This large population-based study among four-year-olds showed that young children’s eating patterns and feeding practices of parents are strongly associated with children’s BMI. Not only children with overweight but underweight children also had different eating behaviors than children with a healthy weight. Furthermore, analyses with feeding practices of parents showed a fairly graded association across the whole range from children’s underweight to overweight and obesity. The direction of the reported associations are much in line with previous findings among older children in the primary school-ages [15
] suggesting an early age onset of relationships between eating behavior, parental feeding and child BMI. Our observations were largely unaffected by known predictors of unhealthy weight in childhood, such as low socioeconomic background, national origin and parental weight status. As expected, associations between eating behavior and BMI of children attenuated, but persisted after controlling for parental feeding practices. Thus, part of the association between children’s eating behaviors and BMI was due to relations between parental feeding practices and child BMI, suggesting complex associations between these variables. Although we hypothesized that parental feeding practices would be associated with child weight entirely through its effect on child eating behaviors, our results showed that feeding styles were related to offspring BMI even after adjustment for children’s eating behavior. Possibly, parental feeding is associated with child BMI through other dimensions of child eating behavior, such as loss of control or binge eating. However, as a wide range of child eating behaviors was examined, our findings suggest that the behaviors of children and their parents are independently associated with children’s BMI.
Before the results can be discussed, it is essential to consider the reported prevalence rates first. In our study population, approximately one out of ten children was overweight or obese. Although this percentage is lower than the global prevalence estimates of overweight in childhood and adolescence [1
], it is consistent with the general notion that overweight is somewhat less prevalent among preschoolers as compared with older children [2
]. Furthermore, the 2% of children with obesity in our study is very comparable with a recent representative Dutch study reporting obesity prevalence rates of 1% and 3% for 4 year old boys and girls, respectively [57
]. However, the reported overweight rates (11%; boys 8%, girls 14%) in this nation-wide study were somewhat higher than those observed in our study (8%). Possibly, children with low socioeconomic background were somewhat underrepresented in our more urban sample, this may account for a slightly lower prevalence of overweight, as children from families with lower socioeconomic status are at risk of overweight [58
About 13% of the four-year-olds in our study were underweight, which is a higher prevalence estimate than previously shown in school-aged children [4
]. However, the prevalence rates of underweight are less established than overweight and obesity rates, particularly in early childhood. An alternative explanation for differences in prevalence of underweight might lie in the observation that, especially in girls, the prevalence of underweight is increasing [5
As hypothesized, children’s food approach behaviors Food Responsiveness and Enjoyment of Food were positively related to children’s BMI. These scales address children’s general appetite for food with Enjoyment of Food measuring normal variation in general appetite, while the Food Responsiveness scale is designed to detect more dysfunctional levels of appetite such as the tendency to continue eating if given the opportunity. It has been suggested that these food approach behaviors become more apparent as children get older and can make more independent choices about food [44
], but we observed substantial variability in these traits already in preschoolers. This confirms findings from earlier studies focusing on overweight [15
], and adds to the current literature that underweight children also had relatively low levels on Enjoyment of Food and Food Responsiveness. These relations between appetite and the BMI spectrum might be explained by genetic variants contributing to both children’s weight status and their susceptibility to eating in response to the presence of foods [59
]. Genes can exert a direct influence on child behavior and weight status, but can also work indirectly through the early food environment that is primarily provided by the parents. For instance, overweight parents might provide an obesogenic eating environment which stimulates appetite and food intake in the offspring, while relatively lean parents may discourage the overconsumption of food. However, behavioral genetics of child eating patterns are relatively understudied and future research is needed to clarify the interplay between genetic and environmental influences on children’s eating and weight development [59
In contrast to what we hypothesized, we found no evidence that the food approach scales Desire to Drink or Emotional Overeating were associated with child BMI. The findings with Desire to Drink replicates results from previous studies for which the authors argued that the lack of finding an association between drinking and weight might have been due to limited statistical power [26
]. Apparently, being a thirsty person or the amount of drinking per se is not related with child BMI. The type of beverages, i.e. the consumption of high-energy drinks, probably has more influence on weight status [60
]. Regarding eating in response to emotional cues, it has been suggested that the food approach scale Emotional Overeating reflects the opposite of the food avoidant Emotional Undereating [46
]. However, in our study, we found a positive correlation between the two scales. Moreover, we showed an association between emotional undereating and BMI, while surprisingly, emotional overeating was not related with weight status in these very young children. Emotional distress may lead to inhibition of appetite, but does not result in food craving in young children, suggesting that these emotional eating behaviors cannot be simply seen as two extremes of the same continuum. Alternatively, the children in the present study might have been too young to exhibit excessive eating and snacking, as they probably do not have free access to foods yet. This hypothesis is substantiated by previous studies showing that increasing BMI was associated with progressively higher levels of emotional overeating among school-aged children [16
