Application of anthropometric indices in childhood obesity.

OBJECTIVES: The purpose of this study was to determine which anthropometric index can best predict childhood obesity and to investigate the interrelationship between these anthropometric indices and metabolic abnormalities.

METHODS: This was a cross-sectional study. In March 2000, 2005 children aged 7 to 8 from the first grade of primary schools in Taichung City in Taiwan were enrolled in this study. Data were obtained on the children's anthropometry, blood pressure, and serum lipid profiles. Body mass index (cutoff points by international age- and sex-specific body mass index [BMI]) and weight-length index (WLI) were measured. The t test, the chi2 test, and stepwise multivariate logistic regression were used. All subjects were divided into four groups: Group 1, nonobese by BMI definition, WLI < 1.2; group 2, obese by BMI definition, WLI < 1.2; group 3, nonobese by BMI definition, WLI > or = 1.2; and group 4, obese by BMI definition, WLI > or = 1.2.

RESULTS: There were 951 girls (47.43%) and 1,054 boys (52.57%). The mean age was 7.3 +/- 0.4 years. The prevalence of obesity was 4.21% in girls and 7.87% in boys using BMI definition, and 12.83% in girls and 14.14% in boys using WLI definition. The prevalence of obesity revealed an increased trend with age in both sexes, whether by BMI or WLI definition. Group 4 had the highest prevalence rate of all metabolic abnormalities, except hypertriglyceridemia. After controlling for age and sex, and with group 1 as a reference, odds ratios were considerably higher in group 3 for the risk of hypertension (2.73; 95% confidence interval [CI], 1.67-4.44) and hypertriglyceridemia (5.83; 95% CI, 1.42-23.95). Odds ratios were considerably higher in group 4 for the risk of hypertension (3.75; 95% CI, 2.31-6.07) and high level of low-density lipoprotein cholesterol (1.93; 95% CI, 1.32-2.82).

CONCLUSIONS: The prevalence of obesity apparently differs depending on the definition of estimation. Childhood obesity exhibits significant correlation with some metabolic abnormalities. We hypothesize that, at present, both indices, BMI and WLI, should be used together to define childhood obesity in clinical practice until a more appropriate and excellent index can be established.