A prospective study of weight development and behavior problems in toddlers: the Norwegian Mother and Child Cohort Study

In recent decades, childhood overweight has been increasing rapidly worldwide. Between 1975 and 2000 in Norway, body weight in childhood increased evenly across the entire weight spectrum of weight by 1 unit of the body mass index (BMI) (kg/m²). In 2000, close to 22% of all newborns in Norway weighed more than 4 kg [1]. Today, around 19% of Norwegian children and adolescents age 4-15 years are above the 90^th weight-for-height percentile of the growth references for the years 1971-1974 [2]. However, for children younger than age 4, exact overweight rates are still lacking. The rapid increase of overweight rates in children is alarming because of the serious consequences of overweight for health [3, 4]. First, obese children are at high risk of obesity in adulthood [3]. Moreover, obesity has adverse effects on many health parameters, such as blood pressure, cholesterol, triglycerides, and insulin resistance. These parameters in turn increase the risk for cardiovascular disorders and type 2 diabetes later in life, among other things [4]. In addition, thickening of the arteries, high blood pressure, fatty streaks in the arteries, and even diabetes type 2, formerly considered exclusively an adult disease, have been observed increasingly during childhood [3, 5].

Increasing evidence suggests that child temperament and behavior problems are associated with overweight [6‐12]. There are several mechanisms that may account for this association. Children with behavior problems in the externalizing domain may have a relatively underdeveloped capacity for self-regulation [13]. This self-regulatory capacity may in turn impact children's health behaviors (e.g., their diet). Poor self-regulation may prevent children from attending to satiety cues and may prevent them from adequately moderating their eating behaviors. Indirectly, poor self-regulation may prevent children from modulating negative emotionality or emotional reactivity, which may contribute to the development of maladaptive behaviors, such as higher consumption of highly palatable foods - foods containing high amounts of sugar, salt, and fat - and overeating. This "comfort eating" may eventually lead to the emergence of child overweight or obesity [12, 14]. In support of this, we found recently that 18-month-old children scoring higher on externalizing problems consumed more sweets and sweet drinks [15]. Especially with infants and toddlers, parents may overfeed difficult children in a misguided effort to calm them or to reduce their emotional intensity [9, 16]. Also, a recent study showed that maternal insensitivity, i.e., not being well attuned to the child's wants and needs, was associated with significantly higher risks for the child being overweight or obese [12].

Another mechanism implies that overweight leads to emotional and behavior problems, as overweight individuals may experience a number of frustrations. For instance, obese toddlers may experience that they are physically less able to participate in common activities together with other children or that they are excluded or rejected by them. Indeed, rejection of obese children by their peers has been shown as early as age 3 years [17]. Obese children may also experience that their mothers try to control and restrict their food intake, particularly the intake of highly palatable but unhealthy foods [18]. Mothers have been shown to use controlling feeding practices with children as young as age 1 or 2 years [19, 20].

A third reason for the association between behavior problems and overweight may be shared, biologically based constitutional factors. Temperament/behavior traits, taste preferences, and the regulation of appetite are all steered by the dopaminergic system [21‐25]. Accumulating evidence indicates that the pathways in the brain governing appetite and emotion regulation are interrelated. Consequently, a common set of factors contribute to both a child's risk for overweight and the degree to which a child displays behavior problems [26].

The mechanisms presented here do not exclude each other: It is possible that there is a reciprocal relationship between behavior problems and weight over time, where behavior problems cause overweight, and overweight in turn causes behavior problems.

To date, however, we know little about the age at which the relationship between behavior and weight first appears. For children below 5 years, there are only four community studies available that examine the cross-sectional association between overweight and behavior [8, 10, 27, 28]. Three of these studies found associations [8, 10, 28] (although one of them found rather small effect sizes [28]), and one study found none [27]. There are also some prospective studies investigating this relationship. One small laboratory study examined the association between infant temperament, a construct tapping the stable aspects of behavior [29], and early childhood overweight over time. That study found that the temperament dimension labeled "distress to limitations" was positively correlated with percentage of body fat in early childhood [11]. Also Anderson et al. [6] recently demonstrated that externalizing behavior problems were associated with higher BMI and obesity in children as young as age 24 months. This average difference in BMI remained stable through age 12 among the white children in the study. On the other hand, another study following children from 6 months onwards found that children with difficult temperament and insensitive mothers had significantly higher risks for being overweight-or-obese during the school age phase but not during early childhood [12]. Similarly, a different study conducted longitudinal analyses of the relationship between BMI and behavior problems from age 2 years through the 6^th grade and found that higher BMI was associated with an increased likelihood for internalizing problems beginning in the 1^st grade [7]. One more study showed that boys with a shorter attention span and girls who were difficult to soothe and had more negative mood at age 1 year, had greater increases in standardized weight and were more likely to be overweight or obese at the age of 6 years [9]. No information was given, however, at which age these associations first emerged.

To summarize, the onset of the relationship between behavior and weight remains unclear. Community studies on the association of weight development and temperament or behavior problems in infants and toddlers are sparse. One exception is our previous study of close to 30 000 infants in Norway [30]. That study was the first to examine the association between difficult child temperament (that is, distress, crying, and fussing) and overweight in 6-month-old infants. At that early age, we found no clinically significant relationship between difficultness and overweight.

