APPLIED NUTRITIONAL INVESTIGATION

Body Composition Interpretation: Contributions of

the Fat-Free Mass Index and the Body Fat Mass Index
Ursula G. Kyle, MS, RD, Yves Schutz, PhD, Yves M. Dupertuis, PhD, and
Claude Pichard, MD, PhD
From the Department of Clinical Nutrition, Geneva University Hospital, Geneva, Switzerland;
and the Institute of Physiology, University of Lausanne, Lausanne, Switzerland

OBJECTIVE: Low and high body mass index (BMI) values have been shown to increase health risks and
mortality and result in variations in fat-free mass (FFM) and body fat mass (BF). Currently, there are no
published ranges for a fat-free mass index (FFMI; kg/m2), a body fat mass index (BFMI; kg/m2), and
percentage of body fat (%BF). The purpose of this population study was to determine predicted FFMI and
BFMI values in subjects with low, normal, overweight, and obese BMI.
METHODS: FFM and BF were determined in 2986 healthy white men and 2649 white women, age 15 to
98 y, by a previously validated 50-kHz bioelectrical impedance analysis equation. FFMI, BFMI, and %BF
were calculated.
RESULTS: FFMI values were 16.7 to 19.8 kg/m2 for men and 14.6 to 16.8 kg/m2 for women within the
normal BMI ranges. BFMI values were 1.8 to 5.2 kg/m2 for men and 3.9 to 8.2 kg/m2 for women within
the normal BMI ranges. BFMI values were 8.3 and 11.8 kg/m2 in men and women, respectively, for obese
BMI (⬎30 kg/m2). Normal ranges for %BF were 13.4 to 21.7 and 24.6 to 33.2 for men and women,
respectively.
CONCLUSION: BMI alone cannot provide information about the respective contribution of FFM or fat
mass to body weight. This study presents FFMI and BFMI values that correspond to low, normal,
overweight, and obese BMIs. FFMI and BFMI provide information about body compartments, regardless
of height. Nutrition 2003;19:597– 604. ©Elsevier Inc. 2003

KEY WORDS: bioelectrical impedance analysis, fat-free mass, body fat, fat-free mass index, fat mass index,
body composition, sex

INTRODUCTION increased likelihood of being functionally obese, not everyone will

Malnutrition and obesity have been shown to increase morbidity
and mortality.1,2 Epidemiologic studies have also shown that low
and high body mass index (BMI) values increase morbidity and
mortality.3,4 Research also has indicated that body composition,
more than BMI, is a primary determinant of health5 and a better
predictor of mortality risk than BMI.6
Over half of the adults in the United States7 are overweight, and
an increasing number of Europeans are becoming overweight.
Obesity is normally defined by BMI. BMI values between 18.5 and
24.9 kg/m2 are considered normal; individuals with BMI values
from 25 to 29.9 kg/m2 are considered overweight and those with
values greater than 30 kg/m2 are considered obese.8 Underweight
subjects (BMI ⬍ 18.5) are also at nutritional risk.9 Thus, preven-
tion of weight gain and low body mass are a public health priority.
Recent studies also have shown that weight and BMI alone are
not adequate measurements of underlying changes in fat-free mass
(FFM) and body fat mass (BF) during menopause,10 aging,11 and
illness in general.12,13
Little is known about what constitutes “normal” FFM and BF.
“Healthy” body fat percentages (%BF) have been suggested to be
in the ranges of 12 to 20 for men and 20 to 30 for women.14 Even
though individuals with %BF above these cutoff values have an
We thank the Foundation Nutrition 2000Plus for its financial support.
Correspondence to: Claude Pichard, MD, PhD, Head, Clinical Nutrition,
Geneva University Hospital, 1211 Geneva, Switzerland. E-mail: claude.
pichard@medecine.unige.ch
be. Higher %BF values have been reported in apparently healthy
populations.15,16 We also found that hospitalized patients are more
likely to have low FFM and low or excess BF,12,17 suggesting
increased health risks in patients.
Until now, it has been customary to use absolute FFM (kg) and
%BF to evaluate nutrition status.18 Because FFM and BF change
with height, weight, and age, it is difficult to determine whether
individual subjects have low or high FFM or BF. The use of FFM
in absolute terms complicates interpretation because FFM is
closely related to height and decreases with age. The use of
percentage of FFM does not adequately reflect nutrition status
because FFM automatically decreases in proportion to %BF in-
creases. The colinearity between the two parameters limits the
interpretation of absolute FFM and relative changes of FFM.
Variations of FFM and BF with age and height also make it
difficult to establish desirable ranges.
Just as BMI is useful in evaluating the excess or deficit in body
weight of individuals of different heights, the fat-free mass index
(FFMI; kg/m2) and body fat mass index (BFMI; kg/m2) are po-
tentially useful in evaluating body composition parameters by
effectively eliminating differences in FFM and BF associated with
height.19 Height-independent body composition parameters (FFMI
and BFMI) allow height-independent interpretation of nutrition
status, comparisons of results between studies, and the develop-
ment of body composition percentile tables.
The use of FFMI and BFMI allows the tracking of the effects
of illness, treatment, or aging in individuals and in groups because
they can be interpreted as absolute values or by classifying indi-
viduals or groups of individuals as normal or abnormal (low and
high). FFMI also permits one to determine whether lower FFM

Nutrition 19:597– 604, 2003 0899-9007/03/$30.00

©Elsevier Inc., 2003. Printed in the United States. All rights reserved. doi:10.1016/S0899-9007(03)00061-3
598 Kyle et al. Nutrition Volume 19, Numbers 7/8, 2003

TABLE I.

