Ad libitum food intake on a "cafeteria diet" in Native American women: Relations with body composition and 24-h energy expenditure

Abstract

Epidemiologic studies consistently report associations between obesity and dietary fat but not total energy intake. We measured ad libitum food intake in a laboratory setting and evaluated its relation to body weight and composition, energy expenditure, and macronutrient utilization in 28 women of Pima-Papago heritage (aged 27 +/- 7 y, 85.3 +/- 19.0 kg, 44 +/- 6% body fat; means +/- SD). All women were studied during the follicular phase of the menstrual cycle. After a 4-d weight-maintenance period, the volunteers selected their food for 5 d from computerized vending machines offering a variety of familiar and preferred foods, ie, a "cafeteria diet". Twenty-four-hour energy expenditure and substrate oxidation were measured in a respiratory chamber on the 4th d o weight maintenance and the 5th d of ad libitum intake. Average ad libitum intake was 13,732 +/- 4238 kJ/d (11 +/- 1% protein, 40 +/- 1% fat, 49 +/- 4% carbohydrate), ie, moderate overeating by 27 +/- 37% above weight maintenance requirements (range: -27% to 124%). Percent body fat correlated with daily energy intake (r = 0.53, P < 0.01), the degree of overeating (r = 0.41, P < 0.05), and the selection of a diet higher in fat and lower in carbohydrate (r = 0.70 and r = -0.63, respectively, P < 0.001). Excess carbohydrate intake caused an increase in carbohydrate oxidation (r = 0.51, P < 0.01), whereas excess fat intake resulted in a decrease in fat oxidation (r = -0.53, P < 0.01) and thus a positive fat balance of 85 +/- 65 g/d. The positive relations among degrees of obesity, dietary fat intake and overeating, and the fact that dietary fat does not induce fat oxidation, support the hypothesis that dietary fat promotes obesity in women.

Full text

Am J Clin Nutr 1995;62:911-7. Printed in USA. © 1995 American Society for Clinical Nutrition 911
Ad libitum food intake on a “cafeteria diet” in Native
American women: relations with body composition and
24-h energy expenditure1’2
D Enette Larson, Pietro A Tataranni, Robert T Ferraro, and Eric Ravussin
ABSTRACT Epidemiologic studies consistently report asso-
ciations between obesity and dietary fat but not total energy intake.
We measured ad libitum food intake in a laboratory setting and
evaluated its relation to body weight and composition, energy
expenditure, and macronutrient utilization in 28 women of Pima-
Papago heritage (aged 27 ±7 y, 85.3 ± 19.0 kg, 44 ± 6% body
fat; i ±SD). All women were studied during the follicular phase
of the menstrual cycle. After a 4-d weight-maintenance period, the
volunteers selected their food for 5d from computerized vending
machines offering a variety of familiar and preferred foods, ie, a
“cafeteria diet”. Twenty-four-hour energy expenditure and sub-
strate oxidation were measured in a respiratory chamber on the 4th
d of weight maintenance and the 5th d of ad libitum intake.
Average ad libitum intake was 13 732 ±4238 kJ/d (11 ± 1%
protein, 40 ±1% fat, 49 ±4% carbohydrate), ie, moderate
overeating by 27 ±37% above weight maintenance requirements
(range: -27% to 124%). Percent body fat correlated with daily
energy intake (n =0.53, P<0.01), the degree of overeating (r =
0.41, P<0.05), and the selection of a diet higher in fat and lower
in carbohydrate (r =0.70 and r-0.63, respectively, P<
0.001). Excess carbohydrate intake caused an increase in carbo-
hydrate oxidation (r =0.51, P<0.01), whereas excess fat intake
resulted in a decrease in fat oxidation (r =-0.53, P<0.01) and
thus a positive fat balance of 85 ±65 g/d. The positive relations
among degrees of obesity, dietary fat intake and overeating, and
the fact that dietary fat does not induce fat oxidation, support the
hypothesis that dietary fat promotes obesity in women. Am J
Clin Nutr 1995;62:911-7.
KEY WORDS Obesity, macronutrient intake and oxida-
tion, respiratory chamber, food-selection system
INTRODUCTION
It is well accepted that energy intake in excess of energy
expenditure is the primary cause of obesity. A more recent
approach to the study of the etiology of obesity, however, is to
consider separately the dietary balances of protein, carbohy-
drate, and fat (1, 2). Flatt (1), a pioneer in this area, proposed
that the difficulty in achieving fat balance on a Western fat-rich
diet may be a key factor in promoting body weight gain and
obesity. According to Flatt, fluctuations in carbohydrate and
protein intakes are compensated for by rapid, parallel fluctua-
tions in carbohydrate and protein oxidation, whereas most
excess dietary fat is not oxidized but stored in adipose tissue.
