Content uploaded by Kristi Reynolds
Author content
All content in this area was uploaded by Kristi Reynolds on Sep 05, 2014
Content may be subject to copyright.
ORIGINAL INVESTIGATION
The Obese Without Cardiometabolic Risk Factor
Clustering and the Normal Weight With
Cardiometabolic Risk Factor Clustering
Prevalence and Correlates of 2 Phenotypes Among the US Population
(NHANES 1999-2004)
Rachel P. Wildman, PhD; Paul Muntner, PhD; Kristi Reynolds, PhD; Aileen P. McGinn, PhD;
Swapnil Rajpathak, MD, DrPH; Judith Wylie-Rosett, EdD; MaryFran R. Sowers, PhD
Background:The prevalence and correlates of obese in-
dividuals who are resistant to the development of the adi-
posity-associated cardiometabolic abnormalities and nor-
mal-weight individuals who display cardiometabolic risk
factor clustering are not well known.
Methods:The prevalence and correlates of combined
body mass index (normal weight, ⬍25.0; overweight,
25.0-29.9; and obese, ⱖ30.0 [calculated as weight in ki-
lograms divided by height in meters squared]) and car-
diometabolic groups (metabolically healthy, 0 or 1 car-
diometabolic abnormalities; and metabolically abnormal,
ⱖ2 cardiometabolic abnormalities) were assessed in a
cross-sectional sample of 5440 participants of the Na-
tional Health and Nutrition Examination Surveys 1999-
2004. Cardiometabolic abnormalities included elevated
blood pressure; elevated levels of triglycerides, fasting
plasma glucose, and C-reactive protein; elevated homeo-
stasis model assessment of insulin resistance value; and
low high-density lipoprotein cholesterol level.
Results:Among US adults 20 years and older, 23.5%
(approximately 16.3 million adults) of normal-weight
adults were metabolically abnormal, whereas 51.3% (ap-
proximately 35.9 million adults) of overweight adults
and 31.7% (approximately 19.5 million adults) of obese
adults were metabolically healthy. The independent cor-
relates of clustering of cardiometabolic abnormalities
among normal-weight individuals were older age, lower
physical activity levels, and larger waist circumference.
The independent correlates of 0 or 1 cardiometabolic
abnormalities among overweight and obese individuals
were younger age, non-Hispanic black race/ethnicity,
higher physical activity levels, and smaller waist cir-
cumference.
Conclusions:Among US adults, there is a high preva-
lence of clustering of cardiometabolic abnormalities
among normal-weight individuals and a high preva-
lence of overweight and obese individuals who are meta-
bolically healthy. Further study into the physiologic
mechanisms underlying these different phenotypes and
their impact on health is needed.
Arch Intern Med. 2008;168(15):1617-1624
THE VARIATION IN META-
bolic and cardiovascular
disease (CVD) risk factors
observed among individu-
als of similar body mass in-
dex (BMI), and recent studies indicating
that individuals’ CVD risk may depend
jointly on their body size and metabolic
profile1-3 has led to increasing recognition
that the disease risks associated with obe-
sity may not be uniform. This has resulted
in the investigation of body size pheno-
types. One recognized body size pheno-
type is the metabolically healthy but obese
individual, sometimes referred to as “un-
complicated” obesity.4Although obese (BMI
ⱖ30 [calculated as weight in kilograms
divided by height in meters squared]), this
subset of individuals appears to be rela-
tively resistant to the development of the
adiposity-associated cardiometabolic ab-
normalities that increase CVD risk.5,6 A sec-
ond body size phenotype includes indi-
viduals with normal weight (BMI ⬍25),
who express cardiometabolic abnormali-
ties often associated with being over-
weight and obese.7
CME available online at
www.jamaarchivescme.com
and questions on page 1603
For editorial comment
see page 1607
Author Affiliations:
Department of Epidemiology
and Population Health, Albert
Einstein College of Medicine,
Bronx, New York
(Drs Wildman, McGinn,
Rajpathak, and Wylie-Rosett);
Department of Community and
Preventive Medicine, Mount
Sinai School of Medicine, New
York, New York (Dr Muntner);
Research and Evaluation, Kaiser
Permanente Southern
California, Pasadena
(Dr Reynolds); and Department
of Epidemiology, University of
Michigan, Ann Arbor
(Dr Sowers).
(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 15), AUG 11/25, 2008 WWW.ARCHINTERNMED.COM
1617
©2008 American Medical Association. All rights reserved.
at Albert Einstein College of Medicine, on August 18, 2008 www.archinternmed.comDownloaded from
Despite the potential implications of these pheno-
types for disease risk, little is known regarding their
prevalence and correlates. Therefore, the purpose of
the present study was 3-fold: (1) to determine the
prevalence of each of 6 body size phenotypes (normal
weight with and without cardiometabolic abnormali-
ties, overweight with and without cardiometabolic
abnormalities, and obese with and without cardio-
metabolic abnormalities) among a nationally represen-
tative sample of the US adult population, (2) to exam-
ine the demographic and behavioral correlates of
expressing clustered cardiometabolic abnormalities if
normal weight, and (3) to examine the demographic
and behavioral correlates of being metabolically
healthy (no cardiometabolic risk factor clustering) if
overweight or obese.
METHODS
STUDY PARTICIPANTS
The National Health and Nutrition Examination Surveys
(NHANES) 1999-2004 included a nationally representative
sample of the US noninstitutionalized, civilian population iden-
tified through a stratified, multistage probability sampling de-
sign. As described in detail on the Web site of the National Cen-
ter for Health Statistics,8non-Hispanic blacks and Mexican
Americans were oversampled in NHANES 1999-2004 to pro-
vide stable estimates for these groups. The present analyses are
limited to individuals 20 years and older.
Of the 6036 participants of the NHANES 1999-2004 who
were 20 years and older, and from whom blood samples were
obtained after an overnight fast of 8 or more hours, 409 were
excluded because of a positive history of self-reported CVD at
baseline, and 187 were excluded for being underweight (BMI
⬍18.5). Therefore, the final sample for these analyses in-
cluded 5440 participants.