ANTHROPOMETRIC AND BODY COMPOSITION DATA IN HEALTHY CAUCASIAN ADULTS

Age (y) 18–98 18–39 40–59 60⫹

Men (n) 2982 1487 1098 397

Height (cm) 175.7 ⫾ 7.1 177.7 ⫾ 6.7 174.7 ⫾ 6.6† 171.2 ⫾ 7.0§
Weight (kg) 74.1 ⫾ 9.1 73.7 ⫾ 8.9 74.8 ⫾ 9.1* 74.0 ⫾ 9.7
Body mass index (kg/m2) 24.0 ⫾ 2.6 23.3 ⫾ 2.4 24.5 ⫾ 2.6† 25.2 ⫾ 2.9§
Fat-free mass (kg) 59.1 ⫾ 5.6 59.9 ⫾ 5.5 59.1 ⫾ 5.4† 56.0 ⫾ 5.7§
Fat-free mass index (kg/m2) 19.1 ⫾ 1.4 19.0 ⫾ 1.3 19.4 ⫾ 1.4† 19.1 ⫾ 1.6§
Body fat (kg) 15.0 ⫾ 5.5 13.7 ⫾ 4.9 15.6 ⫾ 5.5† 18.0 ⫾ 5.7§
Body fat (%) 19.8 ⫾ 5.4 18.3 ⫾ 4.8 20.5 ⫾ 5.3† 24.0 ⫾ 5.3§
Body fat mass index (kg/m2) 4.9 ⫾ 1.8 4.3 ⫾ 1.5 5.1 ⫾ 1.7† 6.2 ⫾ 1.9§
Women (n) 2647 1325 843 479
Height (cm) 163.3 ⫾ 6.8 165.6 ⫾ 6.2 162.8 ⫾ 5.9† 158.1 ⫾ 6.5§
Weight (kg) 60.0 ⫾ 8.7 58.5 ⫾ 7.5 60.5 ⫾ 8.7† 63.2 ⫾ 10.5§
Body mass index (kg/m2) 22.5 ⫾ 3.3 21.3 ⫾ 2.4 22.8 ⫾ 3.0† 25.3 ⫾ 4.0§
Fat-free mass (kg) 42.4 ⫾ 4.3 42.7 ⫾ 4.0 43.0 ⫾ 4.2 40.5 ⫾ 5.0§
Fat-free mass index (kg/m2) 15.9 ⫾ 1.3 15.6 ⫾ 1.1 16.2 ⫾ 1.3† 16.2 ⫾ 1.7
Body fat (kg) 17.6 ⫾ 6.1 15.8 ⫾ 4.7 17.5 ⫾ 5.8† 22.7 ⫾ 7.0§
Body fat (%) 28.7 ⫾ 6.4 26.5 ⫾ 5.1 28.3 ⫾ 5.8† 35.2 ⫾ 6.3§
Body fat mass index (kg/m2) 6.6 ⫾ 2.4 5.7 ⫾ 1.7 6.6 ⫾ 2.2† 9.1 ⫾ 2.9§

*P ⬍ 0.05, 18 –39 versus 40 –59, analysis of variance.

†P ⬍ 0.001, 40 –59 versus 60⫹, analysis of variance.
‡P ⬍ 0.05.
§P ⬍ 0.001.