The development of obesity, therefore, may be related to ex-
cess fat intake, as consistently reported in large-scale studies
(3-8), or because of a reduced ability of obesity-prone mdi-
viduals to oxidize fat (2, 9).
The Pima and Papago Indians of the southwestern United
States are members of an obesity-prone population with a
prevalence of obesity approaching 80% in women (10). Pro-
spective studies in nondiabetic Pima Indians showed that a low
metabolic rate, for a given body size and composition (1 1), and
a low ratio of fat to carbohydrate oxidation are risk factors for
body weight gain (9). Little is known, however, about the
relations among ad libitum food intake and energy and macro-
nutrient balances in this and other populations.
The objective of this study was to investigate the associa-
tions between obesity and both energy and fat intakes in
women of Pima-Papago heritage. Recently, we published re-
sults on the use of an automated food-selection system to
measure ad libitum food intake in male volunteers on a meta-
bolic ward (12, 13). This system allows individuals free access
to a wide variety of familiar foods that can be tailored to their
preferences. Using this system, we asked whether 1) ad libitum
energy intake is related to body weight or obesity, 2) ad libitum
fat intake is related to obesity, and 3) changes in ad libitum
energy and macronutrient intakes are balanced by parallel
changes in energy expenditure and macronutrient oxidation?
SUBJECTS AND METHODS
Subjects
Twenty-eight nondiabetic women of Pima-Papago heritage
were admitted for 10 d to the metabolic ward of the Clinical
Diabetes and Nutrition Section of the National Institute of
Diabetes and Digestive and Kidney Diseases (NIDDK). None
of the subjects reported a recent attempt to lose weight. On
IFrom the Clinical Diabetes and Nutrition Section, National Institute of
Diabetes and Digestive and Kidney Diseases, National Institutes of Health,
Phoenix, AZ.
2Reprints not available. Address correspondence to DE Larson, Depart-
ment of Nutrition Sciences, Energy Metabolism Research Unit, The Uni-
versity of Alabama at Birmingham, Susan Mott Webb Nutrition Sciences
Building, 1629 University Boulevard, Birmingham, AL 35294-3360.
Received August 22, 1994
Accepted for publication July 28, 1995
by guest on July 13, 2011www.ajcn.orgDownloaded from

912 LARSON ET AL
‘ ± SD; range in parentheses.
admission, all volunteers were determined to be in good health
by medical history, physical examination, electrocardiogram,
blood screening, and urine test. According to their medical
history, all had regularly occurring menstrual cycles. None
were taking oral contraceptives or other medications. Their
physical characteristics are presented in Table 1. Body com-
position was determined by hydrostatic weighing with simul-
taneous measurement of residual lung volume. Percent body fat
was estimated from body density by using the equation of Sin
(14). Circumferences of the waist (at the level of the umbilicus)
and thigh (at the gluteal fold) were measured supine and
standing, respectively. The ratio of waist to thigh circumfer-
ence is an estimate of centrality of body fat. Seventeen volun-
teens had normal glucose tolerance and 1 1 had impaired glu-
cose tolerance, according to the criteria of the World Health
Organization. None of the volunteers were classified as binge
eaters according to the Gormally Binge Eating Scale (15), on
were clinically depressed according to their medical history.
Volunteers were restricted to the metabolic ward for the
duration of the study. Activity on the metabolic ward was
limited to ambulation in the hallways, playing billiards, and
playing table games. In addition, volunteers had supervised
outings to the hospital countyand two or three times a week for
3()-45 mm each. Sixteen of the volunteers were nonsmokers
and 12 were occasional smokers. They were allowed to smoke
only during scheduled, supervised outings. None of the volun-
teens complained of difficulties from the restrictive smoking
regimen. The experiment was approved by the Institutional
Review Board of the NIDDK and written, informed consent
was obtained.
