American Journal of Epidemiology
Copyright ª 2006 by the Johns Hopkins Bloomberg School of Public Health
All rights reserved; printed in U.S.A.
Vol. 164, No. 5
DOI: 10.1093/aje/kwj217
Advance Access publication July 3, 2006
Original Contribution
The Relations of Body Composition and Adiposity Measures to Ill Health and
Physical Disability in Elderly Men
Sheena E. Ramsay1, Peter H. Whincup2, A. G. Shaper1, and S. G. Wannamethee1
1
Department of Primary Care and Population Sciences, Royal Free Hospital and University College Medical School,
London, United Kingdom.
2
Department of Community Health Sciences, St. George’s University of London, London, United Kingdom.
Received for publication November 1, 2005; accepted for publication February 22, 2006.
Although body build is related to disability and mortality in older people, the independent contributions of
adiposity and lean mass are not fully defined. The authors examined the relations of body composition (fat mass
index, fat-free mass index) and adiposity (body mass index, waist circumference) to ill health and physical disability
in a cross-sectional study of 4,252 British men aged 60–79 years in 1998–2000. Increased body mass index, waist
circumference, and fat mass index were associated with increased prevalence of cardiovascular disease, overall
ill health, and disability. Adjusted odds ratios of cardiovascular disease (top vs. bottom fifth) were 1.58 (95%
confidence interval (CI): 1.23, 2.03) for fat mass index, 1.45 (95% CI: 1.14, 1.86) for body mass index, and 1.27
(95% CI: 0.99, 1.62) for waist circumference. For overall ‘‘poor/fair’’ health, the corresponding odds ratios were 1.71
(95% CI: 1.33, 2.21), 1.49 (95% CI: 1.17, 1.90), and 1.64 (95% CI: 1.28, 2.09) and, for mobility limitation, they were
1.56 (95% CI: 1.17, 2.06), 1.96 (95% CI: 1.48, 2.56), and 1.88 (95% CI: 1.42, 2.49). A high fat-free mass index was
associated with only a decreased prevalence of respiratory problems and cancer (odds ratios ¼ 0.45 (95% CI: 0.33,
0.62) and 0.62 (95% CI: 0.42, 0.94), respectively). Body fatness, not fat-free mass, is associated with cardiovascular disease and disability in older men. Simple measures of overweight, such as body mass index and waist
circumference, are good indicators of the likelihood of morbidity in older men. Prevention of weight gain with
increasing age is likely to reduce morbidity and disability among older men.
body composition; body mass index; chronic disease; mobility limitation
Abbreviations: BNF, British National Formulary; CI, confidence interval; FEV1, forced expiratory volume in 1 second; HDL-C,
high density lipoprotein cholesterol.
Obesity in many developed countries is a major public
health problem. The health problems associated with an increase in the proportion of elderly in the population are
further compounded by the increasing prevalence of obesity
and overweight in older people (1). It is now well established that overweight and obesity are associated with an
increased burden of cardiovascular disease, other chronic diseases, and disability (2, 3). Both body mass index and waist
circumference have been used as markers of obesity and adiposity to study their relation to chronic diseases (2, 4, 5).
The limitation of these measurements is that they fail to
indicate the extent to which body fat and muscle mass independently contribute to disease or disability (6). Recently,
there has been more research on exploring the role of body
composition measures—fat mass and fat-free mass—in disability and mortality (4, 7–12). Fat mass (positively) and fatfree mass (inversely) have been shown to be associated with
disability and mortality. Attention has been drawn to the
opposing effects of these measures, which may help to explain the U-shaped association seen between body mass
Correspondence to Dr. Sheena E. Ramsay, Department of Primary Care and Population Sciences, Royal Free Hospital and University College
Medical School, Rowland Hill Street, London NW3 2PF, United Kingdom (e-mail: [email protected]).
459
Am J Epidemiol 2006;164:459–469
460 Ramsay et al.
index and mortality, particularly in the elderly (13), where
aging is associated with not only a tendency to increased
adiposity but also a loss of muscle mass (sarcopenia) (14,
15). There has thus been growing interest in the role of fat
mass and fat-free mass and their contribution to disease and
disability.
However, few studies have examined the relations of body
composition and adiposity measures to physical disability
and morbidity in older subjects. Previous reports based on
our study population have examined the burden of disease
and disability related to overweight and obesity in the elderly using body mass index (2) and the relation of body
composition to lung function (16). The purpose of this study
was to further explore the extent to which body composition
measures (fat mass and fat-free mass) and markers of adiposity (body mass index and waist circumference) are associated with disability and morbidity in a population-based
study of older British men. We aimed to explore any difference or similarity in these associations to understand which
of these markers are the best indicators of adiposity-related
problems and whether lean mass is also related.
MATERIALS AND METHODS
The British Regional Heart Study is a prospective study
of cardiovascular disease and other outcomes in a socially
and geographically representative sample of 7,735 men
aged 40–59 years from one general practice in each of 24
towns representing all major British regions and who were
initially examined in 1978–1980 (17). Ethical approval was
provided by all relevant local research ethics committees.
All men provided written, informed consent to the investigations, which were carried out in accordance with the Declaration of Helsinki. In 1998–2000, all surviving men, now
aged 60–79 years, were invited to a 20th year reexamination. All men completed a mailed questionnaire providing
information on their lifestyle and medical history, had a
physical examination, and provided a fasting blood sample.
The men were requested to fast for a minimum of 6 hours
to attend a measurement session at a specified time between
8 a.m. and 6 p.m. A total of 4,252 men (77 percent) attended
the examination.
Measurements
Physical examination included anthropometric and physiologic measurements. Details of height, weight, body mass
index, and waist circumference measurements have been
described (18). Fat mass and fat-free mass were calculated
by use of bioelectrical impedance analysis with a Bodystat
model 500 apparatus (Bodystat, Ltd., Douglas, United Kingdom). The equation by Deurenberg et al. (19) designed for
people greater than 60 years of age was used to calculate fat
mass. Fat-free mass was calculated as 6,710 3 height (m)2/
resistance (X) þ 7. Fat mass was calculated as body weight –
fat-free mass. The correlations between measurements taken
1 week apart were 0.67 for fat-free mass and 0.75 for fat
mass. To permit comparison of subjects with different
height, fat mass and fat-free mass measures were normalized
for height by dividing them by (height)2 to obtain the fat
mass index and fat-free mass index (6). Blood pressure
was measured twice in the right arm with a Dinamap model
1846 oscillometric blood pressure recorder (Critikon, Inc.,
Tampa, Florida). Forced expiratory volume in 1 second
(FEV1) was measured as part of lung function tests. FEV1
was recorded for the best test, defined in accordance with
American Thoracic Society recommendations (20). Cole
(21) has shown that dividing by (height)2 is the most appropriate way of standardizing lung function for stature. FEV1
was height standardized to the average height, 1.73 m, in the
study. Thus, height-standardized FEV1 ¼ FEV1 3 (1.73/
height)2. Low FEV1 was defined as being in the lowest quartile of FEV1.
