with all participants. Specifically, all parameter estimates were in the same directions and those that were significant (or not significant) remained significant (or not significant). The one exception occurred with the interaction model of waist circumference controlling for BMI. In this model the parameter estimate for European Americans went from non significant to significant and the parameter estimate for age went from significant to non significant.

Consistent effects were found for Ocdd and/or DDT with BMI. In addition, a relatively consistent association between WC and hpcdd controlling for BMI was found. To the extent that our associations can be speculated to represent causation, on average, one of the toxic effects of these chemicals appears to be weight gain. Unlike the well-known weight loss resulting from high exposure to POPs, this weight gain may occur at much lower levels of exposure, levels which fail to make animals or humans obviously ill [

Because of their previous extensive usage as pesticides, their inherent structural stability, their persistence in body systems and their ability to concentrate in animals that are higher up on the food chain, many POPs are currently present in human fat in relatively high levels. Much of any chemical-induced weight gain may come from increases in the overall proportion of body fat. In one animal study, the pesticide dieldrin more than doubled the total body-fat content of treated mice [

Persistent organic pollutants, synthetic and industrial chemicals, appear to cause weight gain by interfering with most of the different elements that comprise the human weight control system. In particular, these chemicals have been shown to disrupt major weight controlling hormones, such as thyroid hormones, estrogens, testosterone, corticosteroids, insulin, growth hormone, and leptin [

These investigators determined that POPs have a much greater association with these diseases in obese people compared with non-obese people. They hypothesize that the toxicity of POPs related to the risk of metabolic syndrome, insulin resistance, and diabetes substantially increases as people get more obese [

An excellent example of DOHaD that is related to POPs and obesity is recent work done by Retha Newbold and colleagues [

This work was supported in part by T32HL007457, P30DK056336, & R01AR052658 at the University of Alabama at Birmingham and by the Environmental Health Sciences Center in Molecular and Cellular Toxicology with Human Applications Grant P30 ES06639 at Wayne State University, and NIH R01 grants (ES012933 and CA105349) to D.M.R. David B. Allison has received consulting fees, donations, honoraria, royalties, and grants from numerous for-profit and not-for-profit entities with interests in obesity, including companies and litigators interested in effects of environmental toxins on obesity.

Descriptive statistics for detectable persistent organic pollutants.

BMI | Waist Circumference | |||||
---|---|---|---|---|---|---|

n | Mean | 95% CI | n | Mean | 95% CI | |

Male | 1,140 | 27.75 (0.28) | 27.17, 28.33 | 1,133 | 98.53 (0.72) | 97.07, 100.00 |

Female | 1,324 | 27.55 (0.26) | 27.01, 28.09 | 1,315 | 91.82 (0.80) | 90.19, 93.46 |

Mexican American | 758 | 27.77 (0.27) | 27.22, 28.33 | 749 | 92.94 (0.61) | 91.69, 94.19 |

Non-Hispanic Black | 112 | 28.09 (0.83) | 26.39, 29.79 | 112 | 95.19 (2.67) | 89.72, 100.66 |

Non-Hispanic White | 1,022 | 27.47 (0.28) | 26.90, 28.04 | 1,026 | 95.49 (0.80) | 93.85, 97.14 |

Other Hispanic | 485 | 28.91 (0.37) | 28.15, 29.68 | 475 | 94.92 (1.03) | 92.82, 97.02 |

Other Race | 87 | 26.71 (0.68) | 25.32, 28.10 | 86 | 90.34 (1.95) | 86.34, 94.33 |

06–18 yrs | 504 | 23.17 (0.32) | 22.52, 23.82 | 500 | 80.52 (0.81) | 78.86, 82.18 |

19–28 yrs | 397 | 26.71 (0.44) | 25.81, 27.60 | 391 | 90.60 (1.23) | 88.08, 93.12 |

29–39 yrs | 369 | 27.84 (0.49) | 26.84, 28.84 | 367 | 93.64 (1.45) | 90.68, 96.60 |

40+ yrs | 1,194 | 28.42 (0.24) | 27.92, 28.91 | 1,190 | 98.49 (0.70) | 97.07, 99.92 |

Additive regression models for all participants.