Experimental protocol
To avoid any potential influence of menstrual cycle phase on
food intake or energy expenditure, volunteers were studied in
the follicular phase of the menstrual cycle. They reported to the
metabolic ward on the second or third day after the start of
menstruation and were studied during the following 9 d. Sub-
jects’ nude body weights were measured after they voided each
morning at 0530. For the first 4 d, volunteers were placed on a
standardized weight maintenance diet (20% protein, 30% fat,
and 50% carbohydrate) initially calculated by using a meta-
bolic ward prediction equation based on sex, body weight, and
height (unpublished), and subsequently adjusted to maintain
body weight within 1% of the weight on the second day of
admission. If necessary, adjustments were made to accommo-
date food dislikes. In our experience, we have determined that
4 d is sufficient for estimating weight maintenance require-
TABLE 1
Physical characteristics of 28 women of Pima-Papago heritage’
Value
Age (y) 27 ± 7 (19-5 1)
Height (cm) 160 ± 6 (150-173)
Weight (kg) 85.3 ± 19.0 (61.5-131.5)
BMI (kg/m2) 33.1 ± 6.9 (23.()-50.5
Percent body fat (%) 44 ± 6 (29-55)
Fat-free mass (kg) 47.5 ± 7.5 (35.0-64.1)
Fat mass (kg) 38.0 ± 13.0 (19.2-66.7)
Waist-thigh ratio 1.57 ± 0.21 (1.2()-2.23)
ments on a metabolic ward. On the 4th d of weight mainte-
nance, 24-h energy expenditure (24-h EE) was measured in a
respiratory chamber as previously described (16). After this
measurement, volunteers consumed all food ad libitum from an
automated food-selection system for 4 d. The 24-h EE mea-
sunement was repeated on day 5of ad libitum intake with the
volunteer using a food-selection system placed in the respira-
tory chamber. Three blood samples were drawn for measure-
ment of estradiol and progesterone: 1) on admission, 2) the
evening before the first measurement of 24-h EE (weight
maintenance period), and 3) immediately after the last mea-
surement of 24-h EE (ad libitum period). Serum estradiol and
progesterone were measured by nadioimmunoassay in a 10-mL
blood sample drawn from the antecubital vein. Follicular phase
was documented by a progesterone concentration < 1.0 .tWL
throughout the study and verified by estradiol concentration.
Automated food-selection system
The automated food-selection system was described previ-
ously (12, 13). Briefly, it includes two vending machines
(model 3007; U-Select-It, Des Moines, IA) that together con-
tam 40 trays. For this study, 34 trays were stocked with food
items and 6 with beverages. A typical “cafeteria diet” was
available including a variety of foods with varying fat, protein,
and carbohydrate contents. These included low-, medium-, and
high-fat entrees; low- and high-fat meats; cheese; bread; torti-
llas; pinto beans; fruit; vegetables; cereal; French fries; pop-
corn; chips; nuts; and low-and high-fat pastry or desserts. The
selection included food items typical of the Pima diet (17) and
tailored to individual food preferences. The same selection was
offered each day but changed slightly with the four daily loads
(0600, 1000, 1500 and 1800). In general, =20 of the 40 trays
at each load contained low-fat items ( 30% fat) and 10
contained high-fat items ( 50% fat). A selection of low- and
high-fat condiments was also available from a small refniger-
ator and was counted and restocked daily.
During the ad libitum period, volunteers were asked to
follow their typical eating pattern and were instructed to return
empty wrappers and unconsumed food portions to the vending
machines. Weights of all items loaded in and returned to the
vending machines were recorded. Daily energy, protein, fat,
and carbohydrate intakes were calculated from actual weights
of food and condiments consumed. During the 4-d ad libitum
period on the ward, volunteers had 24-h access to the vending
machines, housed in a separate eating area equipped with a
table, chain, television set, microwave oven, and toaster. For
simultaneous measurement of food intake and 24-h EE on day
5, the vending machine in the respiratory chamber was loaded
before the measurement period with the food items most often
selected during the previous 4 d.
Energy expenditure and substrate oxidation
Carbon dioxide production, oxygen consumption, and spon-
taneous physical activity (SPA) were measured continuously in
the respiratory chamber for 23 h from 0800 to 0700 the next
day and values were extrapolated to 24 h (16). Carbohydrate,
fat, and protein oxidation rates were calculated from 24-h
oxygen consumption, carbon dioxide production, and urinary
nitrogen excretion (18). Sleeping metabolic rate (SMR) was
defined as the average energy expenditure during all 15-mm
by guest on July 13, 2011www.ajcn.orgDownloaded from

AD LIBITUM FOOD INTAKE IN WOMEN 913
periods between 2300 and 0500 when the activity measured by
radar was <1.5%. During the weight maintenance measure-
ment, volunteers fasted from 2000 the evening before entering
the chamber. To account for confinement within the chamber,
80-85% of the weight-maintenance energy was given (19). No
restnictjons were applied during the ad libitum period. During
both measurements in the chamber, volunteers were asked not
to exercise in the chamber.
Data analysis
Energy intake is expressed in kJ/d, and as a percentage of
weight-maintenance-energy requirements (mean of 3 d). En-
ergy and macronutrient balances were compared between the
last day of weight maintenance and the last day of ad libitum
intake. To further assess the relation between energy and
macronutrient balances and obesity, the data were also divided
into tertiles of percent body fat.
