Obesity in Korean pre-adolescent school children: comparison of various anthropometric measurements based on bioelectrical impedance analysis

Objectives: To examine the relationships between body mass index (BMI), percentage-weight-for-height (PWH) and percentage body fat (PBF), and to compare their validity based on PBF with the BMI criteria of International Obesity Task Force (IOTF) for childhood obesity in Korean pre-adolescent school children. Design: Statistical comparative analysis for anthropometric measures. Subjects: Korean pre-adolescent children (438 boys and 454 girls, aged 8-12 years, mean BMI 19.5 +/- 3.4 kg/m(2)). Measurements: Body mass index and PWH were calculated as body size indices from height and weight. Hand-to-foot bioelectrical impedance analysis (BIA) was performed to obtain PBF values, with obese children defined as PBF above 35%. Sensitivity and specificity were displayed with the gold standard of PBF, and receiver operating characteristic (ROC) curves were used to assess the performance of BMI and PWH in detecting obesity. Results: The prevalence of obesity varied by the criteria: 18.8% by Korean BMI standards for 95 percentile, 11% by the IOTF-BMI 30 kg/m(2) and 29.4% by Korean PWH cutoffs. Body mass index and PWH were significantly correlated with PBF after adjusting for age, 0.910 (P < 0.01) and 0.915 (P < 0.01), respectively. The sensitivity and specificity of the PWH cutoffs were 0.909 and 0.882, respectively. The local BMI 95 percentiles had lower sensitivity (0.714) and specificity (0.790). The IOTF-BMI definition showed much lower sensitivity (0.457) and higher specificity (0.990). The BMI cutoffs corresponding with the highest accuracy were smaller than IOTF-BMI 30 kg/m(2) for all age groups in both boys and girls. Conclusion: The prevalence of obesity among Korean pre-adolescent children widely varied according to the diagnostic criteria applied. Universally recommended cutoffs for children by IOTF showed the lowest sensitivity among the criteria used, and may therefore underestimate obesity in this population.