Screening indices for pediatric obesity

• Published in: Nutrition research (New York, N.Y.) Vol. 19; no. 6; pp. 805 - 815
• Main Authors: Ebbeling, Cara B., Backstrand, Jeffrey R., Rodriguez, Nancy R.
• Format: Journal Article
• Language: English
• Published: New York, NY Elsevier Inc 01.06.1999; Elsevier Science
• Subjects: Assessment; Biological and medical sciences; Body Mass Index; body weight; diagnostic techniques; disease diagnosis; evaluation; height; Medical sciences; Metabolic diseases; methodology; Obesity; Pediatric Obesity; school children; screening; skinfold thickness; Skinfolds; Weight-for-Height; Body Mass Index; Weight-for-Height; Skinfolds; Assessment; Pediatric Obesity; Human; Body mass index; Obesity; School age; Methodology; Anthropometry; Estimation; Nutrition disorder; Corporal biometry; Technique; Child; Nutritional status
• ISSN: 0271-5317; 1879-0739
• DOI: 10.1016/S0271-5317(99)00042-1

Conventional techniques which employ weight-for-height (WH), triceps skinfold (TSF), and body mass index (BMI) for assessing pediatric obesity were compared in the present investigation. Subjects were children aged 6 to 9 years (585 boys, 586 girls) from the NHANESII sample. Criterion-referenced standards consisting of morbidity-related cutoff points of 25% and 30% fat for boys and girls, respectively, were used to define obesity and served as the basic screening criteria for comparison among conventional techniques. The conventional techniques used to identify obese children which employ specified norm-referenced cutoff points were 1) WH ≥ 120 of median, 2) TSF ≥ 85th percentile, 3) WH ≥ 120% of median and TSF ≥ 85th percentile, and 4) BMI ≥ 85th percentile. Sensitivities (Se), specificities (Sp), and positive (V+) and negative (V−) predictive values were calculated with respect to screening criteria for comparison among conventional techniques. To explore alternative cutoff points, receiver operator characteristic (ROC) curves were used to illustrate trade-offs between improving Se or Sp for a series of potential TSF and BMI cutoff points. While the method employing both WH and TSF yielded the highest V+ among the conventional techniques, the necessary exclusion of children from analyses incorporating WH limits the utility of this technique. Classification of subjects according to obesity status using 85th percentiles for TSF and BMI furnished V+ ≤ 0.5. The cutoff points derived from ROC curves with respect to TSF (17 mm for boys; 20 mm for girls) and BMI (19.5 kg/m 2 for both genders) were comparable to values associated with respective 95th percentiles and provided favorable V+. Accurate, yet practical, techniques for identifying children at risk for obesity-related diseases are needed. Cutoff points which correspond to 95 th percentiles should be applied when using a norm-referenced approach. When using a criterion-referenced approach, a BMI of 19.5 kg/m 2 for both genders and TSF values of 17 mm and 20 mm for boys and girls, respectively, are suggested at the present time. Additional research is warranted to validate the clinical utility of these cutoff points.