Statistical analyses were performed with the programs of the
SAS Insthute Inc (Cary, NC). All data are expressed as mean
±SD. Comparisons of mean daily energy intake, macronutni-
ent composition, and energy metabolism between the weight
maintenance and ad libitum phases were performed by using
paired Student’s ttests. Comparisons between the upper and
lower tertiles of body fat were performed by using unpaired
Student’s ttests. During the ad libitum period, an analysis of
variance (ANOVA) with repeated measures was used to detect
the effect of time on energy or macronutrient consumption.
Relations between variables were assessed by Spearman cor-
relation or by linear-regression analyses.
RESULTS
Ad libitum food intake
Mean energy intake required for weight maintenance on our
metabolic ward was 9360 ± 1510 kJ/d (4.184 kJ =1 kcal).
During the ad libitum period, energy intake was increased to an
average of 13 732 ±4238 kJ/d (P <0.0001). Compared with
weight maintenance, ad libitum intake resulted in moderate
overeating for the group by 27 ± 37% (P <0.001) varying
from -27% to 124%. On average, the selected diet was com-
posed of 1 1 ±1% protein, 40 ±3% fat, and 49 ±4%
carbohydrate, which is higher in fat and lower in protein than
the imposed weight-maintenance diet (P <0.0001). The food
quotient of the ad libitum diet or the theoretical ratio of carbon
dioxide production to oxygen consumption was 0.846 ± 0.01 1.
By experimental design, the food quotient of the weight-main-
tenance diet was 0.866. In the ad libitum period, by repeated-
TABLE2
measures ANOVA, there was no effect of time (day of study)
on energy or macronutnient consumption.
After 5d of ad libitum intake, body weight increased by 0.4
±0.9 kg (P <0.05) over mean baseline body weight. The
change in body weight varied widely among individuals, from
-1.1 to 2.8 kg, and correlated with energy balance (deficit or
excess) over the 5-d period (r =0.60, P<0.001).
The relations between ad libitum energy intake, macronutri-
ent composition, and body weight and composition are re-
ported in Table 2. As shown in Figure 1, percent body fat was
associated with the percentage of fat in the diet (r =0.70, P<
0.00 1 )and the degree of overeating above weight maintenance
(r 0.41, P<0.05). By multiple-regression analysis, how-
ever, percent body fat was no longer associated with the degree
of overeating after the percent of fat in the diet was adjusted
for. This was due to the fact that percentage fat in the diet and
the degree of overeating were correlated (r =0.41 ,P<0.05).
There was no difference in food intake and composition of the
diet between the 17 subjects with normal glucose tolerance and
the 1 1 with impaired glucose tolerance.
Energy expenditure and macronutrient oxidation
Intake, oxidation, and balance values on day 4 of the weight-
maintenance period and day 5of the ad libitum period are
reported in Table 3. Overeating during the ad libitum period
was associated with significant increases (P <0.001) in SMR
(from 5994 ± 870 to 6464 ± I 126 kJ/d) and spontaneous
physical activity (from 6.7 ± 1.8 to 7.6 ± 3.1%/mm, P<
0.05), but not 24-h EE (NS). The higher respiratory quotient
during ad libitum intake reflects increased carbohydrate oxida-
tion and decreased fat oxidation. These changes in intake and
oxidation during the ad libitum period resulted in positive
balances for each macronutnient (Table 3).
During the ad libitum period, both carbohydrate and protein
oxidation correlated with their respective intakes (r =0.77 and
r0.71, respectively, P<0.0001). Fat oxidation, however,
was not positively correlated with fat intake, but was positively
correlated with the difference or “gap” between 24-h EE and
the sum of the energy ingested as carbohydrate and protein (r
=0.73, P<0.0001).
As shown in Figure 2, the change in 24-h EE between the ad
libitum and weight-maintenance periods correlated with
change in energy intake (r =0.51, P<0.01). The change in
carbohydrate oxidation correlated with the change in carbohy-
drate intake (r =0.51, P<0.01) whereas the change in fat
oxidation correlated negatively with the change in fat intake
(r -0.53, P<0.01).
Spearman rank correlation coefficients (r between body weight and body composition. and ad lihitum energy intake and macronutrient composition’
Energy (kJ) Overeating2
(%) Protein
(%) Fat
(e/() Carbohydrate
(C/e) Food quotient’
Weight 0.41 (0.03) 0.27 0.26 0.30 -0.32 -0.33
Percent body fat 0.53 (0.01) 0.41 (0.03) 0.27 0.70 (0(1(X)!) -0.63 (0.(X)l) -0.68 (0.()001)
Waist-thigh ratio 0.52 (0.01) 0.44 (0.02) -0.15 0.23 -0.09 -0.14
,Pvalue in parentheses.
2Defined as the percent above or below weight-maintenance-energy requirements.
3Theoretical respiratory quotient of the diet.
by guest on July 13, 2011www.ajcn.orgDownloaded from

U-
E
2
IL
>1
a)
5-
0
C
Lu
0’j’
)5 0
015
 .2
. 
