Waist circumference percentile charts

Given the findings of the research into intra-abdominal fat accumulation in children, its association with metabolic risk factors and the ability of WC to predict abdominal fatness, WC centile charts should prove to be a useful addition or alternative to BMI centile charts in the clinical assessment of childhood obesity. Compared with BMI, WC would better identify children with high abdominal fatness and hence risk for MS. This premise is especially true given the known limitations of BMI and BMI centile charts in children for identifying those who may be at risk for the MS. Up until 2001, there had been no obesity-related reference data for WC in UK children, but WC centile charts had been developed for the Italian (aged 6–14 years; Zannolli & Morgese, Reference Zannolli and Morgese1996) and Spanish (aged 6–15 years; Moreno et al. Reference Moreno, Fleta, Mur, Rodriguez, Sarria and Bueno1999) paediatric populations. Waist:hip ratio centiles had also been developed for Cuban children and adolescents (Martinez et al. Reference Martinez, Devesa, Bacallao and Amador1994). For equivalent UK charts it was essential that historical data were utilised and collected before the emergence of the current obesity epidemic. If contemporary data had been used, then comparison with the UK references for height, weight and BMI would be difficult (Cole et al. Reference Cole, Freeman and Preece1995). One key issue in the development of the WC centile chart and the use of WC measurement in paediatric obesity research is that there is no universal agreement on the definition of the WC landmark. Most research groups have taken the WC as being mid-way between the 10th rib and the top of the iliac crest (for example, see Freedman et al. Reference Freedman, Serdula, Srinivasan and Berenson1999; Moreno et al. Reference Moreno, Fleta, Mur, Rodriguez, Sarria and Bueno1999). Others, such as those who developed the Italian charts, have measured WC at the level of the umbilicus (Zannolli & Morgese, Reference Zannolli and Morgese1996). This disparity makes comparisons between populations difficult.

Data used for the development of the UK WC centile charts were collected in 1977 for boys and 1987 for girls and the measurement site was the midpoint between the 10th rib and the top of the iliac crest (British Standards Institute, 1990). Using the LMS curve-fitting software developed by Cole and colleagues (Cole, Reference Cole1990), gender-specific and age-related smoothed centile charts have been constructed (McCarthy et al. Reference McCarthy, Jarrett and Crawley2001). In this dataset there is an age-related increase in mean WC up to age 17 years. Gender differences are evident from age 9 years onwards, with boys having the greater mean WC at each age; the difference in mean WC between boys and girls at age 17 years is 6·3 cm. At present, the published WC centile charts cover the age-range between 3 and 17 years, and include the standard nine centile lines, with a channel width of 0·66 sd score. The 91st and 98th centiles are used to define overweight and obese, in line with the UK BMI charts.

Since the development of WC centile charts for UK children, other nations have produced similar charts, including Cyprus (ages 6–17 years; Savva et al. Reference Savva, Kourides, Tornaritis, Epiphaniou-Savva, Tafouna and Kafatos2001), USA (ages 2–18 years; Fernandez et al. Reference Fernandez, Redden, Pietrobelli and Allison2004), Canada (ages 11–18 years; Katzmarzyk, Reference Katzmarzyk2004), Australia (ages 7–15 years, Eisenmann, Reference Eisenmann2005), The Netherlands (ages birth to 21 years; Fredriks et al. Reference Fredriks, van Buuren, Fekkes, Verloove-Vanhorick and Wit2005) and Mexico (ages 6–10 years; Gomez-Diaz et al. Reference Gomez-Diaz, Martinez-Harnandez, Aquilar-Salinas, Violante, Alarcon, Villarruel, Rodarte and Solorzano-Santos2005). However, there is variation between these studies in the landmark used to define the waist, again leading to potential difficulties when attempting to make cross-cultural comparisons. Furthermore, most of the WC data were collected from contemporary children and in countries with known increased prevalences of overweight and obesity.

