id 0.851 respec-
i"
cients, although
HTER (No. 5) for
> substitution of
ula (No. 4), also
zight. For comt with anthropo-
¢
a
4
a In accordance
¥
Q
enificant for all
iy
8
o. 7), based on
groups but was
OsaZs a
aM
higher body fat
il body fat from
at determination
were no error in
W-bodydensity
ody weight [22].
tin kg
tin kg
tinkg
tinkg
juivalents have
uivalents. The
769,
Be
. and age have
The number of
The equations.
eth Lae 2 BIAS hare
TEASURE
sadskc GS Lo es eS
.
Ta Tee
itS Mia MR SG tie!
wheEAa.
2 formulas from
tween predicted
letric equations
surements have
ments and the
as taken these
\opulation, was
up with a wide
heterogeneous-
1305
Reproducibiltty of the anthropometric measurements was determined by comparing results obtained by two examiners. For those measurements entering the
equations, the discrepancy between examiners amounted generally to less than 2 per
cent. However, despite careful attention to technique, certain of the measurements,
Le., thigh length and antero-posterior diameter of thigh and iliac crest, were consistently taken by the two examiners with discrepancies greater than 2 per cent.
Of these, only the thigh length entered cne of the equations.
Analysis of the accuracy of the laboratory measurements has been presented [22].
The combined TBW-body density technique is considered the most accurate in vivo
measurement of body fat: the standard error for fat measurement by this method is
+ 1.7 per cent of total body weight. In addition fewer assumptions of body composi-
tion relationships are necessary with the combined technique than are required for
the estimation of body fat from the single determination of TBW or of body density.
Although K*’ whole-body counting was not used as the reference laboratory
method for measuring body composition, an analysis of the error of this method is
pertinent. The error of the method, as used here, was 9.7 per cent. Others [23]
using an Argonne type whole-body counter have achieved apparently lower errors,
but the inherent error in counting statistics is always at least 2-3 per cent. With
the more favorable geometry of the 4= scintillation counter, the errors related to
height and weight of the subject are lower. Although these relationships may exist
[23], we were unable to find a correction factor for height and weight to apply to the
data presented. An additional source of error for methods using K‘" analysis may
be associated with the use of the results of chemical analysis of potassium available
for only 4 adult human cadavers.
Comparison of calculations of LBW*"° with LBM calculated from results of the
TBW-body density technique gave correlation coefficients of 0.861 for 44 men and
0.798 for 24 women. These correlations, while high, are less than those for four
anthropometric variables with the TBW-body density technique. Correlation of
LBW*"*" with TBW wasslightly higher than with LBM calculated from TBW-—body
density measurements. Since the ratio of muscle to boneis not accounted for in the
TBW-body density technique and little K*° is in bone. TBW measurements with
K*° measurements may be a better indicator of muscle as ANDERSON has suggested
[24]. Our data would indicate, however, that when the estimated essential lipids are
subtracted from LBM the resultant estimates of ‘fat-free body’ differ on the average
only by 1.31 kg for men and by 3.07 kg for women fromsimilar estimates based on
ALLEN’s boneprediction [14] plus LBW*"°.
Damon and GOLDMAN [25] have compared results of 10 anthropometric formulas
predicting body composition to measurements of 13 athletic young men, aged
18-29 years, who were also measured densitometrically. Their work showed best
correlation with formulas using two standard skinfold thicknesses. Unsatisfactory
:
a
d
i
a
#4
4
$
correlations were obtained with other formulas for several possible reasons among
which are selection of reference population, size of reference population or too
small a validating population.
Similar comparisons of seven anthropometric formulas predicting total body
weight, body fat or lean body weight and one estimating K equivalents have been
made with our data. Our results have been similar to those of DAMONin that the
CHINN and ALLEN [16] equation provided poor correlation with laboratory results.
ART cay geen pee epee
Measures of Body Fat and Related Factors in Nermal Adults-- JI