|Home | About | Journals | Submit | Contact Us | Français|
Recent data suggest that prenatal exposure to p,p’-DDE may reduce height and increase body mass index (BMI) in childhood, thus potentially raising the risk of adult health problems. The association between prenatal DDE exposure and growth was evaluated in 788 boys from Chiapas, an area of Mexico where DDT was recently used. The median DDE levels in maternal serum at birth (2002-2003) were 2.7 μg/g lipid. 2,633 measurements of height (cm) and weight (kg) were obtained in 2004-2005. The median age of the children during follow-up was 18 months (quartiles 14 and 22 months). Height and body mass index (kg/m2) were age-standardized and expressed as standard deviation scores (SDS). Multivariate random-effect models for longitudinal data were fitted and predicted height and BMI SDS were estimated from the adjusted models. Overall, associations between prenatal DDE level and height or BMI SDS at any given age were not observed. For example, the predicted values showed that children with the highest exposure (DDE: >9.00 μg/g) compared to those least exposed (DDE: < 3.01 μg/g) grew similarly and they had a BMI SDS similar to the referent group. The results do not support the prior findings of an association of DDE exposure with childhood height or BMI.
DDT (1,1,1-trichloro-2,2-bis(4-chlorophenyl)ethane) is an organochlorine pesticide used worldwide in the past for agriculture, and use continues in some countries for vector-borne disease control (UNEP, 2009). The cost-benefit of DDT use for public health purposes in developing countries is under debate, and adverse effects of high exposure are not definite (Roberts et al., 2000; Rogan and Chen, 2005; Sadasivaiah et al., 2007; Schapira, 2006). Thus, epidemiologic studies of populations with high exposure may be helpful to those making decisions about vector-borne disease control policy.
DDE (1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene), the major and most persistent degradation product of DDT, is generally found at higher concentrations than DDT in human tissues (Turusov et al., 2002). Previous reports suggest that higher levels of prenatal DDE exposure may be associated with altered postnatal growth. Studies evaluating such an association have reported decreased height at ages 1, 4 and 7 years (Ribas-Fito et al., 2006), and increased body mass index (BMI) at ages 1 and 3 years (Verhulst et al., 2009), and in adult women (Karmaus et al., 2009). Other studies however, have not confirmed associations between prenatal DDE exposure and reduced height or increased BMI during infancy, childhood, or adolescence (Gladen et al., 2004; Gladen et al., 2000; Jusko et al., 2006; Rogan et al., 1987). One study reported an association between postnatal exposure to DDE and decreased height in girls at age 8 years (Karmaus et al., 2002). Lower postnatal growth has been related to higher infant mortality in developing countries (Pelletier et al., 1993) and with increased morbidity and mortality in adults (Adair, 2008; Fisher et al., 2006; Gluckman et al., 2005; Hardy et al., 2004; Skidmore et al., 2007; Wells, 2007); increased BMI with its attendant risks also tracks from childhood to adulthood (Martin et al., 2006; Viner et al., 2008).
The objective of the present study was to evaluate the association of prenatal exposure to DDE and longitudinal measurements of height and body mass index at early ages in boys from Tapachula, Chiapas, an area of Mexico where DDT was recently used. Although this country stopped using DDT for agriculture in 1991 and for malaria control in 1998, the population in the Tapachula area experienced unusually high exposures for almost 40 years.
Newly-delivered male infants and their mothers (n=872) were recruited during 2002 – 2003 at both of the city hospitals in Tapachula, which also serve the surrounding areas. The participation rate was 95 percent (Longnecker et al., 2007; Romano-Riquer et al., 2007). Due to an initial study hypothesis regarding effects on outcomes mediated by androgen-blocking, only boys were enrolled. Between January 20th, 2004 and June 29th, 2005 we conducted a relatively short follow-up study to evaluate the association of maternal DDE exposure with duration of lactation (Cupul-Uicab et al., 2008) and other outcomes; during follow-up anthropometric measurements of the children were obtained.
According to the eligibility criteria for the original study, we excluded women younger than 15 or older than 35 years, and those with certain medical conditions described previously (Cupul-Uicab et al., 2008; Longnecker et al., 2007). In addition, we excluded children born when less than 36 weeks of gestational age, or under 2,500 g of birthweight (Gluckman et al., 2005; Longnecker et al., 2007; Romano-Riquer et al., 2007). One child who was given up for adoption after being recruited was excluded from the follow-up (Cupul-Uicab et al., 2008). The study was approved by the Institutional Review Boards at the Instituto Nacional de Salud Pública in México and the National Institute of Environmental Health Sciences in the United States. All mothers gave written informed consent.
