Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
PAPER
Fat-free mass index and fat mass index percentiles in
Caucasians aged 18 98 y
Y Schutz1*, UUG Kyle2 and C Pichard2
1
Institute of Physiology, Faculty of Medicine, University of Lausanne, Lausanne, Switzerland; and 2Nutrition Division, Geneva
University Hospital, Geneva, Switzerland
OBJECTIVE: To determine reference values for fat-free mass index (FFMI) and fat mass index (FMI) in a large Caucasian group of
apparently healthy subjects, as a function of age and gender and to develop percentile distribution for these two parameters.
DESIGN: Cross-sectional study in which bioelectrical impedance analysis (50 kHz) was measured (using tetrapolar electrodes
and cross-validated formulae by dual-energy X-ray absorptiometry in order to calculate FFMI (fat-free mass=height squared) and
FMI (fat mass=height squared).
SUBJECTS: A total of 5635 apparently healthy adults from a mixed non-randomly selected Caucasian population in Switzerland
(2986 men and 2649 women), varying in age from 24 to 98 y.
RESULTS: The median FFMI (18 34 y) were 18.9 kg=m2 in young males and 15.4 kg=m2 in young females. No difference with
age in males and a modest increase in females were observed. The median FMI was 4.0 kg=m2 in males and 5.5 kg=m2 in
females. From young to elderly age categories, FMI progressively rose by an average of 55% in males and 62% in females,
compared to an increase in body mass index (BMI) of 9 and 19% respectively.
CONCLUSIONS: Reference intervals for FFMI and FMI could be of practical value for the clinical evaluation of a deficit in fat-free
mass with or without excess fat mass (sarcopenic obesity) for a given age category, complementing the classical concept of body
mass index (BMI) in a more qualitative manner. In contrast to BMI, similar reference ranges seems to be utilizable for FFMI with
advancing age, in particular in men.
International Journal of Obesity (2002) 26, 953 960. doi:10.1038=sj.ijo.0802037
Keywords: BMI; fat-free mass; fat mass; body composition; gender; aging
Introduction
In the last decade a dramatic increase in overweight and
obesity has been reported in both developed and underdeveloped countries.1 Associated with this excess of body
weight and body fat, there is an increased risk of developing
heart diseases, diabetes and cancer, so that obesity was
declared a disease more than 15 y ago.2
Prevention of weight gain, which ultimately leads to
obesity, is becoming a priority in public health policy. The
degree of obesity is simply defined in most epidemiological
studies by means of the body mass index (BMI).
Nowadays, BMI has progressively replaced the concept of
ideal body weight since the latter had the drawback of being
954
as excess mass. On the other hand, a deficit of BMI may be
due to a fat-free mass (FFM) deficit (sarcopenia) or a mobilization of adipose tissue or both combined.20
The concept of fat-free mass index (FFMI) and fat mass
index (FMI), in analogy to the BMI but using a two compartment model, merits a reappraisal and appears to be of
interest in the classification of overweight=overfat patients
(respectively underweight=underlean).
The partitioning of BMI into FFMI and FMI is obviously
not possible without associated measurements of body composition. Note that the original idea of calculating the FFM
and fat mass (FM) indexes, in analogy to the BMI, was
proposed several years ago.21 The potential advantage is
that only one component of body weight, ie FFM or FM, is
related to the height squared. Surprisingly, these indexes
have not found a wide application yet, probably because
appropriate reference standards have yet to be defined. By
determining these indexes, quantification of the amount of
excess (or deficit) of FFM, respectively FM, can be calculated
for each individual.
In the present study, we have attempted to describe, in a
large number of apparently healthy Caucasian subjects, the
percentile values for FFMI and FMI as a function of gender
and age.
