Our results suggest that externalizing behavior problems are associated with higher BMI and obesity in children as young as 24 months old. Among two-year-old children, irrespective of race, based on our analyses we would predict an average difference of three-quarters of a BMI unit between children with high levels of externalizing behavior and children with low levels of externalizing behavior. This average difference in BMI was stable through age 12 among white children, and gradually doubled with age to about 1.5 BMI units among children who were black or of another race. These predicted average differences in BMI for a child with high levels of externalizing behavior compared to a child with low levels of externalizing behavior correspond to differences in weight of less than 2 pounds at 24 months of age, increasing to between 3 and 15 pounds at age 11 depending upon race and height.
It is well-established that weight status tends to track [31, 32]; children who have a high BMI-for-age early in childhood are more likely than children with a low BMI-for-age to later be obese. Our analyses suggest that externalizing behaviors in early childhood are associated with children's weight status early in childhood and throughout the elementary school years, though the magnitude of the effect is modest. We are unaware of previously published analyses of externalizing behavior and weight status in children younger than age four.
A few studies of non-clinical populations have examined the relationship between obesity and externalizing behavior in preadolescent children and findings from these studies have not been consistent [3, 5, 6, 8, 11, 12]. Some of the heterogeneity in results, summarized below, may be due to differences in assessment of externalizing behavior, analytic approach, and the population studied. In the Longitudinal Study of Australian Children, overweight 4- and 5-year-old children (born 1999-2000), had higher teacher-reported conduct problems than did non-overweight children, but differences were small . In an earlier Australian cohort of children born 1981-1984, no association between behavior problems and obesity was observed for boys or girls cross-sectionally at age 5, however at age 14, compared to girls of normal weight status, girls who were overweight had over two times the odds of concurrent high levels of total behavior problems (externalizing behaviors were not examined separately); no association was seen for boys cross-sectionally at age 14 . In a separate analysis of this same cohort, behavior problems at age 5 and age 14 years in both boys and girls were associated with higher BMI and obesity at age 21 years . In the Netherlands, conduct problems were not related to overweight among 5- to 6-year-old children studied during 2004-2005 . In the United States, evidence linking externalizing behavior problems in children to obesity is somewhat more consistent, with associations observed among 8- to 11-year-old children in the 1998 National Longitudinal Survey of Youth , and 5-year-old girls (but not boys) in the Early Childhood Longitudinal Study - Kindergarten class 1998-1999 . In addition, oppositional defiant disorder was reported to be more prevalent among boys and girls in the Great Smoky Mountain Study (1993-2000) who were chronically obese between 9 and 16 years of age .
Using a different analytic approach than we have employed, Bradley et al. have also studied relationships between children's externalizing behavior, internalizing behavior, and BMI in the SECCYD . They used cross-lagged structural equation models to address the directionality of associations between BMI and children's behavior. Our analyses of the same data were designed to address different research questions. We sought to understand the extent to which externalizing behavior early in childhood was associated with BMI early in childhood and change in BMI through age 12, to quantify the magnitude of any association, and to determine the extent to which associations were different in nonwhite children. The structural equation models used by Bradley et al. assume an association between BMI at 24 months, and behavior (externalizing and internalizing) at 24 months, and then controlling for that association and the tracking of BMI and behavior with age, ask whether BMI is associated with subsequent internalizing or externalizing behavior, and whether internalizing behavior or externalizing behavior are associated with subsequent BMI . Although the results of the structural equation models they present do not allow for quantification of the magnitude of any associations, our findings and those of Bradley et al. are consistent. We found that externalizing behavior in early childhood was associated with higher weight status cross-sectionally, and that, among whites, the association between externalizing behavior at 24 months and BMI was maintained through childhood. This is consistent with Bradley et al.'s statement: "controlling for stability,... externalizing behavior was not related to subsequent BMI" .
Our analyses provide evidence of a modest association between externalizing behavior early in childhood and BMI. We have quantified the magnitude of this association using linear mixed effects models, and also present odds ratios for the cross-sectional association at 24 months. We found that externalizing behavior and BMI were associated in two-year-old children. Our analysis does not allow us to identify whether the higher BMI came before, after, or developed concurrently with the higher levels of externalizing behavior. However that they are associated so early in children's lives suggests the need for further research. One might speculate that higher BMI could result from child behavior management techniques employed by parents and caregivers, such as use of food as a reward, to deal with challenges presented by toddlers with higher levels of externalizing behaviors. In an attempt to avoid a child's difficult or disruptive behavior, adults may acquiesce to demands for foods or activities (such as television viewing) associated with energy imbalance. We think the other direction of association - that externalizing behavior results from a child's high BMI - is less plausible. Perhaps more likely than either BMI or externalizing behavior being the cause or consequence of the other, is that a third factor or factors contributes to both a child's weight status/risk for obesity, and the degree to which they display externalizing behaviors. It is biologically plausible that such a factor or set of factors could impact both weight and behavior by influencing young children's brain development . Identification of such modifiable risk or protective factors has potential to simultaneously help children achieve a healthy weight and decrease problematic behavior.
We provide evidence that the association between early childhood externalizing behaviors and BMI was not confounded by the sociodemographic covariates of sex, maternal education, or household poverty status, the association was as strong or stronger in children who were not white, and was not attenuated by adjustment for child internalizing behavior or maternal depression. We were not able to adjust for parental obesity, however, which is strongly related to child obesity and may be related to child behavior problems or maternal reports of child behavior problems. Our data suggest that the association between externalizing behavior and obesity at 24 months may differ by level of maternal education and household poverty status, in addition to race, but the pattern to these findings is not consistent, and further investigation is needed.
An additional limitation of our analyses is that we relied upon mother's reports of their child's behavior. The CBCL is not a diagnostic instrument and our analyses do not take into account the extent to which externalizing behaviors were problematic for the mother or the child. It is possible that ratings of children's behavior are influenced by the child's weight status. Considerable evidence indicates that obese children may be stigmatized and viewed negatively by peers and teachers , however the extent to which mother's reports of their toddler's externalizing behavior is influenced by weight status is not known.