MEASUREMENT OF DEMOGRAPHIC,
HEALTH BEHAVIOR, AND PHYSICAL FACTORS
Age, sex, race/ethnicity, smoking status, physical activity, and al-
cohol intake were assessed by self-report. Participants who had
not smoked 100 or more cigarettes in their lifetimes were con-
sidered never smokers; participants who had smoked 100 or more
cigarettes in their lifetimes were considered current smokers if
they answered “some days” or “every day” to the question “Do
you smoke now?” and former smokers if they answered “not at
all.” Moderate or vigorous leisure-time physical activity was as-
sessed via assessment of the number of times an activity was per-
formed per day, per week, or per month, depending on the ac-
tivity, the number of minutes the activity was performed each time,
and the level of exertion reported. Metabolic equivalents task
(MET) scores were then assigned.8,9 If moderate or vigorous physi-
cal activity was not reported, then a 0 value was assigned. Owing
to a large prevalence of MET scores of 0, physical activity was
categorized into a 5-level variable representing an MET score of
0 plus quartiles of METs scores for those reporting moderate or
vigorous activity. Alcohol intake was split into 4 categories (non-
drinkers, ⬍1 drink per day, 1-2 drinks per day, and ⬎2 drinks
per day). Nondrinkers were classified as those who reported con-
suming less than 12 alcoholic drinks in their lifetime, and drink-
ers reported the frequency of drinking alcoholic beverages in the
past 12 months and the mean number of drinks consumed on
those occasions.8The use of antihypertensive, lipid-lowering, and
antidiabetic medications were also assessed by self-report.
Height was measured using a fixed stadiometer with a ver-
tical backboard and movable headboard, with participants stand-
ing on the floor. Weight was taken by asking each participant to
stand on the center of the platform of a Toledo digital scale (Met-
tler-Toledo Inc, Columbus, Ohio) while wearing underwear, a
disposable gown, and foam slippers. Based on their BMI, indi-
viduals were classified as being normal weight (BMI, 18.5-
24.9), overweight (BMI, 25.0-29.9), or obese (BMI, ⱖ30.0). Waist
circumference was measured to the nearest 0.1 cm at the level
of the iliac crest at the end of normal respiration.8
MEASUREMENT OF
CARDIOMETABOLIC COMPONENTS
The 6 metabolic components measured include elevated blood
pressure; elevated levels of triglycerides, fasting glucose, and
high-sensitivity C-reactive protein; elevated homeostasis model
assessment of insulin resistance value; and reduced high-
density lipoprotein cholesterol (HDL-C) level. Seated systolic
and diastolic blood pressures were measured using a mercury
sphygmomanometer according to the American Heart Asso-
ciation’s recommendations.10 Up to 3 measurements were av-
eraged for systolic and diastolic blood pressures. High-density
lipoprotein cholesterol and triglycerides were measured enzy-
matically, and glucose was also measured enzymatically via a
hexokinase reaction.8Insulin was measured by immunoenzy-
matic assay.8Homeostasis model assessment was used to evalu-
ate insulin resistance using the following formula:
Fasting Serum Insulin Level (Microunits per Milliliter)
⫻Fasting Plasma Glucose Level (Millimoles per Liter)/22.5.
High-sensitivity C-reactive protein was measured by latex-
enhanced nephelometry.
BODY SIZE PHENOTYPE DEFINITIONS
There is not yet a standardized definition of body size pheno-
types. For the present analyses, 6 metabolic abnormalities were
considered (elevated blood pressure; elevated triglyceride and
glucose levels; insulin resistance; systemic inflammation; and
decreased HDL-C level; Figure 1A). Body size phenotypes were
Cardiometabolic abnormalities considered:
1. Elevated blood pressure
: Systolic/diastolic blood pressure ≥130/85 mm Hg
or antihypertensive medication use
2. Elevated triglyceride level: Fasting triglyceride level ≥150 mg/dL
3. Decreased HDL-C level: HDL-C level <
40 mg/dL in men or <
50 mg/dL in
women or lipid-lowering medication use
4. Elevated glucose level: Fasting glucose level ≥100 mg/dL or antidiabetic
medication use
5. Insulin resistance: HOMA-IR >
5.13 (ie, the 90th percentile)
6. Systemic inflammation: hsCRP level >0.1 mg/L (ie, the 90th percentile)
Criteria for body size phenotypes:
Normal weight, metabolically healthy: BMI <
25.0 and <
2 cardiometabolic
abnormalities
Normal weight, metabolically abnormal: BMI <
25.0 and ≥
2 cardiometabolic
abnormalities
Overweight, metabolically healthy: BMI 25.0-29.9 and <
2 cardiometabolic
abnormalities
Overweight, metabolically abnormal: BMI 25.0-29.9 and ≥
2 cardiometabolic
abnormalities
Obese,
metabolically healthy
: BMI ≥
30.0 and <
2
cardiometabolic
abnormalities
Obese,
metabolically abnormal
: BMI ≥
30.0 and ≥
2
cardiometabolic
abnormalities
A
B
Figure 1. Definition of body size phenotypes. A, Cardiometabolic
abnormalities considered; B, criteria for body size phenotypes. BMI indicates
body mass index (calculated as weight in kilograms divided by height in
meters squared); HDL-C, high-density lipoprotein cholesterol;
HOMA-IR, homeostasis model assessment of insulin resistance; and
hsCRP, high-sensitivity C-reactive protein. To convert to millimoles per liter,
multiply by 0.0259 for HDL-C, by 0.0113 for triglycerides, and by 0.0555 for
glucose; to convert hsCRP to nanomoles per liter, multiply by 9.524.
(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 15), AUG 11/25, 2008 WWW.ARCHINTERNMED.COM
1618
©2008 American Medical Association. All rights reserved.
at Albert Einstein College of Medicine, on August 18, 2008 www.archinternmed.comDownloaded from
defined based on the combined consideration of BMI category
(normal weight, overweight, and obesity) and having 0 to 1
(metabolically healthy) or 2 or more (metabolically abnor-
mal) cardiometabolic abnormalities (Figure 1B). Sensitivity
analyses were performed using definitions with more strin-
gent (metabolically healthy, 0 cardiometabolic abnormalities;
metabolically abnormal, ⱖ1 metabolic abnormalities) and less
stringent criteria for definition of the “metabolically healthy”
phenotype (using Adult Treatment Panel III [ATP-III] criteria
for metabolic syndrome,11 with metabolically healthy classi-
fied as ⱕ2 metabolic abnormalities and metabolically abnor-
mal as ⱖ3 metabolic abnormalities).
