Cross-Sectional Comparisons of BMI and Waist Circumference in British Children: Mixed Public Health Messages
Research suggests that there has been a leveling off in obesity prevalence occurring in the child population. However, a concern with the evidence base is that all of the studies have relied upon the use of BMI. The purpose of this study was to compare waist circumference (WC), BMI, and waist-to-height ratio (WHtR) data in three different sample of children (total number: 14,697) typically aged 11–12 years. Obesity prevalence defined by BMI did not change significantly between measurement years (2005 boys 20.6%, girls 18.0%; 2006 boys 19.3%, girls 17.3%; 2007 boys 19.8%, girls 16.4%). Obesity prevalence defined by WC was considerably higher especially, in girls (2005 boys 26.3%, girls 35.6%; 2006 boys 20.3%, girls 28.2%; 2007 boys 22.1%, girls 30.1%). The prevalence of children defined as “at risk” according to WHtR (2005 boys 23.3%, girls 21.1%; 2006 boys 16.7%, girls 15.6%; 2007 boys 17.6%, girls 17.2%) was found to be between obesity prevalence, estimated using BMI and WC. This data are the most up to date collection that includes BMI and WC in three large samples of children and clearly demonstrates inconsistencies between different measurements based on current classification systems. There is a need to understand the relationship between BMI and WC, with growth and health risk to establish a consistent public health message that is easily understood by the public.
Recent publications from a number of countries, including the United Kingdom (1) suggest that there has been a leveling off in overweight and obesity prevalence occurring in the child population as a whole, or in certain subgroups of it. Although this stabilization is, on the face of it, good news, a major concern with the evidence base is that all of the studies have relied upon the use of BMI as either their sole or primary measure of obesity.
Although the choice of BMI as a measure of obesity in children is well established (2), there is an emerging evidence base to suggest that central adiposity in children is more relevant to health outcomes than overall adiposity estimated by BMI (3,4) and waist circumference (WC) has been advocated as a good indicator of central adiposity (5). A matter of concern, however, is the influence of size on WC, therefore it has also been proposed that a measure of WC in relation to height is more appropriate than WC alone (6). Thus the waist-to-height ratio (WHtR) has also been proposed as simple indicator of abdominal obesity (5,6).
The purpose of this study was to compare BMI, WC, and WHtR data in three cohorts of children (total number: 14,697) typically aged 11–12 years over three consecutive years to investigate if the prevalence of obesity shows the same trends with all measures and whether there is evidence to support a leveling off in all three measures in consecutive years.
Methods and Procedures
BMI and WC were collected as part of the Rugby League and Athletics Development Scheme, which is a collaboration between Leeds City Council, Leeds Metropolitan University and the Education authority (Education Leeds). Ethical clearance was granted by the ethics committee of the Carnegie Faculty, Leeds Metropolitan University.
Data were collected over 3 years (2005 n = 5,447, 2006 n = 5,525 and 2007 n = 4,869) and after validation data on 14,697 children (2005 n = 5,143; 2006 n = 5,094; 2007 n = 4,460) were used in the analysis. The participation rates are in line with those of National surveys (1).
All testing took place on school premises during scheduled physical education lessons. The anthropometric measures taken were height, weight, and WC. The same protocols were followed in each measurement year, and all measurements were carried out by the same person (C.G.). Height was measured to the nearest 0.1 cm using a floor-standing Leicester height measure (model 220) with children standing erect without shoes. Weight was measured to the nearest 0.01 kg using manually calibrated electronic scales (Tanita TBF-310; Tanita, Tokyo, Japan), without shoes. WC was measured mid way between the 10th rib and the iliac crest in a horizontal (transverse) plane to the nearest 0.1 cm using an inelastic anthropometric tape, with the child in a standing position at the end of a normal expiration (7). To allow for clothing (a thin t-shirt or vest) 0.5 cm was subtracted from the WC measurement (8). BMI was calculated as weight (kg)/height2 (m). WHtR was calculated by dividing WC (in cm) by height (in cm).
The technical error of measurement and coefficient of variation (9) for WC in 250 pupils (selected from one school) measured twice on the same day were 0.74 cm and 0.98%, respectively. The technical error of measurement (coefficient of variation) for height and weight (in the same 250 pupils) was 0.21 cm (0.14%) and 0.05 kg (0.13%), respectively. These figures demonstrate appropriate reliability (10).
BMI and WC measurements were converted to standard deviation scores using the British 1990 growth reference charts for BMI (UK90 (11)), and the 2001 McCarthy references for WC (7). Children were classified as overweight or obese based on their standardized scores to allow comparison while accounting for normal growth. The 85th and 95th centiles (standardized score = 1.04 and 1.64, respectively) were used to define overweight and obesity respectively for BMI and WC.
