Kelly et al. (NHANES DXA)
About this reference
Kelly and colleagues analysed dual-energy X-ray absorptiometry (DXA) body composition data from NHANES 1999-2004, a nationally representative study of the US population. DXA is considered one of the most accurate methods for measuring body fat percentage in large populations. The percentiles shown here are computed from the LMS curve parameters (median, standard deviation, and skewness) that the paper publishes for percent body fat by age and sex in supplementary Table S2, using the White reference group as the Western comparator.
Known limitations
- Measured by DXA (Hologic scanners, NHANES calibration); DXA, bioelectrical-impedance, and skinfold methods are not directly interchangeable and can differ by several percentage points, so these values should not be compared straight against scale-based or caliper-based body-fat charts
- Percentiles are computed from the paper's published LMS parameters (Table S2, White group), not from raw microdata. Table S2 reports smooth curves evaluated at exact ages in 5-year steps; each bracket carries the value at its youngest age (age 20 for bracket 20-24, age 25 for 25-29, and so on), so figures are age-point values at the bracket start rather than aggregate statistics over the whole bracket
- US population only; Americans tend to have 5-10 percentage points higher body fat than European or Asian populations at the same age and BMI, so these percentiles skew higher than they would in most other countries
- Cross-sectional design means values reflect population averages, not individual changes over time
- NHANES participants are non-institutionalised civilians, excluding certain populations