When BMI–WC relationships in children and adolescents are examined, a strong positive relationship between these two variables exists. However WC can vary greatly for a given BMI at a particular age. As a result, this relationship can give an insight into how combining age-related BMI and WC measurements could improve the identification of those children most at risk of MS, and conversely how it could assist in avoiding the incorrect classification of some children. Table 1 illustrates this issue in four boys of approximately similar age, and the BMI–WC relationships highlighted can be demonstrated across all ages and in both genders.

Table 1.

Combining age-related BMI and waist circumference (WC) measurements could assist in avoiding the incorrect classification of some children, as illustrated for four boys of similar ageFootnote *

Approx, approximately.

* Child A can be considered a reference child with an average BMI for age and gender. Child B, based on BMI would also be considered average for age. However, when WC is taken into account, this child would be classified as obese. The latter is an example of where there is not excess body mass, but excess central body fat accumulation (furthermore, it is likely that this child would be shorter than child A). Child C would be considered overweight based on his BMI, but his WC is average. This example could be a case where the child is more muscular (and possibly taller, leading to a greater fat-free mass) for his age. Finally, child D is considered obese based on his BMI and obese based on his WC.

As two separate measures of fatness BMI and WC interact in a complex fashion, especially in childhood, and are both influenced by age, gender and pubertal stage. In a recent cross-sectional study in 474 healthy adolescents aged 17 years (Neovius et al. Reference Neovius, Linne and Rossner2005) both BMI and WC have been shown to perform well as diagnostic tests for fatness (air-displacement plethysmography was used as the reference method and diagnostic accuracy was evaluated through receiver operating characteristics). However, as measures of risk for obesity-related morbidity, age- and gender-related BMI and WC can lead to different interpretations. As a result, it may be more useful in clinical practice (and possibly in epidemiological studies) to perform both measurements and to make an evaluation on the child based on the combined observations. Indeed, such a recommendation has been supported by findings in children for whom outcome measures of HDL-cholesterol level, triacylglycerol level, high glucose level, high insulin level and high blood pressure were predicted from age-specific BMI and WC values (Katzmarzyk et al. Reference Katzmarzyk, Srinivasan, Chen, Malina, Bouchard and Berenson2004; Janssen et al. Reference Janssen, Katzmarzyk, Srinivasan, Chen, Malina, Bouchard and Berenson2005). From these studies the authors have concluded that a combination of BMI and WC should be recommended for use clinically to predict risk factor clustering and elevated health risk among children and adolescents.

