We compared the assessment of overweight among 5-year-old children using BMI versus WC and BMI versus the WHtR. The results of the agreement analyses show that overall agreement between BMI classification versus WC and BMI classification versus the WHtR was high but the positive agreement between the measures was moderate to substantial. This indicated that BMI versus WC and BMI versus the WHtR agree moderately to substantially about the presence of overweight among 5-year-old children. It appeared that more than one third of the total group of children classified as overweight according to BMI was not classified as overweight according to WC or the WHtR. This was the same in the group of children classified as overweight according to WC. In the group of children classified as overweight according to the WHtR, more than half of the total group was classified as overweight only according to the WHtR.
Compared to the overweight BMI only group, children classified as overweight according to WC only were relatively taller and more boys were of Dutch ethnic background. When comparing BMI to WHtR, children classified as overweight according to WHtR only were relatively younger, shorter, and lighter. Approximately 2 in 3 of the children classified as overweight according to BMI only were also clinically judged as overweight by a healthcare professional during a well-child visit. In the subgroup classified as overweight according to WC only, this ratio was approximately 1 in 7 and in the subgroup classified as overweight according to WHtR only, this ratio was approximately 1 in 12.
Our data comes from a large population-based study of young children. Because of the small age range, our results are specific to the 5-year-old age group. The weight status of the children according to BMI was assessed using the IOTF’s age-specific and sex-specific cut-off points [21
]. These cut-offs were chosen because they are used by the healthcare professionals at the municipal health services in the Netherlands [19
]. By using these cut-off points, international comparisons of the prevalence of childhood overweight are also possible. Cut-off points for WC are not used in general by the healthcare professionals to assess children’s weight status during well-child visits. Also, no international validated cut-off points for WC are available. We used the available age-specific and sex-specific cut-off points for WC for Dutch children as presented by Fredriks et al. [22
]. These cut-off points were based on data of 14,500 children aged 0–21 years in the Fourth Dutch Growth Study [22
]. Internationally accepted cut-off points are also not available for the WHtR, and we used the 90th
percentile within our total study population at baseline (n
8784) as the cut-off point for classification of overweight (obesity included). We also investigated the agreement between BMI and WHtR classification by using the cut-off point of 0.5 for the WHtR [16
] instead of the 90th
percentile, and results were similar.
As reported in previous studies, more girls were classified as overweight compared to boys in our study [2
]. As indicated in literature, a contribution to this may have been made by a significant difference in the sensitivity of the IOTF BMI cut-off points [2
]. Children of parents with a low educational level, as an indicator of low socio-economic status [33
], and children of non-Dutch ethnic background [34
] are at increased risk for overweight and this is also reflected in our results. A relatively large number of the children in our study population were of non-Dutch ethnic background (main ethnicities: Moroccan, Turkish, Surinamese and Dutch Antillean), which allowed us to investigate differences in the distribution of ethnic background across the subgroups classified as overweight. The number of non-Dutch children was higher in the overweight subgroups but only among boys there was a statistically significant difference in ethnic background between the subgroups classified as overweight according to BMI only and WC only.
We compared the characteristics of the children in the study population with the characteristics of the children who were excluded from analyses due to missing data. Among boys, we found no statistically significant differences between those groups in age, height, weight, WC, ethnic background, parental educational level, or weight status of the children. Among girls, we found that weight, WC, and the prevalence of overweight according to BMI was higher among girls in the study population compared to girls who were excluded from analyses. However, we assume that the results of our analyses would be similar in the subgroup of girls with missing data.
Previous studies indicated that measuring only BMI results in an underestimation of health risk. To our knowledge, our study is the first study comparing classification results between BMI and WC and between BMI and the WHtR among 5-year-old children. The study by Fredriks et al. [22
] found a strong correlation between BMI and WC. In additional analyses we also investigated the overall correlation in our study population, and we found comparable results. However, when we divided the BMI-group into quartiles, it appeared that the correlation was high only among children with a BMI in the highest quartile (see Additional file 2
). So these findings indicate that BMI and WC merely agree among children with excess body fat in the highest percentile groups. This is also reflected in the results of the analyses in which we compared the children’s characteristics between the overweight subgroups; the children classified as overweight according to both BMI and WC had the highest amount of overall body fat (as estimated by BMI) and abdominal fat (as estimated by WC).
Further, we found that children classified as overweight according to both BMI and WC, and according to WC only, appeared to be relatively taller than the group classified as overweight according to BMI only. On the other hand, when comparing BMI and the WHtR, we found that children classified as overweight according to BMI only were relatively tall, and children classified as overweight according to WHtR only were short. There could be several explanations for these findings. First, it has been suggested that BMI is a measure of excess weight relative to height and not of excess body fat [14
]. Therefore, BMI might not be a sensitive measure of body fat among children who are particularly short or tall or who have an unusual body fat distribution [36
]. Second, children classified as overweight according to WC only might have a high WC as a consequence of being relatively tall for their age; this subgroup might include ‘overall large children’. We recommend future studies to investigate whether the WHtR can be used to assess overweight among relatively short children. Specifically, future studies should assess which cut-off points for WC or the WHtR are best to classify overweight among young children.