STATISTICAL ANALYSIS
Participant characteristics were calculated as means, geomet-
ric means, and percentages, overall and by body size pheno-
type. Differences in these characteristics across BMI category,
for metabolically healthy and metabolically abnormal individu-
als separately, were tested using linear orthogonal contrasts and
adjusted Wald 2tests. The age-standardized prevalence of hav-
ing 0 to 1 and 2 or more metabolic abnormalities was calcu-
lated by body size phenotype and sex using direct standard-
ization, with the age distribution of the US population as the
standard. Also, the age- and sex-standardized prevalence of hav-
Table 1. Demographic and Metabolic Characteristics of the Study Population by Body Size Phenotypea
Demographic and
Behavioral Characteristic Overall
Metabolically Healthy Metabolically Abnormal
Normal Weight Overweight Obese Normal Weight Overweight Obese
Prevalence, % (population
frequency)
100 (200 690 825) 26.4 (52 982 378) 17.9 (35 923 658) 9.7 (19 467 010) 8.1 (16 255 957) 17.0 (34 117 440) 20.9 (41 944 382)
Age, y 45.0 (0.4) 39.7 (0.6) 42.0 (0.6)c40.1 (0.8) 54.7 (1.0) 50.8 (0.6)c48.2 (0.6) c
Men, % 47.9 (0.6) 40.3 (1.4) 54.4 (2.2)c39.2 (2.5) 49.8 (3.1) 59.8 (2.1)d45.8 (1.4)
Race/ethnicity, %
Non-Hispanic white 71.3 (1.9) 74.9 (1.8) 69.3 (2.4) 61.4 (3.1) 74.2 (3.6) 71.3 (3.2) 71.9 (2.2)
Non-Hispanic black 10.8 (1.1) 9.9 (1.2) 11.5 (1.5) 19.7 (2.3) 6.1 (1.2) 6.5 (1.0) 12.6 (1.5)
Mexican American 7.7 (0.9) 6.0 (0.7) 8.8 (1.3) 10.3 (1.8) 4.9 (0.9) 9.1 (1.5) 7.7 (1.0)
Other 10.2 (1.4) 9.2 (1.3) 10.4 (2.0)d8.7 (2.1) c14.8 (3.5) 13.2 (2.5)e7.7 (1.7) c
Smoking, %
Never 50.9 (1.2) 53.5 (1.8) 53.5 (2.2) 55.4 (3.3) 42.2 (3.7) 45.1 (2.3) 51.3 (1.8)
Former 25.5 (1.0) 19.5 (1.4) 25.1 (2.0) 22.3 (2.6) 27.5 (2.8) 32.0 (2.0) 28.7 (1.6)
Current 23.7 (1.0) 27.1 (1.9) 21.4 (1.8)e22.3 (2.4) 30.3 (2.7) 23.0 (1.7)d20.0 (1.6)c
Alcohol intake, %
Nondrinkers 47.2 (1.6) 37.7 (1.9) 43.2 (2.3) 53.4 (3.4) 49.6 (3.5) 49.2 (2.6) 57.4 (2.4)
⬍1 drink per day 39.4 (1.3) 48.6 (1.6) 41.9 (2.5) 35.3 (2.7) 35.3 (2.8) 34.7 (2.2) 32.9 (2.0)
1-2 drink per day 7.7 (0.5) 8.9 (1.0) 9.4 (1.2) 7.0 (1.3) 7.2 (1.8) 7.7 (1.0) 5.1 (0.8)
⬎2 drink per day 5.7 (0.5) 4.7 (0.7) 5.5 (0.9) 4.3 (1.1)c7.9 (1.2) 8.4 (1.3) 4.6 (0.7)
Leisure time physical activity, %
0 METs/d 36.3 (1.2) 30.3 (2.1) 30.1 (1.8) 36.1 (2.8) 46.6 (3.3) 40.7 (2.0) 41.5 (1.8)
1.0-49.9 METs/d 15.8 (0.8) 14.4 (1.6) 15.6 (1.2) 16.6 (1.7) 10.6 (2.0) 15.5 (1.3) 19.7 (1.8)
50.0-131.9 METs/d 16.0 (0.8) 17.4 (1.4) 18.0 (1.4) 16.5 (2.2) 13.1 (1.7) 13.8 (1.6) 15.3 (1.3)
132.0-279.9 METs/d 15.9 (0.8) 18.5 (1.6) 16.1 (1.6) 16.1 (2.3) 17.0 (2.8) 14.8 (1.3) 13.0 (1.3)
ⱖ280.0 METs/d 16.0 (0.8) 19.3 (1.3) 20.3 (1.8) 14.7 (1.9) 12.7 (2.1) 15.2 (1.3) 10.6 (1.0)e
SBP, mm Hg 122.4 (0.4) 114.7 (0.6) 117.2 (0.7) e119.2 (0.7)c131.3 (1.4) 128.9 (0.7) 129.3 (0.6)
DBP, mm Hg 72.0 (0.3) 69.4 (0.4) 70.7 (0.5)d72.3 (0.6) c70.7 (0.9) 73.7 (0.4) c75.4 (0.5)c
Elevated blood pressure (SBP ⱖ130
mm Hg and/or DBP ⱖ85 mm Hg
and/or medication use), %
39.1 (1.0) 14.6 (1.0) 18.8 (1.7)d21.6 (2.4) e65.2 (3.7) 61.0 (1.8) 66.9 (1.8)
HDL-C, mg/dL 52.0 (0.4) 60.5 (0.6) 54.1 (0.5)c53.4 (0.6) c50.2 (1.2) 46.0 (0.5) c44.2 (0.5)c
HDL-C ⬍40 mg/dL for men or ⬍50
mg/dL for women, %
33.5 (0.9) 12.0 (1.6) 16.7 (1.5)d18.8 (1.7) e46.3 (2.8) 53.0 (2.2) 60.7 (1.9)c
Triglycerides, mg/dLb121.1 (1.5) 84.7 (1.3) 95.3 (1.6)c100.0 (2.3)c158.2 (6.7) 176.8 (3.6) d168.3 (3.8)
Triglycerides ⱖ150 mg/dL, % 33.0 (0.8) 6.3 (0.8) 10.2 (1.3)d10.1 (1.5)d59.4 (2.8) 66.5 (1.8)d59.0 (1.9)
Glucose, mg/dLb98.1 (0.4) 90.4 (0.3) 92.0 (0.4) c93.0 (0.5)c103.7 (1.2) 106.9 (0.9)d107.8 (0.9) e
Glucose ⱖ100 mg/dL and/or
antidiabetic medication use, %
34.1 (1.1) 8.7 (1.0) 10.6 (1.4) 11.2 (1.4) 54.9 (3.0) 66.3 (1.7)c62.4 (1.6)d
Insulin, µU/mLb9.2 (0.1) 5.6 (0.1) 7.5 (0.2)c11.1 (0.3) c7.7 (0.3) 11.0 (0.3)c17.0 (0.4) c
HOMA-IRb1.8 (0.04) 0.9 (0.02) 1.4 (0.03) c2.2 (0.06)c1.6 (0.07) 2.5 (0.07)c4.1 (0.1)c
HOMA-IR ⬎5.13, % 10.2 (0.7) 0 0.1 (0.1) 1.7 (0.7) 6.2 (1.6) 13.3 (1.4) e34.4 (2.0)c
BMI 28.2 (0.1) 22.4 (0.05) 27.3 (0.06)c34.2 (0.18) c22.9 (0.08) 27.6 (0.06)c36.2 (0.26) c
Waist circumference, cm 96.4 (0.3) 81.2 (0.2) 94.2 (0.3)c107.9 (0.6)c86.8 (0.5) 98.1 (0.3)c115.0 (0.6)c
hsCRP, mg/Lb0.02 (⬍0.001) 0.009 (⬍0.001) 0.016 (0.001)c0.026 (0.001)c0.023 (0.002) 0.025 (0.001) 0.045 (0.001) c
hsCRP ⬎0.1 mg/L, % 9.7 (0.5) 1.6 (0.4) 1.9 (0.4) 5.2 (1.2)d16.2 (2.0) 11.4 (1.5) 25.0 (1.4)e
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); DBP, diastolic blood pressure; HDL-C, high-density
lipoprotein cholesterol; HOMA-IR, homeostasis model assessment of insulin resistance; hsCRP, high-sensitivity C-reactive protein; METs, metabolic equivalent tasks;
SBP, systolic blood pressure.
