Identification of the obese child: adequacy of the body mass index for clinical practice and epidemiology

Abstract

OBJECTIVE: To assess the ability of simple de®nitions of BMI to successfully screen for children with high body fatness. DESIGN: We determined the sensitivity and speci®city of the body mass index (BMI) by testing its ability to correctly identify children with high body fat percentage. Receiver operator characteristic (ROC) analyses were carried out using the top 5% of body fat percentage to de®ne children as obese (true positives). SUBJECTS: Representative sample of 4175 7 y-old (88 ± 92 month-old) children (2120 boys; 2055 girls) participating in the Avon Longitudinal Study of Pregnancy and Childhood (ALSPAC). RESULTS: The current obesity de®nition based on BMI (95th centile) had moderately high sensitivity (88%) and high speci®city (94%). Sensitivity and speci®city did not differ signi®cantly between boys and girls. The ROC analysis showed that lower cut-offs applied to the BMI improved sensitivity with no marked loss of speci®city: the optimum combination of sensitivity (92%) and speci®city (92%) was at a BMI cut-off equivalent to the 92nd centile. Sensitivity of BMI using the new International Obesity Task Force (IOTF) cut-off for obesity was much lower, and differed signi®cantly (P`0.001) between boys (46%) and girls (72%). CONCLUSIONS: Screening for childhood obesity using the BMI is speci®c, and can have moderately high sensitivity if an appropriate cut-off is chosen. New recommendations based on the IOTF approach to de®ning childhood obesity are associated with lower sensitivity, and sensitivity differs between boys and girls.

Diagnostic accuracy of the BMI for age in paediatrics

“…11 Diagnostic accuracy of national and 'international' obesity definitions Diagnostic accuracy of BMI for age depends on whether it is defined relative to national BMI reference data, or to the 'international' or 'IOTF' definitions of paediatric overweight and obesity. 12 Four published studies [13][14][15][16] have now made a direct comparison of the diagnostic ability of cutoffs based on national reference data (from the US, UK, and Singapore) with those recommended by the IOTF. All four found that the sensitivity of the IOTF obesity definition was much lower than that of definitions based on national reference data (such as BMI X95th percentile).…”

“…Up to 40-50% of the fattest children were not identified as obese by the IOTF approach. [13][14][15][16] These studies also reported that the sensitivity of the IOTF definition was highly sex-specific. In the UK, for example, use of the IOTF approach has much lower sensitivity in boys than girls, 13 leading to the erroneous conclusion that obesity is more prevalent in girls than boys.…”

“…[13][14][15][16] These studies also reported that the sensitivity of the IOTF definition was highly sex-specific. In the UK, for example, use of the IOTF approach has much lower sensitivity in boys than girls, 13 leading to the erroneous conclusion that obesity is more prevalent in girls than boys. 11 Chinn and Rona 17 have shown that this sexdifference in sensitivity results from the IOTF definitions being derived by forcing the BMI curves through 'adult' values of 25 and 30 kg/m 2 at age 18 years.…”

Mexican anthropometric percentiles for ages 10–18

“…It prepared BMI percentiles from databases of six countries with low prevalence of overweight and obtained the percentile corresponding to a BMI of 25 kg/m 2 for different age groups and both genders. In general, compared to other references, WHO's standards produce similar estimates of overweight but lower estimates of obesity (Reilly et al, 2000;Kain et al, 2002;Abrantes et al, 2003). Chilean investigators have developed local percentiles of BMI for age and have proposed the use of percentiles based not on age but on Tanner stage (Burrows and Muzzo, 1999;Burrows, 2001).…”

The development of waist circumference percentiles in British children aged 5.0–16.9 y

“…Estimates of the prevalence of overweight and obesity in population groups are typically based on body mass index (BMI), and BMI centile curves have been developed for use in the paediatric population for clinical and possibly epidemiological purposes (Cole et al, 1995). It has been suggested, however, that BMI may be a less sensitive indicator of fatness amongst children (Reilly et al, 2000), and BMI gives no indication about fat distribution. Until more studies include ethnic groups other than whites however, BMI should be used cautiously in assessing fatness across populations (Dietz & Bellizzi, 1999).…”