(r=O.70, P=O.0001)
3 0 25 30 35 40 45 50 55 6
I50
0
100
0
:. 
(r=O.41 ,P<O.05)
-50 25 30 35 40 45 50 55 60
%Body Fat
FIGURE 1. The degree of overeating (lower panel) and the selection of
a high-fat diet (upper panel) correlated positively with percent body fat
(it =28). (After the exclusion of one outlier, r=0.51, P<0.01 for the
relation between the degree of overeating and percent body fat).
914 LARSON ET AL
Most compared with least obese volunteers
When individuals in the upper and lower tertiles for obesity
were grouped (n 9 per group), energy intake for the most
obese group (101.8 ± 17.6 kg, 50 ± 3% body fat) was 16 117
±3845 compared with 1 1 334 ± 4586 kJ/d (P <0.05) in the
least obese group (74. 1 ±1 1.1 kg, 36 ± 4% body fat). On
average, the selected diet was composed of 12 ±1% protein,
44 ±2% fat, and 44 ±2% carbohydrate for the most obese
group and 1 1 ±1% protein, 38 ± 2% fat, and 51 ±3%
carbohydrate for the least obese group (P <0.0001 for fat and
carbohydrate). These differences were associated with a higher
24-h EE in the most obese compared with the least obese group
(10376 ± 1502 and 7401 ± 749 kJ/d, respectively) but no
difference in 24-h respiratory quotient (0.892 ± 0.033 and
0.881 ± 0.041, respectively). As shown in Figure 3, this
resulted in a more positive fat balance in the most obese group.
DISCUSSION
Using an automated food-selection system, we investigated
whether ad libitum food intake was related to body weight and
obesity in women with a genetic predisposition to obesity.
When offered unlimited access to a variety of palatable and
familiar foods for 5d, there was a wide range of under- and
)overeating that for the group averaged “‘27% above weight-
maintenance-energy requirements. Obesity was associated with
both overeating and the selection of a high-fat diet. Such
overeating resulted in an increase in carbohydrate oxidation,
related to the increase in carbohydrate intake, but a decrease in
fat oxidation. This combination of overeating and decreased fat
oxidation led to a more positive fat balance in the most obese
individuals.
Little is known about the influence of ad libitum food intake
on the pathogenesis of obesity in the Pima, Papago, or other
populations. One likely reason is because accurate collection of
food intake data and interpretation of its relation to obesity are
difficult in free-living conditions. It is well established that
obese individuals have higher energy expenditures than do
normal-weight individuals (16, 18, 19), yet food intake studies
have failed to find higher energy intakes among obese individ-
uals (3-6, 20). The expected associations between reported
energy intake and body weight and obesity, however, may be
obscured by error in the use of imprecise proxies for estimating
food intake and body composition (21) or in systematic under-
reporting of intake by heavier individuals. Recent studies mea-
suning energy expenditure in free-living conditions (doubly
labeled water) simultaneously with estimation of food intake,
provide evidence that obese individuals underreport food in-
take to a greater extent than do nonobese individuals (22, 23).
Still, despite this evidence, some investigators continue to
question the role of overeating in the pathogenesis of obesity
(5, 24). In the present study, we measured food intake on a
metabolic ward to avoid the problems associated with recorded
or recalled food intake data. In the controlled setting of a
metabolic ward, our finding that energy intake was correlated
with both body weight and obesity supports the intuitive con-
cept that excess energy intake is required to induce and main-
tam obesity.
Despite the errors associated with the collection of food
intake data, a growing number of studies have found links
between obesity and fat intake. Epidemiologic reports from the
TABLE 3
Energy metabolism, macronutrient intakes, oxidation, and balances on day 4 of the weight-maintenance period and day 5of the ad libitum periods’
Weight maintenance Ad libitum
Intake Oxidation Balance Intake Oxidation Balance
Energy(kJ/d) 7912± 1372 8569± 1443 -649±962 13949±42472 8832± 1753 5117±362323
Protein (kJ/d) 1632 ± 268 862 ± 247 770 ± 3353 1598 ± 556 787 ±351 787 ± 351
Fat (KJ/d) 2356 ± 397 3527 ± 1280 -1172 ± 1280’ 5707 ±18702 2515 ±12302 3197 ± 243923
Carbohydrate (kJ/d) 4029 ± 774 4117 ±916 88 ± 623 6720 ± 20922 5473 ± 15272 1247 ± 143523
I: ±SD.
2Significantly different from weight-maintenance period, P<0.0001.
