Content uploaded by Barbara Jean Rolls
Author content
All content in this area was uploaded by Barbara Jean Rolls on Feb 10, 2014
Content may be subject to copyright.
ABSTRACT This study examined the effect of energy den-
sity, independent of fat content and palatability, on food and
energy intakes. With use of a within-subjects design, normal-
weight women (n = 18) were provided with meals for 2 d during
each of three test sessions. During lunch, dinner, and an evening
snack, subjects were given free access to a main entree varying
in energy density (low, medium, or high). The manipulated main
entrees were similar in palatability to their counterparts across
conditions. Low-energy compulsory (consumption required) side
dishes accompanied each meal. Subjects also consumed a stan-
dard, compulsory breakfast. Results showed that subjects con-
sumed a similar amount of food (by weight) across the three con-
ditions of energy density. Thus, significantly more energy was
consumed in the condition of high energy density (7532 ± 363
kJ, or 1800 ± 86 kcal) than in the medium- (6356 ± 281 kJ, or
1519 ± 67 kcal) and low- (5756 ± 178 kJ, or 1376 ± 43 kcal)
energy-density conditions (P < 0.0001). There were no differ-
ences in hunger or fullness before meals, after meals, or over the
2 d across conditions. The results from this study indicate that
energy density affects energy intake independent of macronutri-
ent content or palatability, suggesting that the overconsumption
of high-fat foods may be due to their high energy density rather
than to their fat content. Am J Clin Nutr 1998;67:412–20.
KEY WORDS Caloric density, energy density, energy
intake, fat, human food intake, macronutrients, satiation, women
INTRODUCTION
The high consumption of dietary fat has been linked to
adverse health outcomes, including obesity and associated disor-
ders such as type 2 diabetes mellitus, cardiovascular disease, and
certain forms of cancer (1). Approximately 34% of energy intake
in the United States is derived from fat, despite recommenda-
tions that this be reduced to 30% (2). Dietary fat may be
overeaten for several reasons, including high palatability, high
energy density, and insensitivity to the satiety value of fat (3).
Dietary fat may affect satiation, the processes involved in the
termination of a meal (4). Some investigators suggest that fat is
overeaten because it is not as satiating as the other macronutrients,
especially carbohydrate (5, 6). On the other hand, it has been sug-
gested that dietary fat may be overeaten because of its high energy
density (the energy content per weight of food) (7). Dietary fat pro-
vides more than twice as much energy per gram as either carbohy-
drate or protein. A high-fat food or meal is usually much smaller in
weight than an isoenergetic high-carbohydrate food or meal.
Therefore, if individuals consume a constant amount of food, the
high energy density of fat, rather than the fat content, could be a
key factor in its overconsumption (8). Fat can also improve the tex-
ture and flavor of foods, thereby increasing palatability (9). This
influence of fat on the sensory properties of foods could be an
important determinant of satiation and food intake (10).
Previous studies have investigated the effects of dietary fat on
food intake by manipulating the ratio of fat to carbohydrate in diets
while holding energy density constant. Thus, the effects of
macronutrient composition were studied apart from the effects of
energy density on food intake. In one of the first studies, van Stra-
tum et al (11) used liquid diets varying in fat content (24% and 47%
of energy) but similar in energy density. Subjects were allowed
25% of their energy from standardized snacks. Over 2 wk with
each diet, a constant weight of the liquid diet was consumed and
energy intake remained constant, suggesting that energy intake was
not affected by the fat content. In another study, Stubbs et al (12)
compared energy intakes from diets containing solid foods varying
in fat content (20%, 40%, and 60%) but similar in energy density.
Six men consumed ad libitum each of the three diets for 14 d.
Within each diet, all foods were of the same macronutrient compo-
sition. Energy intakes from the three diets were similar. Although
results from both of these studies imply that fat content, indepen-
dent of its effects on energy density, does not affect energy intake,
the interpretation and application of the findings are limited
because liquid diets were used in one study and in both studies
palatability of foods was not shown to be well-matched.