For the present analysis, one child was excluded because his growth had been greatly affected by severe malnutrition as reported by the mother and confirmed by anthropometric measurements. Of the eligible participants (n=870), 90.7 percent (n=789) had at least one follow-up visit and 9.3 percent (n=81) were lost to follow-up. The main reasons for the loss to follow-up were: recorded address did not exist (n=10), the address was found but the mother was not (n=59), refused participation (n=4), or the mother or child had died (n=8). The mean age of the children when the follow-up began was 13.0 months (range, 3.3 to 25.1 months) and at the end of the follow-up was 30.5 months (range, 20.8 to 42.7 months).
A total of 2,667 observations were available for the 789 subjects with at least one follow-up visit. The median age during the follow-up period was 18.1 months (quartiles 14 and 22 months). The number of follow-up visits varied for logistic reasons and ranged from 1 to 7 (median 3); 5 percent (n=39) had only one visit and 11 percent (n=86) had 6 or more visits. The gap between visits ranged from 0.8 to 12.9 months (median 2.5). For those with only one visit the median interval between the beginning of the follow-up and their visit was 11 months (quartiles 2.8 and 15.5); for those with two or more visits, the median interval between the beginning and first visit was 3.1 months (quartiles 0.7 and 5.2).
Recumbent length was measured to the nearest millimeter using portable fiberglass measuring boards for children up to 1.3 meters (locally designed). Hereafter we will refer to length as height. Weight was measured using portable electronic pediatric scales (Tanita© Models 1583 or BD-585, Tokyo, Japan) with readable increments of 10 g for children weighing up to 10 kg and 20 g for those weighing > 10 kg. The scales were calibrated before each use. All measurements were performed using standardized methods (Habicht, 1974) during home visits by trained personnel, who also received periodic retraining throughout the study.
To evaluate the quality of the data and the measurement error (Malina et al., 1974; Ulijaszek and Kerr, 1999), measurements of height (n=232) and weight (n=233) were obtained by a second observer. The coefficient of reliability was 0.998 for height and 0.997 for weight.
Body mass index (kg/m2) was used as an indicator of excess weight relative to height to facilitate comparisons of the present study with previous results. Standard deviation scores (SDS) for height, weight and BMI were calculated using internal means and standard deviations (SD) for each month of age by subtracting the mean from a given measurement and dividing by the SD of the corresponding age group. Height and weight were also transformed into age-specific z-scores (length-for-age, 0 to 36.5 months; weight-for-age, 0 to 20 years) using the 2000 CDC growth reference for boys (Kuczmarski et al., 2000) to facilitate comparisons. Weight was transformed into height-specific SDS and z-scores (weight-for-length, 45 to 103.5 cm) for use in sensitivity analyses (Kuczmarski et al., 2000). Biologically implausible values for height (n=31) and weight (n=2) were excluded, leaving a total of 788 subjects with 2,634 observations for the present analysis. After these exclusions, the number of subjects with only one available measurement for anthropometry increased from 39 to 43.
DDE and DDT were measured in maternal serum obtained within a day of delivery. DDE and DDT were quantified after solid phase extraction, using gas chromatography with mass spectrometry detection (Saady and Poklis, 1990; Smith, 1991). The limit of detection for both analytes was 0.2 μg/L. The between-assay coefficients of variation were 7 percent for DDE (at 10 μg/L) and 6 percent for DDT (at 2.5 μg/L). The recovery was 97 percent for both analytes. Total serum lipid content was estimated based on levels of free and total serum cholesterol, triglycerides and phospholipids (Patterson et al., 1991), measured using standard enzymatic methods. DDE and DDT concentrations were expressed as micrograms per gram of lipid (μg/g). For the analysis, DDE categories (μg/g) were defined as: ≤3.00, 3.01-6.00, 6.01-9.00, and >9.00, which were chosen before examining the DDE-outcome relationships. As described previously these categories resulted in a large exposure difference between those in the high and low categories while maintaining adequate numbers for analysis (Cupul-Uicab et al., 2008; Longnecker et al., 2007). For DDT (μg/g), we used the following categories: ≤0.25, 0.26-0.75, 0.76-1.99, and ≥2.00 (Cupul-Uicab et al., 2008).