BMI | WC | WC|BMI | ||||
---|---|---|---|---|---|---|

Variable | Beta | p-value | Beta | p-value | Beta | p-value |

Intercept | 13.26 (2.83) | 0.0001 | 51.75 (7.43) | <0.00005 | 7.46 (4.15) | 0.0827 |

Mexican American | 0.69 (0.86) | 0.4263 | 1.04 (2.37) | 0.6629 | −0.58 (0.77) | 0.4590 |

Other Hispanic | 1.27 (1.08) | 0.2518 | 3.42 (3.30) | 0.33088 | −0.12 (1.00) | 0.9028 |

White | 0.38 (0.82) | 0.6442 | 2.16 (2.43) | 0.3815 | 1.56 (0.92) | 0.0982 |

African American | 1.89 (0.91) | 0.0466 | 2.93 (2.46) | 0.2431 | −1.28 (0.90) | 0.1688 |

Male | 0.63 (0.39) | 0.1192 | 8.01 (1.05) | <0.00005 | 6.40 (0.37) | <.00005 |

Age | 0.33 (0.04) | <0.00005 | 0.89 (0.10) | <0.00005 | 0.11 (0.04) | 0.0056 |

Age-2 | −0.00 (0.005) | <0.00005 | −0.01 (0.005) | <0.00005 | 0.00 (0.005) | 0.6518 |

BMI | . | . | . | . | 3.33 (0.20) | <0.00005 |

BMI-2 | . | . | . | . | −0.02 (0.005) | <0.00005 |

ln(hpcdd) | 0.44 (0.29) | 0.1370 | 0.24 (0.74) | 0.7476 | −0.68 (0.24) | 0.0091 |

ln(Ocdd) | 0.83 (0.30) | 0.0099 | 2.31 (0.76) | 0.0052 | 0.31 (0.25) | 0.2184 |

ln(oxychlordane) | 0.58 (0.64) | 0.3716 | 1.59 (1.63) | 0.3359 | 0.16 (0.48) | 0.7404 |

ln(trans-nonachlor) | −0.74 (0.60) | 0.2258 | −1.72 (1.52) | 0.2697 | 0.04 (0.33) | 0.9047 |

ln(DDT) | 0.48 (0.22) | 0.0353 | 0.81 (0.52) | 0.1283 | −0.17 (0.22) | 0.4610 |

Joint Test p-value | 0.0001 | 0.0008 | 0.0798 | |||

R-Squared | 0.0793 | 0.1579 | 0.8762 | |||

R-Squared (No POPs) | 0.0591 | 0.1449 | 0.8756 | |||

N | 2,464 | 2,448 | 2,428 |

Gender Interaction Regression Models for All Participants.

BMI | WC | WC|BMI | ||||
---|---|---|---|---|---|---|

Variable | Beta | p-value | Beta | p-value | Beta | p-value |

Intercept | 13.46 (3.19) | 0.0002 | 52.91 (9.37) | <0.00005 | 8.58 (5.19) | 0.1090 |

Mexican American | 0.95 (0.85) | 0.2744 | 1.78 (2.33) | 0.4514 | −0.39 (0.77) | 0.6143 |

Other Hispanic | 1.43 (1.04) | 0.1792 | 3.84 (3.15) | 0.2316 | −0.01 (1.00) | 0.9921 |

European American | 0.54 (0.81) | 0.5068 | 2.60 (2.35) | 0.2787 | 1.67 (0.89) | 0.0711 |

African American | 2.08 (0.90) | 0.0286 | 3.39 (2.39) | 0.1674 | −1.15 (0.88) | 0.2019 |

Male | −0.97 (5.27) | 0.8550 | 2.41 (12.12) | 0.8439 | 3.90 (4.36) | 0.3782 |

Age | 0.31 (0.04) | <0.00005 | 0.86 (0.10) | <0.00005 | 0.10 (0.04) | 0.0084 |

Age-2 | −0.003 (0.005) | <0.00005 | −0.01 (.005) | <0.00005 | 0.00 (0.005) | 0.5574 |