In two recent comprehensive reviews (7, 13), the authors con-
cluded that the overconsumption of energy from a high-fat diet
is a result of the diet’s high energy density rather than the diet’s
fat content. Most conclusions related to energy density and its
effects on food intake are based on experiments in which indi-
viduals were provided with low- or high-fat diets varying in
energy density. In studies by Duncan et al (14), Kendall et al
(15), Lissner et al (16), Stubbs et al (17, 18), and Tremblay et al
(19), subjects consumed significantly more energy from the diets
high in fat content and energy density. One may conclude from
these studies that diets high in both fat and energy density result
Energy density of foods affects energy intake in normal-weight
women
1–3
Elizabeth A Bell, Victoria H Castellanos, Christine L Pelkman, Michelle L Thorwart, and Barbara J Rolls
1
From the Nutrition Department, The Pennsylvania State University, Uni-
versity Park, and Florida International University, Miami.
2
Supported by NIH grants DK39177 and DK50156.
3
Address reprints requests to BJ Rolls, The Pennsylvania State Universi-
ty, Nutrition Department, 226 Henderson Building, University Park, PA
16802. E-mail: bjr4@psu.edu.
Received August 12, 1997.
Accepted for publication November 9, 1997.
Am J Clin Nutr 1998;67:412–20. Printed in USA. © 1998 American Society for Clinical Nutrition
412
by guest on July 12, 2011www.ajcn.orgDownloaded from
in greater energy intake. However, it is not possible to separate
the effects of energy density from the effects of macronutrient
content (fat) on food or energy intake because fat content and
energy density were related. The possibility also exists that
palatability might have affected food intake because palatability
tends to increase in direct relation to fat content (9).
The present experiment investigates the effect of energy den-
sity on food intake while specifically controlling for the poten-
tial effects of macronutrient content (ie, fat) and palatability on
food and energy intakes. This study was designed to test the
hypothesis that individuals are not sensitive to energy density
cues. That is, individuals will consume a constant weight or
amount of food, regardless of changes in energy density, thus
resulting in variations in energy intake.
SUBJECTS AND METHODS
Subjects
Eighteen healthy, normal-weight women successfully com-
pleted this study. We chose to conduct this study in women
because a greater percentage of the American adult female pop-
ulation than of the adult male population is overweight (2). Vol-
unteers were recruited via advertisement from The Pennsylvania
State University community. On response to the advertisement,
individuals completed a standard telephone interview to ensure
that they met the general criteria for inclusion in the study:
20–45 y of age, nonsmoking, in good health, not dieting, not in
athletic training, not pregnant or lactating, no chronic health
problems, not using medications known to affect food intake or
appetite, consuming meals (including breakfast) at regularly
scheduled intervals, no weight gain or loss (±10 lbs or 4.5 kg) in
the previous 6 mo, and no food allergies or food restrictions.
After the initial telephone interview, potential subjects were
measured for weight and height and were asked to complete sev-
eral screening questionnaires, including the Eating Inventory
(EI) (20), which assesses dietary restraint, disinhibition, and
hunger; the Eating Attitudes Test (EAT) (21), which detects aber-
rant attitudes toward food and eating; and the Beck Depression
Inventory (BDI) (22), which detects cognitive indicators of
depression. Only normal-weight women [body mass index
(BMI; in kg/m
2
): 19–26] who scored <30 on the EAT and <10
on the BDI were included in the study. Scores on the EI were not
used as inclusion or exclusion criteria.
Each subject signed a consent form acknowledging receipt of
a description of the experimental procedures and participation in
the study. Each subject received financial compensation for her
participation in the study. To prevent experimental bias, the con-
sent form indicated that the experiment sought to examine peo-
ple’s perceptions of different types of food products. The study
was approved by The Pennsylvania State University Institutional
Review Board.
