Lifestyle factors associated with age-related differences in body composition: the Florey Adelaide Male Aging Study.

Abstract

Age-related change in body composition is associated with adverse health outcomes, including functional decline, disability, morbidity, and early mortality. Prevention of age-related changes requires a greater understanding of the associations among age, lifestyle factors, and body composition.
We aimed to comprehensively determine lifestyle factors associated with age-related differences in body composition assessed by using dual-energy X-ray absorptiometry.
We analyzed baseline (cross-sectional) data collected from 2002 to 2005 for approximately 1200 men in the Florey Adelaide Male Aging Study, a regionally representative cohort of Australian men aged 35-81 y.
Mean values for whole-body lean mass (LM) and areal bone mineral density (aBMD) decreased, whereas mean values for abdominal fat mass (FM) and whole-body and abdominal percentage FM (%FM) increased with age. No significant age-related differences were found for whole-body FM. Multiple adjusted odds of being in the highest tertiles for whole-body and abdominal %FM decreased for smokers (63-71%) but increased with age group and for lowest energy (43-50%), carbohydrate (92-107%), and fiber (107%) intake tertiles. Multiple adjusted odds of being in the highest aBMD tertile decreased for lowest body mass (92%) and carbohydrate intake (63%) tertiles and for men aged > or = 75 y (78%) but increased for Australian birth (58%) and for participation in vigorous physical activities (82%).
Age-related differences in body composition indicate that whole-body FM remains stable but increases viscerally and that whole-body %FM is confounded by LM, whereas aBMD decreases with age. Age-related differences in %FM and aBMD are associated with demographic and lifestyle factors.

Full text

Page 1

Lifestyle factors associated with age-related differences in body
composition: the Florey Adelaide Male Aging Study1–3
Evan Atlantis, Sean A Martin, Matthew T Haren, Anne W Taylor, and Gary A Wittert for the
Florey Adelaide Male Aging Study
ABSTRACT
Background:Age-relatedchangeinbodycompositionisassociated
withadversehealthoutcomes,includingfunctionaldecline,disabil-
ity, morbidity, and early mortality. Prevention of age-related
changes requires a greater understanding of the associations among
age, lifestyle factors, and body composition.
Objective: We aimed to comprehensively determine lifestyle fac-
tors associated with age-related differences in body composition
assessed by using dual-energy X-ray absorptiometry.
Design: We analyzed baseline (cross-sectional) data collected from
2002to2005for?1200menintheFloreyAdelaideMaleAgingStudy,
a regionally representative cohort of Australian men aged 35–81 y.
Results: Meanvaluesforwhole-bodyleanmass(LM)andarealbone
mineraldensity(aBMD)decreased,whereasmeanvaluesforabdomi-
nal fat mass (FM) and whole-body and abdominal percentage FM
(%FM)increasedwithage.Nosignificantage-relateddifferenceswere
foundforwhole-bodyFM.Multipleadjustedoddsofbeinginthehigh-
esttertilesforwhole-bodyandabdominal%FMdecreasedforsmokers
(63–71%) but increased with age group and for lowest energy (43–
50%),carbohydrate(92–107%),andfiber(107%)intaketertiles.Mul-
tiple adjusted odds of being in the highest aBMD tertile decreased for
lowestbodymass(92%)andcarbohydrateintake(63%)tertilesandfor
menaged?75y(78%)butincreasedforAustralianbirth(58%)andfor
participation in vigorous physical activities (82%).
Conclusions:Age-relateddifferencesinbodycompositionindicate
thatwhole-bodyFMremainsstablebutincreasesviscerallyandthat
whole-body %FM is confounded by LM, whereas aBMD decreases
withage.Age-relateddifferencesin%FMandaBMDareassociated
with demographic and lifestyle factors.
88:95–104.
Am J Clin Nutr 2008;
INTRODUCTION
Obesity is a well-known cause of cardiovascular disease (1)
and early mortality (2) and is associated with functional decline
anddisabilityamongolderadults(3).Theetiologicmechanisms
of excess body weight are not well known. Age-related changes
in body composition, namely reductions in skeletal tissue (lean
body mass) and relative increases in adipose tissue (percentage
body fat mass), particularly in the abdominal region (often
termed central obesity), are believed to be key risk factors for
functionaldecline,disability,morbidity,andearlymortality(4).
Studies have shown that relative lean body mass is inversely
associated with functional impairment (5), that central obesity
predicts early mortality far better than does body mass index (6,
7), and that elevated blood pressure and cholesterol concentra-
tions are associated with higher percentage body fat and lower
leanbodymassinoverweightmen(8).Thus,age-relatedchanges
in whole-body and regional body composition suggest several
plausibleexplanationsfortheheightenedriskofdisease,disabil-
ity, and early mortality found with obesity.
Describing the aging effect on body composition provides
important epidemiologic information for identifying possible
etiologic mechanisms as targets for treatment and prevention of
suchadversehealthoutcomes,giventheworldwidegrowthrates
of obesity (9) and the increase of the aging population (10). Our
present knowledge of age-related differences in body composi-
tion is insufficient, because the existing body of evidence does
not comprehensively describe regional as well as whole-body
composition, and unreliable, because results are mostly drawn
from study samples consisting of participants who were not ran-
domly selected. We aimed to comprehensively define age-
related differences in body composition in a random sample of
men aged 35–81 y from the Florey Adelaide Male Aging Study
(FAMAS). Furthermore, we tested whether demographic and
lifestyle variables were associated with age-related differences
in whole-body and regional body composition, as potential risk
factors.
SUBJECTS AND METHODS
Study sample and design
We analyzed baseline (cross-sectional) data from the FAMAS,
aregionallyrepresentativecohortstudyof1195menaged35–80
1From the Discipline of Medicine, University of Adelaide, Adelaide,
Australia (SAM, MTH, AWT, and GAW); the Spencer Gulf Rural Health
School&CentreforRuralHealthandCommunityDevelopment,University
ofSouthAustralia,Whyalla,Australia(MTH);andtheExercise,Healthand
Performance Faculty Research Group, Faculty of Health Sciences, Univer-
sity of Sydney, Sydney, Australia (EA).
2Supported by the University of Adelaide Florey Foundation, the South
AustralianPremier’sScienceandResearchFund,andaNationalHealthand
MedicalResearchCouncilpostdoctoralfellowship(PublicHealth)award(to
MTH).
3Reprints not available. Address correspondence to E Atlantis, Exercise,
Health and Performance Faculty Research Group, Faculty of Health Sci-
ences, The University of Sydney, 75 East Street, Lidcombe, NSW 2141,
Australia. E-mail: e.atlantis@usyd.edu.au.
Received January 14, 2008.
Accepted for publication March 31, 2008.
95
Am J Clin Nutr 2008;88:95–104. Printed in USA. © 2008 American Society for Nutrition
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 2