BMI | . | . | . | . | 3.31 (0.21) | <0.00005 |

BMI-2 | . | . | . | . | −0.02 (0.005) | 0.0001 |

ln(hpcdd) | −0.37 (0.46) | 0.4172 | −2.00 (1.25) | 0.1211 | −1.16 (0.39) | 0.0060 |

ln(Ocdd) | 1.23 (0.44) | 0.0091 | 3.38 (1.26) | 0.0120 | 0.50 (0.46) | 0.2839 |

ln(oxychlordane) | −0.21 (0.88) | 0.8135 | −0.51 (2.26) | 0.8235 | −0.29 (0.78) | 0.7081 |

ln(trans-nonachlor) | −0.55 (0.80) | 0.4971 | −1.49 (2.01) | 0.4650 | 0.05 (0.65) | 0.9414 |

ln(DDT) | 1.08 (0.32) | 0.0020 | 2.73 (0.75) | 0.0010 | 0.34 (0.33) | 0.3147 |

male* ln(hpcdd) | 1.77 (0.55) | 0.0033 | 4.85 (1.63) | 0.0059 | 1.03 (0.67) | 0.1384 |

male* ln(Ocdd) | −0.69 (0.67) | 0.3103 | −1.82 (1.75) | 0.3087 | −0.26 (0.69) | 0.7141 |

male* ln(oxychlordane) | 1.68 (0.96) | 0.0896 | 4.59 (2.46) | 0.0718 | 1.03 (0.82) | 0.2229 |

male* ln(trans-nonachlor) | −0.47 (0.80) | 0.5611 | −0.78 (2.05) | 0.7054 | −0.12 (0.87) | 0.8908 |

male* ln(DDT) | −1.24 (0.42) | 0.0060 | −3.97 (0.97) | 0.0003 | −1.04 (0.35) | 0.0056 |

Joint Test p-value | 0.0005 | <.00005 | 0.0050 | |||

R-Squared | 0.0953 | 0.1793 | 0.8775 | |||

N | 2464 | 2448 | 2428 |

Gender Interaction Equations for All Participants.

BMI | WC | WC|BMI | ||||
---|---|---|---|---|---|---|

Variable | Male | Female | Male | Female | Male | Female |

Intercept | 12.493 (4.493) | 13.464 (3.194) | 55.315 (9.411) | 52.907 (9.369) | 12.481 (4.146) | 8.578 (5.188) |

Mexican American | 0.948 (0.851) | 0.948 (0.851) | 17.780 (2.331) | 1.779 (2.331) | −0.390 (0.766) | −0.390 (0.766) |

Other Hispanic | 1.426 (1.036) | 1.426 (1.036) | 3.844 (3.146) | 3.844 (3.146) | −0.010 (1.000) | −0.010 (1.000) |

European American | 0.541 (0.806) | 0.541 (0.806) | 2.596 (2.352) | 2.596 (2.352) | 1.672 (0.892) | 1.672 (0.892) |

African American | 2.081 (0.903) | 2.081 (0.903) | 3.391 (2.395) | 3.391 (2.395) | −1.145 (0.877) | −1.145 (0.877) |

Age | 0.314 (0.041) | 0.314 (0.041) | 0.858 (0.099) | 0.858 (0.099) | 0.101 (0.036) | 0.101 (0.036) |

Age-2 | −0.003 (0.0004) | −0.003 (0.0004) | −0.007 (0.001) | −0.007 (0.001) | 0.0002 (0.0004) | 0.0002 (0.0004) |

BMI | . | . | . | . | 3.311 (0.210) | 3.311 (0.210) |

BMI-2 | . | . | . | . | −0.0165 (0.003) | −0.0165 (0.003) |

ln(hpcdd) | 1.391 (0.297) | −0.375 (0.455) | 2.849 (0.809) | −2.00 (1.252) | −0.132 (0.434) | −1.161 (0.392) |

ln(Ocdd) | 0.540 (0.483) | 1.230 (0.440) | 1.559 (1.091) | 3.77 (1.259) | 0.243 (0.391) | 0.499 (0.457) |

ln(oxychlordane) | 1.471 (0.684) | −0.209 (0.878) | 4.084 (1.651) | −0.508 (2.258) | 0.732 (0.440) | −0.293 (0.776) |

ln(trans-nonachlor) | −1.02 (0.630) | −0.552 (0.802) | −2.270 (1.635) | −1.487 (2.008) | −0.072 (0.432) | 0.048 (0.647) |

ln(DDT) | −0.163 (0.307) | 1.081 (0.318) | −1.242 (0.686) | 2.728 (0.748) | −0.704 (0.170) | 0.337 (0.329) |

Additive Regression Models for Participants with Detectable Persistent Organic Pollutants.