Ill health
Men were asked to describe their health status as excellent, good, or fair/poor. In addition, subjects were asked to
report any physician diagnosis of the following conditions:
history of cardiovascular disease (heart attack (i.e., coronary
thrombosis or myocardial infarction), angina, or stroke); diabetes; cancer; and history of taking medication for respiratory or musculoskeletal disease. Reported medication use
was coded according to British National Formulary (BNF)
codes (22). Respiratory medication included BNF codes
3.1–3.10, and musculoskeletal medication included BNF
codes 10.1–10.3 (all drugs used in the treatment of rheumatic diseases, neuromuscular disorders, and soft tissue
inflammation).
Physical disability
Information on three different aspects of physical
disability—problems with mobility, carrying out usual activities, and self-care—was collected using a questionnaire.
Mobility limitation was determined by asking subjects
whether they currently had difficulty carrying out any of
the following activities on their own as a result of a longterm health problem: 1) difficulty going up or down stairs or
2) difficulty walking for a quarter of a mile (0.40 km) on the
level. Those who answered positively for any of these questions were classified as having mobility limitation. Subjects
were asked if they had some problems with or were unable
to carry out their usual activities. To ascertain problems with
self-care, subjects were asked if they had some problems
with or were unable to wash and dress themselves.
Lifestyle factors
Subjects were asked detailed questions on their smoking
and drinking habits. The men were classified into groups
based on their alcohol intake—none, occasional, light, moderate, and heavy. Heavy drinking was defined as drinking
more than 6 units (1 United Kingdom unit ¼ 10 g) of alcohol
daily or on most days in the week. On the questionnaire,
subjects were also asked to report their pattern of physical
activity, such as walking, cycling, and other sporting activities. Physical activity scores were assigned on the basis of
frequency and type of activity, and the men were divided
Am J Epidemiol 2006;164:459–469
Relations of Body Composition and Adiposity to Morbidity
into six groups: none, occasional, light, moderate, moderately vigorous, and vigorous. Subjects who reported none or
occasional activity were classified as ‘‘inactive.’’
TABLE 1. Correlation of adiposity and body composition
measures in a cross-sectional study of British men aged 60–79
years in 1998–2000*
Body
mass
index
Social class
Social class was derived from the longest held occupation
recorded at the time of the baseline questionnaire (1978–
1980) using the Registrar General’s classification of occupations, with categories grouped as nonmanual (I, II, and III
nonmanual) and manual (III manual, IV, and V).
461
Waist
circumference
Fat
mass
index
Fat-free
mass
index
Body mass index
1.00
0.87
0.80
0.43
Waist circumference
0.87
1.00
0.76
0.30
Fat mass index
0.80
0.76
1.00
0.07
Fat-free mass index
0.43
0.30
0.07
1.00
* For all: p < 0.001.
Metabolic risk factors
Details of blood lipid, blood glucose, and insulin measurements have been described (2, 23). Insulin was adjusted
for the effects of fasting duration and time of day (23).
Blood pressure was adjusted for observer variation (24).
The International Society for Hypertension guidelines were
used to identify patients with hypertension (25). Hypertensive patients were those with systolic blood pressure of
greater than or equal to 160 mmHg or diastolic blood pressure of greater than or equal to 90 mmHg or those taking
antihypertensive treatment. High cholesterol was defined as
total cholesterol of greater than or equal to 6.2 mmol/liter,
and low high-density lipoprotein cholesterol (HDL-C) was
defined as levels of less than or equal to 1.0 mmol/liter.
Insulin resistance was estimated by use of homeostasis
model assessment as the product of fasting glucose
(mmol/liter) and insulin (lU/ml) divided by the constant
22.5 (26). High homeostasis model assessment was defined
as being in the top fifth of the distribution.
Statistical analysis
Correlation coefficients were calculated for body mass
index, waist circumference, fat mass index, and fat-free
mass index. Descriptive statistics (age, inactivity, smoking,
alcohol intake, and social class) for subjects were computed
according to fifths of body mass index, waist circumference,
fat mass index, and fat-free mass index. Multiple logistic
regression was used to assess the relation of these measures
to metabolic risk factors (hypertension, high cholesterol,
low HDL-C, and high homeostasis model assessment) and
low FEV1; ill health (fair/poor health, cardiovascular disease, diabetes, cancer, and use of respiratory or musculoskeletal medication); and physical disability (mobility
limitation, problem with usual activities, or problem with
self-care). Odds ratios with 95 percent confidence intervals
for these outcome measures were obtained for the adiposity
and body composition measures by use of the lowest fifth as
the referent category.
Models for the various cardiovascular disease risk factors
and chronic diseases were adjusted for age, social class,
smoking, alcohol intake, and physical activity. For the adjustment, age was fitted as a continuous variable; social class
(six levels), smoking (six levels), alcohol intake (five levels),
and physical activity (five levels) were fitted as categorical
variables. The odds ratios for physical disability were adjusted for age, social class, smoking, alcohol intake, physAm J Epidemiol 2006;164:459–469
ical activity, and morbidity. Tests for trends were carried out
fitting body mass index, waist circumference, fat mass index, and fat-free mass index as continuous variables. To
explore the effect of fat mass index on the relation of body
mass index to physical disability, the odds ratios for body
mass index were additionally adjusted for fat mass index.
Statistical analyses were performed using SAS, version 8.2,
software (SAS Institute, Inc., Cary, North Carolina).
Receiver operating characteristic curve analysis
Receiver operating characteristic analysis was used to
compare the association of adiposity measures (body mass
index, waist circumference, and fat mass index) with measures of ill health and physical disability. Tests for differences
between the curves were performed by using STATA, version
7.0, software (Stata Corp., College Station, Texas). The receiver operating characteristic curve tests the ability of a
variable to predict an outcome by plotting sensitivity against
1 – specificity, and it simultaneously compares this in different variables. The area under the curve is the summary statistics from the receiver operating characteristic curve that
ranges from 0 to 1, with 0.5 indicating no predictive power
and 1 indicating perfect predictive power. The area under the
curve with 95 percent confidence intervals was calculated.
RESULTS
Table 1 shows the correlation of the adiposity and body
composition measures. Body mass index, waist circumference, and fat mass index had a strong positive correlation
with each other. The fat-free mass index was positively
correlated with body mass index and waist circumference
and showed a small but significant inverse association with
fat mass index.
