Receiver operating characteristic (ROC) curve analysis in 5–10-year-old Bengalee girls from Calcutta, India

Abstract

Summary. Objective: The purpose of the present cross-sectional study was to evaluate the sensitivity (SN) and specificity (SP) of body mass index (BMI) and skinfolds thickness: BMI (subcutaneous to overall fat) in detecting excess adiposity in pre-menarcheal Bengalee girls.

Methods: Four hundred and fifty Bengalee girls aged 7.4 ± 1.16 years (mean ± SD; range: 5–10 years) from Calcutta were studied. Anthropometric measures – namely height, weight, circumference of mid upper arm, waist and hip, and skinfold thickness at biceps, triceps, subscapular and suprailiac – were taken from all participants using standard protocols. BMI and the log-transformed sum of four (biceps, triceps, subscapular, suprailiac) skinfolds (log₁₀ SF4) were computed subsequently. Values of log₁₀ SF4 were then converted into seven percentiles category (5th, 10th, 25th, 50th, 75th, 85th and 95th). Excess adiposity was defined as a level of log₁₀ SF4 greater than the internally derived 85th percentile (log₁₀ SF4>85th percentile). SN and SP of each internally derived percentile of BMI and log₁₀ SF4:BMI in detecting excess adiposity were then computed.

Results: SN and SP were 0.49 and 0.94 for the 95th percentile of BMI, and 0.76 and 0.82 for the 95th percentile of log₁₀ SF4:BMI. Moreover, there was a considerable decrease in overall misclassification with the use of log₁₀ SF4:BMI instead of BMI at the 95th percentile (11.2% vs 32%).

Conclusion: Percentiles of BMI in the study had higher SP but low SN in detecting excess adiposity. The use of log₁₀ SF4:BMI, on the other hand, had the merit of increasing SN in a screening programme to evaluate excess adiposity in Bengalee children aged 5–10 years.