BMI | WC | WC|BMI | ||||
---|---|---|---|---|---|---|

Variable | Beta | p-value | Beta | p-value | Beta | p-value |

Intercept | 12.54 (3.61) | 0.0016 | 53.02 (8.35) | <0.00005 | 11.45 (6.11) | 0.0711 |

Mexican American | 0.21 (0.96) | 0.8309 | −0.62 (2.72) | 0.8196 | −0.99 (0.93) | 0.2926 |

Other Hispanic | −0.13 (1.25) | 0.9146 | −1.63 (3.38) | 0.6329 | −1.11 (1.11) | 0.3234 |

European American | 0.45 (1.00) | 0.6523 | 1.11 (2.84) | 0.6990 | 0.56 (1.01) | 0.5870 |

African American | 2.43 (1.16) | 0.0450 | 3.26 (3.11) | 0.3039 | −1.99 (0.92) | 0.0388 |

Male | 0.35 (0.36) | 0.3361 | 6.96 (1.06) | <0.00005 | 5.85 (0.45) | <0.00005 |

Age | 0.44 (0.05) | <0.00005 | 1.15 (0.12) | <0.00005 | 0.12 (0.04) | 0.0045 |

Age-2 | −0.00 (0.005) | <0.00005 | −0.01 (0.005) | <0.00005 | 0.00 (0.005) | 0.8474 |

BMI | . | . | . | . | 3.25 (0.36) | <0.00005 |

BMI-2 | . | . | . | . | −0.02 (0.01) | 0.0167 |

ln(hpcdd) | 0.62 (0.43) | 0.1554 | 0.51 (1.17) | 0.6677 | −0.77 (0.38) | 0.0527 |

ln(Ocdd) | 0.62 (0.40) | 0.1344 | 1.69 (0.94) | 0.0843 | 0.13 (0.30) | 0.6732 |

ln(oxychlordane) | 1.56 (0.91) | 0.0979 | 3.91 (2.41) | 0.1147 | 0.08 (1.06) | 0.9401 |

ln(trans-nonachlor) | −1.76 (0.86) | 0.0483 | −3.94 (2.28) | 0.0940 | 0.26 (0.68) | 0.7114 |

ln(DDT) | 0.55 (0.25) | 0.0340 | 0.98 (0.53) | 0.0736 | −0.14 (0.24) | 0.5575 |

Joint Test p-value | 0.0017 | 0.0129 | 0.2207 | |||

R-Squared | 0.1310 | 0.2030 | 0.8736 | |||

R-Squared (No POPs) | 0.1026 | 0.1881 | 0.8727 | |||

N | 1,727 | 1,709 | 1,696 |

Gender Interaction Regression Models for Detectable Persistent Organic Pollutants.

BMI | WC | WC|BMI | ||||
---|---|---|---|---|---|---|

Variable | Beta | p-value | Beta | p-value | Beta | p-value |

Intercept | 6.53 (4.57) | 0.1637 | 39.17 (10.98) | 0.0013 | 10.60 (7.06) | 0.1440 |

Mexican American | 0.21 (0.94) | 0.8204 | −0.54 (2.60) | 0.8371 | −0.94 (0.94) | 0.3258 |

Other Hispanic | −0.23 (1.24) | 0.8539 | −1.85 (3.30) | 0.5788 | −0.10 (1.12) | 0.3329 |

European American | 0.38 (0.95) | 0.6932 | 0.91 (2.68) | 0.7380 | 0.49 (1.01) | 0.6314 |

African American | 2.28 (1.14) | 0.0550 | 2.81 (2.99) | 0.3553 | −2.06 (0.90) | 0.0299 |

Male | 12.35 (5.50) | 0.3250 | 34.31 (12.66) | 0.0112 | 8.37 (5.68) | 0.1513 |

Age | 0.42 (0.05) | <0.00005 | 1.10 (0.12) | <0.00005 | 0.11 (0.04) | 0.0105 |

Age-2 | −0.00 (0.005) | <0.00005 | −0.01 (0.005) | <0.00005 | 0.00 (0.005) | 0.6216 |