3Balance significantly different from zero, P<0.01.
by guest on July 13, 2011www.ajcn.orgDownloaded from

C
J1
C
0
15
.15
Sc
o-
o
5-15
-a)
0Prot&n
E: CHO
aFat
Least Obese Most Obese
Fat
2000
0
-2000
AD LIBITUM FOOD INTAKE IN WOMEN 915
y7
2000 o
o0
-2000
-4000 (raO.51. P<O.O1)
-4000 0 4000 8000 12000
4000 o CHO
o’
-2000
-4000 (rO.51, P<O.O1)
-4000 0 4000 8000 12000
4000
-4000
S
S (r=..O.53, P<O.O1)
-4000 0 4000 8000 12000
Change In Energy or Macronutrlent Intake
(Ad Libltum -Weight Maintenance, kJ/d)
FIGURE 2. Changes in 24-h energy expenditure and carbohydrate
(CHO) and fat oxidation compared with the changes in energy, carbohy-
drate, and fat intakes from the weight maintenance to the ad libitum period.
The small increase in 24-h energy expenditure (upper panel) represents the
balance between the increase in carbohydrate oxidation (middle panel) and
the decrease in fat oxidation (lower panel). (After the exclusion of one
outlier, r=0.54, P<0.01 for the relationship between the change in fat
intake and the change in fat oxidation).
early 1970s suggest that the prevalence of obesity is higher in
populations with a diet rich in fat (25). Results from large-scale
and population-based studies show that obesity is characterized
by consumption of a high-fat, low-carbohydrate diet (3-8, 20).
Longitudinal studies have determined that a high fat intake
predicts weight gain in both men and women (26), particularly
women predisposed to obesity (27). Also, the sensory and
hedonic studies of Drewnowski et al (28) and Mela and Sac-
chetti (29) have demonstrated that obese individuals may have
a greater preference for dietary fat than do nonobese individ-
uals. In the present study, women of Pima or Papago heritage
selected a diet composed of 1 1 % protein, 40% fat, and 49%
carbohydrate, similar to the national average for women, which
is 16% protein, 37% fat, and 46% carbohydrate (30). Consis-
tent with previous findings, the most obese individuals selected
a diet richer in fat. Interestingly, our correlations between
macronutnient intake and obesity are stronger than those re-
ported in large-scale studies, possibly reflecting more precise
FIGURE 3. Protein, carbohydrate (CHO), and fat balances on day 5 of
ad lihitum intake for the lower (least obese) and upper (most obese) tertiles
of obesity (ii =9 in each group). Corresponding energy balances were
3690 ± 3217 in the least obese and 6673 ± 3920 kJ/d in the most obese
subjects (NS). Error bars are SD for macronutrient balance. 5Significantly
different from least obese subjects, P<0.05.
measurements of food intake on a metabolic ward; better
control of other interfering factors such as activity, alcohol
consumption, or smoking, or the genetic homogeneity of this
obesity-prone population. On the other hand, we recognize that
the artificial conditions of the study, ie, free food, restricted
habitual activity, and boredom may have influenced food in-
take in these women.
What are the possible mechanisms responsible for the role of
dietary fat in the pathogenesis of obesity? Clinical studies have
clearly shown that diets rich in fat induce hyperphagia (7, 31),
possibly because fat is more energy dense (32) and less satiating
than is carbohydrate (33). Because fats determine the texture,
flavor, and odor of many foods (34), it could be argued that
fat-rich foods are easy to overeat because of their hedonic appeal
and “mouthfeel”. Flatt has also proposed that high-fat diets pro-
mote hyperphagia to maintain carbohydrate balance (1). In this
study, obesity was associated with both overeating and the selec-
tion of a high-fat diet. Although we cannot imply cause and effect,
our analyses suggest that overeating by many of our volunteers
was primarily the result of selecting a high-fat diet.
Another explanation for the role of dietary fat in promoting
obesity may be related to differences in the metabolism of the
macronutnients (1, 19). It is fairly well established that although
the consumption of excess carbohydrate is met with increased
carbohydrate oxidation (12, 13), the consumption of excess fat
does not increase fat oxidation (35). According to Flatt’s stud-
ies in mice fed ad libitum, fat oxidation correlates with the
“gap” between total energy expenditure and energy ingested as
carbohydrate and protein (36). In our female volunteers, fat
oxidation correlated with the “gap” described by Flatt, provid-
ing further support for the hypothesis that fat oxidation depends
on the intakes of carbohydrate and protein, and total energy
expenditure (1, 35, 36). In agreement with our studies in white
(12) and Pima and Papago men (13), the relation between
changes in carbohydrate oxidation and changes in carbohydrate
intake support the close regulation of carbohydrate balance,
which is likely due to the limited storage capacity for carbo-
hydrate (1, 19). The negative relation between fat oxidation and
excess fat intake in the present study most likely reflects the
suppression of fat oxidation via the antilipolytic effect of an
insulin response to the carbohydrate load.