SI conversion factors: To convert to millimoles per liter, multiply by 0.0259 for HDL-C, by 0.0113 for triglycerides, and by 0.0555 for glucose; to convert insulin to
picomoles per liter, multiply by 6.945; and hsCRP to nanomoles per liter, multiply by 9.524.
aData are given as mean (SE) value or percentage (SE) of participants unless otherwise specified.
bGeometric mean (SE of the geometric mean).
cP⬍.001 vs normal weight within metabolic subgroup.
dP⬍.05 vs normal weight within metabolic subgroup.
eP⬍.01 vs normal weight within metabolic subgroup.
(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 15), AUG 11/25, 2008 WWW.ARCHINTERNMED.COM
1619
©2008 American Medical Association. All rights reserved.
at Albert Einstein College of Medicine, on August 18, 2008 www.archinternmed.comDownloaded from
ing 0 to 1 and 2 or more metabolic abnormalities by body size
phenotype was calculated by race/ethnicity using direction stan-
dardization, with the joint age and sex distribution of the US
population as the standard. Among normal-weight individu-
als, prevalence ratios of expressing 2 or more metabolic abnor-
malities associated with demographic and behavioral charac-
teristics were calculated, while among overweight or obese
individuals, prevalence ratios of expressing 0 to 1 metabolic
abnormality associated with demographic and behavioral char-
acteristics were calculated. Unadjusted prevalence ratios were
calculated initially, followed by multivariate-adjusted models
including all demographic and behavioral factors simulta-
neously. Finally, prevalence ratios were calculated after fur-
ther adjustment for waist circumference. All analyses were con-
ducted using SUDAAN 9.0 (Research Triangle Institute, Research
Triangle Park, North Carolina) statistical software and used tech-
niques appropriate to the complex survey design of NHANES
1999-2004.
RESULTS
PREVALENCE OF BODY SIZE PHENOTYPES
Among the overall US population 20 years and older, 17.9%
(approximately 35.9 million adults) were overweight yet
metabolically healthy (0 or 1 metabolic abnormalities) and
9.7% (approximately 19.5 million adults) were obese yet
metabolically healthy, whereas 8.1% (approximately 16.3
million adults) were normal weight but metabolically ab-
normal (ⱖ2 metabolic abnormalities) (Table 1). As a per-
centage of each BMI group, 51.3% of overweight individu-
als were metabolically healthy, 31.7% of obese individuals
were metabolically healthy, and 23.5% of normal-weight
individuals were metabolically abnormal.
Compared with normal-weight men and women, the
age-standardized prevalence of the metabolically abnor-
mal phenotype was significantly higher among over-
weight and obese men and women (Figure 2). Despite
this, 30.1% of normal-weight men and 21.1% of normal-
weight women were metabolically abnormal, whereas
29.2% of obese men and 35.4% of obese women were meta-
bolically healthy. Similar patterns were seen when preva-
lence estimates were stratified by race/ethnicity (Figure 3).
The prevalence of the metabolically abnormal phe-
notype among normal-weight individuals was 10.3%
among those aged between 20 and 34 years, 16.9% among
those aged between 35 and 49 years, 41.7% among those
aged between 50 and 64 years, 54.7% among those aged
between 65 and 79 years, and 56.2% among those 80 years
and older. The prevalence of the metabolically healthy
phenotype among obese individuals was 47.7% among
those aged between 20 and 34 years, 31.1% among those
aged between 35 and 49 years, 20.4% among those aged
between 50 and 64 years, 14.3% among those aged be-
tween 65 and 79 years, and 22.1% among those 80 years
and older.
100
69.9
30.1
48.8 51.2∗
29.2
70.8∗
60
80
40
20
90
50
70
30
10
0
Normal Weight Overweight Obese
Prevalence, %
100
78.9
21.1
57.0
43.0∗
35.4
64.6∗
60
80
40
20
90
50
70
30
10
0
Normal Weight Overweight Obese
Prevalence, %
Metabolically healthy Metabolically abnormal
A
B
Figure 2. Age-standardized prevalence of cardiometabolic abnormalities by
body size and sex (A, men; B, women). *P⬍.001 for proportion
metabolically abnormal vs normal weight.
100
75.6
24.4
54.2
45.8∗
30.8
69.2∗
60
80
40
20
90
50
70
30
10
0
Normal Weight
Metabolically healthy Metabolically abnormal
Overweight Obese
Prevalence, %
100
79.3
20.7
63.1
36.9∗38.9
61.1∗
60
80
40
20
90
50
70
30
10
0
Normal Weight Overweight Obese
Prevalence, %
100
67.2
32.8
43.6
56.4∗
33.8
66.2∗
60
80
40
20
90
50
70
30
10
0
Normal Weight Overweight Obese
Prevalence, %
A
B
C
Figure 3. Age- and sex-standardized prevalence of cardiometabolic
abnormalities by body size and race/ethnicity. A, Non-Hispanic whites;
B, non-Hispanic blacks; and C, Mexican Americans. *P⬍.001 for proportion
metabolically abnormal vs normal weight.
(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 15), AUG 11/25, 2008 WWW.ARCHINTERNMED.COM
1620
©2008 American Medical Association. All rights reserved.
at Albert Einstein College of Medicine, on August 18, 2008 www.archinternmed.comDownloaded from
Among those with 2 or more metabolic abnormali-
ties, the 2 most common cardiometabolic risk factor com-
binations within all body size groupings were high tri-
glyceride level/low HDL-C level and high blood pressure/
high glucose level.