with age is due to shorter height in older subjects or to changes in
body composition. FFMI and BFMI also identify individuals with
excess muscle mass but without excess BF. These changes are of
interest at the group or population level.
The purpose of this population study was to determine FFMI,
BFMI, and %BF ranges in subjects with low, normal, overweight,
and obese BMI. Our study presents FFMI and BFMI in 5629
healthy men and women across the age span, from 15 to 98 y. The
ranges of FFMI and BFMI for a given BMI category can be used
to evaluate whether FFM is low and BF is low or high in adults.
MATERIALS AND METHODS
Subjects
Healthy adults (2982 men and 2647 women), age 18 to 98 y, were
non-randomly recruited through advertisements in local newspa-
pers with an offer of free bioelectrical impedance analyses (BIAs),
at exhibitions at trade fairs and fun runs, among public adminis-
tration staffs, and by invitations sent to leisure clubs for the
elderly. The anthropometric and BIA data and the number of
healthy subjects are shown in Table I. All subjects were ambula-
tory whites (Western European) without known pathologies or
physical handicap. Subjects were questioned on their use of med-
ications and reasons for visits to their physicians within the past 6
mo to eliminate subjects with acute or chronic diseases. Subjects
with water or electrolyte imbalances (edema or ascites), skin
abnormalities (e.g., pachydermia secondary to hypothyroidism),
and abnormal body geometry (e.g., amputation, limb atrophy, or
BMI ⬎ 38 kg/m2) that might interfere with BIA measurements
were excluded.
The study to validate BIA methods in healthy subjects was
approved by the Geneva University Hospital Ethics Committee.
All subjects volunteered for the study.
Anthropometric Measurements and BIA
Body height was measured to the nearest 0.5 cm and body weight
to the nearest 0.1 kg on a balance beam scale. Elderly subjects who
were not able to stand or whose height could not be measured were
excluded. Subjects wore indoor clothing without shoes, heavy
sweaters, or jackets. One kilogram was deducted for pants and
shirts.
Whole-body resistance and reactance were measured with four
surface electrodes placed on the right wrist and ankle, as previ-
ously described.20 Briefly, an electrical current of 50 kHz and 0.8
mA was produced by a generator (Bio-Z2, Spengler, Paris, France)
and applied to the skin with adhesive electrodes (3M Red Dot T,
3M Health Care, Borken, Germany) with the subject lying su-
pine.21 The skin was cleaned with 70% alcohol. The Bio-Z2
generators were cross-validated at 50 kHz against the RJL-109 and
101 analyzers (RJL Systems, Clinton Township, MI, USA) and
against the Xitron 4000B analyzer (Xitron Technologies, San
Diego, CA, USA). No difference (P ⬎ 0.05) was found for
resistance at 50 kHz across the Xitron, Bio-Z2, and RJL 101
devices. Earthman et al.22 reported no significant differences be-
tween the Xitron 4000B and RJL 101 devices.
FFM was calculated by the following multiple regression equa-
tion on a Xitron 4000B analyzer (Xitron Technologies) that had
been previously validated23 against dual-energy x-ray absorptiom-
etry (DEXA; Hologic QDR-4500 instrument, Hologic, Waltham,
MA, USA) in 343 healthy subjects between 18 and 94 y with BMIs
between 17.0 and 33.8 kg/m2:
FFM ⫽ ⫺4.104 ⫹ (0.518 ⫻ height2/resistance) ⫹ (0.231
⫻ weight) ⫹ (0.130 ⫻ reactance) ⫹ (4.229
⫻ sex [men ⫽ 1, women ⫽ 0])
DEXA-measured FFM was 54.0 ⫾ 10.7 kg (mean ⫾ standard
deviation). BIA-predicted FFM was 54.0 ⫾ 10.5 kg (bias, 0.03 ⫾
Nutrition Volume 19, Numbers 7/8, 2003
1.7 kg; r ⫽ 0.986; standard error of the estimate [SEE] ⫽ 1.72 kg;
technical error, 1.74 kg; coefficient of variance, 3.6%). The equa-
tion also was validated in elderly subjects.24 We also validated the
Bio-Z2 in 250 of the 343 subjects included in the above study
(unpublished data) with the above BIA equation and found that
FFM as measured by DEXA was 54.2 ⫾ 10.7 kg and that as
measured by Bio-Z2 was 54.1 ⫾ 10.5; the mean difference be-
tween DEXA and the Bio-Z2 was 0.13 ⫾ 1.7 kg (r ⫽ 0.99, SEE ⫽
1.6 kg; P ⫽ 0.22, unpaired t test), which was not different from the
bias between DEXA and the Xitron.
Short- and long-term reproducibilities of resistance measure-
ments produced coefficients of variance of 1.8% to 2.9%.25,26 In
our data, reproducibilities were 0.999 (r) for measurements taken
in the same subject within 1 wk (n ⫽ 29) and 0.977 for repeat
measurements within 1 mo (n ⫽ 40), or a 2.5% variance.
The FFMI and BFMI were derived as FFM (kg) and BF (kg),
respectively, and divided by height (m) squared (kg/m2).
Statistics
The statistical analysis program StatView 5 (Abacus Concepts,
Berkeley, CA, USA) was used for statistical analysis. The results
are expressed as mean ⫾ standard deviation.
The differences between age groups were analyzed by analysis
of variance with Fisher’s protected least-significant difference
comparison. Statistical significance was set at P ⬍ 0.05 for all
tests.
Polynomial regressions were used to determine FFMI, BFMI,
and %BF for each of the following BMIs: 18.5, 20, 25, 27.8, 27.3
(for men and women), and 30 kg/m2. Goodness of fit was verified.
Lines were plotted on the regression graph to show the values for
low, normal, overweight, and obese BMI values on the x axis and