Height and BMI as SDS were the main outcomes for the present analysis. SDS were used to facilitate the comparison of effects at different ages (Cole, 2007). The association of prenatal DDE exposure with height and BMI were each evaluated using a multivariate multilevel model for change (linear mixed model), where the longitudinal measurements were nested within children (Singer and Willett, 2003). Subject-specific random intercepts and random slopes for age (months) were included in the models assuming an unstructured covariance matrix for these random effects (Rabe-Hesketh and Skrondal, 2008; Singer and Willett, 2003). Age of the children (months) was also included as a covariate in all the models to account for different timing of measurements for each child. To give a meaningful interpretation of the intercept, age was centered to the median at first measurement (15.7 months). The effect on height or BMI SDS of each variable included as predictor in the models was estimated with two terms, a main effect term (initial status at 15.7 months) and an interaction term with child’s age (monthly rate of change) (Rabe-Hesketh and Skrondal, 2008; Singer and Willett, 2003). For instance, when interpreting the effect of the highest exposure group (DDE >9.00 μg/g) compared to the least exposed (DDE ≤3.00 μg/g), the main effect term corresponds to the population average difference in SDS (of height or BMI) between highest versus lowest exposure groups at the median age of the children at first follow-up and the corresponding interaction term is the average monthly change in this difference. Predicted values for height and BMI SDS were also estimated from the adjusted models; differences across categories of DDE exposure at different ages of the children during follow-up (other than 15.7 months) were also tested using the post-estimation tools for mixed models available in Stata (linear combination of estimators [command lincom]). See the online supplementary data for detailed information about the fitted models. All analyses were done using Stata 10.1 (Stata Corporation, 2009).
Predictors of childhood growth considered a priori as variables to include in the models besides DDE exposure and child’s age at measurement, were mother’s smoking status during pregnancy, hospital of recruitment, rural/urban residence, gestational age at birth (median centered), and maternal education. Maternal height (median centered) was included a priori only for height models and pre-pregnancy BMI only for BMI models. Maternal weight at the first follow-up visit was used to calculate BMI if the pre-pregnancy weight was missing (n=128, 16.2 percent). Other maternal, paternal and child characteristics listed in Table 1 (and table 1a of supplementary data) were considered as potential confounders.
The SDS of height and BMI were modeled separately. After adjustment for the variables selected a priori, the P values for the potential confounders were evaluated using models that included one potential confounder at a time; those with a P ≤ 0.20 were given further consideration. For both models (height and BMI SDS) these variables were maternal age at childbirth and work status during follow-up; paternal smoking status during pregnancy and body habitus at childbirth (Osuna-Ramirez et al., 2006; Sorensen et al., 1983); child’s number of older siblings, duration of any breastfeeding, age at introduction of formula or cow’s milk and age at introduction of grains and cereal. Additional variables that met the criteria (P ≤ 0.20) in the height model only were maternal alcohol consumption during pregnancy, poverty index during follow-up and pre-pregnancy BMI; paternal age at childbirth, education and smoking status during follow-up; child’s birth season and ever exclusively breastfed. For the BMI model the additional variables with P ≤ 0.20 were maternal poverty index during pregnancy; father’s presence at home during pregnancy and follow-up; having a younger sibling during the study period. Although parity may be a predictor of growth (Prentice et al., 1987) and met the criteria, it was not included as a potential confounder in the full models because it was highly correlated with number of older siblings (r = 0.95), which showed greater effects in the individual models and has been suggested as an important predictor of early growth (Lawson and Mace, 2008; Lei et al., 1999).
All potential confounders (P ≤ 0.20) were introduced in full models (for height and BMI), and removed in a stepwise manner deleting variables one by one using the change in estimate method for confounding assessment (Greenland, 1989). Initially the interest was to evaluate the effect of high prenatal DDE exposure, therefore to facilitate the selection process we considered only the change in the two coefficients (main effect and its corresponding interaction term with child’s age) that corresponded to the highest exposure category (DDE >9.00 μg/g). The change in estimate criteria used are described in the online supplement. None of the potential confounders met the criteria for either outcome; results from models that included all potential confounders (P ≤ 0.20) gave results that were nearly identical to those presented. Using DDE as a continuous variable for the selection instead of the highest exposure category terms gave the same results.
Thus, the final models of height and BMI (SDS) included only the variables selected a priori. A similar analysis with categories of DDT as the main exposure variable instead of DDE was done using the same explanatory variables as the final DDE models. Effect modification by duration of lactation and maternal smoking (P < 0.20) was evaluated with likelihood-ratio tests. We defined maternal smoking as having ever smoked >100 cigarettes, similar to the analysis by Verhulst et al. (2009). The power to test interactions between DDE and smoking during pregnancy in the present study however, was limited.