Details of the characteristics of the subjects are given in
table 2. With increasing age, the fat-free mass index, fat
mass index, body mass index, and waist circumference decreased significantly. The percentage of manual group and
inactive subjects increased with increasing body mass index,
waist circumference, and fat mass index. The percentage of
current smokers decreased with increasing fat-free mass
index, fat mass index, body mass index, and waist circumference. The prevalence of never smokers decreased with
increasing body mass index, waist circumference, and fat mass
462 Ramsay et al.
TABLE 2. Characteristics (%) of body mass index, waist circumference, fat mass index, and fat-free mass
index in a cross-sectional study of British men aged 60–79 years in 1998–2000
Characteristics according
to fifths (lowest to
highest: 1–5)
Aged
70 years
(n ¼ 1,709)
Inactive
(n ¼ 1,428)
Never
smokers
(n ¼ 1,233)
Current
smokers
(n ¼ 548)
Heavy
drinking
(n ¼ 125)
Manual
social class
(n ¼ 2,166)
Body mass index (kg/m2)
1 (14–23), n ¼ 846
46
30
32
19
3
48
2 (24–25), n ¼ 847
42
30
31
13
2
46
3 (26), n ¼ 846
41
32
29
11
3
48
4 (27–29), n ¼ 847
36
32
28
11
3
53
5 (30–48), n ¼ 846
36
43
25
9
4
60
<0.001
<0.001
0.09
<0.001
ptrend
0.002
<0.001
Waist circumference (cm)
1 (57–88), n ¼ 845
43
27
36
17
2
48
2 (89–94), n ¼ 845
40
29
30
13
2
47
3 (95–98), n ¼ 845
39
32
29
12
2
49
4 (99–104), n ¼ 845
40
35
28
11
4
52
5 (105–149), n ¼ 845
39
ptrend
0.05
44
22
11
<0.001
<0.001
<0.001
4
<0.001
58
<0.001
Fat mass index (kg)
1 (57–88), n ¼ 822
44
30
34
17
3
46
2 (89–94), n ¼ 822
42
31
31
13
3
48
3 (95–98), n ¼ 823
42
31
31
12
3
49
4 (99–104), n ¼ 822
38
34
26
12
2
52
5 (105–149), n ¼ 822
ptrend
35
41
23
10
4
61
<0.001
<0.001
<0.001
<0.001
0.07
<0.001
43
35
25
17
3
51
Fat-free mass index (kg)
1 (0.14–6.9), n ¼ 832
2 (7–7.9), n ¼ 833
42
34
29
14
4
49
3 (8–9.9), n ¼ 833
37
32
31
12
2
51
4 (10–11.9), n ¼ 833
39
30
30
11
2
53
5 (12–28.9), n ¼ 833
39
35
31
11
3
51
<0.001
0.70
ptrend
0.04
0.10
index (ptrend < 0.001 for all) but increased with increasing
fat-free mass index (ptrend < 0.001). The prevalence of
heavy drinking rose with increasing waist circumference
but varied little for other body measures.
Table 3 shows the relation of body mass index, waist
circumference, fat mass index, and fat-free mass index to
metabolic risk factors and FEV1. The likelihood of having
hypertension was similar for increasing fifths of fat mass
index, body mass index, and waist circumference, with a
progressive increase in the odds of hypertension (ptrend <
0.001). Fat mass index, body mass index, and waist circumference all showed strong positive associations with insulin
resistance (high homeostasis model assessment; ptrend <
0.001 for all). Those in the higher fifths of fat-free mass
index showed significantly increased odds of having hypertension, low HDL-C, and high homeostasis model assessment compared with those in the bottom fifth. However,
since those with a high fat-free mass index tended to have
0.02
0.51
higher waist circumferences, we further adjusted for waist
circumference in the model for fat-free mass index to explore whether waist circumference explained these relations
of fat-free mass index. Adjustment for waist circumference
abolished the relations of fat-free mass index to hypertension, low HDL-C, and high homeostasis model assessment.
The odds ratios of those in the top fifth compared with those
in the bottom fifth of fat-free mass index were 1.16 (95
percent confidence interval (CI): 0.94, 1.45) for hypertension, 1.22 (95 percent CI: 0.91, 1.65) for low HDL-C, and
0.90 (95 percent CI: 0.68, 1.18) for high homeostasis model
assessment. No association was seen between the measures
of adiposity or body composition and high cholesterol.
Waist circumference was positively associated with low
lung function, but little association was seen among fat mass
index, body mass index, and low lung function. By contrast,
a low fat-free mass index was associated with increased
odds of having low lung function (ptrend < 0.001).
Am J Epidemiol 2006;164:459–469
Relations of Body Composition and Adiposity to Morbidity
463
TABLE 3. Prevalence of metabolic risk factors and low forced expiratory volume in 1 second and odds ratios with 95% confidence
intervals according to fifths of body mass index, waist circumference, fat mass index, and fat-free mass index in a cross-sectional
study of British men aged 60–79 years in 1998–2000
Hypertension
(n ¼ 2,030)
Characteristics according
to fifths (lowest to
highest: 1–5)
Low high
density lipoprotein
cholesterol
(n ¼ 692)
High cholesterol
(n ¼ 1,730)
Odds
ratio*
9
1.00
1.41
1.03, 1.95
14
1.61
1.16, 2.22
14
1.66
1.21, 2.28
18
2.32
1.71, 3.16
0.84, 1.26
18
2.24
1.66, 3.03
27
3.86
2.88, 5.19
24
0.83
0.65, 1.05
0.78, 1.19
27
3.66
2.73, 4.92
50 11.84
8.84, 15.86
26
0.82
0.64, 1.05
%
Odds
ratio*
38
1.00
10
1.00
44
1.23
1.00, 1.49
12
1.28, 1.92
44
1.18
0.96, 1.44
1.74
1.42, 2.14
40
1.03
2.41
1.95, 2.98
38
0.97
1 (14–23), n ¼ 846
38
1.00
2 (24–25), n ¼ 847
44
1.32
1.07, 1.62
3 (26), n ¼ 846
48
1.57
4 (27–29), n ¼ 847
50
5 (30–48), n ¼ 846
58
95%
confidence
interval
%
95%
Odds
confidence
ratio*
interval
Odds
ratio*
95%
confidence
interval
95%
confidence
interval
%