CORRELATES OF THE METABOLICALLY
ABNORMAL PHENOTYPE IF NORMAL WEIGHT
Among normal-weight individuals, the prevalence of
cardiometabolic risk factor clustering was higher in older
age groups, men, former and current smokers, and those
with larger waist circumference and was lower in non-
Hispanic blacks and in moderate alcohol drinkers (⬍1
drink per day) (Table 2). In a multivariate adjustment
regression model of normal-weight individuals, older
age, male sex, and moderate physical activity remained
independently associated with the metabolically abnor-
mal phenotype. After further adjustment for waist cir-
cumference, male sex was no longer statistically signifi-
cantly associated with the metabolically abnormal
phenotype, but the associations with older age and physi-
cal activity remained.
CORRELATES OF THE
METABOLICALLY HEALTHY PHENOTYPE
IF OVERWEIGHT OR OBESE
Among overweight and obese individuals, older adults,
former smokers, and those with greater waist circumfer-
ences were less likely to express the metabolically healthy
phenotype (Table 3). In contrast, non-Hispanic blacks,
moderate alcohol drinkers, and those with greater lev-
els of physical activity were more likely to express the
Table 2. Unadjusted and Multivariate-Adjusted
Prevalence Ratios of the Metabolically Abnormal Phenotype
(ⱖ2 Cardiometabolic Abnormalities) Associated
With Demographic and Behavioral Characteristics
Among Normal-Weight Individuals
Demographic
and Behavioral
Characteristic
Prevalence Ratio (95% CI)
Unadjusted
Multivariate
Adjusteda
Multivariate-
AdjustedⴙWaist
Circumferenceb
Age group, y
20-34 1 [Reference] 1 [Reference] 1 [Reference]
35-49 1.58 (1.00-2.48) 1.57 (1.00-2.47) 1.44 (0.91-2.26)
50-64 4.01 (2.81-5.72) 3.80 (2.57-5.60) 3.01 (2.03-4.47)
65-79 5.49 (3.82-7.90) 5.11 (3.54-7.37) 3.86 (2.66-5.60)
ⱖ80 5.95 (4.14-8.54) 5.90 (4.01-8.67) 4.16 (2.69-6.43)
Sex
Women 1 [Reference] 1 [Reference] 1 [Reference]
Men 1.34 (1.08-1.66) 1.37 (1.10-1.71) 1.04 (0.84-1.28)
Race/ethnicity
Non-Hispanic
white
1 [Reference] 1 [Reference] 1 [Reference]
Non-Hispanic
black
0.68 (0.49-0.95) 0.74 (0.52-1.05) 0.84 (0.59-1.19)
Mexican American 0.86 (0.61-1.21) 1.14 (0.83-1.57) 1.20 (0.86-1.67)
Smoking
Never 1 [Reference] 1 [Reference] 1 [Reference]
Former 1.55 (1.17-2.07) 1.11 (0.85-1.43) 1.10 (0.87-1.40)
Current 1.31 (0.98-1.76) 1.31 (0.94-1.83) 1.31 (0.96-1.79)
Alcohol intake,
drinks/d
Nondrinkers 1 [Reference] 1 [Reference] 1 [Reference]
⬍1 0.63 (0.51-0.79) 0.83 (0.69-1.00) 0.84 (0.71-1.01)
1-2 0.69 (0.42-1.13) 0.84 (0.50-1.40) 0.85 (0.52-1.40)
⬎2 1.17 (0.83-1.66) 1.08 (0.73-1.58) 0.97 (0.67-1.41)
Log leisure time
physical activity,
METs/d
0 1 [Reference] 1 [Reference] 1 [Reference]
1.0-49.9 0.57 (0.39-0.85) 0.70 (0.51-0.96) 0.71 (0.51-0.98)
50.0-131.9 0.59 (0.45-0.77) 0.71 (0.54-0.93) 0.71 (0.54-0.93)
132.0-279.9 0.69 (0.46-1.03) 0.82 (0.56-1.20) 0.86 (0.61-1.22)
ⱖ280.0 0.52 (0.38-0.72) 0.73 (0.52-1.02) 0.78 (0.56-1.10)
Waist circumference,
per5cm
1.42 (1.33-1.52) . . . 1.23 (1.15-1.32)
Abbreviations: CI, confidence interval; METs, metabolic equivalent tasks.
aEach factor in the table is adjusted for every other factor in the table, except
waist circumference.
bEach factor in the table is adjusted for every other factor in the table,
including waist circumference.
Table 3. Unadjusted and Multivariate-Adjusted
Prevalence Ratios of the Metabolically Healthy Phenotype
(0-1 Cardiometabolic Abnormalities) Associated
With Demographic and Behavioral Characteristics
Among Overweight and Obese Individuals
Demographic
and Behavioral
Characteristic
Prevalence Ratio (95% CI)
Unadjusted
Multivariate
Adjusteda
Multivariate
AdjustedⴙWaist
Circumferenceb
Age group, y
20-34 1 [Reference] 1 [Reference] 1 [Reference]
35-49 0.75 (0.65-0.87) 0.76 (0.65-0.87) 0.80 (0.70-0.93)
50-64 0.53 (0.46-0.62) 0.55 (0.47-0.64) 0.61 (0.54-0.70)
65-79 0.38 (0.31-0.47) 0.38 (0.31-0.48) 0.44 (0.35-0.54)
ⱖ80 0.49 (0.37-0.64) 0.50 (0.39-0.65) 0.53 (0.40-0.69)
Sex
Women 1 [Reference] 1 [Reference] 1 [Reference]
Men 0.93 (0.84-1.04) 0.85 (0.76-0.97) 1.00 (0.89-1.11)
Race/ethnicity
Non-Hispanic
white
1 [Reference] 1 [Reference] 1 [Reference]
Non-Hispanic
black
1.28 (1.16-1.40) 1.17 (1.06-1.29) 1.18 (1.08-1.29)
Mexican American 1.11 (0.99-1.24) 0.97 (0.87-1.08) 0.90 (0.80-1.00)
Smoking
Never 1 [Reference] 1 [Reference] 1 [Reference]
Former 0.82 (0.72-0.93) 0.98 (0.87-1.12) 1.01 (0.89-1.15)
Current 0.95 (0.83-1.08) 0.92 (0.81-1.04) 0.91 (0.80-1.04)
Alcohol intake,
drinks/d
Nondrinkers 1 [Reference] 1 [Reference] 1 [Reference]
⬍1 1.19 (1.05-1.34) 1.11 (0.97-1.28) 1.05 (0.92-1.19)
1-2 1.28 (1.04-1.57) 1.29 (1.04-1.59) 1.17 (0.94-1.45)
⬎2 0.96 (0.79-1.16) 1.02 (0.82-1.28) 0.96 (0.78-1.19)
Log leisure time
physical activity,
METs/d
0 1 [Reference] 1 [Reference] 1 [Reference]
1.0-49.9 1.09 (0.92-1.29) 1.03 (0.87-1.20) 1.06 (0.90-1.24)
50.0-131.9 1.28 (1.14-1.45) 1.23 (1.08-1.40) 1.16 (1.02-1.33)
132.0-279.9 1.26 (1.07-1.49) 1.25 (1.06-1.48) 1.16 (0.97-1.38)
ⱖ280 1.41 (1.23-1.62) 1.29 (1.12-1.49) 1.13 (0.98-1.30)
Waist circumference,
per5cm
0.84 (0.82-0.86) . . . 0.85 (0.83-0.88)
Abbreviations: CI, confidence interval, METs, metabolic equivalent tasks.
