Introduction to Human Nutrition

Body Composition 29

In physiological studies where, for example,
energy metabolism has to be corrected for body
composition, a systematic bias in body composition
results may lead to completely wrong conclusions.
When a body composition formula or method sys-
tematically overestimates the FFM in obese subjects,
a “normal” resting metabolic rate in obese subjects
would be underestimated after “correction” for dif-
ferences in body composition (expressed as kJ/kg
FFM). This may lead to the conclusion that obese
subjects have lower metabolic rates, which might
have caused their excess adipose tissue stores. This
argument does not even imply whether the FFM can
be treated as one entity in different populations
instead of different components such as the heart,
liver, or muscles (Gallagher et al., 1998). These con-
siderations are important for comparative studies
between groups (e.g., obese versus nonobese, elderly
versus young or middle-aged), but also for longitudi-
nal studies (e.g., weight-loss studies).
The use of bioelectrical impedance to predict
changes in FFM, TBW, or body fat during weight loss
is questionable. The difference in body water distri-
bution (anatomically as well as intracellular/extracel-
lular) before and after weight loss may be considerable,
leading to a different and systematic bias of the pre-
diction equation. In such a study it may be better to

report changes in impedance values instead of changes

in predicted body composition. Another example is

the very low change in impedance after water removal

in patients with ascites. Despite the fact that a consid-

erable amount of water is removed from the body, the

change in total body impedance is low as the trunk

hardly contributes to total body impedance.

All methods have their limitations, doubly indirect

more than indirect methods. Figure 2.7 shows the

individual errors in body fat percentage from den-

sitometry, DXA, deuterium dilution, a chemical

three-compartment model, skinfold thicknesses, and

bioelectrical impedance in young and elderly women

compared with body fat percentage obtained from a

four-compartment model. It is obvious that errors

can be considerable, both at a group level and espe-

cially at an individual level in the elderly. Many of

these errors can be explained by the violation of

assumptions. This clearly shows that information on

body composition must be used with an awareness of

the limitations of the techniques.

2.8 Perspectives on the future

Given the importance of body composition to evalu-

ate nutritional status and to gain information on

certain disease processes, it can be expected that more

20

10

0

–10

Young females

(22 years)

Elderly females

(72 years)

Bias in body fat percentage (% body weight)

Figure 2.7 Individual differences in mea-

sured body fat percentage compared with a

chemical four-compartment model in 20

young and 20 elderly females using various

techniques. y-axis, BF% from four-compart-

ment model minus: , densitometry (Siri);

, deuterium dilution; , DXA; +, three-

compartment model (Siri); , skinfold thick-

ness; , bioelectrical impedance.