A WHtR boundary value of 0.5 has been proposed as a simple means of indicating whether in adults, the amount of upper body fat accumulation is excessive and poses a risk to health (6). This measure has not been closely studied in children, however studies that have measured WHtR in children have applied the same boundary (12,13).
Prevalence (percentages) of overweight and obesity are reported and obesity prevalence or “at risk” was modeled using a main effects binary logistic regression model (outcome = obese or not, exceeding or equal to the UK90 95th centile) controlling for gender. Statistical analysis was performed on SPSS (version 18)
Using the 85th (s.d. = 1.04) and 95th (s.d. = 1.64) centiles for overweight and obesity (based on BMIsd and WCsd), respectively, we would expect 5% of children to be classified as obese and a further 10% classified as overweight (but not obese) in the Rugby League and Athletics Development Scheme sample, if there was no difference in the distribution of scores compared with when the reference data were established. Interestingly, the proportion of children classified as overweight has not changed significantly for either of the measures, however the proportion of children classified as obese has increased fourfold. Furthermore, the prevalence of overweight relative to the prevalence of obese children in the Rugby League and Athletics Development Scheme sample is lower using both BMIsd and WCsd (Table 1).
Table 1. Prevalence of overweight and obesity by gender according to BMIsd and WCsd
Logistic regression suggested that the odds of being obese is not changing between measurement years (compared to the baseline, 2005) when defined by BMIsd 2006, odds ratio (OR) 0.935 95% confidence interval (CI) (0.845–1.04) 2007, OR 0.921, 95% (0.829–1.02). The odds of being obese according to WCsd did change over time, 2006, OR 0.712, 95% CI (0.652–0.778) 2007, OR 0.785, 95% CI (0.717–0.859) as did the prevalence of those “at risk” by WHtR 2006, OR 0.675 95% CI (0.611–0.740) 2007, OR 0.735, 95% CI (0.664–0.815). The prevalence according to WC and WHtR decreased between 2005 and 2006 and then increased between 2006 and 2007 (Table 1).
The odds of being obese according to BMIsd is significantly lower for girls; OR 0.840 95% CI (0.772–0.915). However, according to WCsd the odds of girls being obese is 1.54 times that of boys; OR 95% CI (1.43–1.65). Girls are less likely to be “at risk” according to WHtR; OR 0.921, 95% CI (0.847–1.00).
These results are consistent with reports of a leveling off in overweight and obesity in children of a similar age according to BMI (1). However, these results also show that the prevalence of central obesity is considerably higher than estimates based on BMI especially in girls. Publications reporting a stabilization in the prevalence of obesity in children are well documented, however, there is limited information on the changes in obesity measured by WC in children. Furthermore, methodological differences described in similar studies (7,8) are overcome in the present study with one individual measurer (C.G.) providing a low technical error of measurement and coefficient of variation, ensuring confidence in collected data.
The findings raise three major concerns. First is the inconsistency in public health messages. On the one hand BMI is a well established methodology to determine weight status in children and is associated with adverse risk factors (14). Furthermore the current evidence base suggests that the prevalence of obesity according to BMI in children is leveling off. On the other hand evidence suggests that increased central adiposity, measured by WC, is associated with greater health risk in children (3,4) and that obesity prevalence according to WC is increasing especially in girls (8). Other research suggests that expressing WC relative to height, is the best measure of risk (15) and that WHtR also shows increases in prevalence over time, especially in girls (12). Although the data presented here are cross-sectional, to the best of our knowledge they are the most up to date collection that includes measured BMI, WC, and WHtR in a large sample of the same children. Based on these results, which are in agreement with those emerging from the evidence base, it is not surprising that the current public health message regarding trends in children's obesity is unclear.
The second concern is the potential misclassification of children; this data provide a strong case for questioning current interpretation and use of BMI and WC and highlight the need to better understand the relationship between changes in BMI and WC associated with growth during childhood and health risk. Finally, the complexities associated with the standardization process for BMI and WC are potentially confusing to persons outside of the heath context, for example parents and children. The limitations associated with the reference data for WC have already been acknowledged (7) and so these data should be interpreted with caution at this stage. Although a lot could be gained from the use of suggested simple public health messages such as “keep your waist circumference to half your height”, there is limited data investigating the changes in WHtR with age. Further research evaluating all measures from a longitudinal perspective is needed before any message can be quoted without caution in children.
We would like to thank Leeds City Council and Carnegie Weight Management for their initial collaboration in establishing the programme. In addition we would like to thank Leeds City Council and the secondary schools that participated in the programme for providing the data from the Rugby League and Athletics Development Scheme. All authors had full access to the all of the data and can take responsibility for the integrity of the data and the accuracy of the data analysis.
The authors declared no conflict of interest.