y at recruitment and living in the northwest section of Adelaide,
Australia. The FAMAS aims to investigate the health of aging
meninAustralia.Asamplesizeof1200wasinitiallytargetedfor
chronicdiseases,whichallowsforaminimumdetectablerelative
risk of disease of ?1 of 1.52 and ?1 of 0.61 between groups of
participantswithandthosewithoutriskfactorexposure,byusing
the following assumptions: 1) a sample power of 80%, 2) a
2-tailed test; 3) a 5% chance of type 1 error; 4) an estimated risk
factor prevalence of 30% (conservative estimate for behavioral
risk factors such as physical inactivity); and 5) an estimated
disease prevalence in the reference group of 15% (conservative
estimate for prevalence of ?1 main chronic disease).
Details of the FAMAS design, procedures, and participants
werepublishedelsewhere(11).Briefly,studyeligibilityrequired
participants to be male, between 35 and 80 y old at the time of
recruitment, and living in the defined catchment area (the Aus-
tralianBureauofStatisticsNorthandWestStatisticalDivisionof
Adelaide)andtohaveaconnectedtelephonenumberlistedinthe
ElectronicWhitePages.Telephonenumbersfornoncommunity-
dwelling units (ie, businesses, institutions, and residential-care
facilities) were excluded from the sampling frame. An introduc-
tory mailing to randomly selected households was followed by
computer-assisted telephone interview (CATI) recruitment by
trained interviewers. The adult male (aged ?35 y) in the house-
hold to last have a birthday was invited to attend the clinic for
baseline assessment of a variety of biomedical and sociodemo-
graphic factors. A response rate of 45.1% (1195/2650) was cal-
culated by dividing the number of households of participants
who attended clinical assessments by the total number of ran-
domly sampled households with eligible participants (12). Re-
cruitmentofparticipantsoccurredduring2phases,fromAugust
2002toJuly2003andfromJune2004toMay2005,andcomplete
clinical follow-up is scheduled to recur every 5 y from baseline
throughout the life of the cohort. Follow-up self-reported infor-
mation from questionnaires is collected annually.
All participants gave written informed consent. All protocols
and procedures were approved by the Royal Adelaide Hospital
Research Ethics committee and, when appropriate, by the Ab-
original Health Research Ethics Committee of South Australia.
Anthropometric variables
Anthropometricmeasurementswereperformedbyusingstan-
dard protocols (13) in the morning, before subjects broke an
overnight fast and while they were barefoot and wearing light
clothing. Height was measured by using a wall-mounted stadi-
ometer (model no. 220; SECA, Hamburg, Germany). Body
weight was obtained by using digital platform scales (Wedder-
burn UW OFWB Series—Digital Platform Scale, Taipei, Tai-
wan, China), which were maintained by on-site engineering ser-
vices. Waist and hip circumferences were assessed by using a
fiberglass tape measure (Gulik II; Country Technology, Gays
Mills, WI). Waist circumference was measured 3 times, taken at
the level of the narrowest point (or midway) between the lower
costal border and the top of the iliac crest, and it was read in the
midaxillaryline.Hipcircumferencewasmeasured3timesatthe
level of the greatest posterior protuberance of the buttocks with
the tape placed in the horizontal plane. The mean of the 3 mea-
surements was used in analyses. CVs for triplicate waist and hip
circumferencemeasurementswere?2.2%for99.7%ofthesam-
ple. Waist-to-hip ratio was computed as waist circumference
divided by hip circumference.
Body composition variables
Whole-bodyandregionalbodycompositionwasmeasuredby
using dual-energy X-ray absorptiometry (DXA) performed ei-
ther on a fan-beam (Prodigy DF ? 14759; ENCORE software,
version 9.15) or a pencil-beam (DPX?; LUNAR software, ver-
sion 4.7e) densitometer (both machines and software programs:
GE Lunar, Madison, WI). System accuracy and stability checks
were performed regularly throughout the study by using a cus-
tomizedphantom,andquality-assurancecheckswereperformed
at the beginning of each scan day to calibrate and verify correct
operationofthescanner.Across-calibrationanalysispreviously
showed no significant differences between densitometers (14).
Fatmass(FM),leanmass(LM),bonemineralcontent(BMC;
in kg), and areal bone mineral density (aBMD; in g/m2) were
defined for whole body, arms, legs, trunk, and spine regions by
using default settings. For assessment of soft tissue composition
in the abdominal region, the top of lumbar vertebrae L2 to the
bottom of L4 and extending outwards to a vertical line touching
the inner edges of the rib cage were adopted as customized
anatomical settings. Percentage FM (%FM) was computed with
the following equation:
%FM ? ?FM/?FM ? LM ? BMC?? ? 100(1)
Demographic and lifestyle variables
Demographic and lifestyle information was obtained by self-
report questionnaires. Smoking status was determined by using
question items from recent Australian National Health Surveys.
Leisure-time physical activity during the past 2 wk was deter-
mined by using question items from the 1999 National Physical
ActivitySurvey(15).Questionitemsanddescriptorsusedinthe
1999 National Physical Activity Survey to define physical ac-
tivityduration,frequency,andintensityaresimilartothoseofthe
International Physical Activity Questionnaire (16), which
showed good repeatability coefficients and criterion validity for
classifying those who achieved sufficient physical activity for
health benefits compared with accelerometer methodology. To-
tal time spent in leisure-time physical activities was multiplied
by intensity weights (3.5 for walking and 5.0 for moderate- and
7.5forvigorous-intensityexercise)tocomputemetabolicequiv-
alent hours (MET-h). The MET is a proxy estimate of total
energy expenditure during exercise expressed as multiples of
standard resting energy expenditure (equivalent to 1 MET unit)
(17). The semiquantitative food-frequency questionnaire (FFQ)
developed by the Cancer Council of Victoria was used to esti-
mate usual energy-providing macronutrient intakes (18). The
FFQ has 74 items with 10 frequency response options, and it
contains3photographsofscaledportionsfor4foods.Question-
naires were processed by the Cancer Council of Victoria’s Nu-
trition Assessment Office to compute macronutrient intakes.
Correlation coefficients for energy, protein, fat, carbohydrate,
fiber, and alcohol intakes ranged from 0.23 to 0.60 for relative
validityassessmentoftheFFQcomparedwith7-dweighedfood
records methodology (19).
Statistical analysis
Statistical analyses were performed with SPSS software (ver-
sion 15.0; SPSS Inc, Chicago, IL). Data are presented as means
96
ATLANTIS ET AL
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 3

andSDsunlessotherwisespecified.Agewascategorizedintothe
following groups: 35–44, 45–54, 55–64, 65–74, and ?75 y old.
Between-group differences for categorical variables were com-
pared by using chi-square analysis. Between-group differences
for continuous variables were compared by using analysis of
variance (ANOVA). If overall ANOVAs were statistically sig-
nificant, Dunnett’s 2-tailed tests for post hoc contrasts were per-
formed by using the youngest age group (35–44 y old) as the
reference.Becausemanycomparisonswerenecessary,aconser-
vative P ? 0.001 was adopted for determining statistical signif-
icance, as recommended, to decrease the probability of type 1
errors(20).Prevalenceoddratios(ORs)(and95%CIs)forhigh-
est to lowest tertile (reference) for whole-body and abdominal
%FM and for whole-body aBMD were determined by using
unadjusted, age-adjusted, and multiple risk factor–adjusted (in-
cludes demographic and lifestyle variables) logistic regression
analyses.Subtypesoffat(saturated,polyunsaturated,andmono-
unsaturated) and carbohydrate (sugar and starch) intake were
excludedfromthesemodelsbecauseofconcernsregardingmul-
ticollinearity. Body mass was included as a covariate in logistic
regression models to adjust for confounding effects of weight
status on aBMD (21).
RESULTS
Age-related differences in means by age group for anthropo-
metricandbodycompositionvariablesappearinTable1.Com-
pared with the values in men aged 35–44 y, mean values were
smallerforheightinmenaged?55yandforweightinmenaged
?75 y, and there was a trend showing smaller mean values for
BMI in men aged ?65 y, with increasing age group. Compared
with men aged 35–44 y, mean values were larger for waist
circumference and waist-to-hip ratio in men aged 55–64 y, and
65–74y,and,althoughthedifferenceswerenotsignificant,mean
valuesforhipcircumferencetendedtobelargerineacholderage
group.Largermeanvaluesforwaist-to-hipratioacrosstheseage
groups therefore are mostly due to differences in waist circum-
ference, which increases with age category, whereas the null
finding for BMI between age groups is likely confounded by
age-related differences in height.
Mean values were smaller for whole-body LM in men aged
?65yandlargerfor%FMinmenaged65–74ythaninmenaged
35–44 y. Because no significant differences across age groups
werefoundforFM,largermeanvaluesforwhole-body%FMper
decade are due mostly to smaller mean values for LM and to a
lesser degree to those for BMC, considering their relative con-
tributiontobodymass.Resultsforabdominalbodycomposition
were contrasting. Mean values for abdominal FM were larger in
men aged 55–64 y and 65–74 y and for LM in men aged 55–64
y than in men aged 35–34 y, which indicates that larger mean
values for abdominal %FM actually reflect larger values for
abdominalFM.Therefore,whole-bodyFMremainsstablebutis
redistributed viscerally, and larger mean values for whole-body
%FM are confounded by smaller mean values for whole-body
LM with increasing age group.
Comparedwithmenaged35–44y,meanvaluesweresmaller
for whole-body BMC and aBMD in men aged ?55 and 65 y.
Age-relateddifferencesforwhole-bodyBMCandaBMDappear
tobemostlyduetosmallermeanvaluesforthelegsandarmsand
partlyforthetrunk.Nosignificantdifferencesacrossagegroups
were found for aBMD values for the spine. No significant
age-related differences were observed for FM in other body
regions,andmeanvaluesforLMweresmallerforarmsinmen
aged ?65 y and smaller for legs in men aged ?55 y than in
men aged 35–44 y.
Age-relateddifferencesinmeansperdecadefordemographic
andlifestylevariablesappearinTable2.Increasingagecategory
was associated with lower household income, non-Australian
birth, and nonsmoking status categories. Compared with values
inmenaged35–44y,meanvaluesfortotalenergyintakeandfor
monounsaturated and saturated fat intakes as proportions of ma-
cronutrient intakes were smaller in men aged ?65 y and 55–74
y, respectively. In contrast, mean values for sugars and fiber
intakeasproportionsofmacronutrientintakewerelargerinmen
aged 65–74 y and ?55 y, respectively, than in men aged 35–44
y. Although no significant differences between age groups were
foundforphysicalactivity,protein,totalfat,polyunsaturatedfat,
totalcarbohydratestarch,andalcoholvariables,trendssuggested
thatmeanvaluesweresmallerfortotalfatandalcoholandlarger
for total carbohydrate intake with increasing age group.
Prevalence ORs for highest to lowest (reference) whole-body
%FM tertile associated with demographic and lifestyle factors
are shown in Table 3. The odds of being in the highest %FM
tertileincreasedwithagecategoryandformenborninAustralia
(61%) but decreased for current smokers (63%) in multiple ad-
justed analyses. For macronutrients, the odds of being in the
highest %FM tertile increased for men in the lowest energy
(50%), carbohydrate (92%), and fiber (107%) intake tertiles in
multiple-adjusted analyses. No significant associations were
found for household income, physical activity, protein, fat, and
alcohol variables in multiple-adjusted analyses.
Prevalence ORs for highest to lowest (reference) abdominal
%FM associated tertile with demographic and lifestyle factors
are shown in Table 4. As with whole-body %FM, the odds of
being in the highest abdominal %FM tertile increased with age
and decreased for current smokers (71%) in multiple-adjusted
analyses. For macronutrients, the odds of being in the highest
abdominal %FM tertile increased for men in the lowest energy
(57%), carbohydrate (107%), and fiber (107%) intake tertiles in
multiple-adjusted analyses. No significant associations were
found for household income, Australian birth, physical activity,
protein, fat, and alcohol variables in multiple-adjusted analyses.
Prevalence ORs for highest to lowest (reference) whole-body
aBMD tertile associated with demographic and lifestyle factors
are shown in Table 5. The odds of being in the highest aBMD
tertile were most consistently decreased for men in the lowest
bodymasstertile(91–92%)acrossunadjustedandadjustedanal-
yses but were also decreased for men aged ?75 y (78%) and for
thoseinthelowestcarbohydrateintaketertile(63%).Incontrast,
the odds of being in the highest aBMD tertile increased for men
borninAustralia(58%)andforthosewhoreportedparticipating
in vigorous physical activities (82%) in multiple-adjusted anal-
yses.
DISCUSSION
Thisreportcomesfromthemostcomprehensivestudytohave
investigated all of the important lifestyle factors associated with
age-related differences in whole-body and regional body com-
position in a large random sample that was regionally represen-
tative of Australian men aged 35–81 y. We first defined age-
related differences in anthropometric and body composition
LIFESTYLE- AND AGE-RELATED DIFFERENCES IN BODY COMPOSITION
97
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 4

