High intake of added sugar and sodium among US preschool-aged children is a public health concern; recent nationally-representative surveys indicate that US preschool-aged children on average obtain between 13 and 16% of total Calories from added sugar [1] and over 90% of children ages 4–8 years (y) old exceed recommended levels of sodium intake [2] This is concerning because high consumption of sodium as well as added sugar in early childhood has been implicated in the development of cardiometabolic risk factors during childhood and adolescence [3, 4]. In addition, there is evidence that free sugar intake >10% of total Calories is positively associated with dental caries among children [5]. Thus, elucidating the predictors of higher added sugar and sodium intake among young children could provide valuable information for intervention.

One potential predictor of these dietary risk factors is externalizing behavior. Externalizing behavior during early childhood is marked by angry, aggressive, selfish and oppositional conduct; it includes fighting with or bullying other children, physically hitting, and taking things from others [6]. Externalizing behaviors have been linked with higher consumption of salty and/or sugary foods in a few cross-sectional studies [7, 8]. For example, a study of 13,486 Iranian children and adolescents found that those with higher self-reported aggressive behaviors also reported higher consumption of salty snacks [7]. Another study among 1324 adolescents in Australia found that those with more externalizing behaviors had higher intake of fast foods, confectionary items, and red meat [8]. There has also been one longitudinal study examining externalizing behavior and frequency of sugary food consumption, although not sodium. In this study of 6997 Norwegian children, researchers found that children with higher externalizing behavior at age 1.5 y had higher intake of sugary foods at age 3 y and 7 y [9]. A longitudinal study design is important as it separates the exposure from the outcome in time, lending stronger support to the notion that behavior leads to consumption patterns. There are some plausible mechanisms to support this pathway. For example, externalizing behavior during early childhood could be a marker of stress [10], and stress has been linked to higher consumption of snacks and sweet foods in young children [11–13]. Also, children with more externalizing behavior may have greater control over dietary choices because caregivers impose fewer restrictions and/or use food to calm them [14, 15].

The aim of this study was to examine the longitudinal association between externalizing behavior and intake of added sugar (% of Calories) and sodium (mg/1000 Calories) among low socioeconomic status (SES) preschoolers [16]. We hypothesized that children with higher levels of externalizing behavior at baseline would have higher percentage of Calories from added sugar intake and sodium intake (mg/1000 Calories) at follow-up. All analyses were stratified by sex, a potential effect modifier when examining associations between behavior and diet [13].

The mean age ± SD of the children at baseline was 4.1 ± 0.5 y; 48% were boys (Table 1).

There was not a statistically significant association between externalizing behavior and mg sodium/1000 Calories among boys (Table 4). However, after adjustment for potential confounding, there was a marginally statistically significant inverse association between externalizing behavior at baseline and sodium intake at follow up. For every 5 points lower externalizing T-score, there was a 22 mg/1000 Calories lower sodium intake at follow-up (95% CI -45 to 1; P value = 0.06). In contrast, there was a positive association between externalizing behavior at baseline and sodium intake at follow-up among girls. In adjusted analysis, every 5 points lower externalizing T-score was related to 24 mg/1000 Calories greater sodium intake at follow-up (95% CI 1 to 46; P value = 0.04). Further adjustment for intervention arm did not alter the estimates in any of the models.

In this group of low-SES Michigan preschoolers, we found that preschool boys with higher levels of externalizing behavior had higher percentage of intake from added sugar but lower mg sodium/1000 Calories (after confounder adjustment). In contrast, girls with higher levels of externalizing behavior in the fall of one school year had higher intake of sodium/1000 Calories but lower intake of added sugar as a percentage of total Calories during the spring of the same school year.