BMI
(kg=m2) FFMI
18 98 y
18 34 y
35 54 y
55 74 y
75 98 y
Men
Height
Weight
Body mass index
Fat-free mass
Fat mass
Fat mass
n
cm
kg
2
kg=m
kg
kg
%
2986
175.7 7.1
74.2 9.2
24.0 2.7
59.1 5.6
15.0 5.5
19.9 5.4
(150 203)
(47 117.6)
(17.4 40.1)
(38.9 81.6)
(4.4 49.1)
(6.8 42.7)
1088
177.9 6.6
73.3 8.7
23.1 2.3
59.9 5.4
13.4 4.8
17.9 4.7
1323
175.6 6.8**
74.7 9.2**
24.2 2.5**
59.4 5.5*
15.2 5.3**
20.0 5.1**
448
172.4 6.6**
75.1 10.2
25.3 3.1**
57.7 5.5**
17.5 6.2**
22.7 5.6**
127
169.3 7.5**
72.2 8.7*
25.2 3.0
53.6 5.5**
18.6 5.2*
25.4 5.1**
Women
Height
Weight
Body mass index
Fat-free mass
Fat mass
Fat mass
n
cm
kg
kg=m2
kg
kg
%
2649
163.3 6.8
60.0 8.8
22.5 3.3
42.4 4.4
17.6 6.2
28.7 6.4
(139 186)
(37.5 127.4)
(16.0 47.1)
(26.6 67.2)
(4.8 60.2)
(11.6 53.7)
1019
165.9 6.2
58.6 7.5
21.3 2.4
42.7 4.0
15.9 4.7
26.6 5.0
1033
163.5 6.0**
59.6 8.6**
22.3 2.9**
43.0 4.1
16.6 5.7*
27.3 5.7**
386
160.1 6.1**
63.9 10.3**
24.9 3.8**
42.1 4.7**
21.8 6.9**
33.4 6**
211
156.3 6.6**
61.8 10.2**
25.3 4.2
38.8 4.7**
22.9 7.0*
36.4 6.4**
ANOVA comparison between age groups *P < 0.05, **P < 0.001.
a
Range
955
Whole-body resistance (R) was measured with four surface
electrodes placed on the right wrist and ankle, as previously
described.29 Short-term reproducibility of resistance measurements indicates coefficients of variation ranging from
1.8 to 2.9%.26,27
Briefly, the principle was based on the application of an
electrical current of 50 kHz and 0.8 mA produced by a generator (Bio-Z21, Spengler, Paris, France) and applied to the
skin using adhesive electrodes (3M Red Dot T, 3M Health
Care, Borken, Germany) with the subject lying supine.30 The
skin was cleaned with 70% alcohol.
In order to permit inclusion of a large number of subjects,
several BIA instruments, which were cross-validated, were
used. The limit of tolerance between instruments was 5O at
50 kHz using a calibration jig. In vivo comparative measurements were also performed. The Bio-Z21 generators were
cross-validated at 50 kHz against the RJL-1091 and 1011
analyzers (RJL Systems Inc. Clinton Twp, MI, USA) and
against the Xitron 4000B1 analyzer (Xitron Technologies
Inc., San Diego, CA, USA). No substantial difference
(P > 0.05) was found between the Xitron, the Bio-Z instrument and the RJL 101 device. Earthman et al31 have also
reported no significant differences between the Xitron 4000B
and the RJL 101 devices.
All investigations were subsequent to a standardized training in order to minimize errors due to multiple operators.
FFM derived from BIA has been validated previously32
against dual-energy X-ray absorptiometry (DXA; Hologic
QDR-45001 instrument, Hologic Inc. Waltham, MA, USA)
in 343 apparently healthy subjects between 18 and 94 y with
a BMI ranging from 17.0 to 33.8 kg=m2.
The following multiple regression equations were found:
Statistics
The statistical analysis program StatView, version 5.0 (Abacus
Concepts, Berkeley, CA, USA) was used for statistical analysis.
The results are expressed as mean standard deviation
(x s.d.). Age- and sex-specific percentile distributions were
calculated for each of the following parameters: FFMI and
FMI. The data were stratified by steps of 10 y as reported for
BMI and anthropometric data in NHANES study33,34 and
Canada.35 The ith percentile (Pi) was the value at or below
which there was i% of the sample. For example, the 50th
percentile (P50) was the value at or below which there were
50% of the observations for a given variable. Given a total of
n ordered values for each parameter (X1, X2, X3, . . . Xn) the ith
percentile (Pi) in any of the calculated distribution was
computed as follows: Pi (1 7 A)(Xb) (A)(Xb 1), using
the Statview1 4.1 statistical program.