variables.Comparedwithvaluesinmenaged35–44y,thevalues
for height decrease linearly with age, as reported previously for
menandwomen(22–25),whereasthevaluesforweighttendfirst
to increase and then to decrease, notably in older men, in whom
itdisplaysaninverted“U”curve,whichwaspreviouslyreported
alsoinmenandwomen(22–25).Waistcircumferenceandwaist-
to-hip ratio tend to increase and then stabilize or decrease mar-
ginally,consistentwithsomestudiesinmenandwomen(23,26,
27), whereas other studies (24, 26, 28, 29) showed continuous
increases with age. The BMI tends to stabilize or decrease mar-
ginallyamongoldermen,consistentwithseveralotherstudiesin
men (22, 23, 25, 28, 29) and women (22, 23, 25, 26), but it may
continuetoincreaseamongwomeninotherpopulations(24,28,
29), whereas the hip circumference tends to stabilize with age,
consistent with other studies in men and women (23, 24, 26).
Inconsistent findings compared with some studies suggest that
the between-study variation in age-related differences in some
anthropometric variables are partly due to sex and ethnicity ef-
fects.Nonetheless,thefindingsofthepresentstudyandthoseof
theseotherstudiessuggestthatage-relatedchangesinheightmay
weaken the BMI’s ability to adequately assess the risk factor
statusofexcesswhole-bodyweight,particularlyinolderadults,and
that the waist circumference may be a better anthropometric measure
thanwaist-to-hipratioorBMIforassessingtheriskfactorstatusof
TABLE 1
Age-related differences per decade for anthropometric and body-composition variables obtained by using dual-energy X-ray absorptiometry in Australian
men aged 35–81 y1
Variable 35–44 y old 45–54 y old55–64 y old 65–74 y old
?75 y old
P2
Anthropometric
Subjects (n)
Height (cm)
Weight (kg)
BMI (kg/m2)
Waist circumference (cm)
Hip circumference
Waist-to-hip ratio
Body composition
Subjects (n)
Whole-body
FM (kg)
%FM
LM (kg)
BMC (kg)
aBMD (g/m2)
Trunk
FM (kg)
%FM
LM (kg)
BMC (kg)
aBMD (g/m2)
Abdominal
FM (kg)
%FM
LM (kg)
Arms
FM (kg)
%FM
LM (kg)
BMC (kg)
aBMD (g/m2)
Legs
FM (kg)
%FM
LM (kg)
BMC (kg)
aBMD (g/m2)
Spine
aBMD (g/m2)
271325304 22271
176.7 ? 7.23
87.3 ? 15.5
27.9 ? 4.2
97.3 ? 10.9
103.6 ? 8.3
0.94 ? 0.06
175.1 ? 6.4
88.6 ? 14.9
28.9 ? 4.5
100.8 ? 11.9
105.3 ? 8.5
0.96 ? 0.06
173.8 ? 6.64
87.9 ? 17.0
29.0 ? 4.9
102.4 ? 13.44,5
105.5 ? 9.45
0.97 ? 0.064,5
172.2 ? 6.54
84.3 ? 13.5
28.4 ? 4.4
102.6 ? 11.34
106.2 ? 8.5
0.96 ? 0.064
170.6 ? 6.54
79.2 ? 9.64
27.2 ? 3.1
101.0 ? 8.9
105.0 ? 6.0
0.96 ? 0.05
?0.001
?0.001
NS
?0.001
NS
?0.001
228288 275 21266
22.6 ? 8.7
25.6 ? 6.7
60.1 ? 7.9
3.29 ? 0.48
1.29 ? 0.10
24.4 ? 8.8
27.3 ? 6.4
59.4 ? 7.0
3.24 ? 0.42
1.28 ? 0.09
23.8 ? 9.1
26.8 ? 6.4
59.1 ? 8.2
3.14 ? 0.454
1.27 ? 0.11
23.9 ? 8.2
28.0 ? 6.34
56.5 ? 6.84
3.09 ? 0.454
1.25 ? 0.114
22.8 ? 6.9
28.6 ? 6.1
52.8 ? 4.84
2.90 ? 0.444
1.19 ? 0.114
NS
?0.001
?0.001
?0.001
?0.001
13.1 ? 5.1
30.1 ? 7.6
28.2 ? 4.0
0.98 ? 0.19
1.02 ? 0.10
14.2 ? 5.0
31.9 ? 6.9
28.3 ? 3.6
0.95 ? 0.18
1.01 ? 0.09
14.0 ? 5.1
31.2 ? 7.1
28.7 ? 4.4
0.90 ? 0.214
0.99 ? 0.10
14.4 ? 4.9
32.8 ? 6.94
27.6 ? 3.7
0.90 ? 0.194
0.99 ? 0.11
13.8 ? 4.3
33.3 ? 7.2
26.0 ? 2.44
0.86 ? 0.194
0.96 ? 0.104
NS
?0.001
?0.001
?0.001
?0.001
2.2 ? 0.9
31.8 ? 8.5
4.4 ? 0.7
2.5 ? 0.9
34.0 ? 7.9
4.6 ? 0.8
2.5 ? 1.04
33.7 ? 8.1
4.7 ? 0.94
2.6 ? 1.24
35.7 ? 7.94
4.5 ? 0.8
2.6 ? 0.8
36.5 ? 8.94
4.4 ? 0.7
?0.001
?0.001
?0.001
2.3 ? 1.2
20.7 ? 7.7
7.8 ? 1.3
0.49 ? 0.08
1.08 ? 0.13
2.6 ? 1.5
23.2 ? 8.3
7.7 ? 1.1
0.49 ? 0.08
1.07 ? 0.12
2.6 ? 1.5
23.5 ? 8.94
7.5 ? 1.2
0.48 ? 0.08
1.07 ? 0.13
2.6 ? 1.2
25.2 ? 8.04
7.0 ? 1.04
0.49 ? 0.08
1.04 ? 0.13
2.3 ? 0.9
24.7 ? 6.6
6.4 ? 0.84
0.44 ? 0.074
0.96 ? 0.124
NS
?0.001
?0.001
?0.001
?0.001
6.3 ? 2.7
22.3 ? 6.4
19.9 ? 3.2
1.29 ? 0.22
1.45 ? 0.13
6.5 ? 2.7
23.4 ? 6.3
19.4 ? 2.7
1.28 ? 0.20
1.44 ? 0.12
6.2 ? 2.8
22.6 ? 6.5
18.9 ? 2.94
1.25 ? 0.21
1.42 ? 0.14
6.0 ? 2.3
23.6 ? 6.1
17.9 ? 2.54
1.22 ? 0.20
1.41 ? 0.154
5.8 ? 2.1
24.3 ? 6.1
16.6 ? 1.94
1.13 ? 0.174
1.33 ? 0.134
NS
NS
?0.001
?0.001
?0.001
1.15 ? 0.141.16 ? 0.141.14 ? 0.161.18 ? 0.17 1.18 ? 0.19 NS
1FM, fat mass; FM, percentage FM; LM, lean mass; BMC, bone mineral content; aBMD, areal bone mineral density. FM was computed as FM ?
?FM/(FM ? LM ? BMC)? ? 100.
2Overall ANOVA.
3x ? ? SD (all such values).
4Significantly different from 35–44-y-old group (reference), P ? 0.001 (Dunnett’s 2-tailed post hoc test).
5Data are missing for 2 participants in this age group.
98
ATLANTIS ET AL
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 5