BMI | . | . | . | . | 3.21 (0.37) | <0.00005 |

BMI-2 | . | . | . | . | −0.02 (0.01) | 0.0211 |

ln(hpcdd) | 0.18 (0.65) | 0.7877 | −1.18 (1.87) | 0.5323 | −1.49 (0.61) | 0.0214 |

ln(Ocdd) | 1.42 (0.61) | 0.0277 | 3.84 (1.58) | 0.0213 | 0.58 (0.48) | 0.2329 |

ln(oxychlordane) | −0.24 (1.37) | 0.8638 | −1.04 (3.64) | 0.7770 | −1.26 (1.43) | 0.3847 |

ln(trans-nonachlor) | −1.21 (1.18) | 0.3148 | −2.47 (3.21) | 0.4480 | 0.88 (1.18) | 0.4614 |

ln(DDT) | 0.98 (0.44) | 0.0340 | 2.38 (0.95) | 0.0184 | 0.17 (0.37) | 0.6565 |

male* ln(hpcdd) | 0.96 (0.76) | 0.2172 | 3.83 (2.34) | 0.1129 | 1.80 (0.78) | 0.0283 |

male* ln(Ocdd) | −1.55 (0.80) | 0.0637 | −4.18 (2.21) | 0.0681 | −0.89 (0.85) | 0.3067 |

male* ln(oxychlordane) | 3.74 (1.52) | 0.0204 | 10.43 (3.89) | 0.0119 | 2.94 (1.23) | 0.0231 |

male* ln(trans-nonachlor) | −1.34 (1.06) | 0.3316 | −3.68 (3.51) | 0.3037 | −1.55 (1.30) | 0.2423 |

male* ln(DDT) | −0.83 (0.56) | 0.1503 | −2.65 (1.24) | 0.0409 | −0.53 (0.43) | 0.2285 |

Joint Test p-value | 0.0121 | 0.0011 | 0.0012 | |||

R-Squared | 0.1480 | 0.2259 | 0.8757 | |||

N | 1727 | 1709 | 1696 |

Gender interaction equations for detectable persistent organic pollutants.

BMI | WC | WC|BMI | ||||
---|---|---|---|---|---|---|

Variable | Male | Female | Male | Female | Male | Female |

Intercept | 18.877 (4.107) | 6.530 (4.569) | 73.476 (8.789) | 39.167 (10.985) | 18.966 (6.742) | 10.596 (7.056) |

Mexican American | 0.214 (0.936) | 0.214 (0.936) | −0.540 (2.602) | −0.540 (2.602) | −0.941 (0.942) | −0.941 (0.942) |

Other Hispanic | −0.230 (1.237) | −0.230 (1.237) | −1.850 (3.295) | −1.850 (3.295) | −1.105 (1.122) | −1.105 (1.122) |

European American | 0.379 (0.951) | 0.379 (0.951) | 0.905 (2.680) | 0.905 (2.680) | 0.488 (1.006) | 0.486 (1.006) |

African American | 2.282 (1.141) | 2.282 (1.141) | 2.810 (2.991) | 2.810 (2.991) | −2.061 (0.902) | −2.061 (0.902) |

Age | 0.423 (0.050) | 0.423 (0.050) | 1.095 (0.118) | 1.095 (0.118) | 0.002 (0.041) | 0.112 (0.041) |

Age-2 | −0.004 (0.0005) | −0.004 (0.0005) | −0.009 (0.001) | −0.009 (0.001) | 0.0002 (0.0005) | 0.0002 (0.0005) |

BMI | . | . | . | . | 3.205 (0.369) | 3.205 (0.369) |

BMI-2 | . | . | . | . | −0.0152 (0.006) | −0.015 (0.006) |

ln(hpcdd) | 1.134 (0.423) | 0.176 (0.647) | 2.654 (1.207) | −1.179 (1.865) | 0.305 (0.423) | −1.493 (0.614) |

ln(Ocdd) | −0.129 (0.481) | 1.418 (0.611) | −0.342 (1.164) | 3.839 (1.577) | −0.302 (0.588) | 0.585 (0.480) |

ln(oxychlordane) | 3.504 (0.994) | −0.236 (1.367) | 9.392 (2.443) | −1.042 (3.645) | 1.683 (0.937) | −1.258 (1.425) |

ln(trans-nonachlor) | −2.550 (0.991) | −1.212 (1.185) | −6.152 (2.439) | −2.473 (3.215) | −0.665 (0.581) | 0.880 (1.179) |

ln(DDT) | 0.154 (0.324) | 0.982 (0.441) | −0.278 (0.699) | 2.377 (0.951) | −0.366 (0.215) | 0.165 (0.368) |