The differences among age groups were analyzed by
analysis of variance (ANOVA) with Fisher protected least
significant difference comparison.
Results
Table 1 presents the anthropometric characteristics of the
men and women. It is apparent that in both genders, the
mean BMI of the apparently healthy elderly individuals
(25.2 kg=m2 in males and 25.3 kg=m2 in females) was higher
as compared with the young individuals, but still on the
borderline of the normal reference range defined in young
individuals.
The results of the FFMI categorized by gender and age are
given in Table 2, where they are distributed into different
percentiles values. In the young male subjects, FFMI was
Table 2
Percentiles values for FFM and FM index in men and women by different age categories
P5
Age (y)
P10
F
P25
P50
P75
P90
P95
17.2
17.6
17.6
16.9
14.1
14.7
14.6
13.7
18.0
18.3
18.4
17.6
14.7
15.3
15.4
14.7
18.9
19.2
19.4
18.5
15.4
15.9
16.2
15.9
19.8
20.1
20.3
19.4
16.2
16.7
17.4
17.0
20.5
21.1
21.1
20.9
17.1
17.5
18.4
18.1
21.1
21.7
22.1
21.2
17.6
18.0
19.0
18.7
2.5
2.9
3.4
4.3
3.9
3.9
5.4
5.6
3.2
3.7
4.3
5.2
4.6
4.8
6.5
7.5
4.0
4.8
5.7
6.4
5.5
5.9
8.3
9.3
5.0
6.0
7.2
7.6
6.6
7.3
10.3
11.4
6.1
7.2
8.4
9.0
7.8
8.8
12.0
13.5
7.0
7.9
9.3
10.1
8.7
9.9
13.5
14.3
3.5
3.4
4.5
4.9
956
around 19 kg=m2 (P50 18.9 kg=m2, range 5th 95th percentile: 16.8 21.1 kg=m2) and did not change significantly in
the higher age category. In young women, FFMI was around
15 kg=m2, ie 20% lower than in males (P50 15.4 kg=m2,
range 5th 95th percentile: 13.8 17.6 kg=m2). FFMI in
women tended to be modestly but significantly higher
(P 0.001) in the advanced age category.
The FMI results are presented in Table 2, where they are
classified into different percentiles: the average FMI for
young men was 4.0 kg=m2 (range 5th 95th percentile:
2.2 7.0 kg=m2) and was higher by about 2 units in the
higher age category. In young women, FMI averaged
5.5 kg=m2 (range 5th 95th percentile: 3.5 8.7 kg=m2) ie
38% higher than in males, with a significantly greater
value (P 0.0001) of almost 4 units in the advanced age
category.
An overview of the effect of age and gender on FFMI and
FMI values is given in Figures 1 and 2.
Table 3 Fat-free mass index (FFMI) and fat mass index (FMI) for the
25th and 75th percentile and various BMIs in apparently healthy
Caucasian adultsa aged 18 98 y
2
Percentiles
BMI (kg=m )
P25
P75
18.5
20.0
25.0
Men
FFMI
FMI
kg=m2
kg=m2
18.2
3.5
20.0
5.9
16.7a
1.8a
17.5a
2.5a
19.8a
5.2a
Women
FFMI
FMI
kg=m2
kg=m2
15.0
4.9
16.6
7.8
14.6a
3.9a
15.1a
4.9a
16.7a
8.3a
a
FFMI and FMI predicted from the following regression equations: for FFMI
prediction, Men y 4.809 0.773* x 7 0.007* x2; r2 0.619, P < 0.001;
women y 7.127 0.459* x 7 0.003* x2; r2 0.606, P < 0.001; where y is
the FFMI (kg) and x is BMI (kg=m2). For FMI prediction: Men
2
2
y 7 4.74 0.222*x 0.007*x ;
r 0.772,
P < 0.001;
women
2
2
y 7 7.12 0.54* x 0.003* x ; r 0.885, P < 0.001; where y is the FMI
2
(kg) and x is BMI (kg=m ).
Discussion
Figure 2 Fat mass index (FMI) men (n 2986) and women (n 2649)
by age category. Mean s.d.