95%
confidence
interval
%
Low forced
expiratory volume
in 1 second
(n ¼ 1,051)
High homeostasis
model assessment
(n ¼ 1,001)
%
Odds
ratio*
29
1.00
23
0.79
0.62, 1.01
22
0.69
0.55, 0.89
Body mass index (kg/m2)
ptrend
0.13
<0.001
<0.001
0.15
<0.001
Waist circumference (cm)
1 (57–88), n ¼ 845
39
1.00
38
1.00
9
1.00
8
1.00
25
1.00
2 (89–94), n ¼ 845
45
1.28
1.04, 1.56
44
1.22
0.99, 1.49
13
1.5
1.08, 2.07
13
1.58
1.14, 2.19
21
0.76
0.59, 0.98
3 (95–98), n ¼ 845
45
1.26
1.03, 1.55
42
1.15
0.94, 1.41
15
1.89
1.38, 2.59
20
2.61
1.91, 3.55
22
0.84
0.65, 1.08
4 (99–104), n ¼ 845
50
1.50
1.23, 1.85
41
1.08
0.89, 1.33
18
2.29
1.67, 3.12
30
4.65
3.45, 6.25
25
1.00
0.78, 1.28
5 (105–149), n ¼ 845
58
2.04
1.66, 2.52
39
0.99
0.81, 1.22
27
3.51
2.60, 4.74
47 10.17
7.57, 13.67
31
1.21
0.95, 1.55
ptrend
0.32
<0.001
<0.001
0.007
<0.001
Fat mass index (kg)
1 (57–88), n ¼ 822
38
1.00
2 (89–94), n ¼ 822
44
1.32
3 (95–98), n ¼ 823
48
1.57
4 (99–104), n ¼ 822
51
5 (105–149), n ¼ 822
58
ptrend
37
1.00
11
1.00
1.07, 1.62
41
1.12
1.28, 1.93
46
1.36
1.72
1.39, 2.12
42
2.38
1.92, 2.94
38
0.91, 1.37
12
1.21
1.11, 1.67
14
1.42
1.19
0.97, 1.46
19
1.00
0.81, 1.24
25
0.73
<0.001
10
1.00
0.89, 1.65
13
1.26
1.05, 1.93
18
1.84
1.96
1.46, 2.63
30
2.75
2.06, 3.68
48
<0.001
25
1.00
0.92, 1.73
23
0.86
0.67, 1.11
1.36, 2.48
24
0.99
0.78, 1.28
3.73
2.81, 4.95
23
0.89
0.69, 1.15
8.6
6.49, 11.41
29
1.14
0.88, 1.46
0.31
<0.001
Fat-free mass index (kg)
1 (0.14–6.9), n ¼ 832
44
1.00
41
1.00
13
1.00
18
1.00
32
1.00
2 (7–7.9), n ¼ 833
45
1.04
0.85, 1.28
43
1.02
0.84, 1.25
14
1.21
0.91, 1.63
20
1.14
0.88, 1.47
25
0.77
0.61, 0.98
3 (8–9.9), n ¼ 833
48
1.23
1.00, 1.51
43
1.06
0.87, 1.29
16
1.29
0.97, 1.73
21
1.27
0.98, 1.63
22
0.64
0.50, 0.82
4 (10–11.9), n ¼ 833
49
1.28
1.05, 1.58
40
0.91
0.74, 1.11
18
1.50
1.13, 1.99
28
1.78
1.39, 2.28
23
0.70
0.55, 0.89
5 (12–28.9), n ¼ 833
52
1.43
1.17, 1.76
39
0.89
0.72, 1.09
21
1.88
1.43, 2.49
31
2.1
1.65, 2.68
22
0.65
0.51, 0.83
ptrend
ptrendy
<0.001
0.11
<0.001
<0.001
0.01
0.56
0.29
0.69
0.36
<0.001
* Adjusted for age, social class, smoking, alcohol intake, and physical activity.
y Adjusted for the above in addition to waist circumference.
Table 4 shows the prevalence and adjusted odds ratios of
having ill health according to fifths of body mass index,
waist circumference, fat mass index, and fat-free mass index. The odds of reporting fair/poor health increased with
body mass index, waist circumference, and fat mass index
levels, although these increases were statistically significant
only in the top fifths of the body size measures (about 50
percent or more increased odds of reporting fair/poor health;
ptrend < 0.001 for all). The odds of cardiovascular disease
increased with fat mass index and body mass index levels
(ptrend < 0.001). Subjects in the fourth and top fifth of fat
mass index were 33 percent and 58 percent, respectively,
Am J Epidemiol 2006;164:459–469
more likely to have cardiovascular disease compared with
those in the bottom fifth of fat mass index. Fat-free mass
index was not significantly associated with having cardiovascular disease or reporting fair/poor health.
With increasing body mass index, waist circumference,
fat mass index, and fat-free mass index, subjects were more
likely to be taking musculoskeletal medication and to have
diabetes (table 4). To explore whether the higher waist circumference explained these relations between fat-free mass
index and morbidity indicators, we further adjusted for waist
circumference in the model for fat-free mass index. The
relations of fat-free mass index to use of musculoskeletal
Cardiovascular
disease
(n ¼ 962)
Fair/poor
health
(n ¼ 1,093)
Characteristics according
to fifths (lowest to
highest: 1–5)
95%
confidence
interval
%
Odds
ratio*
20
1.00
0.78, 1.29
21
1.05
0.59, 1.00
22
1.13
1.09
0.85, 1.40
22
1.49
1.17, 1.90
28
%
Odds
ratio
1 (14–23), n ¼ 846
24
1.00
2 (24–25), n ¼ 847
22
1.00
3 (26), n ¼ 846
20
0.77
4 (27–29), n ¼ 847
25
5 (30–48), n ¼ 846
37
95%
confidence
interval
Diabetes
(n ¼ 555)
%
Odds
ratio*
95%
confidence
interval
%
Odds
ratio*
7
1.00
95%
confidence
interval
Musculoskeletal
medication
(n ¼ 537)
Respiratory
medication
(n ¼ 487)
Cancer
(n ¼ 252)
%
Odds
ratio*
14
1.00
95%
confidence
interval
%
Odds
ratio*
8
1.00
95%
confidence
interval
Body mass index (kg/m2)
ptrend
9
1.00
0.82, 1.35
14
1.49
1.09, 2.04
5
0.77
0.51, 1.16
11
0.81
0.59, 1.09
11
1.51
1.08, 2.13
0.88, 1.45
12
1.25
0.91, 1.72
5
0.78
0.52, 1.18
11
0.74
0.55, 1.01
11
1.54
1.09, 2.17
1.10
0.86, 1.41
14
1.42
1.04, 1.94
6
0.94
0.63, 1.39
9
0.61
0.45, 0.85
17
2.47
1.79, 3.41
1.45
1.14, 1.86
16
1.66
1.21, 2.26
6
0.89
0.59, 1.34
12
0.67
0.49, 0.92
15
1.92
1.38, 2.69
0.004
<0.001
0.008
0.91
0.003
<0.001
Waist circumference (cm)
1 (57–88), n ¼ 845
22
1.00
20
1.00
11
1.00
6
1.00
13
1.00
9
1.00
2 (89–94), n ¼ 845
21
0.98
0.76, 1.27
20
1.04
0.81, 1.34
12
1.07
0.79, 1.46
5
0.84
0.55, 1.29
10
0.77
0.57, 1.06
9
1.06
0.75, 1.49
3 (95–98), n ¼ 845
21
0.92
0.71, 1.19
23
1.24
0.97, 1.59
13
1.23
0.91, 1.66
6
1.12
0.74, 1.68
8
0.57
0.40, 0.79
13
1.57
1.14, 2.17
4 (99–104), n ¼ 845
27
1.29
1.00, 1.65
23
1.23
0.96, 1.58
12
1.09
0.81, 1.49
6
0.99
0.65, 1.51
12
0.91
0.67, 1.23
15
1.65
1.19, 2.27
5 (105–149), n ¼ 845
37
1.64
1.28, 2.09
27
1.27
