Page 111 - Go4it

A total of 55 studies were retrieved examining BIA-FFM equations. Our first search provided 24 citations. Another 31 citations were obtained by screening reference lists of relevant publications. Forty-two papers were excluded due to the mean age <11 year (n=16); one gender (n=4); patients (n=6), insufficient information (n=6); specific ethnic group (n=2); small sample size (n<50) (n=1); only based on elderly (>60 y) (n=1); unusual variable (e.g. body density, deuterium dilution) (n=3); non-standard (standard=50kH) method (1MHz) or no hand to foot measurement (n=3). Of the thirteen included studies (see Table 2) we selected the best equation per study based on explained variance in regression analysis. Five equations were based on only children <19 years (8,9,18-20), only 2 equations were based on adolescents in the age range of 10-18 years (9,19). One equation was based on obese adolescents only (9) and 2 studies were based on Dutch adolescents (18,21).
The FFM average for the entire study group measured by DXA was 56.1±9.8 kg. Table 3 provides the FFM data as mean measured FFM in kg, the percentage of accurate under- and overestimation, the percentage bias, the maximum values found for negative error (underestimation) and positive error (overestimation) and the RMSE in kg. The percentage accurate estimations varied between equations from 0 to 68%. The bias for equations varied from
-21.5% to +21.6% and RMSE varied from 2.9 to 13.5 kg. Individual errors were much higher as shown by maximum negative and maximum positive error. Figure 1 shows the percentage of accurate predictions (based on an FFM-BIA within ±5% of FFM-DXA), percentage bias, and RMSE for the total group of adolescents by sex. For the total group of adolescents the Deurenberg’90 equation had the smallest RMSE (2.9 kg), 68% accurate predictions (with 4% underprediction and 28% over-prediction) and a bias of 2.5%. The Gray equation had an RMSE of 3.2 kg, 63% accurate prediction (with 18% underprediction and 18% over prediction) and the smallest bias (-0.1%). The Kyle equation had an RMSE of 3.1 kg, 61% accurate prediction (16% underprediction, 23% over prediction and a bias of 1.2%). When split by sex, the Gray equation had the narrowest range in accurate predictions, bias, and RMSE. Subjects were adolescents and therefore FFM measured at 6 months after baseline had increased by 1.49±2.72 kg as observed by DXA. The best performing BIA equation (the Gray equation) underestimated the FFM change with 0.98±2.90 kg, (p=0.037).
Fat-free mass
111