low, normal, high, and very high FFMI, BFMI, and %BF values on
the y axis. Thus, measured FFMI, BFMI, and %BF values falling
below the values for a BMI of 18.5 kg/m2 were defined as low;
measured FFMI, BFMI, and %BF values falling in the range for
BMIs between 18.5 and 25 kg/m2 were considered normal, and
values above that range were considered high and very high.
RESULTS
Table I shows the anthropometric and body composition measure-
ments of healthy men and women between 18 and 98 y. Height
was highest in the youngest men and women and progressively
decreased thereafter. Weight was progressively greater in older
than in younger subjects. BMI (Table I) was progressively greater
in older than in younger men and women.
Effects of Age on FFMI and BMI
FFM was lower in older men and women. Conversely, FFMI was
highest in men 40 to 59 y old and women older than 40 y in
comparison with younger subjects (Table I). FFMI was signifi-
cantly higher in women older than 60 y than in women 20 to 39 y
and not significantly different between men 20 to 39 y and older
than 60 y and women 40 to 59 y and older than 60 y. The relation
between FFMI and age was curvilinear, with the highest predicted
values in the age category of 30 to 59 y for men and women and
lower values noted in younger and older subjects (Fig. 1). Thirteen
percent of men and 25% of women 60 y and older were in the low
FFMI range (defined in Table II and Fig. 2) compared with 11%
and 28% of men and women, respectively, 60 y and older. Thus,
the small differences in FFMI (0.4 kg/m2 in men and 0.6 kg/m2 in
women) noted between older and younger subjects did not appear
to invalidate the use of the same values for all subjects, including
elderly subjects.
Fat-Free and Body Fat Mass Indices in Healthy Adults 599
Fat mass, BFMI, and %BF were progressively and significantly
higher in older men and women (Table I and Fig. 1). BMI was
higher in older subjects due to higher BFMI.
FFMI and BMI
Figure 2 shows a significant relation between BMI and FFMI for
men (top) and women (bottom). Overall, the sensitivity was 69.0%
and specificity was 70.6% for FFMI versus BMI. Few men in this
study were in the low BMI, low FFMI category. Most subjects (n
⫽ 3533) with normal BMI had normal FFMI (n ⫽ 2502, 71%); the
remaining 29% of subjects had low or high FFMI. Incidence of
low FFMI was more predominant in women (n ⫽ 728, 27.4%)
than in men (n ⫽ 331, 11.1%). Only three men and two women
who were considered obese were in the normal FFMI range, and
none was in the low FFMI range. Thus, higher BMIs resulted not
only in higher BFMI but also in higher FFMI in healthy subjects,
regardless of age. Figure 2 and Table II show that FFMI for
subjects in the normal BMI ranges were 16.7 to 19.8 kg/m2 for
men and 14.6 to 16.8 kg/m2 for women (Table II).
BFMI, %BF, and BMI
Figure 3 shows the significant positive relation between BMI and
BFMI, and Figure 4 shows the same between BMI and %BF, i.e.,
the higher the BMI, the higher the BFMI and %BF. Overall, for
BFMI versus BMI, sensitivity was 77.4% and specificity was
84.0%. Subjects with normal BMIs generally were in the normal
FMI category, and overweight subjects were in the high BFMI
category. For example, 77% (n ⫽ 1498) of men and 79.6% (n ⫽
1263) of women with normal BMIs were in the normal BFMI
category, and 77.4% of men (n ⫽ 658) and 76.8% of women (n ⫽
312) with overweight BMI were in the high BFMI category. Based
on Figure 3, the “normal” or “desirable” BFMI values (Table II)
for subjects in the BMI range of 18.5 and 24.9 kg/m2 were 1.8 to
5.2 kg/m2 for men and 3.9 to 8.2 kg/m2 for women. “High” BFMIs
(Table II) for subjects in the overweight BMI range (25 to 29.9
kg/m2) were 5.2 to 8.3 kg/m2 for men and 8.2 to 11.8 kg/m2 for
women. BFMIs exceeding these limits would be considered very
high and would correspond to BMIs in the obese range.
The “normal” %BF values (Table II and Fig. 4) for subjects in
the BMI range between 18.5 and 24.9 kg/m2 were 10.8% to 21.6%
for men and 21.7% to 33.1% for women. “High” %BF values
(Table II and Fig. 4) for subjects in the overweight BMI range (25
to 29.9 kg/m2) were 21.7% to 28.7% for men and 33.2% to 39.9%
for women, and values exceeding these limits would be considered
very high and would correspond to obese BMIs. Further, Figure 5
shows that the %BF in the normal BFMI ranges of 1.8 to 5.2 kg/m2
in men and 3.9 to 8.2 kg/m2 in women corresponded to %BF
values of 9.0 to 21.2 and 20.6 to 33.7, respectively. The %BF
determined from BFMI was very similar to the %BF determined
from BMI (Table II).
Table III shows the comparison between %BF for BMIs of
18.5, 25, and 30 kg/m2 as determined by Gallagher et al.27 in white
men and women of various age groups and %BF values found in
our study. The %BF ranges in our subjects were 1% to 2% lower
in men and 2% to 3% lower in women older than 40 y with a BMIs
of 25 and 30 kg/m2, respectively.
DISCUSSION
BMI does not separate body compartments into FFM and BF.
Because research has indicated that body composition is a primary
determinant of health,5 FFM and BF compartments should be
determined as part of a health assessment. FFM and BF change
with height, weight, and age. It is therefore difficult to determine
whether individual subjects have low or high FFM or BF. FFMI
600 Kyle et al. Nutrition Volume 19, Numbers 7/8, 2003