0.99, 1.62
16
1.45
1.07, 1.96
7
1.08
0.71, 1.64
14
0.89
0.66, 1.21
16
1.83
1.33, 2.52
ptrend
0.03
<0.001
0.02
0.51
0.89
<0.001
Fat mass index (kg)
1 (57–88), n ¼ 822
20
1.00
18
1.00
12
1.00
6
1.00
12
1.00
11
1.00
2 (89–94), n ¼ 822
22
1.11
0.86, 1.45
21
1.14
0.88, 1.47
12
0.98
0.72, 1.34
5
0.79
0.52, 1.22
11
0.93
0.67, 1.28
10
0.89
0.64, 1.25
3 (95–98), n ¼ 823
23
1.12
0.86, 1.46
21
1.15
0.89, 1.49
12
1.05
0.77, 1.42
6
1.05
0.69, 1.57
11
0.88
0.64, 1.22
12
1.09
0.79, 1.51
4 (99–104), n ¼ 822
26
1.24
0.96, 1.61
24
1.33
1.03, 1.71
13
1.00
0.74, 1.36
5
0.86
0.56, 1.32
10
0.77
0.55, 1.07
13
1.30
0.95, 1.79
5 (105–149), n ¼ 822
37
1.71
1.33, 2.21
28
1.58
1.23, 2.03
16
1.24
0.92, 1.68
7
1.19
0.79, 1.79
14
0.98
0.72, 1.35
17
1.56
1.15, 2.13
Am J Epidemiol 2006;164:459–469
ptrend
<0.001
0.18
<0.001
0.35
0.59
<0.001
Fat-free mass index (kg)
1 (0.14–6.9), n ¼ 832
28
1.00
2 (7–7.9), n ¼ 833
25
1.01
3 (8–9.9), n ¼ 833
22
0.84
4 (10–11.9), n ¼ 833
26
5 (12–28.9), n ¼ 833
26
23
1.00
11
1.00
0.79, 1.28
23
1.07
0.65, 1.08
21
0.96
1.01
0.79, 1.29
23
1.00
0.79, 1.28
22
8
1.00
0.84, 1.37
12
1.16
0.75, 1.23
13
1.16
1.13
0.88, 1.43
13
0.97
0.76, 1.24
15
17
1.00
0.85, 1.59
6
0.69
0.85, 1.59
6
0.74
1.25
0.92, 1.71
5
1.44
1.07, 1.95
5
11
1.00
0.47, 1.04
12
0.71
0.50, 1.09
11
0.62
0.53, 0.95
12
1.04
0.76, 1.43
0.46, 0.83
11
1.06
0.6
0.39, 0.91
9
0.77, 1.46
0.52
0.38, 0.71
14
1.23
0.62
0.42, 0.94
8
0.90, 1.67
0.45
0.33, 0.62
16
1.44
1.07, 1.95
ptrend
0.97
0.98
0.02
0.02
<0.001
0.008
ptrendy
0.04
0.25
0.11
0.006
<0.001
0.16
* Adjusted for age, social class, smoking, alcohol intake, and physical activity.
y Adjusted for the above in addition to waist circumference.
464 Ramsay et al.
TABLE 4. Prevalence of ill health and odds ratios with 95% confidence intervals according to fifths of body mass index, waist circumference, fat mass index, and fat-free
mass index in a cross-sectional study of British men aged 60–79 years in 1998–2000
Relations of Body Composition and Adiposity to Morbidity
465
TABLE 5. Prevalence and odds ratios with 95% confidence intervals for disability according to fifths of
body mass index, waist circumference, fat mass index, and fat-free mass index in a cross-sectional study
of British men aged 60–79 years in 1998–2000
Characteristics according
to fifths (lowest to
highest: 1–5)
Mobility limitation,
n ¼ 837 (20%)
%
Odds
ratio*
Problem with usual
activities, n ¼ 983 (23%)
95%
confidence
interval
%
21
1.00
0.71, 1.30
20
0.96
Odds
ratio*
Problem with self-care,
n ¼ 238 (6%)
95%
confidence
interval
%
4
1.00
0.74, 1.24
5
1.40
Odds
ratio*
95%
confidence
interval
Body mass index (kg/m2)
1 (14–23), n ¼ 846
16
1.00
2 (24–25), n ¼ 847
14
0.96
0.84, 2.35
3 (26), n ¼ 846
15
1.05
0.77, 1.42
18
0.86
0.66, 1.11
4
1.09
0.63, 1.87
4 (27–29), n ¼ 847
20
1.44
1.08, 1.92
24
1.20
0.93, 1.55
6
1.43
0.87, 2.37
5 (30–48), n ¼ 846
32
1.99
1.50, 2.64
31
1.32
1.03, 1.69
9
1.59
0.99, 2.57
ptrend
0.005
<0.001
0.07
Waist circumference (cm)
1 (57–88), n ¼ 845
15
1.00
20
1.00
4
1.00
2 (89–94), n ¼ 845
15
1.12
0.82, 1.52
20
1.07
0.89, 1.39
5
1.47
0.86, 2.53
3 (95–98), n ¼ 845
17
1.20
0.89, 1.63
22
1.14
0.88, 1.47
6
1.79
1.06, 3.01
4 (99–104), n ¼ 845
20
1.35
1.00, 1.82
22
1.04
0.80, 1.35
5
1.48
0.88, 2.51
5 (105–149), n ¼ 845
31
1.95
1.47, 2.59
32
1.49
1.16, 1.91
8
1.61
0.98, 2.66
ptrend
0.005
<0.001
0.13
Fat mass index (kg)
1 (57–88), n ¼ 822
16
1.00
20
1.00
4
1.00
2 (89–94), n ¼ 822
14
0.73
0.54, 1.00
19
0.93
0.71, 1.21
4
0.99
0.57, 1.70
3 (95–98), n ¼ 823
16
0.99
0.73, 1.34
20
0.96
0.74, 1.25
5
1.17
0.69, 1.98
4 (99–104), n ¼ 822
22
1.29
0.97, 1.74
25
1.16
0.90, 1.51
7
1.53
0.93, 2.52
5 (105–149), n ¼ 822
30
1.59
1.20, 2.12
30
1.24
0.96, 1.59
8
1.28
0.78, 2.09
ptrend
0.02
<0.001
0.11
Fat-free mass index (kg)
1 (0.14–6.9), n ¼ 832
20
1.00
2 (7–7.9), n ¼ 833
18
1.03
0.77, 1.37
24
1.00
23
1.04
0.81, 1.34
5
1.00
5
1.45
0.89, 2.35
3 (8–9.9), n ¼ 833
17
0.99
0.74, 1.33
21
0.99
0.76, 1.27
5
1.27
0.78, 2.07
4 (10–11.9), n ¼ 833
21
1.22
0.92, 1.62
22
1.01
0.78, 1.29
6
1.54
0.96, 2.47
5 (12–28.9), n ¼ 833
21
1.26
0.95, 1.67
24
1.11
0.87, 1.43
6
1.36
0.84, 2.20
ptrend
0.05
0.51
0.22
* Adjusted for age, smoking, alcohol intake, physical activity, social class, and morbidity (cardiovascular disease,
cancer, diabetes, and medication for respiratory disease or musculoskeletal disease).
medication and diabetes were attenuated when adjusted for
waist circumference and were no longer significant. The
odds ratios for those in the top fifth compared with those
in the bottom fifth of fat-free mass index were 1.32 (95
percent CI: 0.96, 1.82) for having diabetes and 1.22 (95
percent CI: 0.88, 1.69) for taking musculoskeletal medication. However, with increasing fat-free mass index, there
was a significant reduction in the likelihood of taking respiratory medication (ptrend < 0.001). Those in the bottom
fifth of fat-free mass index were more than twice as likely to
take respiratory medication compared with those in the top
fifth of fat-free mass index. Adjustment for waist circumference slightly strengthened the inverse association seen (odds
ratio ¼ 0.41, 95 percent CI: 0.29, 0.57). With increasing fatAm J Epidemiol 2006;164:459–469
free mass index, the subjects were less likely to have cancer
(ptrend ¼ 0.02).