Secular changes in abdominal fatness in children

Reference data for WC and hence abdominal fatness in UK children exists. Thus, in view of the escalation in childhood overweight and obesity, it was important to determine whether abdominal fatness has increased in British children, thus increasing the numbers at risk for the metabolic consequences of abdominal obesity and the MS. This issue was resolved by comparing the WC reference data (McCarthy et al. 2001) with that collected 10 and 20 years later for 11–16-year-old children participating in the National Diet and Nutrition Survey (Gregory & Lowe, Reference Gregory and Lowe2000). This survey was the first contemporary nationally-representative children's survey in which WC measurements had been collected. Fortunately, the landmark used for WC measurement was the same as that for the earlier survey. In the analysis (McCarthy et al. Reference McCarthy, Ellis and Cole2003) WC was expressed as a standard deviation score, using the first survey (British Standards Institute, 1990) as the reference. Changes in BMI were analysed in the same way but using the UK data as the reference (Cole et al. Reference Cole, Freeman and Preece1995). Overweight and obesity were defined as measurements >91st and 98th centile respectively. The key finding from this study is the sharp increase in WC over the 10-year period in girls (6·2 cm, 1·02 sd score; P<0·0001) and over the 20-year period in boys (6·9 cm, 0·84 sd score; P<0·0001). In centile terms, the increase was found to be greater for girls than for boys. For BMI, the extent of the increase was smaller and was similar for boys and girls (mean 1·5 and 1·6 BMI units respectively, or 0·47 and 0·53 sd score; P<0·0001). In the National Diet and Nutrition Survey (Gregory & Lowe, Reference Gregory and Lowe2000) sample WC was found to exceed the 91st centile in 28% of boys and 38% of girls, with 14% and 17% of boys and girls respectively exceeding the 98th centile (P<0·0001). By comparison, for BMI 17% of boys and 21% of girls were found to exceed the 91st centile in the National Diet and Nutrition Survey (Gregory & Lowe, Reference Gregory and Lowe2000) sample, with the corresponding rates for the 98th centile being 10% for boys and 8% for girls. These findings clearly demonstrate that increases in central fatness over the past 10–20 years (measured by WC) have exceeded increases in overall fatness (measured by the BMI) and that BMI is a poor proxy for central fatness. These findings also indicate that obesity prevalence has been systematically underestimated by measurements based on height and weight. As indicated earlier, BMI measures the sum of fat mass and fat-free mass (Maynard et al. Reference Maynard, Wisemandle, Roche, Chumlea, Guo and Siervogel2001; Wells et al. Reference Wells, Coward, Cole and Davies2002), and it is impossible to know the contribution of each compartment to these increases in BMI in this study. The increase in WC is unlikely to be entirely a result of increases in visceral adipose tissue, and probably reflects subcutaneous and hence total fatness. Taken together, these findings suggest that body composition (and hence body shape) has changed over this period of time, with the increase in (central) fat mass being obscured by decreases in fat-free (and hence probably skeletal muscle) mass. Whether this possible skeletal muscle issue is related to decreases in physical activity levels is unclear at this time. It is known that in obese adolescents excess fat accumulates predominantly in the upper body region rather than in the peripheral region (Brambilla et al. Reference Brambilla, Manzoni, Sironi, Simone, Del Maschio, di Natale and Chiumello1994; Moreno et al. Reference Moreno, Fleta, Mur, Sarria and Bueno1998). These findings have important implications for risk of MS, given the fact that children in the upper distribution of the WC values have a more atherogenic blood lipid profile and raised fasting insulin concentrations. These secular increases in WC and abdominal fatness in adolescents have also been demonstrated outside the UK, but over a shorter period of time, indicating the reproducibility of this phenomenon in other populations (Moreno et al. Reference Moreno, Sarria, Fleta, Marcos and Bueno2005). Also, an earlier study of Spanish children (Moreno et al. Reference Moreno, Fleta, Sarria, Rodriguez and Bueno2001b) has shown that between 1980 and 1995 there was a shift towards a central distribution of fat patterning in children as young as 6 years of age. These observations were based on ratios of skinfold thicknesses at the periphery and trunk and were independent of trends in BMI. Furthermore, conversion of these children's skinfold thickness measurements into predicted percentage body fatness indicates a secular increase in total percentage body fat of between 2·5 and 6·0% that is greater than the increase in BMI, again suggesting a change in body composition over the 15-year period (Moreno et al. Reference Moreno, Fleta, Sarria, Rodriguez and Bueno2001a).

Following on from the novel findings in adolescents (McCarthy et al. Reference McCarthy, Ellis and Cole2003), the next question to be asked was whether these body fatness findings were present in younger children. This question was answered using data from the ALSPAC study. ALSPAC is a longitudinal cohort study being conducted in the West of England, which was initiated in the early 1990s and comprises approximately 14 000 children (Golding et al. Reference Golding, Pembrey and Jones2001). A subset of these children, comprising approximately 1000 children, has been studied in more detail (known as Children in Focus). WC and BMI between the ages of 2 and 5 years were compared with values for similar-aged children measured in 1987. The essential finding of this study is a significantly greater mean WC in the ALSPAC children at equivalent ages compared with those measured in the earlier study (P<0·05; McCarthy et al. Reference McCarthy, Jarrett, Emmett and Rogers2005). Whilst mean BMI was also found to be higher in the contemporary ALSPAC cohort, the proportional increase in WC was again greater than that for BMI. These findings mirror those in the older children, suggesting a greater rate of increase in abdominal fatness than in general fatness, and the occurrence of excess central fat deposition at a very young age. Again, this study demonstrates the shortcomings of BMI measurement, indicating that it can mask the true obesity-related risk in young children.