FIG. 1. FFMI (top) and BFMI (bottom) regressed against age in men (left; FFMI: r ⫽ 0.09, P ⬍ 0.0001, BFMI: r ⫽ 0.385, P ⬍ 0.0001) and women (right;
FFMI: r ⫽ 0.199, P ⬍ 0.0001, BFMI: r ⫽ 0.501, P ⬍ 0.0001). BFMI, body fat mass index; FFMI, fat-free mass index.

and BFMI eliminate differences in FFM and BF due to height and TABLE II.
offer the advantage of having one set of recommended ranges,
regardless of age and height. FFMI and BFMI have been reported FFMI, BFMI, AND %BF VALUES FOR VARIOUS BMI VALUES IN
in studies with small numbers of healthy subjects6,28 and HEALTHY CAUCASIAN ADULTS*
patients.29 –31 We recently published percentiles for FFMI and
BFMI for healthy adults.32 However, these studies have not eval- BMI FFMI BFMI
uated the FFMI and BFMI ranges for various BMI classifications. (kg/m2) (kg/m2) (kg/m2) %BF %BF†
Our current study presents FFMI, BFMI, and %BF values for low,
normal, overweight, and obese BMIs. Men (n ⫽ 2982)
30.0 21.7 8.3 28.8 29.1
27.8 20.9 6.9 25.8 25.9
Effect of Aging on FFMI and BFMI
25.0 19.8 5.2 21.7 21.3
Mean FFMI was higher in men and women older than 60 y than in 20.0 17.5 2.5 13.4 12.0
those 18 to 39 y (Table I). The relation between FFMI and age was 18.5 16.7 1.8 10.8 9.0
curvilinear, with the highest predicted values in the age category of Women (n ⫽ 2647)
30 to 59 y and lower values noted in younger and older subjects 30.0 18.2 11.8 40.0 40.8
(Fig. 1). Forbes33 proposed that a 2.3 kg/decade weight gain (or 0.9 27.3 17.5 9.8 36.5 37.4
kg/m2 BMI increase/decade) is required to counteract the loss of 25.0 16.8 8.2 33.2 33.8
FFM with aging. Our study suggested that mean BMI increases of 20.0 15.1 4.9 24.6 24.0
1.9 kg/m2 in men and 4.0 kg/m2 in women older than 60 y is 18.5 14.6 3.9 21.7 20.6
sufficient to maintain FFMI at or above the levels of those 18 to
39 y. Thus, small BMI increases would prevent FFMI from de- * FFMI, BFMI, and %BF were determined from regression equations
creasing in older subjects. The differences noted between the study shown in Figures 2 to 5.
by Forbes and the current study are likely due to methodologic † %BF predicted from BFMI ranges determined in column 2.
differences (n ⫽ 75 subjects evaluated longitudinally by Forbes %BF, percentage of body fat; BFMI, body fat mass index; BMI, body
versus n ⫽ 5629 subjects evaluated cross-sectionally in our study). mass index; FFMI, fat-free mass index
Nutrition Volume 19, Numbers 7/8, 2003
FIG. 2. FFMI versus BMI for men (top) and women (bottom). Polynomial
relations were observed between FFMI and BMI for men and women (all
P ⬍ 0.001). Predicted values for low, normal, high, and very high FFMI
were plotted on the y axis for low, normal, overweight, and obese BMI
(⬍18.5, 18.5–24.9, 24.9 –29.9, and ⬎30 kg/m2). BMI, body mass index;
FFMI, fat-free mass index; SEE, standard error of the estimate.
Large longitudinal studies will be necessary to determine whether
an increase in weight or BMI is necessary to counteract the
age-related decrease in FFMI. Schutz et al.32 found that the
effects of aging are noticeable only in adults older than 75 y and
that the 25th and 75th percentiles of FFMI are lower in men
older than 75 y than in men 18 to 34 y, whereas the same was
not found in women. Because FFMI remained constant with
aging, an adjustment in FFMI reference values does not appear
to be necessary.
BFMI was progressively higher in older subjects and explained
100% of the higher BMI in men and 85% of the higher BMI in
women. We speculate that 100% of BFMI in men can be explained
by an effect of BMI and BF increasing and FFM remaining stable
with age (and therefore proportionally less of total weight),
whereas BMI, FFMI (although small), and BFMI increase with age
in women.
Fat-Free Mass Index
Greater weights and BMIs lead to higher FFMIs and BFMIs (Figs.
2 and 3). This is expected because, mathematically, a high BMI is
accounted for by a high FFMI, a high BFMI, or both.
Normal FFMIs were 16.7 to 19.8 kg/m2 in men and 14.6 to 16.8
kg/m2 in women for normal BMI (Table II). This result compares
with 18.2 and 20.0 kg/m2 for men and 15.0 and 16.6 kg/m2 for
women in the 25th and 75th percentiles, respectively, reported by
our group32 for subjects with mean BMIs in the normal range (24.0
⫾ 2.7 kg/m2 in men and 22.5 ⫾ 3.3 kg/m2 in women). Thus,
Fat-Free and Body Fat Mass Indices in Healthy Adults 601
FIG. 3. BFMI versus BMI for men (top) and women (bottom). Polynomial
relations were observed between BFMI and BMI for men and women (all
P ⬍ 0.001). Predicted values for low, normal, high, and excess BFMI are
plotted on the y axis for low, normal, overweight, and obese BMI (⬍18.5,
18.5–24.9, 24.9 –29.9, and ⬎30 kg/m2). BFMI, body fat mass index; BMI,
body mass index; SEE, standard error of the estimate.
healthy subjects with BMIs in the normal range were in the normal
FFMI range and in the 25th to 27th percentile range. Van Itallie et
al.,6 using TOBEC, total body electrical conductivity measure-
ments, reported an FFMI of 16.8 kg/m2 for men in the 5th per-
centile in their small study (n ⫽ 124), which is similar to our
results. Upper limits of FFMI are not of interest because high
levels of FFM, assuming normal hydration, do not result in neg-
ative health effects. However, low levels of FFM and FFMI are
important markers of nutrition status in healthy and ill subjects.
Van Itallie et al.6 found that FFMI and BFMI, in addition to the
measurement of oxygen consumption, could be used to diag-
nose protein-energy malnutrition in 27 of 32 subjects after 12