Because anthropometric measurements at birth were expected to be correlated with the outcomes, we conducted separate sensitivity analyses with these variables: the final model for height (SDS) was adjusted for child’s birth length (cm) and the final model for BMI (SDS) was adjusted for weight or BMI (kg/m2) at birth. In addition, height and BMI in the original units, age-specific z-scores for height and weight, and height-specific SDS (and z-scores) for weight were modeled separately with the variables selected a priori as predictors (see online supplement for further details).
Other sensitivity analyses were conducted by refitting the final models of height and BMI after: a) excluding subjects with only one follow-up visit, b) stratifying by the number of visits (< 3 and ≥ 3), c) restricting the analysis to the ages where most of the anthropometric measurements were obtained (at ≤ 30 or ≤ 25 months), d) using deciles of DDE exposure (percentiles: 10th, 0.75 μg/g and 90th, 11.27 μg/g), and e) dividing the lowest category of DDE into finer categories (≤ 1.00, 1.01 - ≤ 2.00, 2.01 - ≤ 3.00 μg/g). Finally, because BMI is not a widely used indicator of excess weight in children younger than 24 months (WHO, 1995), an analysis stratified by child’s age (at < 24 and ≥ 24 months) was conducted for the BMI SDS model. Similar stratification was done with weight for height SDS as an outcome.
Compared with those who were followed, the mothers lost to follow-up were younger, less educated, shorter, and poor; more were recruited at the Ministry of Health hospital (Table 1). A higher proportion of mothers lost to follow-up reported having no prenatal care, absence of the child’s father, and fathers who smoked. Finally, their median levels of DDE and DDT were lower compared with the followed mothers. Comparisons between those with one and two or more follow-up visits are presented in Table 1. Those with only one visit had lower median levels of DDE and DDT.
Overall, the mothers in this study had little education, were young, urban-dwelling and poor or very poor, and most had previously had a child; few reported smoking (n=9) or drinking alcohol (n=23) during pregnancy (Table 1). The fathers of the children were older than the mothers and a higher proportion of them had more than 12 years of school (Table 1a of supplementary data). The percentage of breastfed children was high, the duration of any breastfeeding was long, and the duration of exclusive breastfeeding was short (Table 1).
Because of the exclusion criteria for the original study, the children were delivered at term with birth weights in the normal range (Tables (Tables11 and and2).2). Anthropometry at birth compared with the 2000 CDC growth reference (Kuczmarski et al., 2000) showed that length for age was on average −0.15 SD (quartiles −0.56 and 0.38) below the mean for US boys at birth, and weight for age was on average −0.38 SD (quartiles −0.88 and 0.04) below the mean for US boys at birth (data not shown).
Across maternal serum levels of DDE, anthropometric measurements of the children at birth were similar as were the total number of follow-up visits and the interval between them. However, an increase in the median of children’s age at first and last follow-up visits with levels of DDE was observed, although the difference in ages were on average less than three months (Table 2).
Higher levels of prenatal DDE exposure showed no crude associations with height or BMI SDS (Table 2). With height in its original units (cm), no pattern was observed across categories of DDE at any age (Figure 1). Although after 24 months of age the figure suggests a small effect with higher DDE, the numbers were sparse at those ages. Similarly, no clear pattern of any effect due to DDE exposure on BMI (kg/m2) was observed; at 5 months of age children in the higher exposure categories of DDE had lower BMI than those in low exposure categories (Figure 2) but again the data were sparse in this age range.
Unadjusted and adjusted results from the multivariate models of height SDS generally showed no statistically significant associations with higher levels of DDE, except for the interaction term in the second category of exposure (Table 3). The adjusted coefficients for DDE from the model of height can be interpreted as follows. At 15.7 months of age (median age at first follow-up) the main effect term indicates that, compared to the least exposed group (DDE: < 3.01 μg/g), children in the second exposure category (DDE: 3.01-6.00 μg/g) were shorter by 0.121 SDS; the interaction term however, indicates that their rate of growth was slightly increased by 0.019 SDS a month. Thus, the observed difference in height SDS was reduced until age ~22 months when these children had similar height SDS as those from the referent group (Figure 3). At 15.7 months of age, children in the third exposure category (DDE: 6.01-9.00 μg/g) compared to the referent group were shorter by 0.03 SDS and their rate of growth was decreased by −0.005 a month; these children continued to grow at a slower rate and the observed difference in height SDS increased with time (Figure 3). Unexpectedly, at 15.7 months of age, children with the highest exposure (DDE: >9.00 μg/g) compared to the least exposed were taller by 0.069 SDS (equivalent to +0.21 cm) and their rate of growth was slightly increased by 0.01 SDS a month (equivalent to +0.03 cm). Thus, the difference in height SDS increased with time and they grew taller than the referent group (Figure 3). The observed differences in height before and after age 15.7 months were not significant when comparing the least exposed group with the third (DDE: 6.01-9.00 μg/g) and the highest exposure categories (DDE: >9.00 μg/g); however for the second category (DDE: 3.01-6.00 μg/g) the differences were significant up to 10 months of age and at 38 months (Figure 3, table 2b of supplementary data).