centralobesitywithaging.Evidencesupportingthisnotionstems
from prospective cohort studies that show a stronger and more
consistent association between central obesity measures and
earlymortalitythandoestheBMIinbothmenandwomen(6,7).
Compared with values in men aged 35–44 y, the values for
LM, BMC, and aBMD decrease but %FM increases with age.
Increases in mean values for %FM are clearly confounded by
decreases in LM, because no significant age-related differences
were seen for FM, which is most apparent for older men, con-
sistent with other cross-sectional (24–26, 29) and prospective
cohort studies (30). Thus, the greater risk of early mortality
observed in older adults with lower BMI (2, 31) may be partly
caused by changes in LM. In contrast, increases in abdominal
%FM with age were due to true differences in abdominal FM,
whereasnodifferencesorlargermeanvaluesforabdominalLM
were observed. Although these findings contrast with those for
whole-bodycomposition,overallobesity(BMI)andcentralobe-
sity are hazardous to health even after adjustment for age, and
therefore age-related changes in whole-body and abdominal
body composition are likely to be independent etiologic risk
factors. Well-designed prospective cohort studies are needed to
discover biological causes of age-related changes in body com-
position and the main chronic diseases and the associated mor-
tality burden.
For BMC and aBMD, age-related differences are most appar-
ent in men aged ?65 y, which is similar to findings in other
populations(23),andwhichisclinicallyimportant,giventhatthe
risk of osteoporotic fractures increases with decreasing aBMD
and with age (32). Osteoporotic fractures occur at clinically im-
portant rates for men and women aged ?70 y, but is not a well-
recognized men’s health issue, even though the associated in-
creased risk of early mortality is strongest for men (33). To help
avert the projected economic burden of osteoporotic fractures,
national awareness campaigns targeting healthy aging should
includerecommendationsforaBMDassessmentforallmenaged
?65 y to identify and treat those at risk (34).
We further tested whether demographic and potentially mod-
ifiable(lifestyle)variableswereassociated,aspotentialriskfac-
tors, with age-related differences in body composition. Increas-
ingagecategory,Australianbirth,andlowesttertilesforenergy,
TABLE 2
Age-related differences per decade for demographic and lifestyle risk factors in Australian men aged 35–81 y1
Risk factor35–44 y old 45–54 y old55–64 y old 65–74 y old
?75 y old
P2
Demographic
Subjects (n)
Income in AUD (%)
?$20 000
$20 001–$40 000
$40 001–$60 000
?$60 001
Born in Australia (%)
Lifestyle
Current smoker (%)
MET-h/wk (%)3
Lowest (0)
Middle (0.4–13.5)
Highest (13.6–285)
Vigorous physical activity (%)4
Macronutrients5
Subjects (n)
Energy intake (kJ/d)
Protein (% of macronutrients)
Total fat (% of macronutrients)
Saturated fat (% of macronutrients)
Polyunsaturated fat (% of
macronutrients)
Monounsaturated fat (% of
macronutrients)
Total carbohydrates (% of
macronutrients)
Sugars (% of macronutrients)
Starches (% of macronutrients)
Fiber (% of macronutrients)
Alcohol (% of macronutrients)
271325304 22471
?0.001
8823
27
27
24
62
50
37
48
35
13
17
32
42
76
20
27
46
70
7
64
5665
?0.001
272118 128
?0.001
NS
37
31
32
21.8
38
30
32
17.8
38
28
34
15.5
36
27
37
12.9
35
35
30
14.1 NS
270323 30422471
10 379 ? 33086,7
21.9 ? 3.3
19.4 ? 3.5
7.9 ? 2.0
2.9 ? 0.9
9620 ? 3065
22.0 ? 3.8
19.0 ? 3.4
7.6 ? 1.9
2.9 ? 1.0
9459 ? 3106
22.2 ? 3.8
18.6 ? 3.9
7.3 ? 2.07
3.0 ? 1.1
8590 ? 27537
21.5 ? 3.6
18.2 ? 3.4
7.2 ? 1.97
3.1 ? 1.1
8187 ? 23397
21.6 ? 4.4
18.2 ? 3.8
7.3 ? 2.1
3.1 ? 1.2
?0.001
NS
NS
?0.001
NS
6.9 ? 1.36.8 ? 1.3 6.7 ? 1.6 6.4 ? 1.47
6.3 ? 1.4
?0.001
49.6 ? 5.249.3 ? 6.349.6 ? 6.451.0 ? 6.5 50.7 ? 5.3NS
20.7 ? 5.7
28.7 ? 4.7
4.6 ? 1.1
4.5 ? 4.3
20.8 ? 5.7
28.2 ? 5.1
4.8 ? 1.3
4.9 ? 5.4
21.0 ? 5.8
28.3 ? 4.8
5.3 ? 1.57
4.3 ? 4.9
22.8 ? 6.37
27.9 ? 4.9
5.5 ? 1.67
3.7 ? 4.7
23.1 ? 5.7
27.3 ? 4.6
5.7 ? 1.67
3.7 ? 4.1
?0.001
NS
?0.001
NS
1AUD, Australian dollars; MET, metabolic equivalent.
2Overall ANOVA or chi-square analysis.
3Computedastotalleisure-timephysicalactivitymultipliedbyintensityweights(3.5forwalking,5.0formoderate-intensity,and7.5forvigorous-intensity
exercise) to generate tertile groups.
4Proportion who reported participating in any vigorous leisure-time physical activity.
5Macronutrient intakes (in g/d) were computed as a percentage of total energy-providing macronutrients listed.
6x ? ? SD (all such values).
7Significantly different from 35–44-y-old group (reference), P ? 0.001 (Dunnett’s 2-tailed post hoc test).
LIFESTYLE- AND AGE-RELATED DIFFERENCES IN BODY COMPOSITION
99
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 6

