Table 3. Linear regression equation comparison of BMI and RFM against FM% obtained from the body composition methods.
| Method | Precision | Accuracy | |||||
|---|---|---|---|---|---|---|---|
| Pearson´s r | p-value | R2 | RMSE (%) | Intercept | p-value | ||
|
DXA (FM%) |
BMI | 0.51 | <0.001 | 0.32 | 7.01 | 10.1 (1.85 to 18.3) | 0.0172 |
| RFM | 0.91 | 0.84 | 3.43 | 1.12 (-2.44 to 4.68) | 0.5328 | ||
|
ADP (FM%) |
BMI | 0.66 | <0.01 | 0.43 | 8.10 | -5.60 (-15.1 to 3.89) | 0.5328 |
| RFM | 0.85 | 0.73 | 5.60 | -9.95 (-15.7 to -4.14) | 0.0011 | ||
|
BIAa (FM%) |
BMI | 0.49 | 0.001 | 0.49 | 8.00 | -7.80 (-17.2 to 1.60) | 0.1019 |
| RFM | 0.82 | 0.82 | 4.69 | -12.6 (-17.5 to -7.74) | <0.0001 | ||
|
4C model (FM%) |
BMI | 0.45 | <0.001 | 0.45 | 8.22 | -7.99 (-17.62 to 1.64) | 0.1020 |
| RFM | 0.90 | 0.81 | 4.85 | -13.63 (-18.6 to -8.60) | <0.0001 | ||
Abbreviations: ADP; air displacement plethysmography; BIA, bioelectrical impedance analysis; BMI, body mass index; DXA; dual- energy X-ray absorptiometry; RFM, relative fat mass; RMSE, root mean squared error; 4C model, 4-compartment model. Regression equation comparisons by body composition methods, was done using a Fisher’s Z transformation for correlation coefficient´s (Pearson´s r). Precision: improvement of precision is given by the significant increase in Pearson’s r, with simultaneous decrease in RMSE %. Accuracy: improvement of accuracy is given by a non-significant difference from the zero intercept of each regression.
aBIA was estimated by bioimpedance prediction equation proposed by Kushner & Schoeller [26].