Overall, 30 percent of the men reported some form of
physical disability. Table 5 shows the prevalence and adjusted odds ratios of having physical disability by fifths of
body mass index, waist circumference, fat mass index, and
fat-free mass index. Fat-free mass index demonstrated
a weak but significant positive relation with mobility limitation, but this was attenuated and no longer significant after
adjusting for waist circumference (odds ratio ¼ 0.92, 95
percent CI: 0.68, 1.24; ptrend ¼ 0.46). Body mass index,
waist circumference, and fat mass index showed a significant positive relation with mobility limitation, with those in
the top fifth showing substantially increased risk compared
466 Ramsay et al.
TABLE 6. Analysis results of receiver operating characteristics with area under the curve and 95% confidence intervals, p values, v2,
and degrees of freedom in a cross-sectional study of British men aged 60–79 years in 1998–2000
Fair/poor
health
(n ¼ 1,093)
Cardiovascular
disease
(n ¼ 962)
Diabetes
(n ¼ 555)
Musculoskeletal
medication
(n ¼ 537)
Mobility
limitation
(n ¼ 837)
Problem with
usual activities
(n ¼ 983)
Area
95%
under
confidence
the
interval
curve
Area
under
the
curve
95%
confidence
interval
Area
95%
under
confidence
the
interval
curve
Area
95%
under
confidence
the
interval
curve
Area
95%
under
confidence
the
interval
curve
Area
under
the
curve
95%
confidence
interval
Body mass index
0.56 0.54, 0.58
0.54 0.52, 0.56
0.54 0.52, 0.57
0.57 0.55, 0.59
0.59 0.57, 0.62
0.55
0.53, 0.57
Waist
circumference
0.58 0.56, 0.60
0.54 0.52, 0.56
0.54 0.51, 0.56
0.57 0.54, 0.59
0.59 0.57, 0.62
0.56
0.54, 0.58
Fat mass index
0.58 0.56, 0.60
0.55 0.53, 0.58
0.53 0.50, 0.55
0.56 0.53, 0.58
0.59 0.57, 0.62
0.56
0.54, 0.58
0.001
0.12
0.18
0.18
0.79
0.3
4.3
3.44
3.42
0.46
2.37
2
2
2
2
2
p value
v2
df
13.69
2
with those in the bottom fifth. Increased body mass index
and waist circumference, and to a lesser extent fat mass
index, were significantly associated with problems of usual
activities. To assess whether fat mass index was responsible
for the associations of body mass index and disability, we
examined these associations adjusting in addition for fat
mass index. The positive association of body mass index
with physical disability was slightly attenuated when it
was further adjusted for fat mass index (for the top fifth
of body mass index, the odds ratios were 1.45 (95 percent
CI: 0.96, 2.18) for mobility limitation and 1.13 (95 percent
CI: 0.78, 1.62) for problems with usual activities).
Receiver operating characteristic analyses for the relations of body mass index, waist circumference, and fat
mass index to ill health and physical disability are shown in
table 6. Areas under the curve for cardiovascular disease,
diabetes, musculoskeletal medication use, mobility limitations, and problems with usual activities were very similar
for body mass index, waist circumference, and fat mass
index. The receiver operating characteristic analyses give
results similar to those already observed in the logistic regression analyses. However, waist circumference and fat
mass index had slightly higher areas under the curve than
did body mass index for fair/poor health.
DISCUSSION
This study in a cohort of British men aged 60–79 years
examines the relation of body composition and body fat
measures to ill health and physical disability. Fat mass index, body mass index, and waist circumference had similar
patterns of relations to ill health and physical disability. Our
results suggest that increasing fat mass, body mass index,
and waist circumference are associated with physical disability, ill health (cardiovascular disease, diabetes, taking musculoskeletal medication, and reporting fair/poor health), and
metabolic risk factors, that is, hypertension, low HDL-C,
and insulin resistance (high homeostasis model assessment).
Overall, the fat mass index contributed to a wider range of
disease than did the fat-free mass index. Low fat-free mass
was associated with poor respiratory function and cancer but
was not independently associated with metabolic risk factors, reporting of poor/fair health, cardiovascular disease,
taking musculoskeletal medication, or physical disability.
Body composition and ill health
It has been suggested that the relation of body mass index
to chronic disease masks the underlying opposing relations
of fat mass and fat-free mass to disease risk (13, 27). The
opposing effects of fat mass and fat-free mass on mortality
were seen to explain the U-shaped association of body mass
index with mortality (27). Studies by Allison et al. (27) and
others (7) demonstrate a relation between fat mass (positive)
and fat-free mass (inverse) with mortality. To our knowledge, the current literature on the relations of fat mass and
fat-free mass to morbidity, however, is very limited. The
positive relation of fat mass index to cardiovascular disease
and other chronic diseases and the inverse association between fat-free mass and respiratory disease observed in our
study could explain the mortality pattern of body composition reported by other studies, in particular, the opposing
effects of fat-free mass and fat mass on all-cause mortality.
The inverse relation of fat-free mass index and cancer observed in our results could reflect the loss of muscle mass
among subjects with cancer. Such a loss of muscle mass in
cancer patients or cachexia has been reported (28, 29).
Therefore, loss of muscle mass could have been a consequence of cancer rather than being a cause of it.
Adiposity measures and morbidity
There has been controversy as to whether body mass
index is an adequate marker of adiposity-related problems
and whether indicators such as waist circumference or
body composition should be used instead (5, 14). We have
shown previously that body mass index is a good indicator
of adiposity-related problems and that body mass index
and waist circumference were the adiposity measures most
strongly associated with metabolic risk factors (hypertension,
Am J Epidemiol 2006;164:459–469
Relations of Body Composition and Adiposity to Morbidity
dyslipidemia, and glucose), insulin resistance, and the metabolic syndrome (2, 18). In this report, we extend our findings
to examining the associations between waist circumference,
body mass index, and fat mass and indicators of morbidity
and disability. The measures of adiposity in this study (body
mass index, waist circumference, and fat mass index) all
showed strong associations with metabolic risk factors and
many of the chronic diseases and disability. Our study shows
that the pattern of relation of body mass index and waist
circumference to indicators of ill health is similar to that of
fat mass index. In particular, all showed significant positive
associations with cardiovascular disease morbidity, musculoskeletal problems, and overall ill health, although fat mass
index showed a more consistent positive association with
cardiovascular disease morbidity but weaker association
with diabetes than did waist circumference and body mass
index. Overall, body mass index and waist circumference
were good indicators of obesity-related problems, and both
showed similar relations.