by guest on July 13, 2011www.ajcn.orgDownloaded from

916 LARSON El AL
Of further interest is the possibility that some individuals
may be more susceptible to obesity with the consumption of a
high-fat diet (2). In Swedish women, high-fat intake was as-
sociated with a 6-y increase in BMI only in women defined to
be genetically predisposed to obesity (27). In nondiabetic Pima
Indians, a low ratio of fat to carbohydrate oxidation, measured
under standardized dietary conditions, is associated with sub-
sequent weight gain (1 1). In whites, Thomas et al (31) found
that after 7 d of a high-fat diet, lean men and women were
better able to match fat oxidation to fat intake and were
therefore in less positive fat balance than were their obese
counterparts. In the present study, we found no evidence to
support that the most obese women were less able to stimulate
fat oxidation in response to fat intake than were the least obese
women. Although ad libitum fat balance was significantly
more positive among the more obese women, this was largely
due to higher fat intake. It is worth noting, however, that the
least obese group in our study was “fatter” than the women
studied by Thomas et al (37 ± 4% compared with 26 ± 1%
body fat) and that our experimental period was 2 d shorter (5
compared with 7 d). A cross-sectional study of a population
with a high prevalence of obesity, however, may not be ideal
for detecting defects in fat oxidation among individuals.
The findings in this study support the argument that body
weight gain occurs through a combination of overeating and
the selection of a high-fat diet. Our results, collected in a
controlled setting, agree with large-scale studies reporting a
link between obesity and dietary fat intake. The moderate
overeating induced by exposure to a variety of palatable foods,
particularly in more obese women, may be a consequence of a
high-fat diet. Such moderate overeating leads to increased
carbohydrate oxidation parallel with an increased carbohydrate
intake and suppression of fat oxidation, leading to fat deposi-
tion. These findings support the growing evidence that excess
dietary fat promotes obesity. LI
We thank John Graves, Emma Thin Elk, Ella Mae Enos, Tony Brown,
and Carson Mineer of the metabolic kitchen for their assistance in main-
taming the automated food-selection system. We thank Carol Massengill
and the nursing staff for their professional care of the volunteers. We thank
Tom Anderson for his help with the hydrostatic weighing. Most impor-
tantly. we thank the volunteers.
REFERENCES
1. Flatt JP. The difference in the storage capacities for carbohydrate and
for fat, and its implication in the regulation of body weight. Ann N Y
Acad Sci 1987;499:1()4-2l.
2. Astrup A, Raben A. Obesity: inherited metabolic deficiency in the
control of macronutrient balance? Eur I Clin Nutr 1992;46:6l 1-20.
3. Dreon DM, Frey-Hewitt B, Ellsworth N, Williams PT, Terry RB,
Wood PD. Dietary fat:carhohydrate ratio and obesity in middle-aged
men. Am J Clin Nutr l988;47:995-l000.
4. Kromhout D. Energy and macronutrient intake in lean and obese
middle-aged men (the Zutphen study). Am J Clin Nutr 1983;37:295-9.
5. Miller WC, Niederpruem MG, Wallace JP, Lindeman AK. Dietary fat,
sugar, and fiber predict body fat content. J Am Diet Assoc l994;94:
612-5.
6. Romicu I, Willett WC. Stampfer MJ, Ct al. Energy intake and other
determinants of relative weight. Am J Clin Nutr 1988;47:406-12.
7. Tremblay A, Plourde G, Despres J-P, Bouchard C. Impact of dietary
fat content and fat oxidation on energy intake in humans. Am J Clin
Nutr 1989;49:799-805.
8. Tucker LA, Kano MJ. Dietary fat and body fat: a multivariate study of
205 adult females. Am J Clin Nutr I992;56:616-22.
9. Zurlo F, Lillioja 5, Esposito-Del Puente A. Ct al. Low ratio of fat to
carbohydrate oxidation as predictor of weight gain: study of 24-h RQ.
Am J Physiol 1990;259:E650-7.
10. Knowler WC, Peltitt DJ, Saad MF, et al. Obesity in the Pima Indians:
its magnitude and relationship with diabetes. Am J Clin Nutr 1991;
53(suppl): 15435-515.
11. Ravussin E, Lillioja 5, Knowler WC, et al. Reduced rate of energy
expenditure as a risk factor for body-weight gain. N Engl J Med
1988;3I8:467-72.
12. Rising R, Alger 5, Boyce V. et al. Food intake measured by an
automated food-selection system: relationship to energy expenditure.
Am J Clin Nutr 1992;55:343-9.
13. Larson DE, Rising R, Ferraro RT, Ravussin E. Spontaneous overfeed-
ing with a “cafeteria diet” in men: effects on 24-hour energy expen-
diture and substrate oxidation. tnt J Obes 1995; 19:331-7.