Partition of BMI
Considering that BMI is the sum of FFMI FMI, an increase
(or a decrease) in BMI could be accounted for by a rise (or a
drop) in one component, in the other or in both components. Note that, for a given BMI, if FFMI increases then FMI
should decrease, since, at a constant BMI, there is an inverse
mathematical relationship between the two.
Therefore, the advantage of the combined use of these
indices is that one can judge whether the deficit or excess of
body weight is selectively due to a change in FFM vs FM or
both combined. For example, an individual of 1.85 m and
100 kg has a BMI of 29.2 kg=m2 and would be judged as
largely overweight and even borderline obese. This would
be true if his FMI is higher than the reference values and
conversely if his FFMI is not simultaneously elevated.
957
Expression of FFM
An issue which has plagued nutritionists and body composition specialists is the expression of body composition results
when inter-individual comparison are made: comparison in
absolute value (kg) vs in relative value (ie percentage of body
weight) or normalized value for size (ie typically height
squared in FFMI concept or occasionally adjustment for body
surface area).
Since FFM is related to height, it seems inappropriate to
give, for any individual, a cut-off point of FFM in absolute
value (kg) below which FFM is judged as low. For example, a
short individual would be penalized since his absolute FFM is
expected to be lower than that of a tall individual. Indeed a
healthy and well-nourished young man would have a FFM
expressed in absolute terms virtually the same as that of a
similarly aged but taller individual suffering from proteinenergy malnutrition.21
Bartlett et al36 examined the relationship of FFM-to-height
ratio (not height squared, which is not strictly equivalent to
the FFMI) in 1103 people aged 6 86 y and confirmed the
potential interest of an index related to height. We have
chosen percentile values for evaluating the relative excess
FM vs deficit of FFM. However, expressing the percentage
deficit (or excess) with regard to the mean value (or median
value, P50) would also be possible, in particular when the
subjects are below P5 or above P95.
Effect of aging
As expected, FMI were significantly higher in elderly subjects
as compared to younger ones. Previous results confirmed the
evolution of the index with age found in the present study in
men but not in women, in whom FFM index was found to
be slightly but significantly higher after 60 y of age.36 The
cross-sectional nature of these studies may explain this
discrepancy.
Since body weight increases with aging in industrialized
countries, BMI requires an adjustment for age. This is not the
case with FFMI, since the increase in body weight observed
with aging in industrialized countries allows the net decrease
in FFM accompanying aging to be partly offset. This is
observed even when body weight is constant over the
years. During aging, the weight gain is mostly explained by
a gain in body fat, but this is linked to a slight rise in FFM.
Forbes38 stated that weight gain of about 2 kg per decade
was required to counteract the loss of FFM with aging. This
corresponds to an increase in BMI slightly lower than 1 unit.
During menopause and aging39,40 changes in FFM and FM
are not adequately picked up by changes in BMI since, as
explained above, the two components of BMI (FFMI and
FMI) can vary in divergent directions, the former increasing
while the latter may be decreasing.
It is of interest that FFMI remained relatively constant
with aging, at least in men, so that this does not require an
age adjustment in the reference value, as does BMI.
958
her FMI (and FFMI) by 10%, despite no change in relative
body fat. It is true that the relative drop will be identical to
the loss of body fat expressed in absolute value (10%).
However, the FMI value obtained will allow a more appropriate comparison of the decrease in fatness with other
patients of different heights who have lost the same
amount of weight but have a different initial BMI.
We could take an opposite example during a slimming
program, where BMI may fail to change substantially
because of an associated physical activity program (strength
training), but body fat mass may substantially decrease and
hence FMI as well.
The high sensitivity of FMI (respectively FFMI) to a slight
change of body fat stores (respectively lean tissue mass),
compared to the use of BMI or percentage body fat as factors,
make it an index of potential interest for assessing static and
dynamic nutritional status and energy reserves endpoints.
Acknowledgements
We thank the Foundation Nutrition 2000 Plus for its financial
support. We are indebted to the dietitians at the Geneva
University Hospital for their help.
References
1 Kolata G. Obesity declared a disease. Science 1985; 227: 1019
1020.
2 James WPT, Ralph A. New understanding in obesity research. Proc
Nutr Soc 1999; 58: 385 393.
959
960