Twenty-three subjects began participation in this study. Two
subjects were selected for the study and were dropped when it
was recognized that they had elevated scores on the BDI. Two
subjects consumed the entire amount of food offered during the
evening snack and were dropped from the study because we
could not be certain that the amount of food offered did not limit
consumption. One subject withdrew from the study for personal
reasons. Therefore, a total of 18 subjects (x
–
± SEM: aged 25.5 ±
1.4 y with a BMI of 23.0 ± 0.4) completed the study. Mean (±
SEM) scores on the EI for cognitive restraint, disinhibition, and
hunger were 7.2 ± 1.1, 4.4 ± 0.6, and 4.6 ± 0.6, respectively.
Procedures
A within-subjects, repeated-measures design was used in this
experiment. Subjects participated in three, 3-d test sessions.
Order of presentation of the conditions was counterbalanced
across subjects.
During the first 2 d of the test sessions, subjects ate breakfast,
lunch, and dinner in our laboratory and were provided with take-
home evening snacks. On the third day subjects completed a
brief questionnaire. Breakfast was standard across conditions
and was required to be consumed in full. Lunch, dinner, and the
evening snack consisted of one main entree that varied across
three conditions of energy density and several low-energy side
dishes that were standard (in type and amount) across dietary
treatments. Subjects consumed the main entrees ad libitum and
were required to consume the side dishes in full. Large servings
(lunch: >1000 g; dinner: > 1500 g; snack: >600 g) of the main
entrees were provided so that food intake was not limited by por-
tion size. Small amounts of the compulsory side dishes were
served (Table 1). Therefore, the study was designed so that the
majority of both daily food and energy intake was derived from
the manipulated main entrees. Four different 2-d menus were
created using combinations of the two possible lunches and two
possible dinners (Table 1). These menus were randomly assigned
to the subjects.
Study sessions were scheduled on the same days of alternate
weeks for 5 wk (ie, Tuesday, Wednesday, and Thursday of every
other week). During the test days, subjects were instructed to
consume only foods and beverages provided by the laboratory,
with the exception of nonenergetic beverages. Beverage intake
was recorded by subjects and verified for compliance with the
experiment protocol. Subjects were asked to refrain from drink-
ing alcohol during the 24 h preceding and throughout test ses-
sions and to maintain similar exercise schedules during test ses-
sions. On arrival at the laboratory in the morning of each test
day, subjects completed a brief questionnaire to assess compli-
ance with the experimental protocol. Meals were scheduled at
approximately the same time during the day across all conditions
with a minimum of 3.5 h between breakfast and lunch, and lunch
and dinner.
Subjects were seated in individual cubicles and were not
allowed to read during meals. Before service of each meal and
again after the consumption of the meal, subjects completed a
series of 100-mm visual analogue scales (VAS) rating their
degree of hunger, thirst, fullness, perception of how much they
could eat (prospective consumption), and nausea. Also, before
and after lunch and dinner, subjects were presented with a sam-
ple (10 g) of the main entree and were asked to complete the fol-
lowing questions on 100-mm VAS: “How pleasant is the taste of
this food right now?”, “How pleasant is the texture of this food
right now?”, “How much of this food do you think you could eat
right now?”, and “How many calories do you think this food
has?”. After dinner, subjects were provided with their evening
snack. Subjects returned food packages, including uneaten food,
from the evening snack to the laboratory the following morning.
On the third day of each session, subjects completed a brief
questionnaire with open-ended questions about their experiences
in the laboratory during the previous 2 d. Subjects also retro-
spectively rated their hunger between meals and their fullness
BELL ET AL 413
by guest on July 12, 2011www.ajcn.orgDownloaded from
over the previous 2 d on the VAS. At the end of the study, sub-
jects completed a discharge questionnaire. This questionnaire
asked subjects to state what they believed the purpose of the
study to be and gave subjects the opportunity to share other com-
ments relevant to the study.