aEach factor in the table is adjusted for every other factor in the table, except
waist circumference.
bEach factor in the table is adjusted for every other factor in the table,
including waist circumference.
(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 15), AUG 11/25, 2008 WWW.ARCHINTERNMED.COM
1621
©2008 American Medical Association. All rights reserved.
at Albert Einstein College of Medicine, on August 18, 2008 www.archinternmed.comDownloaded from
metabolically healthy phenotype. In a multivariate ad-
justment regression model restricted to overweight and
obese individuals, age, non-Hispanic black race/
ethnicity, moderate alcohol intake, and higher leisure-
time physical activity remained independently associ-
ated with expressing the metabolically healthy phenotype.
After further adjustment for waist circumference, mod-
erate alcohol intake was no longer significantly associ-
ated with the metabolically healthy phenotype, but age,
non-Hispanic black race/ethnicity, and moderate leisure-
time physical activity level remained independently as-
sociated with the metabolically healthy phenotype.
SENSITIVITY ANALYSES
Overall, 16.6% of obese men and women had 0 cardio-
metabolic abnormalities. Correlates of possessing 0 car-
diometabolic abnormalities among overweight and obese
individuals were similar to those for possessing 0 or 1,
with the following exceptions: race/ethnicity was not sig-
nificantly associated with possessing 0 cardiometabolic
abnormalities after multivariate adjustment, and mod-
erate alcohol intake was associated with an approxi-
mate 60% increased prevalence of possessing 0 cardio-
metabolic abnormalities.
When abdominal obesity (⬎102 cm in men and ⬎88
cm in women) was used in lieu of BMI categories, 28.3%
of individuals without abdominal obesity expressed the
metabolically abnormal phenotype (ⱖ2 metabolic ab-
normalities), whereas 36.4% of individuals with abdomi-
nal obesity expressed the metabolically healthy pheno-
type (0 or 1 metabolic abnormalities).
As expected, when the ATP-III metabolic syndrome
definition was used (ⱖ3 of the following 5 abnormali-
ties: elevated blood pressure, triglyceride level, glucose
level, and waist circumference or decreased HDL-C level),
the prevalence of normal-weight individuals with car-
diometabolic clustering was lower, whereas the preva-
lences of overweight and obese individuals without car-
diometabolic clustering were higher. When the ATP-III
metabolic syndrome definition was used, 8.6% of normal-
weight individuals were metabolically abnormal, whereas
65.8% of overweight individuals and 39.1% of obese in-
dividuals were metabolically healthy.
COMMENT
These data show that a considerable proportion of over-
weight and obese US adults are metabolically healthy,
whereas a considerable proportion of normal-weight adults
express a clustering of cardiometabolic abnormalities.
Among US adults, 29.2% of obese men and 35.4% of obese
women (a total of approximately 19.5 million adults) pos-
sess a healthy profile in terms of the standard cardiometa-
bolic risk factors. In contrast, 30.1% of normal-weight men
and 21.1% of normal-weight women (a total of approxi-
mately 16.3 million adults) exhibit clustering of cardio-
metabolic abnormalities (ie, ⱖ2 cardiometabolic abnor-
malities). High proportions of normal-weight adults with
cardiometabolic clustering and overweight and obese adults
who were metabolically healthy were documented when
more conservative and less conservative definitions of the
metabolically abnormal phenotype were used. This study
also found that older age, smoking, and larger waist cir-
cumference were associated with the metabolically abnor-
mal phenotype, while moderate alcohol intake and leisure-
time physical activity were associated with the metabolically
healthy phenotype.
The prevalence of body size phenotypes has been in-
vestigated in a limited number of studies.4,12,13 Despite
differences in the definitions of “metabolically healthy”
that were used, the prevalence of metabolically healthy
obese individuals is similar between the present and pre-
vious studies. Among a white, Italian, clinic-based popu-
lation (n= 681), 27.5% of obese patients were without car-
diometabolic abnormalities (normal blood pressure, lipid
parameters, and electrocardiograms and low white blood
cell counts and plasma fibrinogen levels),4while among
a sample of 43 obese postmenopausal women, 39.5% were
without cardiometabolic abnormalities (glucose dis-
posal rate ⬎8.0 mg/min/kg of lean body mass).12
The prevalence of individuals who are normal weight
yet have metabolic abnormalities has been far less stud-
ied. Among 96 normal-weight women aged between 18
and 35 years recruited in Montreal, Quebec, Canada, who
were free of acute illness, diabetes, hypertension, and dys-
lipidemia, only 12 (12.5%) were metabolically abnor-
mal, defined as possessing a homeostasis model assess-
ment of insulin resistance value higher than 1.69.13 In
contrast, approximately 21% of normal-weight women
in the present study had clustering of cardiometabolic
abnormalities. The lower prevalence in the Montreal
study is likely owing to the exclusion of women with dia-
betes, hypertension, and dyslipidemia.
In the present study, several demographic and behav-
ioral characteristics were associated with being normal
weight but metabolically abnormal. Although normal-
weight men were 34% more likely than normal-weight
women to have 2 or more metabolic abnormalities, this
was not independent of waist circumference values, sug-
gesting that sex differences in waist circumference was driv-
ing the higher prevalence of cardiometabolic clustering in
men. Data were available in the present study to deter-
mine the prevalence of normal-weight, metabolically ab-
normal individuals and overweight or obese, metaboli-
cally healthy individuals across the adult age span. Although
the prevalence of metabolic abnormalities increased with
age among all body size groups, a substantial proportion
of elderly obese individuals were metabolically healthy,
whereas a substantial proportion of normal-weight young
adults had at least 2 cardiometabolic abnormalities. Spe-
cifically, 22.1% of obese individuals 80 years and older did
not express cardiometabolic clustering and 10.3% of nor-
mal-weight individuals aged between 20 and 34 years had
2 or more cardiometabolic abnormalities. In the present
analyses, among normal-weight individuals there were no
statistically significant race/ethnicity differences in the
prevalence of clustered cardiometabolic abnormalities.