After adjustment, predictors of height SDS at 15.7 months of age were child’s age at measurement (taller), hospital of recruitment (shorter at Ministry of Health), and maternal education (taller) and height (taller). Rural residence was the only predictor (decreased growth) of the monthly rate of growth in the height SDS model (Table 3).
For BMI SDS (Table 4), no statistically significant associations with DDE emerged in the unadjusted and adjusted models. The adjusted coefficients for DDE from the model of BMI are interpreted similarly as those for height. At 15.7 months of age the main effect term indicates that, compared with children in the lowest exposure category, those with the highest exposure had a BMI higher by 0.073 SDS (equivalent to +0.10 kg/m2); the interaction term indicates that they had a slower rate of weight gain by −0.002 a month (equivalent to −0.003 kg/m2). Therefore, the difference in BMI SDS decreased with time and they had similar BMI as those from the referent group by age 38 months (Figure 4). The observed differences in BMI before and after age 15.7 months were not significant when comparing the least exposed group with any of the higher exposure categories (Figure 4, table 3b of supplementary data).
Predictors for BMI SDS at 15.7 months of age after adjustment were: having a mother who smoked during pregnancy (higher BMI), who had no education (lower BMI), and obesity before pregnancy (higher BMI). Years of maternal education was the only predictor (higher BMI) of the average rate of growth in the BMI SDS model.
Evidence of effect modification by duration of breastfeeding or maternal smoking (having ever smoked >100 cigarettes) was not observed in models of either height or BMI SDS. Using DDT as the exposure variable in multivariate models did not show any statistically significant associations with height or BMI SDS (data not shown).
Modeling height and BMI in their original units confirmed the results observed in Tables Tables33 and and4.4. When the age-specific z-scores of height and weight, and the height-specific SDS (and z-scores) of weight were modeled as outcomes, no associations with DDE were seen (data not shown). Stratifying the BMI and weight for height (SDS) models by child’s age (at < 24 and ≥ 24 months) showed similar results; no associations with DDE were observed in either group (data not shown). The results shown in Tables Tables33 and and44 were not materially different after a) adjustment for anthropometry at birth, b) excluding those with only one follow-up visit, c) stratifying by number of follow-up visits (<3 and ≥ 3), d) restricting the analysis by age (at ≤ 30 or ≤ 25 months), e) using deciles of DDE exposure, or f) dividing the lowest category of DDE into finer categories. Removing gestational age from the final models showed essentially the same results as in Tables Tables33 and and44 (data not shown).
Higher prenatal exposure to DDE was not associated with a decrease in growth or an increase in BMI in the first years of life in the male children studied. Although children in the second category (DDE: 3.01-6.00 μg/g) were slightly shorter than the referent group up to age 10 months, they recovered and grew taller by age 38 months. DDT was not associated with any of the outcomes studied.
Previous studies on height have given inconsistent results as shown in Table 5. Prenatal DDE exposure was higher in the two previous studies with null results (Gladen et al., 2004; Jusko et al., 2006) than in those with statistically significant associations (Gladen et al., 2000; Ribas-Fito et al., 2006). Similarly, earlier studies on weight or BMI have shown inconsistent results with DDE exposure (Table 5). Again, associations were seen only in studies with lower (Gladen et al., 2000; Karmaus et al., 2009; Verhulst et al., 2009) but not with higher prenatal exposure to DDE (Gladen et al., 2004). While the reasons for the inconsistent results are not clear, it is possible that in studies finding statistically significant effects associated with DDE, levels are acting as markers of some characteristic of the subject other than DDE (Wolff et al., 2007), which is related to height or BMI.