In line with our hypotheses, not only Emotional Undereating, but the food avoidant scales Satiety Responsiveness and Fussiness were also associated with progressively lower weights in children. This finding for Fussiness contrasts with previous studies [25
]. However, in a population-based sample of 1498 Canadian preschoolers, Dubois and colleagues also reported that picky eaters were more likely to be underweight [34
]. Our findings and this Canadian study suggest that fussiness is indeed associated with a relatively low BMI in the general population. Possibly, underweight of children leads to higher levels of fussiness, for instance through an adverse effect of control or pressure of parents on children’s eating. However, the analyses substantiate this reasoning only to some extent, as the relation between fussiness and BMI attenuated but remained statistically significant after adjusting for parental pressure to eat and monitoring. Thus, it seems likely that at least part of the association is from pickiness leading to insufficient food intake, which eventually hinders adequate weight gain and growth. Although the CEBQ refers to fussiness about food in general, it is also possible that the food intake of picky children is not diverse enough and lacks essential nutrients like vitamins, minerals, proteins and fibres. Clearly, parents and primary health care professionals should carefully monitor fussy children and their food intake, although causal directions have to be ascertained in longitudinal studies.
A negative association between satiety response and BMI in preschoolers was reported once before [15
]. We extend this previous study by showing that both children with overweight and children with underweight have a different satiety response than children with a normal weight. Possibly, some young children have a suboptimal down regulation of their food consumption resulting in excessive weight gain, while other children have a too effective satiety response resulting in underweight. However, that parental feeding practices accounted for part of this association suggests more complex pathways. Parental pressure to eat might reflect a mediation effect, as toddlers who are highly responsive to internal satiety cues and quickly feel full might be pressured by their parents to eat more, which can be counterproductive and actually result in less
]. Alternatively, restrictions of parents regarding food intake could result in a poor responsiveness to internal hunger and satiety cues, thereby influencing children’s food intake and weight gain.
As hypothesized, restrictive parenting during mealtimes was positively and parental pressure to eat was negatively associated with children’s BMI. Within the framework of a child-responsive model, these feeding strategies can be interpreted as a response to child weight: parental efforts to restrict food intake may be a response to children’s overweight, while parents of children with underweight probably pressure their children to eat more. However, the observed associations are probably more complex. It is not possible to infer causality from our cross-sectional study, but the findings suggest a number of explanations. Parental restriction was correlated to children’s Food Responsiveness, and child eating behavior attenuated the association between restriction and child BMI. This suggests that restrictive parenting might stimulate poor intake regulation and overeating at times when access to food is not restricted, eventually resulting in weight gain. Parental pressure to eat may be associated with child weight through a counterproductive effect of lowering children’s enjoyment of food, eventually resulting in eating less and weight loss. Alternatively, pressure to eat might also be a parent’s response to children quickly feeling ‘full’. This explanation is substantiated by the correlation between Pressure to Eat and CEBQ Satiety Responsiveness. Clearly, longitudinal research with repeated measurements of children’s eating behavior, feeding practices of parents and child BMI is needed to further unravel these pathways.
We found no evidence that parental monitoring of children’s food intake is associated with child BMI, which is in line with previous studies using small convenience samples [25
]. Perhaps, keeping track of the amount of sweets, snacks and high-fat food children consume is a very common behavior of parents, not necessarily related to children’s BMI. Alternatively, parents might have provided socially desirable answers on the Monitoring-items.
Some limitations of this study have to be discussed. Firstly, a number of children had no information on eating behavior or BMI. While missing data on BMI was rather random, information on eating behavior was more complete in Dutch children of relatively high educated mothers. However, although this selective response may have influenced the reported prevalence estimates, it probably has had less effect on the associations reported in our study [61
]. Secondly, although children’s anthropometrics were measured objectively, assessments of problematic eating behaviors were based on mothers’ subjective opinions of children’s behavior. Even though the analyses were adjusted for several maternal characteristics, it cannot be ruled out completely that a mother’s well-being and her attitudes about health influenced her ratings of children’s eating behavior. On the other hand, validation studies indicated that parent reports of children’s eating behaviors, such as the CEBQ and CFQ, correlated substantially with children’s actual food intake [34
]. Another limitation is the study’s cross-sectional design which precludes inferences about causation. Longitudinal studies are essential to detect whether feeding practices of parents and children’s eating behaviors predict the development of weight problems, or if they are associated concurrently only.