Our study findings are in line with some other studies, although the sex-dependency of the findings is distinct. Our findings among boys are in agreement with those of a longitudinal study among Norwegian preschoolers [9]. This study showed that children with higher externalizing behavior (based on the Child Behavior Checklist) at 1.5 y of age were more likely to consume sweet drinks and sweet foods daily at 7 y of age. However, this study did not report on sodium intake, and it is also unclear whether the association was sex-specific. Other studies that specifically examined externalizing behaviors in relation to diet were cross-sectional. One study among 453 Belgian children between 5 and 12 y of age found a positive association between parent-reported “problem” behaviors (including emotional problems, conduct problems, hyperactivity problems, peer problems, and antisocial behavior) and sweet and fatty food consumption [12]. Another study among Iranian adolescents found that children with higher self-reported violent behaviors also had higher consumption of salty snacks [7]. There was no association between violent behaviors and sweet food consumption, but it could have been masked by not stratifying on sex. Finally, a cross-sectional population-based survey among Nova Scotian 5th graders found that those who had higher levels of self-reported externalizing behaviors had lower diet quality scores [24]. Lower diet quality often is related to high sodium and/or added sugar levels, although the study did not examine particular nutrients.

Our findings are also in agreement with some of the literature on stress and dietary patterns. For example, one study among 4th through 6th graders from rural Texas found that those with higher perceived stress also reported greater frequency of consumption of salty snacks or high sugar foods [25]. An experimental study among 43 US children aged 5–9 y of age found that greater stress-related cortisol was related to greater energy intake in the absence of hunger among the 8–9 year olds, although not in the younger children [26]. Finally, another US study showed that high stress self-efficacy (i.e., ability to handle stress) among fourth-graders was related to higher fruit and vegetable intake [27]. In summation, it is possible that the associations we observed between externalizing behavior and diet could be related to stress and emotional eating. This is plausible, given that children with externalizing behavior may also have higher levels of stress, anxiety and/or depressive symptoms [28, 29]. However, the lack of specificity of foods in prior studies precludes in-depth comparisons. In several of the studies on stress and eating, “comfort foods” including sugary, salty, and energy-dense foods, were assumed to be eaten together and were not further disentangled into sodium and added sugar intake. Prior assumptions about “comfort foods” are problematic because they may not hold in every population. In addition, it is important to separate added sugar from sodium because high intake of sodium and high intake of added sugar may have different health consequences that should be monitored accordingly (e.g. dental caries for added sugar and high blood pressure for sodium). Finally, one of the limitations in most prior studies was the lack of stratification by sex, so associations in the opposite directions could have been hidden.

There are a few potential mechanisms to explain the findings specifically related to externalizing behavior. One is that children with externalizing behaviors exhibit less effortful control and higher impulsivity [30], and higher impulsivity has been related to eating in the absence of hunger [31]. It could be that girls with externalizing behavior eat foods with higher sodium content in the absence of hunger while boys with externalizing behavior eat foods (or consume drinks) with higher added sugar in the absence of hunger. The notion that boys with externalizing behavior may prefer sweet foods is supported by a few adult studies. For example, in one small laboratory study among adults, men who were put through an acute stressor test reported a higher appetite for sweet foods than the control group, but there was a null association for women [32]. In another US study among young men, after watching a violent video game, there was a tendency towards sweet food consumption, but not salty, savory, or fatty food [33]. One possible physiological explanation for the association between aggression and higher sugar consumption is that glucose is required in order for the brain to control aggressive impulses [34]. In support, an experimental study among US undergraduates showed that those who consumed a glucose beverage immediately before being provoked showed reduced aggression compared to those who consumed a placebo drink, and this was most evident among those who scored higher on a pre-experiment aggressive trait survey. Although the authors did not conduct stratified analysis, males had higher aggressive trait scores than females. In contrast, there may be an explanation for sodium preference among females with externalizing behavior. It has been posited that because the renin-angiotensin-aldosterone system is involved in both sodium regulation and the stress response, sodium intake may help to alleviate the short-term effects of psychosocial stress [35]. Furthermore, this association is thought to be most evident among females because their average absolute salt intake is lower than males’. The main caveat to these potential mechanisms is the fact that the research is mostly from adult population and may not be generalizable to preschool-aged children.

