Skip to main content

Table 3 Evaluation of Fat-Free Mass predictive equations in 103 Dutch obese adolescents, based on bias, RMSE, and percentage accurate prediction, sorted by % accurate prediction

From: Fat-free mass prediction equations for bioelectric impedance analysis compared to dual energy X-ray absorptiometry in obese adolescents: a validation study

REE predictive equation

FFMa

SD

Accurate predictionsb

Under predictionsc

Over predictionsd

Biase

Maximum negative errorf

Maximum positive errorg

RMSEh

Kg/d

%

%

%

%

%

%

kg

FFM-DXA

56.1

9.8

       

Deurenberg’90 [21]

57.1

9.2

68.0

4.9

27.2

2.2

−14.0

12.6

2.8

Gray [22]

55.7

8.3

63.1

18.4

18.4

−0.1

−16.4

12.0

3.2

Kyle [23]

56.4

8.4

61.2

15.5

23.3

1.2

−10.3

17.2

3.1

Lukaski [34]

53.7

8.5

54.4

39.8

5.8

−3.8

−17.3

11.0

3.9

Sun [35]

57.1

9.6

49.5

16.5

34.0

2.2

−11.2

18.8

3.8

Houtkooper [19]

58.9

9.4

44.7

2.9

52.4

5.4

−12.8

19.2

4.0

Deurenberg’91 [18]

58.8

9.3

40.8

2.9

56.3

5.3

−10.7

17.8

3.9

Haroun [10]

59.5

10.9

35.9

7.8

56.3

6.2

−13.0

29.7

5.7

Lazzer [9]

51.2

8.5

28.2

69.9

1.9

−8.3

−23.2

9.0

6.4

Horie

62.5

9.5

17.5

1.0

81.6

12.0

−5.4

29.7

7.4

Wabitsch [8, 36]

49.8

7.6

7.8

92.2

0.0

−10.7

−25.1

3.0

7.1

Schaefer [20]

45.9

6.8

1.9

98.1

0.0

−17.6

−29.3

−0.7

11.0

Suprasongsin [33]

68.0

11.0

1.0

0.0

99.0

21.6

0.1

36.2

12.4

  1. aAs measured
  2. bThe percentage of subjects predicted by this predictive equation within 5 % of the measured value
  3. cThe percentage of subjects predicted by this predictive equation < 5 % of the measured value
  4. dThe percentage of subjects predicted by this predictive equation > 5 % of the measured value
  5. eMean percentage error between predictive equation and the measured value
  6. fThe largest underprediction that was found with this predictive equation as a percentage of the measured value
  7. gThe largest over prediction that was found with this predictive equation as a percentage of the measured value
  8. hRoot mean squared prediction error