wk of semistarvation. Subjects in the lowest quintile of the FFM
percentage were more likely to report overall functional dis-
ability than were those in the highest quintile.34 In lung trans-
plantation patients, Kyle et al.35 reported that 76.2% and 85.7%
of men and 28% and 38% of women had low FFMIs at 1 mo and
2 y after lung transplantation, respectively. We also found that
low FFMI at hospital admission is associated with increased
length of stay.17 These results suggested that FFMI is useful in
determining nutritional risk. Low FFM also has been shown to
correlate with increased mortality.4 Further research is neces-
sary to determine the level of FFMI associated with increased
health risks.
Body Fat Mass Index
The BFMI values were 1.8 to 5.2 kg/m2 in men and 3.9 to 8.2
kg/m2 in women for the normal BMI ranges (Table II). BFMI
602 Kyle et al.
FIG. 4. %BF versus BMI for men (top) and women (bottom). Polynomial
relations were observed between %BF and BMI for men and women (all P
⬍ 0.001). Predicted values for low, normal, high, and very high %BF are
plotted on the y axis for low, normal, overweight, and obese BMI (⬍18.5,
18.5–24.9, 24.9 –29.9, and ⬎30 kg/m2). %BF, percentage of body fat;
BMI, body mass index; SEE, standard error of the estimate.
32
increased linearly with BMI. Schutz et al. reported a BFMIs of
3.5 to 5.9 kg/m2 in men and 4.9 to 7.8 kg/m2 in women for the 25th
and 75th percentiles, respectively. The 5th percentile for BFMI
reported by Van Itallie et al.6 was 2.4 kg/m2, which is slightly
higher than the 5th percentile reported by Schutz et al. (2.2 kg/m2).
The 85th percentile for BFMI reported by Van Itallie et al.6 was
6.8 kg/m2 in men 20 to 39 y compared with 6.6 and 7.8 kg/m2 for
the 75th and 90th percentiles reported in our study. The differences
between the studies by Van Itallie et al. and Schutz et al. would be
expected because of the small number of subjects (n ⫽ 124) used
in the study by Van Itallie et al.
Although misclassification of relative BF is noted for all BMI
categories, it appears to be more common in overweight or obese
subjects. Indeed, 15% of healthy overweight subjects were in the
normal BFMI range and 8.5% were in the very high range. We also
found that 36% and 15% of patients at hospital admission had high
and very high BFMIs, respectively, compared with 26% and 8% of
patients with overweight and obese BMI, respectively; and 34%
and 65% of lung transplantation patients had high and very high
BFMIs compared with 17% and 42% who were overweight or
obese at 2 mo and 2 y after transplantation, respectively.35 Thus,
determination of high and very high BFMI or BF can help evaluate
nutrition status in healthy or ill subjects. High or very high BFMI
or %BF is more likely in subjects with recent weight gain or in
inactive subjects because of the greater gain in BF than in FFM
when there is an imbalance between energy intake and
expenditure.
Nutrition Volume 19, Numbers 7/8, 2003
FIG. 5. %BF versus BFMI for men (top) and women (bottom). Polynomial
relations were observed between %BF and BMI for men and women (all P
⬍ 0.001). Predicted values for low, normal, high, and very high %BF are
plotted on the y axis for low, normal, high, and very BFMI. %BF,
percentage of body fat; BMI, body mass index; SEE, standard error of the
estimate.
Percentage of Fat Mass
Normal ranges for %BF in our study were 13.4 to 21.7 for men and
24.6 to 33.2 for women. Significant weight gains noted in recent
years in North American36 and European populations37 are respon-
sible for large numbers of subjects being above the suggested %BF
ranges of 12 to 20 for men and 20 to 30 for women.14 Forty-five
percent of all men and 38% of all women in a recent study
conducted by us had values above these “desirable” levels.16
The %BF determined from BFMI at various BMI levels (Table
II) differed only slightly from %BF values found at the same BMI
level. Excess fat storage expressed as %BF trails off with increas-
ing BMI (Fig. 4), but this is not the case for BFMI (Fig. 3), because
at the same %BF, an increase in body weight per se generates a
greater absolute BF. However, low and high BFMI levels corre-
spond to low and high %BF levels.
Table III clearly shows that our FM ranges are very similar to
those reported by Gallagher et al.27 for the same BMI, despite
different body composition methods: BIA in the present study and
DEXA, doubly labeled water, and underwater weighing in Gal-
lagher et al.27 It is not known at this time whether the differences
in %BF were due to our older subjects being leaner for the same
BMI than subjects in the United States and United Kingdom or to
differences in measuring body composition. Further, there was
very good correlation between %BF and BFMI (r2 ⫽ 0.96 in men
and 0.95 in women; Fig. 5), which suggests that our BIA mea-
surements fairly reflected actual BF and BFMI ranges.
High BF mass appears to be associated with functional disabil-
ity and mortality. Subjects in the highest quintile of %BF (23.8 to
Nutrition Volume 19, Numbers 7/8, 2003
TABLE III.
COMPARISON OF PREDICTED VALUES FOR PERCENTAGE OF BODY FAT IN HEALTHY CAUCASIAN ADULTS IN
GALLAGHER ET AL.27 AND THE CURRENT STUDY
Men
Body mass
index (kg/m2) Study 20–39 y 40–59 y
30.0 Gallagher et al. 26 29
Current 27.7 27.7
25.0 Gallagher et al. 21 23
Current 20.9 20.9
18.5 Gallagher et al. 8 11
Current 11.1 11.1
46.0 in men and 33.1 to 50.0 in women) were more likely to report
overall functional disability than were those in the lowest quin-
tile.34 Visser et al.38 found high levels of disability in subjects with