The earlier studies were conducted when levels of polychlorinated biphenyls (PCB) were higher; however their findings were similar after taking PCB exposure into account (Gladen et al., 2004; Gladen et al., 2000; Ribas-Fito et al., 2006). In Mexican children PCBs are not a common exposure (Trejo-Acevedo et al., 2009), therefore confounding by these compounds is unlikely in the present study. Information on pesticides is limited, but recent data showed that children from Chiapas have detectable levels of hexachlorobenzene (Trejo-Acevedo et al., 2009), which has been associated with increased childhood BMI in a previous study (Smink et al., 2008). Residual confounding by this or other pesticide exposures cannot be ruled out.
A compelling mechanism to explain an adverse effect of prenatal DDE exposure on postnatal height or weight has not been reported. Conceivably alterations of normal endocrine function due to DDE’s antiandrogenicity (Kelce et al., 1995) could be involved. Males with less androgen are more likely to have higher BMI (Braga-Basaria et al., 2006). Prenatal DDE exposure has been related to thyroid hormone levels during infancy and childhood (Alvarez-Pedrerol et al., 2008), which are relevant for regulation of growth (Reiter and Rosefeld, 2003).
A limitation of the results is that they were based on only males, because this population was derived from a study limited to males. Previous studies suggested that effects of prenatal exposure to DDE on height or weight are sex-specific (Gladen et al., 2000; Ribas-Fito et al., 2006). In the Collaborative Perinatal Project, the decrease in linear growth with high exposure to DDE at age 7 years was greater (−3.32 cm) in African-American girls than in boys (−0.71 cm) (Ribas-Fito et al., 2006). While in the NC Study, significant associations were seen in boys only (Gladen et al., 2000). However other studies did not report sex-specific associations of prenatal exposure (Jusko et al., 2006; Verhulst et al., 2009). If DDE can effect height or BMI at early ages differentially by gender, exclusion of female children may have limited our ability to detect any association.
If prenatal DDE exposure affects postnatal growth via birth weight or preterm birth, excluding these children from our study may have also limited the power to detect associations. However, in the Collaborative Perinatal Project where no limitation on gestational age or birth weight was imposed, prenatal DDE exposure was negatively associated with height even after adjusting for anthropometry at birth (Ribas-Fito et al., 2006). Furthermore, in that study the finding of reduced height among those most exposed to DDE was essentially the same after exclusion of those who were < 36 weeks of gestation or < 2,500 grams at birth ((Ribas-Fito et al., 2006); unpublished results). In addition, prenatal exposure to DDE was positively associated with BMI in a population of children born at term, and the associations were still observed after adjusting for birth weight (Verhulst et al., 2009). These observations from other studies suggest it was unlikely that exclusions in the present study diminished our ability to detect an effect.
Measurement error for the outcome may have little or no effect on the results given the high coefficients of reliability obtained from the repeated measurements (Ulijaszek and Kerr, 1999). DDE levels were determined from serum drawn at delivery, using methods with a low between-assay coefficient of variation. Additionally, a high correlation (r = 0.85) has been reported between DDE levels from maternal serum at delivery and cord serum (Waliszewski et al., 2000). A previous study has shown that DDE levels correlate well during pregnancy and shortly after birth (Longnecker et al., 1999). The present study was double blind, since neither interviewers nor participants knew the DDT or DDE levels.
In the present study other known determinants of height and BMI were confirmed, which supports the validity of the data. Maternal height was a positive predictor of child’s height SDS; and rural residence and low education level of the mother were negative predictors (Hernandez-Diaz et al., 1999; Hindmarsh et al., 2008; Reyes et al., 2004; Reyes et al., 2003; Rivera-Dommarco et al., 1995). Maternal smoking during pregnancy and mother’s pre-pregnancy BMI ≥ 30 kg/m2 were positive predictors of child’s BMI SDS (Chen et al., 2006; Whitaker, 2004).
In summary, in this population of male children with high prenatal exposure to DDE, we did not see an association with height or BMI in the first years of life. The results were consistent across all analyses. DDT was not associated with any of the outcomes studied.
We wish to thank the field work team in Tapachula, Chiapas, México, for their valuable support during data collection.
Source of funding
This work was supported in part by a contract from the National Institute of Environmental Health Sciences, National Institutes of Health [N01-ES-15467], in part by the Intramural Research Program of the National Institutes of Health, National Institute of Environmental Health Sciences, Research Triangle Park, NC, USA; and in part by the Instituto Nacional de Salud Pública, Cuernavaca, Morelos, México.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of interest: