This study has yielded information about the longitudinal change of children’s BMI in relation to parental perception of their child’s weight status. We found that accurate weight status perception at age 5 years was significantly associated with higher BMI in overweight children until the age of 9 years (4 years follow-up), corrected for actual BMIz at age 5 years. Furthermore, children of parents who correctly perceived their child as being overweight showed a consistently higher BMI at different ages, compared to children of parents who underestimated their child’s overweight. A consistently higher BMI probably made it easier for parents to correctly perceive their child’s overweight . Nevertheless, even after correction for actual BMIz, the association between accurate weight status perception and higher BMI at follow-up remained statistically significant. This would imply that parental awareness of the child’s overweight as such is not protective against subsequent overweight development. This contradicts the findings of Kroke and colleagues  who reported that maternal misperception about overweight children aged 6 months had an unfavourable effect on the child’s weight development, compared to mothers who perceived their child’s weight status correctly. However, as the researchers themselves acknowledged,  determination of actual weight status using BMI in children below the age of 2 years may be an inappropriate measure,  and the feeding mode (breastfeeding or bottle-feeding) had a differential effect on early body mass development . Also, their study included relatively small subsamples, impeding external validity.
Eighty-five percent of the parents of overweight children in our study underestimated their child’s weight status. Similar percentages were found in previous studies, with 72 - 90% of the parents underestimating the weight status of their overweight child [4, 8, 10, 11, 13]. Several potential causes may underlie inaccurate parental weight status perception . First, parents may not recognize their child as overweight; they just do not see it or do not have the skills to see it. Another cause may be that parents do not understand what overweight means. This is suggested by the findings of Jain et al.  who performed qualitative interviews with low-income mothers to understand their perceptions of when a child is overweight. The authors found that these mothers did not accept the classifications used by health professionals to determine a child’s overweight. Finally, parents may feel reluctant to admit that their child is overweight . Other studies reported that parental perception of their child’s weight status was related to the mother’s own overweight status, which may influence their norm regarding what is a normal weight [6, 14]. Although a relatively large proportion of the parents in the current study were overweight or obese, we could not find such an association between parental BMI and the parent’s perception of their child’s weight status or the child’s BMIz change.
Predictors of accurate weight status perception, in both overweight and normal weight children aged 5 years, were the children’s actual BMIz at 5 years and the children’s BMIz at 2 years (both predictors being positively associated with accurate weight status perception). Other studies have also shown a positive association between actual cross-sectional BMI(z) and weight status perception, [7, 9] but we are the first to investigate the influence of prior BMI on the parental perception of previous body weight status. In normal weight children, high educational level of the father was an additional significant predictor of accurate perception. In line with this, previous studies [7, 10, 11] found that mothers with a low educational level were more likely to underestimate their child’s weigh status. We did not find any evidence that other demographic variables are significant predictors of accurate weight perception, which appears to contradict what other studies have reported [7–11, 13]. An explanation for the discrepancies between studies may be the use of different instruments for measuring weight status perception. Some studies (including the current study) have used written questionnaires to assess parental perception of their child’s weight status, [9–14] whereas others have used an interview technique, [6, 7, 16] which may make it even more emotionally challenging for parents to admit that their child is overweight compared to an anonymous questionnaire . Another possible cause of discrepancies between study results may be differences in general characteristics of the study populations between earlier studies and ours. The KOALA cohort is relatively healthy; e.g., 9% of the 5-year-old children were overweight or obese, compared to 13% and 18% of 5-year-old boys and girls in the general Dutch population the Dutch boys of the Dutch girls at the age of  The cohort also included relatively highly educated parents: 53% of the parents had a high education level, compared to 26-31% of the general Dutch population . These specific general characteristics of the KOALA cohort may mean that the influence of several predictors was not detected, because of power issues.
In the present study, 5-year-old children whose parents underestimated their child’s overweight status were generally only overweight at that particular measurement, whereas the children whose parents accurately perceived their child’s overweight consistently had a higher average BMI over time. This indicates that children whose parents accurately perceive their weight differ in weight status from the children whose parents underestimate their weight status. We found a similar pattern for parents of normal weight children: children of parents, who correctly perceived their child as normal weight, had a consistently higher BMI compared to children of parents who estimated their child as underweight. This implies that one should be very careful when examining parental perceptions of the child’s weight status in only one cross-sectional weight measurement. It seems that parents can accurately take into account their child’s weight history in assessing his or her weight status at one particular point in time. Alternatively, this phenomenon may be a statistical artefact, caused by measurement errors in either the child’s body weight or parental weight status perception, or regression to the mean.
Accurate weight status perception may be an important prerequisite for involving parents in childhood obesity interventions . In general, parents who are not aware that their child is overweight, will probably not feel the need to become involved in such an intervention [29, 30]. In one adolescent family-based intervention, for example, parental weight status perception proved an important predictor of treatment initiation . Nevertheless, accurate weight status perception alone is not sufficient. Parents may not have sufficient motivation, skills or parenting practices to manage their child’s overweight, or they may even use counterproductive strategies, such as extreme restriction or overcontrolling of unhealthy energy balance-related behaviours, to address their child’s overweight . This may result in further BMI increases in the long run. Parents therefore need to be enabled to develop adequate skills to manage their child’s overweight in family-based interventions [33, 34].
Some limitations should be taken into account when interpreting our results. Almost all data were self-reported by parents, which may have led to bias. However, the weight and height data at age 5 were partly measured by trained research assistants. We tested whether the self-report and measured data led to different results, using interaction terms in the models, and found no difference between the two measurement methods. This is line with the findings by Scholtens et al.  who reported that parental reports of their child’s actual weight and height are relatively valid. However, parents of overweight children tend to underreport their child’s weight [35, 36], indicating that even more parents underestimate their child’s weight status than we could establish in the current study. Unfortunately, we were not able to conduct separate analyses of parental weight status perception for the group of obese children, as the small number of children in this subsample (N = 32) mean there was insufficient statistical power for such analyses. Furthermore, we did not collect any data on parental behaviours, which was beyond the scope of our study. An additional limitation regards the representativeness of the KOALA cohort, as already noted. Another disadvantage of the KOALA cohort might be completing questionnaires on a regular basis, which may increase parental attention for the questioned topics and may influence their behaviour.
To our knowledge, this is the first study to examine parents’ perceptions of their child’s weight status in relation to longitudinal BMIz change, both prior and subsequent to the assessment of parental perception. It would be valuable to further investigate the effect of underestimation on children’s weight status, specifically with regard to causality and underlying mechanisms explaining the associations found in the current study. Future studies with a larger sample of overweight children should investigate what differences there are, within the group of ‘consistently’ overweight children, between parents who accurately estimate their child’s weight status and those who underestimate it.