Longitudinal studies of both total fatness and abdominal fatness in children are scarce. The Nepean study (Garnett et al. Reference Garnett, Cowell, Baur, Shrewsbury, Chan, Crawford, Salmon, Campbell and Boulton2005), which has examined both factors in detail between 7–8 years of age and 12–13 years of age, has found an increase of 0·74 for the mean WC sd score, but an increase 0·18 for BMI sd score. These findings suggest a greater rate of increase in abdominal fatness than in overall fatness in this contemporary cohort of children.

Abdominal fatness, ethnicity and health risk

The findings discussed so far have related almost exclusively to Caucasian children. However, the prevalence of overweight and obesity and its health consequences can be greater for children from other ethnic groups, especially those from south Asian and African and Caribbean backgrounds. A recent UK study (Saxena et al. Reference Saxena, Ambler, Cole and Majeed2004) has demonstrated that British Afro-Caribbean and Pakistani girls have an increased risk of being obese, while Indian and Pakistani boys have an increased risk of being overweight. These findings suggest that these individuals may be at a greater risk of obesity-related morbidity. Studies in the USA (Goran et al. Reference Goran, Nagy, Treuth, Trowbridge and Dezenberg1997; He et al. Reference He, Horlick, Thornton, Wang, Pierson, Heshka and Gallagher2002b) have demonstrated ethnic differences in fat distribution among Asian, African-American and Caucasian prepubertal children. Generally, the relative distribution of adipose tissue in the intra-abdominal region compared with the subcutaneous abdominal region is significantly lower in children from African-American backgrounds compared with Caucasian children. The relationship between intra-abdominal fat and insulin response in African-American children appears to be highly complex (Gower et al. Reference Gower, Nagy, Trowbridge, Dezenberg and Goran1998), although more recent findings in this ethnic group (Lee et al. Reference Lee, Bacha, Gungor and Arslanian2006) have shown that WC is able to predict insulin resistance independently of BMI. The authors further suggest that the prediction of health risks associated with obesity is improved by including WC measurement alongside BMI measurement. They also emphasise the importance of including WC measurement routinely in the assessment of childhood obesity in order to identify those children at increased metabolic risk because of excess abdominal fat. Although data is less abundant that that for Caucasian children, nevertheless the findings generally concur, in that a more central distribution of fat profoundly affects CVD risk factor levels, including HDL, LDL, triacylglycerol levels and systolic and diastolic blood pressure in black boys (Morrison et al. Reference Morrison, Barton, Biro, Daniels and Sprecher1999a) and black girls (Morrison et al. Reference Morrison, Sprecher, Barton, Waclawiw and Daniels1999b).

In relation to insulin resistance and risk for type 2 diabetes in children of south Asian background, recent findings from the UK (Ehtisham et al. Reference Ehtisham, Crabtree, Clark, Shaw and Barrett2005) have demonstrated that south Asian children are less insulin sensitive than white Europeans and have significantly more body fat, with a greater central fat distribution (P<0·005). It was concluded that children from south Asian backgrounds are fourteen times more likely that white European children to develop type 2 diabetes. Further work is required to examine the development of central body fat accumulation in UK children from other ethnic groups, especially those from a south Asian background and African and Caribbean backgrounds. However, British south Asian children as young as 8 years of age have been shown to have higher fasting and glucose-stimulated insulin concentrations and higher triacylglycerol levels, which may reflect an increased sensitivity to adiposity (Whincup et al. Reference Whincup, Gilg, Papacosta, Seymour, Miller, Alberti and Cook2002). Although not observed in the latter study, recent findings have shown that a more truncal distribution of fat is present at birth and in early childhood in south Indian children (Krishnaveni et al. Reference Krishnaveni, Hill, Veena, Leary, Saperia, Chachyarnma, Karat and Fall2005) and that this pattern of fatness may be the forerunner of a diabetogenic adult phenotype.

Waist circumference alone or in combination with height?