large fat mass (⬎32.0% in men and ⬎ 43.7% in women). These
%BF levels would correspond to BMIs in excess of 30 kg/m2.
Only 3.3% of our subjects had BMIs greater than 30 kg/m2, and
none was morbidly obese. Lee et al.39 found that obese men have
a higher relative risk of all-cause and cardiovascular disease mor-
tality than do lean fit men. However, longitudinal studies are
necessary to determine the degree of fatness that poses a real risk
in terms of increased morbidity and mortality.
Interpretation of FFMI and BFMI
The purpose of using FFMI and BFMI is to facilitate interpretation
of body composition parameters regardless of height. FFMI and
BFMI values corresponding to low, normal, high, and very high
BMI categories allow the classification of subjects in low, normal,
high, and very high FFMI and FMI categories.
Van Itallie et al.6 found that FFMI and BFMI are more accurate
than FFM and BF as indicators of nutrition status. As protein-
calorie malnutrition became more severe in the Minnesota sub-
jects, they found that FFM as a percentage of body weight in-
creases, because BF decreased more rapidly than did FFMI and
thus underestimated the decrease in FFM. BF as a percentage of
body weight decreased as expected during semistarvation, but the
extent of the decrease was underestimated because of a concurrent
decline in FFM. Because height explained 45% of the variance in
FFM but only 2% of variance in BF, Van Itallie6 found that values
for FFM and BF unadjusted for height are inadequate indicators of
nutrition status. However, FFMI and BFMI have the advantage of
being independent of height. FFMI and BFMI are therefore po-
tentially useful in nutrition assessment.
It is not necessary to determine FFMI and BFMI in individual
subjects, because the sum of FFMI and BFMI is equal to BMI.
However, this study presented results for both compartments,
because FFMI and BFMI are of interest.
Study Limitations
Body height has steadily increased as a secular process, which
complicates the comparison of subjects of different ages. We
compensated for this by using FFMI and BFMI, which effectively
eliminated differences in FFM and fat mass associated with greater
height in younger adults.
Because body composition changes during aging, a validated
BIA equation (against DEXA) in adults ages 18 to 94 y23 and
further validated in elderly subjects24 was used. The subjects in the
present study were not randomly selected. However, the aim of
this study was not to provide population ranges for FFMI and
Fat-Free and Body Fat Mass Indices in Healthy Adults 603
Women
60–79 y 20–39 y 40–59 y 60–79 y
31 39 41 43
29.9 39.1 39.1 40.4
25 33 35 38
23.8 32.5 32.5 34.9
13 21 23 25
12.7 22 22.1 24.6
BFMI, as might be the goal of epidemiologic studies. The subjects
were volunteers in good health, but they might not be representa-
tive of the general population. We excluded subjects with weight
changes and those who had acute or chronic illness. Regular
physical activity (walking) and the absence of mobility problems
may have aided in maintaining physical function and may have
limited FFM loss and BF gain in subjects older than 60 y.
The BIA methods used in this study may be criticized. How-
ever, validation of the BIA equation used in this study against
DEXA in the same population permitted the determination of FFM
and FM.23 Special care was taken to avoid invalid measurements
due to hydration or electrolyte abnormalities.
The FFMI and BFMI ratios might differ to the same extent that
FFM and FM might differ when different methods are used. The
underlying problem of different results obtained by various body
composition methods and measurements remain and need to be
resolved. FFMI and BFMI could also be derived from skinfold
measurement, but this method is more vulnerable to interobserver
imprecision than BIA is.
Differences in FFM and BF have been reported across DEXA
instruments by different manufacturers, with lower BF values
reported by the Hologic QDR-4500 than by the Lunar Prodigy.
These values might explain the slightly lower %BF levels for the
same BMI reported in our study compared with the study by
Gallagher et al.27 However, it is not know at this time which
DEXA instrument is more accurate in determining FFM and BF.
CONCLUSION
BMI alone cannot provide information about the respective con-
tributions of FFM and FM to body weight. This study presented
the FFMI, BFMI, and %BF values that correspond to low, normal,
overweight, and obese BMIs. FFMI and BFMI can provide mean-
ingful information about body compartments, regardless of height.
REFERENCES
1. Buzby GP, Mullen JL, Matthews DC. Prognostic nutritional index in gastroin-
testinal surgery. Am J Surg 1980;139:160
2. Weinsier RL, Hunker EM, Krumdieck CL, Butterworth CE. A prospective
evaluation of general medical patients during the course of hospitalisation. Am J
Clin Nutr 1979;32:418
3. Allison DB, Gallagher D, Heo M, Pi-Sunyer FX, Heymsfield SB. Body mass
index and all-cause mortality among people age 70 and over: the Longitudinal
Study of Aging. Int J Obes 1997;21:424
4. Heitmann BL, Erikson H, Ellsinger BM, Mikkelsen KL, Larsson B. Mortality
associated with body fat, fat-free mass and body mass index among 60-year-old
Swedish men-a 22-year follow-up. The study of men born in 1913. Int J Obes Rel
Metab Disord 2000;24:33
604 Kyle et al.
5. Segal KR, Dunaif A, Gutin B, et al. Body composition, not body weight, is
related to cardiovascular disease risk factors and sex hormone levels in men.
J Clin Invest 1987;84:1050