14. Sin WE. Body composition from fluid spaces and density: analysis of
methods. In: Brozek J, Henschel A, eds. Techniques for measuring
body composition. Washington, DC: National Academy of Sciences,
1961:233-44.
15. Gormally J, Black 5, Datson 5, Rardin D. The assessment of binge
eating severity among obese persons. Addict Behav l982;7:47-55.
16. Ravussin E, Lillioja 5, Anderson TE, Christin L, Bogardus C. Deter-
minants of 24-hour energy expenditure in man. J Clin Invest 1986;78:
1568-78.
17. Smith Ci, Schakel SF, Nelson RG. Selected traditional and contem-
porary foods currently used by the Pima Indians. J Am Diet Assoc
199 1;91:338-41.
18. J#{233}quierE, Acheson K, Schutz Y. Assessment of energy expenditure
and fuel utilization in man. Annu Rev Nutr 1987;7:187-208.
19. Abbott WGH, Howard BV, Christin L, et al. Short-term energy bal-
ance: relationship with protein, carbohydrate, and fat balances. Am J
Physiol 1988;18:E332-7.
20. Braitman LE, Adlin EV, Stanton IL. Obesity and caloric intake: the
National Health and Nutrition Examination Survey of 1971-1975
(NHANES I). I Chron Dis 1985;38:727-32.
21. Lissner L, Habicht J-P, Strupp BJ, Levitsky DA, Haas ID, Roe DA.
Body composition and energy intake: do overweight women overeat
and underreport? Am J Clin Nutr 1989;49:320-5.
22. Lichtman SW, Pisarska K, Berman, ER, et al. Discrepancy between
self-reported and actual caloric intake and exercise in obese subjects.
N Engl I Med 1992;327:1893-8.
23. Bandini LG, Schoeller DA, Cyr HN. Dietz WH. Validity of reported
energy intake in obese and nonohese adolescents. Am I Clin Nutr
1990;52:421-5.
24. Miller WC. Diet composition, energy intake, and nutritional status in
relation to obesity in men and women. Med Sci Sports Exerc 1991;
23:280-4.
25. West KM. Epidemiology of obesity. In: Vague J, Boyer I, eds. The
regulation of adipose tissue mass. New York: American Elsevier,
1974:201-7.
26. Klesges RC, Kiesges LM, Haddock CK, Eck LH. A longitudinal
analysis of the impact of dietary intake and physical activity on weight
change in adults. Am I Clin Nutr l992;55:818-22.
27. Heitmann BL, Lissner L, Sorensen TIA, Bengtsson C. Dietary fat
intake and weight gain in women genetically predisposed for obesity.
Am I Clin Nutr 1995;61:1213-7.
28. Drewnowski A, Brunzell ID, Sande K, Iverius PH, Greenwood MRC.
Sweet tooth reconsidered: taste responsiveness in human obesity.
Physiol Behav 1985;35:617-22.
29. Mela DJ, Sacchetti DA. Sensory preferences for fats: relationships
with diet and body composition. Am J Clin Nutr 199 l;53:908-15.
30. Peterkin BB. Women’s diets: 1977 and 1985. 1 Nutr Ed 1986;18:
25 1-7.
31. Thomas CD, Peters JC, Reed GW, Ahumrad NN, Sun M, Hill JO.
Nutrient balance and energy expenditure during ad libitum feeding of
by guest on July 13, 2011www.ajcn.orgDownloaded from

AD LIBITUM FOOD INTAKE IN WOMEN 917
high-fat and high-carbohydrate diets in humans. Am I Clin Nutr  34. Drewnowski A. Fats and food texture: sensory and hedonic evalua-
l992;55:934-42. lions. In: Moskowitz HR. ed. Food texture. New York: Marcel Dekker,
32. Duncan KH, Bacon IA, Weinsier RL. The effects of high and low 1987:251-72.
energy density diets on satiety, energy intake, and eating time ofobese 35. Flatt JP, Ravussin E, Acheson K, J#{233}quierE. Effects of dietary fat on
and nonobese subjects. Am J Clin Nutr 1983;37:763-7. postprandial substrate oxidation and on carbohydrate and fat balances.
33. Blundell JE, Burley VJ, Cotton JR, Lawton CL. Dietary fat and the I Clin Invest 1985;76:1019-24.
control of energy intake: evaluating the effects of fat on meal size and 36. Flatt JP. Opposite effects of variations in food intake on carbohydrate and
postmcal satiety. Am I Clin Nutr 1993;57:722S-8S. fat oxidation in ad libitum fed mice. I Nutr Biochem 199l;2:186-92.
by guest on July 13, 2011www.ajcn.orgDownloaded from