The evidence thus far indicates that WC can provide vital information in children in relation to measurement of abdominal fatness and risk for obesity-related ill health. One matter of concern, however, is the influence of stature on WC in both children and adults. There is a strong positive correlation between height and WC throughout growth, throughout childhood and into adulthood, although the precise influence of height on WC remains quantitatively unclear. It has been suggested that a measure of WC in combination with height may partly correct for the effect of height on WC. Thus, WHtR has been proposed as a simple indicator of excess abdominal fat accumulation, with a cut-off or boundary value of WHtR ≥0·50 defining those with excess abdominal fatness (Ashwell et al. Reference Ashwell, LeJeune and McPherson1996). This measure has been proposed to be equally appropriate for use in adults and children, in both boys and girls and at all ages >5 years. Indeed, several studies have demonstrated the WHtR to be superior in its ability to predict CVD risk factors compared with BMI or percentage body fat (Savva et al. Reference Savva, Tornaritis, Savva, Kourides, Panagi, Silikiotou, Georgiou and Kafatos2000; Hara et al. Reference Hara, Saito, Iwata, Okada and Harada2002; Kahn et al. Reference Kahn, Imperatore and Cheng2005). A recent examination of WHtR in British children (McCarthy & Ashwell, Reference McCarthy and Ashwell2006) has shown that the percentage of children with a WHtR above the boundary value of 0·50 has increased in recent years (boys 17 and girls 12 v. 5 and 1·5 respectively 10–20 years earlier).

Too many measures of central obesity in children could be confusing, but the concept of WHtR may be one that is easily grasped by the public, with a simple public health message to ‘keep you waist circumference to less than half your height’. Additional work in this area should help to clarify the 0·50 boundary value and the usefulness of this ratio.

Waist circumference and the metabolic syndrome in children

As stated earlier, the MS is one of the most important public health issues affecting future generations around the world. In order to be able to readily identify those children who are most at risk of developing this disorder a simple and effective screening tool is required, and the WC appears to best fit the requirements of such a tool. While it is generally agreed that obesity is the predominant correlate of risk for MS among the young (Goodman et al. Reference Goodman, Dolan, Morrison and Daniels2005), an increasing number of studies are showing that WC is the best measure for identifying children with insulin resistance and hypertriacylglycerolaemia, and hence those most at risk for MS (Moreno et al. Reference Moreno, Pineda, Rodriguez, Fleta, Sarria and Bueno2002; Hirschler et al. Reference Hirschler, Aranda, Calcagno, Maccalini and Jadzinsky2005; Esmaillzadeh et al. Reference Esmaillzadeh, Mirmiran and Azizi2006). WC has been found to perform better than BMI in this context, and these studies have recommended that WC should be used in paediatric clinical settings. Furthermore, when determining the prevalence of MS in prepubertal children, the classification of MS in children depends strongly on whether hyperinsulinaemia is central to the MS classification (Golley et al. Reference Golley, Magarey, Steinbeck, Baur and Daniels2006). The authors have concluded that there is a need for the establishment of normal insulin ranges in childhood and adolescence and a consistent definition of MS.

Standardisation of waist circumference

WC now appears to be fundamental in determining risk of MS and related disorders in both childhood and adulthood. There is, therefore, an important need for WC to be measured routinely in the clinical setting, and in epidemiological research in which obesity is a component of predictive or outcome measures. At the same time, there is an urgent need for a consensus on the definition of the WC measurement site (Wang, Reference Wang2006). In adults WC measurement certainly differs, depending on the chosen landmark and gender, and there is likely to be a similar difference in children (Wang et al. Reference Wang, Thornton, Bari, Williamson, Gallagher and Heymsfield2003; Bigaard et al. Reference Bigaard, Spanggaard, Lykke, Overvad and Tjonneland2005). Furthermore, whilst WC can be considered a simple measurement technique, this very characteristic can result in erroneous measurements. There is a need for health professionals to become proficient in WC measurement through appropriate training, practice and experience. With increasing experience, measurement error can be reduced (Ulijaszek & Kerr, Reference Ulijaszek and Kerr1999) to an acceptable extent, thus allowing comparison between centres to be more reliable and valid, and for individual children to be confidently assessed in the clinical situation.