6. Van Itallie TB, Yang M-U, Heymsfield SB, Funk RC, Boileau RA. Height-
normalized indices of the body’s fat-free mass and fat mass: potentially useful
indicators of nutritional status. Am J Clin Nutr 1990;52:953
7. Centers for Disease Control and Prevention. Update: prevalence of overweight
among children, adolescents and adults—United States, 1988 –1994. MMWR
Morb Mortal Wkly Rep 1997;46:199
8. World Health Organization. Report of the WHO consultation on obesity. Obesity:
preventing and managing the global epidemic. Geneva: World Health Organi-
zation, 1998
9. Henry CJK. Body mass index and the limits of human survival. Eur J Clin Nutr
1990;44:329
10. Heymsfield SB, Matthews D. Body composition: research and clinical
advances—l993 A.S.P.E.N. Research Workshop. JPEN 1994;18:91
11. Guo SS, Zeller C, Chumlea WC, Siervogel RM. Aging, body composition, and
lifestyle: the Fels Longitudinal Study. Am J Clin Nutr 1999;70:405
12. Kyle U, Morabia A, Unger P, Slosman D, Pichard C. Contribution of body
composition to nutritional assessment at hospital admission in 995 patients: a
controlled population study. Br J Nutr 2001;86:725
13. Kyle U, Unger P, Genton L, Pichard C. Nutrition status in ⬍60 and ⬎60 yr old
patients at hospital admission: a controlled population study in 995 subjects.
Nutrition 2002;18:463
14. Abernathy RP, Black DR. Healthy body weights: an alternative perspective. Am J
Clin Nutr 1996;63:448S
15. Heitmann BL. Body fat in the adult Danish population aged 35– 65 years: an
epidemiological study. Int J Obes 1991;15:535
16. Kyle UG, Genton LC, Slosman DO, Pichard C. Fat-free and fat mass percentiles
in 5225 healthy subjects aged 15 to 98 years. Nutrition 2001;17:534
17. Pichard C, Kyle UG, Morabia A, et al. Nutritional assessment: Lean body mass
depletion at hospital admission is associated with increased length of stay. Am J
Clin Nutr 2003 (in press)
18. Kyle UG, Genton L, Pichard C. Body composition: what’s new? Curr Opin Clin
Nutr Metab Care 2002;5:427
19. Baumgartner RN, Koehler KM, Gallagher D, et al. Epidemiology of sarcopenia
among the elderly in New Mexico. Am J Epidemiol 1998;147:755
20. Lukaski HC, Bolonchuk WW, Hall CB, Siders WA. Validation of tetrapolar
bioelectrical impedance measurements to assess human body composition. J Appl
Physiol 1986;60:1327
21. Houtkooper LB, Lohman TG, Going SB, Howell WH. Why bioelectrical imped-
ance analysis should be used for estimating adiposity. Am J Clin Nutr 1996;64:
436S
22. Earthman CP, Matthie JR, Reid PM, et al. A comparison of bioimpedance
methods for detection of body cell mass change in HIV infection. J Appl Physiol
2000;88:944
Nutrition Volume 19, Numbers 7/8, 2003
23. Kyle UG, Genton L, Karsegard L, Slosman DO, Pichard C. Single prediction
equation for bioelectrical impedance analysis in adults aged 20 –94 yrs. Nutrition
2001;17:248
24. Genton LC, Karsegard VL, Kyle UG, et al. Comparison of four bioelectrical
impedance analysis formulas in healthy elderly adults. Gerontology 2001;47:315
25. Lukaski HC, Johnson PE, Bolonchuk WW, Lykken GL. Assessment of fat-free
mass using bioelectrical impedance measurements of the human body. Am J Clin
Nutr 1985;41:810
26. Jackson AS, Pollock ML, Graves JE, Mahar MT. Reliability and validity of
bioelectrical impedance in determining body composition. J Appl Physiol 1988;
64:529
27. Gallagher D, Heymsfield SB, Heo M, et al. Healthy percentage body fat ranges:
an approach for developing guidelines based on body mass index. Am J Clin Nutr
2000;72:694
28. Westerterp KR, Meijer GA, Kester AD, Wouters L, ten Hoor F. Fat-free mass as
a function of fat mass and habitual activity level. Int J Sports Med 1992;13:163
29. Mostert R, Goris A, Weling-Scheepers C, Wouters EF, Schols AM. Tissue
depletion and health related quality of life in patients with chronic obstructive
pulmonary disease. Respir Med 2000;94:859
30. Engelen MPKJ, Schols AMWJ, Lamers RJS, Wouters EFM. Different patterns of
chronic tissue wasting among patients with chronic obstructive pulmonary dis-
ease. Clin Nutr 1999;18:275
31. Engelen MP, Schols AM, Does JD, Wouters EF. Skeletal muscle weakness is
associated with wasting of extremity fat-free mass but not with airflow obstruc-
tion in patients with chronic obstructive pulmonary disease. Am J Clin Nutr
2000;71:733
32. Schutz Y, Kyle UG, Pichard C. Fat-free mass index and fat mass index percen-
tiles in Caucasians aged 18 –94 y. Int J Obes Rel Metab Disord 2002;26:953
33. Forbes GB. Longitudinal changes in adult fat-free mass: influence of body
weight. Am J Clin Nutr 1999;70:1025
34. Broadwin J, Goodman-Gruen D, Slymen D. Ability of fat and fat-free mass
percentages to predict functional disability in older men and women. J Am
Geriatr Soc 2001;49:1641
35. Kyle UG, Nicod L, Romand J, et al. Four years follow-up of body composition
in lung transplant patients. Transplantation 2003;75(6):821– 828
36. Flegal KM, Carroll MD, Kuczarski RJ, Johnson CL. Overweight and obesity in
the United States: prevalence and trends, l960 –l994. Int J Obes 1998;22:39
37. Seidell JC, Verschuren WMM, Kromhout D. Prevalence and trends of obesity in
the Netherlands 1987–l991. Int J Obes 1998;19:924
38. Visser M, Harris TB, Langlois J, et al. Body fat and skeletal muscle mass in
relation to physical disability in very old men and women of the Framingham
Heart Study. J Gerontol A Biol Sci Med Sci 1998;53:M214
39. Lee CD, Blair SN, Jackson AS. Cardiorespiratory fitness, body composition, and
all-cause and cardiovascular disease mortality in men. Am J Clin Nutr 1999;69:
373