However, among overweight or obese individuals, non-
Hispanic blacks were 18% more likely to be metaboli-
cally healthy compared with non-Hispanic whites. Non-
Hispanic blacks have generally been found to have greater
hypertension prevalence compared with non-Hispanic
(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 15), AUG 11/25, 2008 WWW.ARCHINTERNMED.COM
1622
©2008 American Medical Association. All rights reserved.
at Albert Einstein College of Medicine, on August 18, 2008 www.archinternmed.comDownloaded from
whites, but have a similar or lower prevalence of hyper-
cholesterolemia,14-16 which may underlie differences in the
likelihood of obesity-associated cardiometabolic abnor-
malities demonstrated herein.
In addition to demographic factors, the present analy-
ses also identified a number of behavioral factors asso-
ciated with the normal-weight, metabolically abnormal
phenotype and the overweight or obese, metabolically
healthy phenotype. Cigarette smoking was associated with
cardiometabolic abnormalities in each of these pheno-
types, while leisure-time physical activity and alcohol in-
take were associated with being metabolically healthy.
The beneficial effect of leisure-time physical activity was
statistically significant in both normal-weight and over-
weight and obese individuals and was only somewhat at-
tenuated by adjustment for waist circumference. After
multivariate adjustment, current smoking was not inde-
pendently associated with cardiometabolic abnormali-
ties in either normal-weight or overweight or obese in-
dividuals, primarily due to adjustment for physical activity
levels. Moderate alcohol intake, compared with non-
drinking, was associated with a lower prevalence of hav-
ing clustered metabolic abnormalities in the present analy-
ses, though adjustment for age reduced this association
to nonsignificance in both normal-weight and over-
weight or obese individuals. Since benefits of moderate
alcohol intake on lipid and glucose metabolism have been
identified previously,17-20 it is possible that the wide age
range represented in the present study explained so much
of the variance as to dwarf any possible beneficial effect
of moderate alcohol intake in multivariate regression
analyses. Further research into the potential of moder-
ate alcohol intake to assist obese individuals in main-
taining a healthy cardiometabolic profile is needed.
The role of excess adiposity in CVD risk is unclear. Re-
cent studies have shown that obesity was not associated
with an increased risk of future cardiovascular events
among individuals without the metabolic syndrome, but
that among individuals with the metabolic syndrome, obe-
sity was associated with an increased CVD risk.1-3 Among
the studies that stratified by combined body size and meta-
bolic syndrome phenotypes, obese individuals without car-
diometabolic abnormalities or clustering of cardiometa-
bolic abnormalities appeared not to have increased CVD
risk.1,2 Adipose tissue is now recognized as an endocrine
organ secreting a variety of hormones and cytokines. The
presence of obese individuals, including older adults, who
maintain cardiometabolic factors within the normal range
suggests that certain obese individuals are either less re-
sponsive to the endocrine secretions of excess adipose tis-
sue or that their adipose tissue does not possess the same
endocrine secretory properties of those obese individuals
who develop metabolic derangements. This underscores
the need for future research into the physiologic mecha-
nisms underlying these body size phenotypes.
The interpretation of these data needs to be assessed
within the context of the limitations of the present study.
Body size phenotype definitions have not been standard-
ized, and as demonstrated by our sensitivity analyses,
prevalence estimates are subject to alteration depend-
ing on the number of metabolic abnormalities consid-
ered and the specific cut points of those abnormalities.
In addition, BMI as a measure of obesity has limitations
because it cannot distinguish between fat tissue and lean
tissue. This limitation is especially pertinent for Asian
populations, who have been shown to have a greater per-
centage of body fat per given BMI value compared with
Western populations, and elderly individuals, who have
a greater percentage of body fat per given BMI value com-
pared with younger individuals.21,22 Sarcopenic obesity
is a condition of aging and is characterized by high body
fat in the presence of reduced lean body mass.23 Because
of the simultaneous decrease in lean tissue that accom-
panies the increase in body fat, the BMI of sarcopenic in-
dividuals may underestimate their level of obesity to an
even greater extent than the standard age-related under-
estimation associated with BMI. A similar limitation ex-
ists for waist circumference, whereby certain individu-
als may possess relatively more abdominal visceral fat than
others with the same waist circumference, especially
among older populations.24,25 However, few simple, in-
expensive alternatives to anthropometric indexes exist
for the clinical evaluation of obesity. Bioimpedance analy-
sis is relatively inexpensive and simple to perform, but
it remains unclear whether bioimpedance analysis is sig-
nificantly better at predicting cardiovascular events than
BMI or waist circumference. Further research examin-
ing the effects of different definitions of body size phe-
notypes on the risk of CVD is needed. The NHANES 1999-
2004 data set does not include information on the amount
of visceral and subcutaneous adipose tissue or work-
related physical activity, which may be relevant to de-
fining and evaluating body size phenotypes.
Despite these limitations, our study had a number of
strengths. This study included nationally representative
data on 5440 adults, and women and non-Hispanic black
and Mexican Americans were well represented. In addi-
tion, the majority of previous studies have defined obe-
sity phenotypes based on either solely an insulin resis-
tance cut point or the metabolic syndrome definition. The
present study included not only the components of the
metabolic syndrome, but also insulin resistance and in-
flammation criteria, thereby capturing a wider breadth
of metabolic abnormalities.
In conclusion, the present data suggest a high preva-
lence of cardiometabolic abnormality clustering among
normal-weight individuals, as well as a high prevalence
of obese individuals who are metabolically healthy, ir-
respective of the definition used to define these pheno-
types. Further studies into the behavioral, hormonal or
biochemical, and genetic mechanisms underlying these
differential metabolic responses to body size are needed
and will likely further the identification of possible obe-
sity intervention targets and improve CVD screening tools.
Accepted for Publication: February 11, 2008.
Correspondence: Rachel P. Wildman, PhD, Department
of Epidemiology and Population Health, Albert Einstein
College of Medicine, 1300 Morris Park Ave, Belfer Build-
ing, Room 1309, Bronx, NY 10461 (rwildman@aecom.yu
.edu).
Author Contributions: Dr Wildman had full access to all
of the data in the study and takes responsibility for the in-
tegrity of the data and the accuracy of the data analysis.
(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 15), AUG 11/25, 2008 WWW.ARCHINTERNMED.COM
1623
©2008 American Medical Association. All rights reserved.
at Albert Einstein College of Medicine, on August 18, 2008 www.archinternmed.comDownloaded from
Study concept and design: Wildman, Muntner, and Sowers.