Manipulated main entrees
Only commercially available ingredients were used in the
manipulated entrees. Lunch and dinner entrees were formulated to
vary across three levels of energy density. Lunch entrees consisted
of pasta salad with Italian dressing and pasta salad with yogurt
dressing. Dinner entrees were Italian pasta bake and chicken noo-
dle casserole. Evening snacks consisted of smaller portions of the
entrees served for lunch. Information on the macronutrient and
energy contents of the entrees is provided in Table 2.
Variation in energy density of the entrees was accomplished
primarily by manipulating the proportion of low-fiber vegetables
and pasta, such that the entrees that were low in energy density
contained more vegetables and less pasta than the entrees that
were higher in energy density. In effect, entrees of low energy den-
sity had greater water content than entrees of higher energy den-
sity. The entrees were formulated by using NUTRITIONIST IV
(version 3.5; N-Squared Computing, San Bruno, CA) on the basis
of nutrient composition data from manufacturers and Bowes and
Church’s Food Values of Portions Commonly Used (23). Recipes
for the entrees may be obtained by contacting the first author.
Analysis of the protein, fat, moisture, and ash contents of the
entrees was conducted by The Pennsylvania State University
Crop Quality Laboratory by using methods of the Association of
Official Analytical Chemists (24). Carbohydrate content was cal-
414 ENERGY DENSITY AFFECTS ENERGY INTAKE
TABLE 1
One sample menu order
1
Amount Amount
Day and meal By weight By energy Day and meal By weight By energy
g kJ (kcal) g kJ (kcal)
Day 1 Day 2
Breakfast Breakfast
Bagel
2
57.0 670 (160) Bagel 57.0 670 (160)
Cream cheese
3
16.0 146 (35) Cream cheese 16.0 146 (35)
Peaches, canned
4
124.0 259 (62) Pears, canned
13
124.0 259 (62)
Water, tea, or coffee (ad libitum) Water, tea, or coffee (ad libitum)
Lunch, A Lunch, B
Pasta salad with Italian dressing (ad libitum) Pasta salad with yogurt dressing (ad libitum)
Cracked pepper crackers
5
13.0 218 (52) Golden crackers
14
12.0 216 (52)
Chocolate chip cookies
6
6.7 125 (30) Cinnamon graham snacks
15
6.0 92 (22)
Water (ad libitum) Water (ad libitum)
Dinner, B Dinner, A
Italian pasta bake (ad libitum) Chicken noodle casserole (ad libitum)
Dinner roll
7
35.0 335 (80) Dinner roll 35.0 335 (80)
Carrot sticks
8
18.0 32 (8) Carrot sticks 18.0 32 (8)
Celery sticks
9
16.0 10 (2) Celery sticks 16.0 10 (2)
Devil’s food cookie cake
10
16.0 209 (50) Strawberry gelatin
16
92.0 42 (10)
Water (ad libitum) Whipped topping
17
4.0 42 (10)
Water (ad libitum)
Evening snack Evening snack
Pasta salad (same variety as lunch) Pasta salad (same variety as lunch)
Peaches or pears, canned
11
113.0 236 (57) Peaches or pears, canned 113.0 236 (57)
Wheat crackers
12
15.0 251 (60) Wheat crackers 15.0 251 (60)
1
Although lunch and dinner menus were counterbalanced across subjects, food items in the menus remained constant.
2
Lender’s Bagels, Lender’s Bagel Bakery, Rye Brook, NY.
3
Philadelphia Light Cream Cheese, Kraft, Inc, Glenview, IL.
4
Del Monte Lite Peaches, Del Monte Foods, San Francisco.
5
Snackwell’s Cracked Pepper Crackers, Nabisco Foods, East Hanover, NJ.
6
Snackwell’s Chocolate Chip Cookies, Nabisco Foods.
7
Bread du Jour Italian Roll, Interstate Brands Corporation, Kansas City, MO.
8
Wm Bolthouse Farms, Inc, Bakersfield, CA.
9
Calcel Marketing, Inc, Oxnard, CA.
10
Snackwell’s Devil’s Food Cookie Cakes, Nabisco Foods.
11
Del Monte Fruit Cup, Del Monte Foods.