TABLE 3
Prevalence odds ratios (ORs) for highest to lowest (reference) percentage whole-body fat mass (%FM) tertile associated with demographic and lifestyle
risk factors in Australian men aged 35–81 y1
Risk factors Unadjusted
Prevalence for highest
%FM tertile
OR2
Unadjusted
(n ? 1046–1050)
Age-adjusted
(n ? 1046–1050)
Multiple-adjusted
(n ? 1045)
n%
Age (y)
?75
65–74
55–64
45–54
35–44
Household income (AUD/y)
?$20 000
$20 001–$40 000
$40 001–$60 000
?$60 001
Born in Australia
Yes
No
Smoker
Yes
No
MET-h/wk5
Lowest (0)
Middle (0.4–13.5)
Highest (13.6–285)
Vigorous physical activity6
Yes
No
Energy intake (kJ/d)
Lowest (2032–7917)
Middle (7918–10 319)
Highest (10 320–23 237)
Macronutrients7
Protein (% of macronutrients)
Lowest (12–20)
Middle (21–23)
Highest (24–43)
Fat (% of macronutrients)
Lowest (6–17)
Middle (18–20)
Highest (21–32)
Carbohydrate (% of macronutrients)
Lowest (21–48)
Middle (49–53)
Highest (54–69)
Fiber (% of macronutrients)
Lowest (2–4)
Middle (5–6)
Highest (7–12)
Alcohol (% of macronutrients)
Lowest (0–1)
Middle (2–5)
Highest (6–33)
25/153
76/55
83/98
94/94
59/97
38.5
36.0
31.2
33.1
26.3
2.74 (1.34, 5.61)4
2.27 (1.41, 3.65)
1.39 (0.90, 2.15)
1.64 (1.07, 2.53)
1
3.10 (1.40, 6.88)
2.60 (1.46, 4.63)
1.54 (0.95, 2.50)
1.73 (1.09, 2.73)
1
91/70
81/101
71/92
94/96
38.6
29.7
29.5
31.3
1.33 (0.87, 2.02)
0.82 (0.54, 1.23)
0.79 (0.52, 1.20)
1
0.98 (0.60, 1.58)
0.66 (0.43, 1.03)
0.76 (0.50, 1.17)
1
1.06 (0.63, 1.78)
0.68 (0.43, 1.09)
0.78 (0.50, 1.22)
1
241/227
96/132
34.4
27.5
1.46 (1.06, 2.01)
1
1.58 (1.14, 2.19)
1
1.61 (1.15, 2.27)
1
45/88
292/271
24.2
33.8
0.47 (0.32, 0.70)
1
0.52 (0.35, 0.78)
1
0.37 (0.24, 0.58)
1
136/131
105/96
96/132
34.6
34.7
37.1
1.43 (1.00, 2.04)
1.50 (1.03, 2.20)
1
1.45 (1.01, 2.07)
1.53 (1.04, 2.25)
1
1.30 (0.84, 2.01)
1.32 (0.85, 2.05)
1
48/77
289/282
26.1
33.4
0.61 (0.41, 0.90)
1
0.64 (0.43, 0.95)
1
0.76 (0.47, 1.24)
1
123/117
115/112
98/129
34.4
33.1
28.6
1.38 (0.96, 1.99)
1.35 (0.93, 1.96)
1
1.22 (0.84, 1.78)
1.27 (0.88, 1.85)
1
1.50 (1.00, 2.25)
1.40 (0.95, 2.07)
1
100/143
112/121
124/94
27.9
32.2
36.4
0.53 (0.37, 0.77)
0.70 (0.48, 1.02)
1
0.50 (0.34, 0.72)
0.69 (0.48, 1.01)
1
0.71 (0.43, 1.17)
0.87 (0.56, 1.34)
1
100/131
114/119
122/108
28.0
32.6
35.8
0.68 (0.47, 0.98)
0.85 (0.59, 1.22)
1
0.61 (0.42, 0.88)
0.77 (0.53, 1.12)
1
1.01 (0.62, 1.66)
1.03 (0.68, 1.57)
1
120/93
114/120
102/144
36.3
32.7
27.9
1.82 (1.26, 2.64)
1.34 (0.94, 1.92)
1
2.00 (1.37, 2.92)
1.45 (1.00, 2.10)
1
1.92 (1.02, 3.63)
1.30 (0.82, 2.07)
1
109/107
120/103
107/148
32.2
34.3
29.8
1.41 (0.98, 2.03)
1.61 (1.12, 2.31)
1
1.72 (1.18, 2.52)
1.89 (1.30, 2.74)
1
2.07 (1.32, 3.26)
2.05 (1.35, 3.09)
1
123/124
114/122
99/111
34.6
32.3
29.2
1.11 (0.77, 1.61)
1.05 (0.72, 1.52)
1
1.03 (0.71, 1.50)
1.05 (0.72, 1.53)
1
1.40 (0.82, 2.37)
1.33 (0.82, 2.15)
1
1AUD,Australiandollars;MET,metabolicequivalent.Thevariationsinnin2ofthecategoriesofadjustmentareduetomissingobservationsfor?1risk
factor variable.
2Determined by using unadjusted, age-adjusted, and multiple risk factor–adjusted (includes all demographic and lifestyle risk factor variables tabulated)
logistic regression analyses.
3Highest/lowest (all such values).
495% CI in parentheses (all such values).
5Computedastotalleisure-timephysicalactivitymultipliedbyintensityweights(3.5forwalking,5.0formoderate-intensity,and7.5forvigorous-intensity
exercise) to generate tertile groups.
6Those who reported participating in any vigorous leisure-time physical activity.
7Macronutrient intakes (in g/d) were computed as a percentage of total energy-providing macronutrients listed, to generate tertile groups.
100
ATLANTIS ET AL
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 7

TABLE 4
Prevalence odds ratios (ORs) for highest to lowest (reference) percentage abdominal fat mass (abdominal %FM) tertile associated with demographic and
lifestyle risk factors in Australian men aged 35–81 y1
Risk factors Unadjusted
Prevalence for highest
abdominal %FM tertile
OR2
Unadjusted
(n ? 1064–8)
Age-adjusted
(n ? 1064–8)
Multiple-adjusted
(n ? 1063)
n%
Age (y)
?75
65–74
55–64
45–54
35–44
Household income (AUD/y)
?$20 000
$20 001–$40 000
$40 001–$60 000
?$60 001
Born in Australia
Yes
No
Smoker
Yes
No
MET-h/wk5
Lowest (0)
Middle (0.4–13.5)
Highest (13.6–285)
Vigorous physical activity6
Yes
No
Energy intake (kJ/d)
Lowest (2032–7917)
Middle (7918–10 319)
Highest (10 320–23 237)
Macronutrients7
Protein (% of macronutrients)
Lowest (12–20)
Middle (21–23)
Highest (24–43)
Fat (% of macronutrients)
Lowest (6–17)
Middle (18–20)
Highest (21–32)
Carbohydrate (% of macronutrients)
Lowest (21–48)
Middle (49–53)
Highest (54–69)
Fiber (% of macronutrients)
Lowest (2–4)
Middle (5–6)
Highest (7–12)
Alcohol (% of macronutrients)
Lowest (0–1)
Middle (2–5)
Highest (6–33)
33/163
81/53
86/91
97/96
59/99
50.0
38.2
31.4
33.7
25.9
3.46 (1.76, 6.82)4
2.56 (1.60, 4.12)
1.59 (1.02, 2.45)
1.70 (1.10, 2.60)
1
3.58 (1.68, 7.66)
2.69 (1.51, 4.80)
1.67 (1.03, 2.71)
1.69 (1.07, 2.66)
1
94/75
92/88
69/86
101/106
39.0
33.6
28.2
32.8
1.32 (0.87, 1.98)
1.10 (0.74, 1.64)
0.84 (0.55, 1.28)
1
0.89 (0.55, 1.42)
0.84 (0.55, 1.29)
0.80 (0.53, 1.23)
1
1.05 (0.63, 1.75)
0.94 (0.59, 1.48)
0.85 (0.54, 1.32)
1
246/233
110/122
34.5
31.0
1.17 (0.86, 1.60)
1
1.27 (0.92, 1.76)
1
1.29 (0.92, 1.81)
1
39/89
317/266
20.6
36.1
0.37 (0.24, 0.55)
1
0.41 (0.27, 0.62)
1
0.29 (0.18, 0.45)
1
137/124
112/87
107/144
34.4
36.1
29.7
1.49 (1.05, 2.11)
1.73 (1.19, 2.52)
1
1.52 (1.07, 2.17)
1.78 (1.22, 2.61)
1
1.44 (0.94, 2.21)
1.63 (1.06, 2.50)
1
53/81
303/274
28.3
34.4
0.59 (0.40, 0.87)
1
0.63 (0.43, 0.93)
1
0.79 (0.49, 1.26)
1
134/121
122/101
99/133
36.9
34.5
28.4
1.49 (1.04, 2.13)
1.62 (1.12, 2.35)
1
1.27 (0.88, 1.84)
1.51 (1.04, 2.20)
1
1.57 (1.05, 2.34)
1.71 (1.15, 2.54)
1
107/134
113/122
135/99
29.6
31.7
28.9
0.59 (0.41, 0.84)
0.68 (0.47, 0.98)
1
0.54 (0.38, 0.79)
0.67 (0.46, 0.97)
1
0.79 (0.48, 1.30)
0.86 (0.56, 1.33)
1
107/138
116/105
132/112
29.5
32.7
37.9
0.66 (0.46, 0.94)
0.94 (0.65, 1.35)
1
0.57 (0.40, 0.83)
0.85 (0.59, 1.24)
1
0.92 (0.57, 1.50)
1.15 (0.75, 1.75)
1
133/95
116/114
106/145
39.3
32.5
28.7
1.92 (1.33, 2.75)
1.39 (0.97, 1.99)
1
2.15 (1.48, 3.12)
1.52 (1.05, 2.19)
1
2.07 (1.10, 3.87)
1.41 (0.89, 2.24)
1
120/104
121/111
114/140
34.7
33.8
31.5
1.42 (0.99, 2.03)
1.34 (0.94, 1.91)
1
1.82 (1.24, 2.66)
1.61 (1.11, 2.34)
1
2.07 (1.32, 3.25)
1.67 (1.11, 2.52)
1
125/123
115/117
115/114
34.6
32.0
33.3
1.01 (0.70, 1.44)
0.97 (0.68, 1.40)
1
0.92 (0.64, 1.33)
0.97 (0.67, 1.41)
1
1.21 (0.72, 2.03)
1.22 (0.76, 1.95)
1
1AUD,Australiandollars;MET,metabolicequivalent.Thevariationsinnin2ofthecategoriesofadjustmentareduetomissingobservationsfor?1risk
factor variable.
2Determined by using unadjusted, age-adjusted, and multiple risk factor–adjusted (includes all demographic and lifestyle risk factor variables tabulated)
logistic regression analyses.
3Highest/lowest (all such values).
495% CI in parentheses (all such values).
5Computedastotalleisure-timephysicalactivitymultipliedbyintensityweights(3.5forwalking,5.0formoderate-intensity,and7.5forvigorous-intensity
exercise) to generate tertile groups.
6Those who reported participating in any vigorous leisure-time physical activity.
7Macronutrient intakes (in g/d) were computed as a percentage of total energy-providing macronutrients listed to generate tertile groups.
LIFESTYLE- AND AGE-RELATED DIFFERENCES IN BODY COMPOSITION
101
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 8