Another possible explanation of the study findings is that parent/guardians of externalizing children may try to pacify them with highly-palatable foods [36], which are likely to contain high amounts of added sodium and/or sugar. One reason for the sex difference is that girls could be fed differently than boys due to societal expectations. For example, parents may feel more pressure to shield their daughters from gaining weight than their sons; thus, they may offer foods high in sodium (although not necessarily snack foods) rather than highly sugary foods to daughters due to the perception that sugar causes weight gain. The finding of lower percentage of added sugar intake among girls with higher externalizing behaviors supports this notion. Furthermore, given that the most common sources of sodium consumption were meat dishes, it is possible that girls with externalizing behavior did not have greater salty snack consumption, but rather ate higher proportions of salty entrees at lunch or dinner. Future studies on behavior and diet should consider sex as a potential effect modifier to evaluate the reproducibility of our findings and to further elucidate the potential mechanisms, especially among young children. Understanding mechanisms involved in these associations could inform interventions designed to reduce added sugar and sodium intake among children with externalizing behavior problems. Although parents/guardians would likely be the target of these interventions, teachers may be able to recognize externalizing behaviors that parents/guardians do not perceive; and thus could play a vital role in educating parents/guardians. Future studies investigating the food consumed at school versus home could further elucidate the role that teachers and the school environment could play in establishing healthy eating habits.

Although there were clear differences in added sugar and sodium intake according to externalizing behavior and sex, only about one-third of the children met added sugar recommendations and one-half met sodium guidelines. This is concerning, because the dietary habits and preferences established early in life tend to track into adulthood [37]. Furthermore, both added sugar and sodium intake during childhood are related to adverse health outcomes, even in early life [3, 4]. Thus, although the children with externalizing behavior problems were more likely to consume either added sugar or sodium, the entire population could benefit from reductions in sodium and added sugar in the food supply. This could be achieved by programs such as the National Salt Reduction Initiative, a US program which calls for voluntary reductions in sodium content of packaged foods and restaurants [38]. Policy-level recommendations for reducing added sugar content of products should also be considered. At the parent/guardian level, providing more specific information on specific foods to avoid or limit to reduce consumption of added sugar or sodium could prove useful. This is more pertinent for sodium, as many public health campaigns currently focus on sugar consumption.

This study has strengths and limitations. One strength was the repeated measures of both diet and behavior, which allowed us to examine the directions using longitudinal analysis. The prospective design was also important for reducing potential recall bias, because the behavior of the child at baseline most likely did not affect the perception of diet at follow-up and vice versa. Another was the incorporation of classroom as a clustering variable in our models; this ensures our standard errors were valid. Another strength was that diet was reported by evaluators unaware of the externalizing behavior score of the child. The fact that the evaluators of externalizing behavior were teachers may increase the objectivity of these scores. There were also limitations. We had to rely on multiple 24-h recalls for sodium and added sugar measurements rather than biomarkers (e.g., urinary sodium). Diet measurements are always subject to measurement error, and diet among young children can be especially difficult to measure [39]. These study findings may not be generalizable to other US populations, since all of the families were of low SES. We did not have a measure of the child’s stress in order to evaluate the correlation between child stress and externalizing behavior. Finally, there could be unmeasured confounding that we were not able to account for.

In summary, we found that young children with teacher-reported externalizing behavioral problems had higher proportions of sodium or added sugar in their diet at follow-up in a sex-specific manner. Girls with higher externalizing behavior had higher sodium intake/1000 Calories at follow-up and lower percentage of Calories from added sugar, whereas boys with higher externalizing behavior at baseline had higher percentage of added sugar and lower sodium intake/1000 Calories at follow-up. Further studies are warranted to examine the mechanisms for these associations and to understand the most effective ways to encourage healthy eating habits in boys versus girls with externalizing behaviors. Dietary-based interventions and recommendations directed towards parents/guardians of preschoolers may need to be tailored based on the behavior and sex of the child.