Analysis and interpretation of data: Wildman, Muntner,
Reynolds, McGinn, Rajpathak, Wylie-Rosett, and Sowers.
Drafting of the manuscript: Wildman and Sowers. Critical
revision of the manuscript for important intellectual con-
tent: Wildman, Muntner, Reynolds, McGinn, Rajpathak,
Wylie-Rosett, and Sowers. Statistical analysis: Wildman and
Muntner. Administrative, technical, and material support:
Wylie-Rosett.
Financial Disclosure: None reported.
REFERENCES
1. Kip KE, Marroquin OC, Kelley DE, et al. Clinical importance of obesity versus the
metabolic syndrome in cardiovascular risk in women: a report from the Wom-
en’s Ischemia Syndrome Evaluation (WISE) study. Circulation. 2004;109(6):
706-713.
2. St-Pierre AC, Cantin B, Mauriege P, et al. Insulin resistance syndrome, body mass
index and the risk of ischemic heart disease. CMAJ. 2005;172(10):1301-1305.
3. Katzmarzyk PT, Janssen I, Ross R, Church TS, Blair SN. The importance of waist
circumference in the definition of metabolic syndrome: prospective analyses of
mortality in men. Diabetes Care. 2006;29(2):404-409.
4. Iacobellis G, Ribaudo MC, Zappaterreno A, Iannucci CV, Leonetti F. Prevalence
of uncomplicated obesity in an Italian obese population. Obes Res. 2005;13
(6):1116-1122.
5. Ruderman N, Chisholm D, Pi-Sunyer X, Schneider S. The metabolically obese,
normal-weight individual revisited. Diabetes. 1998;47(5):699-713.
6. Ferrannini E, Natali A, Bell P, Cavallo-Perin P, Lalic N, Mingrone G; European Group
for the Study of Insulin Resistance (EGIR). Insulin resistance and hypersecre-
tion in obesity. J Clin Invest. 1997;100(5):1166-1173.
7. Karelis AD, St Pierre DH, Conus F, Rabasa-Lhoret R, Poehlman ET. Metabolic
and body composition factors in subgroups of obesity: what do we know? J Clin
Endocrinol Metab. 2004;89(6):2569-2575.
8. National Center for Health Statistics. NHANES 1999-2000, 2001-2002, and 2003-
2004 data files: data, docs, codebooks, SAS code. National Center for Health Sta-
tistics. http://www.cdc.gov/nchs/nhanes.htm. Accessed September 27, 2007.
9. 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(9)(suppl):S498-S504.
10. Perloff D, Grim C, Flack J, et al. Human blood pressure determination by
sphygmomanometry. Circulation. 1993;88(5, pt 1):2460-2470.
11. Grundy SM, Cleeman JI, Merz CN, et al. Implications of recent clinical trials for
the National Cholesterol Education Program Adult Treatment Panel III Guidelines.
Circulation. 2004;110(2):227-239.
12. Brochu M, Tchernof A, Dionne IJ, et al. What are the physical characteristics as-
sociated with a normal metabolic profile despite a high level of obesity in post-
menopausal women? J Clin Endocrinol Metab. 2001;86(3):1020-1025.
13. Conus F, Allison DB, Rabasa-Lhoret R, et al. Metabolic and behavioral charac-
teristics of metabolically obese but normal-weight women. J Clin Endocrinol Metab.
2004;89(10):5013-5020.
14. Hyre AD, Muntner P, Menke A, Raggi P, He J. Trends in ATP-III-defined high
blood cholesterol prevalence, awareness, treatment and control among US adults.
Ann Epidemiol. 2007;17(7):548-555.
15. Hozawa A, Folsom AR, Sharrett AR, Chambless LE. Absolute and attributable risks
of cardiovascular disease incidence in relation to optimal and borderline risk fac-
tors: comparison of African American with white subjects—Atherosclerosis Risk
in Communities Study. Arch Intern Med. 2007;167(6):573-579.
16. Diaz VA, Mainous AG III, Koopman RJ, Carek PJ, Geesey ME. Race and diet in
the overweight: association with cardiovascular risk in a nationally representa-
tive sample. Nutrition. 2005;21(6):718-725.
17. Lazarus R, Sparrow D, Weiss ST. Alcohol intake and insulin levels: the Norma-
tive Aging Study. Am J Epidemiol. 1997;145(10):909-916.
18. Kroenke CH, Chu NF, Rifai N, et al. A cross-sectional study of alcohol consump-
tion patterns and biologic markers of glycemic control among 459 women. Dia-
betes Care. 2003;26(7):1971-1978.
19. Koppes LL, Dekker JM, Hendriks HF, Bouter LM, Heine RJ. Meta-analysis of the
relationship between alcohol consumption and coronary heart disease and mor-
tality in type 2 diabetic patients. Diabetologia. 2006;49(4):648-652.
20. Freiberg MS, Cabral HJ, Heeren TC, Vasan RS, Curtis ER. Alcohol consumption
and the prevalence of the Metabolic Syndrome in the US: a cross-sectional analy-
sis of data from the Third National Health and Nutrition Examination Survey. Dia-
betes Care. 2004;27(12):2954-2959.
21. Deurenberg P, Yap M, van Staveren WA. Body mass index and percent body fat:
a meta analysis among different ethnic groups. Int J Obes Relat Metab Disord.
1998;22(12):1164-1171.
22. Gallagher D, Visser M, Sepulveda D, Pierson RN, Harris T, Heymsfield SB. How
useful is body mass index for comparison of body fatness across age, sex, and
ethnic groups? Am J Epidemiol. 1996;143(3):228-239.
23. Roubenoff R. Sarcopenic obesity: the confluence of two epidemics. Obes Res.
2004;12(6):887-888.
24. Hill JO, Sidney S, Lewis CE, Tolan K, Scherzinger AL, Stamm ER. Racial differ-
ences in amounts of visceral adipose tissue in young adults: the CARDIA (Coro-
nary Artery Risk Development in Young Adults) study. Am J Clin Nutr. 1999;
69(3):381-387.
25. Harris TB, Visser M, Everhart J, et al. Waist circumference and sagittal diameter
reflect total body fat better than visceral fat in older men and women: the Health,
Aging and Body Composition Study. Ann N Y Acad Sci. 2000;904:462-473.
(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 15), AUG 11/25, 2008 WWW.ARCHINTERNMED.COM
1624
©2008 American Medical Association. All rights reserved.
at Albert Einstein College of Medicine, on August 18, 2008 www.archinternmed.comDownloaded from