Using receiver operating characteristic curve analysis in detecting excess adiposity in 9-13-year old South African...

Little is known concerning the applicability of receiver operating characteristic (ROC) curve analysis in detecting excess adiposity in preadolescent South African children. Therefore, the purpose of this study was to evaluate the sensitivity and specificity of body mass index (BMI) and skinfold thickness: BMI (subcutaneous to overall fat) in detecting excess adiposity in preadolescent urban South African school children.

This was a cross-sectional survey of 1136 randomly selected children (548 boys and 588 girls) aged 9-13 years old in urban (Pretoria Central) South Africa. Body mass, stature, skinfolds (subscapular, triceps, supraspinale and biceps) and waist circumference were measured. Receiver operating characteristic curve analysis was used to assess the sensitivity and specificity of BMI, and log10 SF4:BMI to detect excess adiposity. Excess adiposity was defined as levels of log10 SF4 greater than the internally derived 85th percentile (log10 SF4 > 85th percentile).

Compared to log10 SF4:BMI, BMI had a high specificity (0.88; 95% CI 0.84, 0.90). The log10 SF4:BMI identified excess adiposity with a sensitivity and specificity of 0.62 (95% CI 0.60, 0.67) and 0.68 (95% CI 0.64, 0.70), respectively. Besides, a decrease in overall misclassification with the use of log10 SF4:BMI instead of BMI at the 95th percentile (9.7% versus 27.1%) was observed.

Similar to other studies, although with varying degrees, the present study confirms that log10 SF4:BMI at conventional cut-off points has a relatively high sensitivity and specificity in detecting excess adiposity, and therefore could be used to identify the excess adiposity in South African children. As such, defining obesity based on population-specific percentiles rather than using cut-off points derived from other geographical settings with contrasting levels of socio-economic development becomes imperative.