Body composition, adiposity measures, and physical
disability
Although studies have reported that overweight or obesity
is associated with disability (1, 2), it was possible in our
study to compare the relation of adiposity measures and
disability to that of body composition measures. Our results
show that fat mass, rather than fat-free mass, was associated
with physical disability. The finding that fat mass was more
strongly associated with physical disability than was fat-free
mass has been reported in other studies (9, 30–33). We also
observed in our study that body mass index and waist circumference had a positive association with physical disability, which was similar to that of fat mass. The attenuation of
the association of body mass index with indicators of physical disability when adjusted for fat mass index suggests
that fat mass contributes to the association with body mass
index. Some studies, on the other hand, have shown fat-free
mass to be inversely related to physical disability (8, 34). In
this study, the fat-free mass index appeared to be positively
associated with mobility limitation, but this was largely due
to the higher waist circumference seen in those with a high
fat-free mass index. Adjustment for waist circumference
abolished this relation, suggesting that fat-free mass index
per se has little association with mobility limitation. Although low muscle mass may lead to frailty in older people,
in this study lower lean mass was not strongly related to
disability, although it was related to an increased prevalence
of respiratory disease, independent of smoking.
Strengths and weaknesses of study
The results of this paper are based on a population-based
study, making the findings largely generalizable to older
men. Our study adds to the current literature on body composition in several ways. We have made comparisons of the
relation of adiposity and body composition measures to
health outcomes in the same sample and at the same time
point. In addition to disability, we have also explored the
relation of these measures to a wide range of conditions
Am J Epidemiol 2006;164:459–469
467
related to quality of life, such as musculoskeletal and respiratory problems, problems with usual activities, and
problems with self-care, which were lacking in previous
reports (35). Finally, we took into account the need to control for body size (14) by dividing fat mass and fat-free mass
by (height)2 (6).
However, there are limitations that need to be addressed.
First, from our study, it is not possible to establish a causeand-effect relation of the associations observed, since the
data used are cross-sectional. Yet, our findings are similar
to those of a prospective study that found fat mass to be more
predictive of disability than fat-free mass (30). Second, there
is also a possibility that disabled or very ill subjects may not
have attended the physical examination or returned the questionnaire. However, this should not have had any marked
effect on the strengths of associations observed; any effect
would have tended to lead to underestimation rather than
overestimation of associations. Third, some studies have
noted inaccuracies and little within-subject variation when
using bioelectrical impedance analysis to assess fat-free
mass in the elderly (36, 37). Fat-free mass measurements
(a surrogate measure of muscle mass) using bioelectrical
impedance analysis may be subject to imprecision because
of variation in hydration status (36), and they may not be
a very accurate measure of muscle mass, which is reflected
in the weak 1-week correlation of fat mass and fat-free mass.
This imprecision in measurement of fat mass and fat-free
mass could have led to underestimation of associations.
However, our bioelectrical impedance analysis was based
on the equation of Deurenberg et al. (19) that has been validated in an elderly population. Studies have also validated
the use of bioelectrical impedance analysis with other measures, such as magnetic resonance imaging (38), dual-energy
x-ray absorptiometry (39), and computerized tomography
(40). We report a strong inverse association with respiratory
disease, which is consistent with previous reports (16, 41).
Last, the presence of chronic diseases was based on the
subject’s reporting of physician diagnosis in the questionnaire, which could be seen as an inaccurate measure of prevalence of disease. However, data from previous studies
(42–46) show that patients’ recall of diagnosis of cardiovascular disease is a valid method of recording diseases. The
kappa statistic in a study on our cohort comparing record
review with patients’ recall of ischemic heart disease was
0.82 (42).
Implications of this study
In this paper, we have attempted to assess the relation of
fat mass, fat-free mass, and adiposity measures to ill health.
Our study shows that body mass index, waist circumference,
and fat mass are strongly related to ill health and disability in
elderly men to a similar degree and suggests that control of
adiposity in the population will help to reduce the burden of
disease and disability. It is clear in our study that measures
of body fat, such as body mass index and waist circumference,
are good indicators of adverse health outcomes. Although
body mass index is considered a surrogate marker of body fat
and fat mass measured by bioelectrical impedance analysis
is a more precise indicator of body fat (47), the relation of the
468 Ramsay et al.
two measures to adverse health outcomes is similar. Low
lean mass, on the other hand, although related to an increase
in respiratory morbidity, was not related to other markers of
ill health or disability function.
Conclusions
Body fatness was observed to be associated with morbidity and physical disability, while fat-free mass was related to
respiratory function and cancer. Body mass index and waist
circumference are adequate measures of overweight and
obesity and the likelihood of ill health in older men. Therefore, body mass index and waist circumference, which are
routine and easily used measures, are of practical value in
identifying people at risk of developing health problems
without the need to undertake complex tests, such as bioelectrical impedance analysis. Our study provides further
evidence in support of the use of body mass index and waist
circumference in clinical and research settings (35, 48, 49).
Using these simple measures of adiposity should be encouraged to reduce the public health burden of obesity and overweight in the elderly, by the promotion of lifestyles that
decrease the weight gain accompanying the aging process.
ACKNOWLEDGMENTS
The British Regional Heart Study is funded by the
British Heart Foundation and receives additional support
from the Department of Health.
The opinions expressed in the paper are those of the
authors and not necessarily those of the funding bodies.
Conflict of interest: none declared.
REFERENCES
1. Inelmen EM, Sergi G, Coin A, et al. Can obesity be a risk
factor in elderly people? Obes Rev 2003;4:147–55.
2. Wannamethee SG, Shaper AG, Whincup PH, et al. Overweight
and obesity and the burden of disease and disability in elderly
men. Int J Obes Relat Metab Disord 2004;28:1374–82.
3. Must A, Spadano J, Coakley EH, et al. The disease burden
associated with overweight and obesity. JAMA 1999;282:
1523–9.
4. Bigaard J, Frederiksen K, Tjonneland A, et al. Waist circumference and body composition in relation to all-cause mortality
in middle-aged men and women. Int J Obes (Lond) 2005;
29:778–84.
5. Janssen I, Katzmarzyk PT, Ross R. Waist circumference and
not body mass index explains obesity-related health risk. Am J
Clin Nutr 2004;79:379–84.
6. Kyle UG, Schutz Y, Dupertuis YM, et al. Body composition
interpretation. Contributions of the fat-free mass index and the
body fat mass index. Nutrition 2003;19:597–604.
7. Bigaard J, Frederiksen K, Tjonneland A, et al. Body fat and fatfree mass and all-cause mortality. Obes Res 2004;12:1042–9.
8. Janssen I, Heymsfield SB, Ross R. Low relative skeletal
muscle mass (sarcopenia) in older persons is associated with
functional impairment and physical disability. J Am Geriatr
Soc 2002;50:889–96.
9. Sternfeld B, Ngo L, Satariano WA, et al. Associations of body
composition with physical performance and self-reported
functional limitation in elderly men and women. Am J Epidemiol 2002;156:110–21.
10. Villareal DT, Banks M, Siener C, et al. Physical frailty and
body composition in obese elderly men and women. Obes Res
2004;12:913–20.
11. Lee CD, Blair SN, Jackson AS. Cardiorespiratory fitness, body
composition, and all-cause and cardiovascular disease mortality in men. Am J Clin Nutr 1999;69:373–80.
12. Oppert JM, Charles MA, Thibult N, et al. Anthropometric
estimates of muscle and fat mass in relation to cardiac and
cancer mortality in men: the Paris Prospective Study. Am J
Clin Nutr 2002;75:1107–13.