In the present study we focus on the concurrent and prospective relationship between BMI on the one hand and behavior problems on the other hand in the same cohort between the age of 18 and 36 months. By examining the cross-sectional associations of weight and behavior problems, we can roughly determine the time point at which this association begins to emerge. Moreover, we will examine the cross-lagged pathways between behavior problems and weight over time. This means that we will compare the association between behavior problems at 18 months and weight at 36 months with the association between weight at 18 months and behavior problems at 36 months. In this way, we can examine their co-development over time and thus shed light on their causal sequence. Figure 1 depicts the conceptual model of the study.

The aim of this study was to investigate the concurrent and prospective relationship between BMI and behavior problems in toddlers aged 18 to 36 months. Based on biological reasoning as well as findings from earlier studies [8, 10, 11, 14‐25, 28, 29, 39‐51], we expected Internalizing and Externalizing behavior problems to occur more often in children with higher weight. Moreover, we expected to find a co-development of weight and behavior problems, where both weight and behavior problems increase each other over time in a prospective vicious circle.

However, we found neither cross-sectional nor prospective associations. The only association that we found between somatic complaints and BMI was the opposite of the association that we expected. Indeed, children with more somatic complaints had a lower BMI than children with fewer complaints. However, this is not very surprising, as the somatic complaint items that are an integral part of Internalizing problems mostly tapped eating behavior and digestion problems. Constipation and stomachaches are likely to reduce food intake and thus to lead in turn to comparatively lower weight.

There may be several explanations for the lack of associations between Internalizing and Externalizing behavior problems (except somatic complaints) and weight. An important reason may be the young age of our sample. Even if children with externalizing or internalizing problems are inclined to eat more obesogenic foods [15], it may take a longer time before a diet high in obesogenic foods leads to overweight or obesity. In a similar vein, negative reactions by mother or peers may occur less frequently at age 18 months, an age when many mothers still prefer a chubby child to a thin child [52]. Indeed, there has only been a single, small study that found negative attitudes and behaviors towards overweight 3-year-olds [17]. Furthermore, whereas the association between overweight/obesity and psychological problems is quite well established in older children and adults [11, 39, 47, 53‐55], the findings for children below 5 years are ambiguous [8, 10, 27, 28]. In addition, one could argue that small children's eating behavior is not under their own control but largely determined by their family and cultural environment [56]. However, although family environment determines the choice of foods a child is exposed to, infants and toddlers eat what they like and spit out what they dislike.

Readers should note some limitations to our findings. Measuring behavior problems using shortened versions of the established questionnaires might threaten the construct validity of the Internalizing and Externalizing scales. Nonetheless, internal consistency was satisfactory for the majority of the subscales [35]. By using latent factors based on confirmatory factor analyses, we eliminated measurement error and improved the internal consistency of the constructs even more. Moreover, demonstrating their discriminant validity we showed in previous publications that the Internalizing and Externalizing scales, measured conventionally, correlated with other variables such as maternal smoking and child diet in a meaningful way [15, 57]. Yet, we cannot exclude that we may have measured somewhat narrower constructs than those assessed by the original CBCL. Particularly the subscale "somatic complaints" can raise questions considering its validity in the given context. The selected items such as "Doesn't eat well" (see Table 1) are closely linked to the child's actual diet, metabolism, and weight. Therefore, a negative association with BMI seems unsurprising. In fact, other studies also cautioned that real physical symptoms may be confused with the symptomatology that reflects psychosocial disturbance when applying the subscale "somatic complaints" [58]. Still, somatic complaints belong to the core symptoms of internalizing problems [59]. Particularly in small children, who have limited cognitive and language abilities, somatic complaints can serve as a crucial hint for internalizing problems. We therefore chose to keep this subscale in our analyses.

Another limitation concerns the measurement of child weight and height. Generally, the most accurate way to determine overweight or obesity is to measure skinfold thickness at different locations (triceps, subscapularis) [2]. However, this type of measurement in parallel with anthropometric measures is not feasible for a population study including 100 000 children. The weight and height measures on which maternal reports based were carried out in a standard fashion by specialized health care staff. However, the large number of missing values suggests that many mothers did not have their child's health chart at hand when filling in the questionnaires. Other mothers may have made errors when copying from the health charts. Yet, we believe that these errors were few and random. When we compared birth weight and height recorded in the Medical Birth Registry with maternal reports of birth weight and height on the 6 months questionnaire, we found correlations of .99 and .96 respectively, indicating that mothers' report of their children's' weight and height in MoBa is accurate.

Another limitation of our study pertains to external validity. With the relative low recruitment of 42.7% in the Norwegian Mother and Child Cohort Study (MoBa), selection bias is likely. Possibly, women of higher socioeconomic status are overrepresented in the study [31]. Thus, the study might not be representative for the entire spectrum of the Norwegian population. However, new results from MoBa suggest that even though selected estimates of exposures and outcomes were biased, exposure-outcome associations were not [60]. Therefore, it is reasonable to assume that our results hold in the entire population, too.

Despite these limitations, this study fills an important gap in the literature by suggesting that behavior problems and weight are still unrelated in toddlers up to age 3. This finding is substantial, as it bases on a large nationwide sample, examines behavior on a fine-grained level, and uses advanced statistical analyses that are suitable to detect even minor associations. Because the association of behavior problems with overweight has been found in 5- to 6-year-olds, further studies should focus on the age span between 3 and 6 years, which also includes the period of adiposity rebound (when the BMI normally increases after having declined to a minimum), which is "a critical period for the development of obesity" [61].