TABLE 5
Prevalence odds ratios (ORs) for highest to lowest (reference) whole-body areal bone mineral density (aBMD) tertile associated with demographic and
lifestyle risk factors in Australian men aged 35–81 y1
Risk factorsUnadjusted
Prevalence for highest
whole-body aBMD tertile
OR2
Unadjusted
(n ? 1063–7)
Age-adjusted
(n ? 1063–7)
Multiple-adjusted
(n ? 1062)
n%
Age (y)
?75
65–74
55–64
45–54
35–44
Household income (AUD/y)
?$20 000
$20 001–$40 000
$40 001–$60 000
?$60 001
Born in Australia
Yes
No
Smoker
Yes
No
MET-h/wk5
Lowest (0)
Middle (0.4–13.5)
Highest (13.6–285)
Vigorous physical activity6
Yes
No
Energy intake (kJ/d)
Lowest (2032–7917)
Middle (7918–10 319)
Highest (10 320–23 237)
Macronutrients7
Protein (% of macronutrients)
Lowest (12–20)
Middle (21–23)
Highest (24–43)
Fat (% of macronutrients)
Lowest (6–17)
Middle (18–20)
Highest (21–32)
Carbohydrate (% of macronutrients)
Lowest (21–48)
Middle (49–53)
Highest (54–69)
Fiber (% of macronutrients)
Lowest (2–4)
Middle (5–6)
Highest (7–12)
Alcohol (% of macronutrients)
Lowest (0–1)
Middle (2–5)
Highest (6–33)
Body mass (kg)
Lowest (46.5–78.2)
Middle (78.3–90.2)
Highest (90.3–140.8)
9/393
58/86
91/97
102/73
90/59
13.6
27.5
33.1
35.5
39.5
0.15 (0.07, 0.34)4
0.44 (0.28, 0.71)
0.62 (0.40, 0.95)
0.92 (0.59, 1.43)
1
0.22 (0.09, 0.53)
0.55 (0.30, 1.01)
0.65 (0.39, 1.09)
0.80 (0.49, 1.32)
1
62/105
74/102
95/74
119/73
25.9
26.9
38.8
38.6
0.36 (0.24, 0.56)
0.45 (0.29, 0.68)
0.79 (0.52, 1.20)
0.51 (0.31, 0.83)
0.57 (0.36, 0.88)
0.85 (0.56, 1.30)
1
0.67 (0.38, 1.17)
0.76 (0.46, 1.25)
0.92 (0.58, 1.47)
11
257/211
93/143
36.1
26.2
1.87 (1.36, 2.57)1.76 (1.27, 2.44)
1
1.58 (1.10, 2.26)
11
55/77
295/277
28.9
33.6
0.67 (0.46, 0.98) 0.56 (0.37, 0.83)
1
0.67 (0.42, 1.05)
11
129/129
103/101
118/124
32.3
33.2
33.0
1.05 (0.74, 1.49)
1.07 (0.74, 1.56)
1.04 (0.73, 1.48)
1.08 (0.74, 1.57)
1
1.35 (0.85, 2.12)
1.18 (0.75, 1.85)
11
72/48
278/306
38.5
31.6
1.65 (1.11, 2.46)1.54 (1.03, 2.32)
1
1.82 (1.09, 3.05)
11
111/138
115/120
123/96
30.5
32.6
35.3
0.63 (0.44, 0.90)
0.75 (0.52, 1.08)
0.76 (0.52, 1.10)
0.82 (0.56, 1.19)
1
0.88 (0.57, 1.35)
0.73 (0.48, 1.11)
11
107/128
125/108
117/118
29.6
35.0
33.8
0.84 (0.59, 1.21)
1.17 (0.81, 1.68)
0.90 (0.62, 1.30)
1.20 (0.83, 1.73)
1
0.70 (0.41, 1.20)
1.09 (0.69, 1.73)
11
117/129
113/116
119/109
32.2
32.0
34.1
0.83 (0.58, 1.19)
0.89 (0.62, 1.29)
0.94 (0.65, 1.36)
0.94 (0.65, 1.37)
1
0.73 (0.44, 1.23)
0.90 (0.58, 1.42)
11
108/123
117/114
124/116
31.9
32.9
33.7
0.82 (0.57, 1.18)
0.96 (0.67, 1.38)
0.74 (0.51, 1.07)
0.92 (0.63, 1.33)
1
0.37 (0.19, 0.72)
0.60 (0.36, 0.98)
11
119/114
120/114
110/126
34.3
33.7
30.4
1.20 (0.83, 1.72)
1.21 (0.84, 1.73)
0.92 (0.63, 1.35)
1.02 (0.70, 1.48)
1
1.04 (0.65, 1.68)
0.95 (0.61, 1.47)
11
120/117
118/118
111/119
33.3
33.0
32.1
1.10 (0.76, 1.58)
1.07 (0.75, 1.54)
1.24 (0.85, 1.80)
1.07 (0.74, 1.55)
1
0.91 (0.52, 1.58)
0.71 (0.43, 1.16)
11
54/196
121/106
175/52
15.3
33.9
49.2
0.08 (0.05, 0.13)
0.34 (0.23, 0.51)
0.09 (0.06, 0.13)
0.38 (0.25, 0.57)
1
0.09 (0.05, 0.14)
0.37 (0.24, 0.56)
11
1AUD,Australiandollars;MET,metabolicequivalent.Thevariationsinnin2ofthecategoriesofadjustmentareduetomissingobservationsfor?1risk
factor variable.
2Determined by using unadjusted, age-adjusted, and multiple risk factor–adjusted (includes all demographic and lifestyle risk factor variables tabulated)
logistic regression analyses.
3Highest/lowest (all such values).
495% CI in parentheses (all such values).
5Computedastotalleisure-timephysicalactivitymultipliedbyintensityweights(3.5forwalking,5.0formoderate-intensity,and7.5forvigorous-intensity
exercise) to generate tertile groups.
6Those who reported participating in any vigorous leisure-time physical activity.
7Macronutrient intakes (in g/d) were computed as a percentage of total energy-providing macronutrients listed, to generate tertile groups.
102
ATLANTIS ET AL
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 9

