Prediction of body composition in elderly men over 75 years of age

A comprehensive number of body composition predictions (involving weight, height, skinfold thicknesses, bioelectrical impedance and near-infrared interactance-NIRI) were evaluated against total body water (TBW from isotope dilution), in 23 randomly selected men over 75 years old, and dual-energy X-ray absorptiometry (DXA), in 15 volunteers from this group. Comparisons were made between anthropometric and impedance methods for estimating limb muscle mass (obtained using DXA). Bias and 95% limits of agreement between measured TBW and DXA estimates were -2.1 kg and 3.1 kg, respectively (for fat, 5.4% and 6.1% body weight). Agreement between TBW predictions and reference measurements was remarkably variable, irrespective of whether TBW was predicted from TBW-specific equations or indirectly from estimates of fat or fat-free mass: for predictions using anthropometry, bias ranged from -4.7 kg to 1.6 kg and 95% limits of agreement from bias +/- 3.8 kg to +/- 5.0 kg; using impedance, bias was -8.8 kg to 3.2 kg and 95% limits of agreement were bias +/- 3.6 kg to +/- 7.8 kg; corresponding values for NIRI were -5.3 kg and +/- 5.4 kg. Although some non-age-specific equations appeared valid, age-specific equations generally predicted TBW better. Limb muscle mass (DXA) was predicted better using the segmental impedance method, from indices of limb muscle area (r = 0.76; SEE = 1.9 kg) and volume (r = 0.86; SEE = 1.6 kg), than by anthropometry alone (r = 0.61 and 0.71; SEE = 2.3 kg and 2.1 kg, respectively). In conclusion, some body composition predictions are unacceptable (at least for TBW) in older men, and care is recommended when selecting from these methods or equations. Also, the segmental impedance method is as good as, if not better than, anthropometry alone in predicting limb muscle mass (DXA) in older men.