Abstract
BACKGROUND: Childhood cancer survivors (CCSs) are at risk for obesity. The purpose of this project was to determine which clinical measures of body composition are most accurate among CCSs in comparison with dual-energy x-ray absorptiometry (DXA). METHODS: The agreement between the body mass index (BMI), skinfold percent body fat, and waist-to-height ratio (WHtR) and DXA was evaluated among 1361 CCSs (mean age, 32.4 ± 7.7 years) 10 or more years after the diagnosis. The sensitivity and specificity of BMI, skinfold, and WHtR obesity classifications were calculated with respect to DXA. Log-binomial regression, stratified by sex, was used to evaluate treatment-related factors for misclassification as nonobese by BMI, skinfolds, and WHtR. RESULTS: The mean body fat values were 23.3% ± 7.7% (males) and 32.3% ± 8.1% (females) for skinfolds and 26.9% ± 7.4% (males) and 38.4% ± 7.7% (females) for DXA. Pearson correlations between skinfolds and DXA were high (R = 0.83 for males, R = 0.84 for females). Skinfolds incorrectly classified 34.5% of obese males and 27.3% of obese females. BMI measures were the least sensitive with false-negative rates of 46.4% (males) and 53.1% (females). Males exposed to abdominal/pelvic radiation were at increased risk for misclassification as nonobese by BMI (relative risk, 1.57; 95% confidence interval, 1.25-1.95). The percentages classified as obese were highest with DXA (males, 63.1%; females, 84.8%) and lowest with BMI (males, 35.7%; females, 39.7%). Although skinfolds and WHtR underestimated the percentage classified as obese in comparison with DXA, the differences were not as large. CONCLUSIONS: Findings suggest that skinfolds and WHtR are better than BMI for obesity classification in CCSs. Clinicians should be aware of the high risk of misclassifying obese CCSs as nonobese.