carbohydrate, and fiber intake in particular were found to be
significantriskfactors,whereassmokingwasastrongprotective
factorforbeinginthehighesttertileforwhole-bodyandabdom-
inal %FM. Because most immigrants were born in Europe (11),
where the diet is inversely associated with obesity (35), obesity
rates in countries of origin (9) or less exposure to the “Western-
ized” Australian culture may partly explain the greater risk as-
sociated with being born in Australia.
Low fiber intake was one of the most reliable factors found to
increasetheriskofbeinginthehighesttertilefor%FM.Although
we found no previous epidemiologic studies that comprehen-
sivelyassessedenergy-providingmacronutrientsandbodycom-
position,thereisevidencethatfiberintakeisassociatedwithless
gain in weight and waist circumference (36, 37). Our finding of
no positive association between energy intake and body compo-
sitionaccordswiththefindingsofothers(38)butcontradictsthe
conventional viewpoint with respect to energy balance and obe-
sity.Experimentalevidenceshowsthatdietaryenergyrestriction
causes substantial loss of both LM and FM (39), which may
explain this apparent inconsistency. Furthermore, the long-term
effectoflowenergyintakeon%FMiscurrentlyunknown.Thus,
we hypothesize that chronic low energy intake may cause unfa-
vorable body composition changes with aging, resulting in
greater %FM due to substantial losses of LM.
The finding of an inverse association between carbohydrate
intake and %FM is consistent with one cohort study that found
highcarbohydrateintaketobeassociatedwithlessannualgainin
BMI and waist circumference (40), and it suggests that a high-
carbohydrate diet may be associated with a lower gain in %FM
with aging. Smoking was found to be a protective factor against
being in the highest tertile %FM, which is consistent with find-
ings for other populations (41, 42), and is a finding that may be
due to greater fat utilization and appetite suppression compared
with not smoking (43). Our findings of no association between
bothtotalphysicalactivityandalcoholintakevariablesandbody
composition are broadly consistent with the existing epidemio-
logic literature (44–46).
Increasing age category, smoking, low carbohydrate intake,
and low body weight in particular were significant risk factors,
whereas Australian birth and vigorous physical activity were
protective factors for being in the highest tertile for aBMD.
Ethnicity effects for those born in other countries may explain
someoftheage-relateddifferencesinaBMD(47).Theincreased
risk of being in the lowest tertile for aBMD associated with
smoking, even after adjustment for body weight, is consistent
with the existing body of evidence (48), although the mecha-
nisms are not yet known. Our finding of a positive association
between carbohydrate intake and aBMD is also consistent with
the findings of others (49). It has been hypothesized that low
carbohydrateintakecausesacidosis,whichstimulatesexcessive
bone resorption to restore the blood’s pH, as a buffer. Further-
more, the lack of evidence for long-term safety (50) and the
potential risk of decreasing aBMD associated with low carbo-
hydrate intake should be of concern to nutritionists and medical
practitioners. Finally, vigorous physical activity was consis-
tently found to be a protective factor associated with aBMD,
which is in broad agreement with the findings of others (49).
Vigorousphysicalactivitymaybemoreeffectivethanmoderate
physical activity in stimulating bone deposition, because of the
greatermechanicalandcompressivestressesontheskeletalsys-
tem, similar to the hypothesized weight-bearing mechanisms of
body weight on aBMD. Thus, several demographic and lifestyle
variables were found to be risks as well as protective factors
associated with age-related differences in body composition
variables.
The strengths of this study include a large random sample
similar in characteristics to men from the general Australian
population, precise measures of body composition, and mea-
suredanthropometry.Further,thestudypopulationandthechar-
acteristicsofdropoutsandnonrespondershavebeensufficiently
described, and statistical models were adjusted for many impor-
tant covariates. Limitations of this study include the cross-
sectional design, the responder bias [although few differences
were found between responders, nonresponders, and drop-outs
(11)], seasonal effects from different periods of data collection,
self-report questionnaires used to collect information on all life-
style factors, generalizations restricted to men, missing risk fac-
tor variables not yet known, and different types of physical ac-
tivity that could not be assessed (eg, aerobic compared with
resistance).
In summary, we defined age-related differences in anthropo-
metric and body composition variables associated with lifestyle
factors in a random sample of men aged 35–81 y. Age-related
differences in whole-body %FM are mostly due to less LM,
whereasdifferencesinabdominal%FMareactuallyduetomore
FM. For whole-body and abdominal %FM, increasing age cat-
egory, Australian birth, and the lowest tertiles for energy, car-
bohydrate, and fiber intake in particular were found to be signif-
icant risk factors, whereas smoking was a strong protective
factor. For aBMD, increasing age category, smoking, and the
lowest tertiles for carbohydrate intake and, in particular, body
mass were significant risk factors, whereas Australian birth and
vigorous physical activity were protective factors. To promote
healthy body composition changes with aging, age-related dif-
ferences in %FM and aBMD are therefore associated with de-
mographic and lifestyle factors that should be targeted for inter-
vention.
Theauthorsacknowledgetheclinicandrecruitmentstaffsfortheirinvalu-
ableefforts.WeextendparticularthankstoJanetGrant,SandyPickering,andthe
staff of the North West Adelaide Health Study for all their assistance. We also
thank Chris Seaborn and Erika Bowden and the staff at the Department of
Nuclear Medicine, The Queen Elizabeth Hospital, for providing expertise and
assistance with dual-energy X-ray absorptiometry procedures.
Theauthors’responsibilitieswereasfollows:KA:dataanalysisandwrit-
ing of the manuscript; SAM, MTH, and AWT: data collection and contribu-
tions to the writing of the manuscript; GAW: the study design, securing of
funding, and contributions to the writing of the manuscript. None of the
authors had a personal or financial conflict of interest.
REFERENCES
1. Bogers RP, Bemelmans WJE, Hoogenveen RT, et al. Association of
overweightwithincreasedriskofcoronaryheartdiseasepartlyindepen-
dent of blood pressure and cholesterol levels: a meta-analysis of 21
cohort studies including more than 300 000 persons. Arch Intern Med
2007;167:1720–8.
2. Flegal KM, Graubard BI, Williamson DF, Gail MH. Cause-specific
excess deaths associated with underweight, overweight, and obesity.
JAMA 2007;298:2028–37.
3. AlleyDE,ChangVW.Thechangingrelationshipofobesityanddisabil-
ity, 1988–2004. JAMA 2007;298:2020–7.
4. Zamboni M, Mazzali G, Zoico E, et al. Health consequences of obesity
in the elderly: a review of four unresolved questions. Int J Obes (Lond)
2005;29:1011–29.
5. Janssen I, Heymsfield SB, Ross R. Low relative skeletal muscle mass
LIFESTYLE- AND AGE-RELATED DIFFERENCES IN BODY COMPOSITION
103
by guest on May 18, 2011
www.ajcn.org
Downloaded from