13. Allison DB, Faith MS, Heo M, et al. Hypothesis concerning
the U-shaped relation between body mass index and mortality.
Am J Epidemiol 1997;146:339–49.
14. Harris TB. Invited commentary: body composition in studies
of aging: new opportunities to better understand health risks
associated with weight. Am J Epidemiol 2002;156:122–4.
15. Roubenoff R. Sarcopenic obesity: the confluence of two epidemics. Obes Res 2004;12:887–8.
16. Wannamethee SG, Shaper AG, Whincup PH. Body fat distribution, body composition and respiratory function in elderly
men. Am J Clin Nutr 2005;82:996–1003.
17. Walker M, Whincup PH, Shaper AG. The British Regional
Heart Study 1975–2004. Int J Epidemiol 2004;33:1185–92.
18. Wannamethee SG, Shaper AG, Morris RW, et al. Measures of
adiposity in the identification of metabolic abnormalities in
elderly men. Am J Clin Nutr 2005;81:1313–21.
19. Deurenberg P, van der Kooij K, Evers P, et al. Assessment of
body composition by bioelectrical impedance in a population
aged greater than 60 y. Am J Clin Nutr 1990;51:3–6.
20. Standardization of spirometry: 1994 update. American Thoracic Society. Am J Respir Crit Care Med 1995;152:1107–36.
21. Cole TJ. Linear and proportional regression models in the
prediction of ventilatory function: with discussion. J R Stat
Soc Ser A 1975;138:297–338.
22. British National Formulary. London, United Kingdom: British
Medical Association and the Royal Pharmaceutical Society
of Great Britain, 1994.
23. Emberson JR, Whincup PH, Walker M, et al. Biochemical
measures in a population-based study: effect of fasting duration and time of day. Ann Clin Biochem 2002;39:493–501.
24. Whincup PH, Bruce NG, Cook DG, et al. The Dinamap
1846SX automated blood pressure recorder: comparison
with the Hawksley random zero sphygmomanometer under
field conditions. J Epidemiol Community Health 1992;46:
164–9.
25. 1999 World Health Organization–International Society of
Hypertension guidelines for the management of hypertension.
Guidelines Subcommittee. J Hypertens 1999;17:151–83.
26. Matthews DR, Hosker JP, Rudenski AS, et al. Homeostasis
model assessment: insulin resistance and beta-cell function
from fasting plasma glucose and insulin concentrations in
man. Diabetologia 1985;28:412–19.
27. Allison DB, Zhu SK, Plankey M, et al. Differential associations of body mass index and adiposity with all-cause mortality among men in the first and second National Health and
Nutrition Examination Surveys (NHANES I and NHANES II)
follow-up studies. Int J Obes Relat Metab Disord 2002;26:
410–16.
28. Baracos VE. Management of muscle wasting in cancerassociated cachexia: understanding gained from experimental
studies. Cancer 2001;92:1669–77.
Am J Epidemiol 2006;164:459–469
Relations of Body Composition and Adiposity to Morbidity
29. Tisdale MJ. Cancer cachexia. Langenbecks Arch Surg 2004;
389:299–305.
30. Visser M, Langlois J, Guralnik JM, et al. High body fatness,
but not low fat-free mass, predicts disability in older men and
women: the Cardiovascular Health Study. Am J Clin Nutr
1998;68:584–90.
31. Zoico E, Di Francesco V, Guralnik JM, et al. Physical disability and muscular strength in relation to obesity and different body composition indexes in a sample of healthy
elderly women. Int J Obes Relat Metab Disord 2004;28:
234–41.
32. 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:M214–21.
33. Zamboni M, Turcato E, Santana H, et al. The relationship
between body composition and physical performance in older
women. J Am Geriatr Soc 1999;47:1403–8.
34. Baumgartner RN, Koehler KM, Gallagher D, et al. Epidemiology of sarcopenia among the elderly in New Mexico. Am J
Epidemiol 1998;147:755–63. Erratum in: Am J Epidemiol
1999;149:1161.
35. Seidell JC, Kahn HS, Williamson DF, et al. Report from
a Centers for Disease Control and Prevention Workshop on use
of adult anthropometry for public health and primary health
care. Am J Clin Nutr 2001;73:123–6.
36. Bussolotto M, Ceccon A, Sergi G, et al. Assessment of body
composition in elderly: accuracy of bioelectric impedance
analysis. Gerontology 1999;45:39–43.
37. Roubenoff R, Baumgartner RN, Harris TB, et al. Application
of bioelectrical impedance analysis to elderly populations.
J Gerontol A Biol Sci Med Sci 1997;52:M129–36.
38. Janssen I, Heymsfield SB, Baumgartner RN, et al. Estimation
of skeletal muscle mass by bioelectrical impedance analysis.
J Appl Physiol 2000;89:465–71.
39. Pietrobelli A, Morini P, Battistini N, et al. Appendicular
skeletal muscle mass: prediction from multiple frequency
Am J Epidemiol 2006;164:459–469
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
469
segmental bioimpedance analysis. Eur J Clin Nutr 1998;52:
507–11.
Brown BH, Karatzas T, Nakielny R, et al. Determination of
upper arm muscle and fat areas using electrical impedance
measurements. Clin Phys Physiol Meas 1988;9:47–55.
Santana H, Zoico E, Turcato E, et al. Relation between body
composition, fat distribution, and lung function in elderly men.
Am J Clin Nutr 2001;73:827–31.
Lampe FC, Walker M, Lennon LT, et al. Validity of a selfreported history of doctor-diagnosed angina. J Clin Epidemiol
1999;52:73–81.
Walker MK, Whincup PH, Shaper AG, et al. Validation of
patient recall of doctor-diagnosed heart attack and stroke:
a postal questionnaire and record review comparison. Am J
Epidemiol 1998;148:355–61.
Bush TL, Miller SR, Golden AL, et al. Self-report and medical
record report agreement of selected medical conditions in the
elderly. Am J Public Health 1989;79:1554–6.
Kriegsman DM, Penninx BW, van Eijk JT, et al. Self-reports
and general practitioner information on the presence of
chronic diseases in community dwelling elderly. A study on
the accuracy of patients’ self-reports and on determinants of
inaccuracy. J Clin Epidemiol 1996;49:1407–17.
Okura Y, Urban LH, Mahoney DW, et al. Agreement between
self-report questionnaires and medical record data was substantial for diabetes, hypertension, myocardial infarction and
stroke but not for heart failure. J Clin Epidemiol 2004;57:
1096–103.
Roubenoff R, Dallal GE, Wilson PW. Predicting body fatness:
the body mass index vs estimation by bioelectrical impedance.
Am J Public Health 1995;85:726–8.
Wang J. Waist circumference: a simple, inexpensive, and reliable tool that should be included as part of physical examinations in the doctor’s office. Am J Clin Nutr 2003;78:902–3.
Wang Y, Rimm EB, Stampfer MJ, et al. Comparison of abdominal adiposity and overall obesity in predicting risk of
type 2 diabetes among men. Am J Clin Nutr 2005;81:555–63.