Page 10

(sarcopenia) in older persons is associated with functional impairment
and physical disability. J Am Geriatr Soc 2002;50:889–96.
6. Simpson JA, MacInnis RJ, Peeters A, Hopper JL, Giles GG, English
DR. A comparison of adiposity measures as predictors of all-cause
mortality:theMelbourneCollaborativeCohortStudy.Obesity2007;15:
994–1003.
7. Lahmann PH, Lissner L, Gullberg B, Berglund G. A prospective study
of adiposity and all-cause mortality: the Malmo Diet and Cancer Study.
Obes Res 2002;10:361–9.
8. Segal KR, Dunaif A, Gutin B, Albu J, Nyman A, Pi-Sunyer FX. Body
composition, not body weight, is related to cardiovascular disease risk
factors and sex hormone levels in men. J Clin Invest 1987;80:1050–5.
9. Silventoinen K, Sans S, Tolonen H, et al. Trends in obesity and energy
supply in the WHO MONICA Project. Int J Obes Relat Metab Disord
2004;28:710–8.
10. UnitedNations.Worldpopulationageing2007.NewYork,NY:United
Nations Publications, 2007. Internet: http://www.un.org/esa/population/
publications/WPA2007/wpp2007.htm (accessed 12 November 2007).
11. Martin SA, Haren MT, Middleton SM, Wittert GA, Members of the
Florey Adelaide Male Ageing S. The Florey Adelaide Male Ageing
Study (FAMAS): design, procedures, participants. BMC Public Health
2007;7:126.
12. American Association for Public Opinion Research. Standard defini-
tions:finaldispositionsofcasescodesandoutcomeratesforsurveys.4th
ed. Lenexa, KS: American Association for Public Opinion Research,
2006.
13. Australian Sports Commission. (Olds T, Norton K, eds.) Anthropomet-
rica: a textbook of body measurement for sports and health courses.
Sydney, Australia: UNSW Press, 1996.
14. Mazess RB, Barden HS. Evaluation of differences between fan-beam
and pencil-beam densitometers. Calc Tissue Int 2000;67:291–6.
15. Armstrong T, Bauman A, Davies J. Physical activity patterns of Aus-
tralianadults.Resultsfromthe1999NationalPhysicalActivitySurvey.
Canberra: Australian Institute of Health and Welfare, 2000.
16. CraigCL,MarshallAL,SjostromM,etal.Internationalphysicalactivity
questionnaire:12-countryreliabilityandvalidity.MedSciSportsExerc
2003;35:1381–95.
17. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical
activities: an update of activity codes and MET intensities. Med Sci
Sports Exerc 2000;32(suppl):S498–504.
18. GilesGG,IrelandPD.Dietaryquestionnaireforepidemiologicalstudies
(version 2). Melbourne, Australia: The Cancer Council of Victoria,
1996.
19. Hodge A, Patterson AJ, Brown WJ, et al. The Anti Cancer Council of
Victoria FFQ: relative validity of nutrient intakes compared with
weighedfoodrecordsinyoungtomiddle-agedwomeninastudyofiron
supplementation.AustrNZPublicHealth2000;24:576–83.(Published
erratum appears in Aust N Z J Public Health 2003;27:468.)
20. Woodward M. Modelling quantitative outcome variables. In: Epidemi-
ology.Studydesignanddataanalysis.2nded.BocaRaton,FL:Chapman
& Hall/CRC, 2005.
21. De Laet C, Kanis JA, Oden A, et al. Body mass index as a predictor of
fracture risk: a meta-analysis. Osteopor Int 2005;16:1330–8.
22. Chumlea WC, Guo SS, Kuczmarski RJ, et al. Body composition esti-
mates from NHANES III bioelectrical impedance data. Int J Obes Relat
Metab Disord 2002;26:1596–609.
23. ItoH,OhshimaA,OhtoN,etal.Relationbetweenbodycompositionand
age in healthy Japanese subjects. Eur J Clin Nutr 2001;55:462–70.
24. HeitmannBL.BodyfatintheadultDanishpopulationaged35–65years:
an epidemiological study. Int J Obes 1991;15:535–45.
25. Kyle UG, Genton L, Hans D, Karsegard L, Slosman DO, Pichard C.
Age-relateddifferencesinfat-freemass,skeletalmuscle,bodycellmass
and fat mass between 18 and 94 years. Eur J Clin Nutr 2001;55:663–72.
26. Shaw KA, Srikanth VK, Fryer JL, Blizzard L, Dwyer T, Venn AJ. Dual
energy X-ray absorptiometry body composition and aging in a
population-based older cohort. Int J Obes 2007;31:279–84.
27. Welch GW, Sowers MR. The interrelationship between body topology
and body composition varies with age among women. J Nutr 2000;130:
2371–7.
28. Thomas GN, Tomlinson B, Hong AWL, Hui SSC. Age-related anthro-
pometric remodelling resulting in increased and redistributed adiposity
is associated with increases in the prevalence of cardiovascular risk
factors in Chinese subjects. Diabetes Metab Res Rev 2006;22:72–8.
29. Gomez JM, Maravall FJ, Gomez N, Navarro MA, Casamitjana R, Soler
J. Interactions between serum leptin, the insulin-like growth factor-I
system,andsex,age,anthropometricandbodycompositionvariablesin
a healthy population randomly selected. Clin Endocrinol 2003;58:
213–9.
30. Woo J, Ho SC, Sham A. Longitudinal changes in body mass index and
body composition over 3 years and relationship to health outcomes in
Hong Kong Chinese age 70 and older. J Am Geriatr Soc 2001;49:737–
46.
31. Wannamethee SG, Shaper AG, Lennon L, Whincup PH. Decreased
musclemassandincreasedcentraladiposityareindependentlyrelatedto
mortality in older men. Am J Clin Nutr 2007;86:1339–46.
32. JohnellO,KanisJA,OdenA,etal.PredictivevalueofBMDforhipand
otherfractures.JBoneMinerRes2005;20:1185–94.(Publishederratum
appears J Bone Miner Res 2007;22:774.)
33. Center JR, Nguyen TV, Schneider D, Sambrook PN, Eisman JA. Mor-
talityafterallmajortypesofosteoporoticfractureinmenandwomen:an
observational study. Lancet 1999;353:878–82.
34. BurgeR,Dawson-HughesB,SolomonDH,WongJB,KingA,Tosteson
A. Incidence and economic burden of osteoporosis-related fractures in
the United States, 2005–2025. J Bone Miner Res 2007;22:465–75.
35. PanagiotakosDB,ChrysohoouC,PitsavosC,StefanadisC.Association
between the prevalence of obesity and adherence to the Mediterranean
diet: the ATTICA study. Nutrition 2006;22:449–56.
36. Koh-BanerjeeP,ChuN-F,SpiegelmanD,etal.Prospectivestudyofthe
association of changes in dietary intake, physical activity, alcohol con-
sumption, and smoking with 9-y gain in waist circumference among
16 587 US men. Am J Clin Nutr 2003;78:719–27.
37. Ludwig DS, Pereira MA, Kroenke CH, et al. Dietary fiber, weight gain,
and cardiovascular disease risk factors in young adults. JAMA 1999;
282:1539–46.
38. MitchellD,HaanMN,SteinbergFM,VisserM.Bodycompositioninthe
elderly: the influence of nutritional factors and physical activity. J Nutr
Health Aging 2003;7:130–9.
39. PritchardJE,NowsonCA,WarkJD.Aworksiteprogramforoverweight
middle-aged men achieves lesser weight loss with exercise than with
dietary change. J Am Dietet Assoc 1997;97:37–42.
40. Newby PK, Muller D, Hallfrisch J, Qiao N, Andres R, Tucker KL.
Dietary patterns and changes in body mass index and waist circumfer-
ence in adults. Am J Clin Nutr 2003;77:1417–25.
41. ElmstahlS,JarnbladG,StavenowL,etal.Bodycompositionanddietary
habits in 80-year-old smoking men without cardiovascular disease. Ag-
ing Clin Exp Res 1991;3:269–77.
42. Samaras K, Kelly PJ, Spector TD, Chiano MN, Campbell LV. Tobacco
smoking and oestrogen replacement are associated with lower total and
central fat in monozygotic twins. Int J Obes Relat Metab Disord 1998;
22:149–56.
43. BishopC,ParkerGC,CoscinaDV.Systemicnicotinealterswhole-body
fat utilization in female rats. Physiol Behav 2004;80:563–7.
44. Raguso CA, Kyle U, Kossovsky MP, et al. A 3-year longitudinal study
on body composition changes in the elderly: role of physical exercise.
Clin Nutr 2006;25:573–80.
45. Hughes VA, Frontera WR, Roubenoff R, Evans WJ, Singh MAF. Lon-
gitudinalchangesinbodycompositioninoldermenandwomen:roleof
bodyweightchangeandphysicalactivity.AmJClinNutr2002;76:473–
81.
46. Starling RD, Ades PA, Poehlman ET. Physical activity, protein intake,
and appendicular skeletal muscle mass in older men. Am J Clin Nutr
1999;70:91–6.
47. AraujoAB,TravisonTG,HarrisSS,HolickMF,TurnerAK,McKinlay
JB. Race/ethnic differences in bone mineral density in men. Osteoporos
Int 2007;18:943–53.
48. WardKD,KlesgesRC.Ameta-analysisoftheeffectsofcigarettesmok-
ing on bone mineral density. Calc Tissue Int 2001;68:259–70.
49. McGuiganFEA,MurrayL,GallagherA,etal.Geneticandenvironmen-
tal determinants of peak bone mass in young men and women. J Bone
Miner Res 2002;17:1273–9.
50. Adam-Perrot A, Clifton P, Brouns F. Low-carbohydrate diets: nutri-
tional and physiological aspects. Obes Rev 2006;7:49–58.
104
ATLANTIS ET AL
by guest on May 18, 2011
www.ajcn.org
Downloaded from