12
Snackwell’s Wheat Crackers, Nabisco Foods.
13
Del Monte Lite Pears, Del Monte Foods.
14
Snackwell’s Classic Golden Crackers, Nabisco Foods.
15
Snackwell’s Cinnamon Graham Snacks, Nabisco Foods.
16
Jell-O Sugar Free Gelatin Snack, Kraft General Foods, Inc, White Plains, NY.
17
Cool Whip Lite Whipped Topping, Kraft General Foods.
by guest on July 12, 2011www.ajcn.orgDownloaded from
culated as the difference between the total weight and the sum of
protein, fat, moisture, and ash contents. The entrees were on aver-
age 22% fat, 59% carbohydrate, and 19% protein. Average values
for the energy density of the entrees were 3.5 kJ/g (0.8 kcal/g),
4.4 kJ/g (1.1 kcal/g), and 5.6 kJ/g (1.3 kcal/g) for the conditions
of low, medium, and high energy density, respectively (Table 2) .
These values reflect reductions in energy density of the entrees of
<20% between conditions such that the entrees of low energy
density contained <40% less energy per gram than the entrees of
high energy density. Dietary fiber content was calculated from
fiber information from manufacturers and from Bowes and
Church’s Food Values of Portions of Commonly Used (23).
To ensure that manipulated entrees were equally well liked
and perceived as having similar energy contents as their counter-
parts in different conditions, a preliminary taste test was con-
ducted. In a within-subjects, repeated-measures design, 22
women rated the pleasantness of taste and perceived energy con-
tent of the entrees. During three test sessions, separated by ≥1 d,
subjects were seated in individual cubicles and presented with
randomly numbered samples (10 g) of both dinner and lunch
entrees with one of the three energy densities. With use of a 100-
mm VAS, subjects were asked to complete the following ques-
tions: “How pleasant is the taste of this food right now?” and
“How many calories do you think this food has?” Results indi-
cated that there were no significant differences in either pleas-
antness of taste or perceived energy content across conditions for
any entree (P > 0.05). Individuals who participated in the taste
test were excluded from participation in the present study.
All foods were weighed (± 0.1 g) before service and then
reweighed after the subjects had eaten to obtain the amount con-
sumed of each food. Energy and macronutrient intakes were cal-
culated by using nutrition information from the proximate analy-
sis described above for the entrees and manufacturers’ nutrition
information for the side dishes.
Data analysis
All data were analyzed by using SAS-PC for Windows (ver-
sion 6.10; SAS Institute, Cary, NC). Results were considered
significant at P < 0.05. Tukey’s honestly significant difference
test was used for post hoc comparisons of significant effects.
Food and energy intakes
Food (g) and energy (kJ) intakes were analyzed by analysis of
variance (ANOVA) using the General Linear Model procedure
with adjusted error terms for the within-subject, repeated- mea-
sures design. Condition (high, medium, or low energy density)
was entered as the within-subjects factor. Where appropriate, the
main effect of test day (day 1 or 2) and the interaction between
test day and condition were also tested in the model. Session (1,
2, or 3) was tested as a covariate on gram intake.
Energy density and macronutrient composition (percentage of
energy) of diets and fiber intake
Energy density and macronutrient composition of the diets
and fiber intake were calculated on the basis of subjects’ food
intake (in g) of main entrees and side dishes.
Analysis of visual analog ratings
Ratings of the manipulated foods (ie, pleasantness of taste)
and ratings of subjective sensations (ie, hunger) were analyzed
by using ANOVA with the energy density as the within-subjects
factor. Where appropriate, the main effect of day (1 or 2) and the
interaction between day and condition were added to the model.
Power analysis
A power analysis (version 2.0; STAT-POWER, Scientific Soft-
ware, Portland, OR) using estimated variances and effect sizes
derived from previous experiments with lean women in our labo-
ratory determined that a sample size of 18 was needed to detect
BELL ET AL 415
TABLE 2
Composition of manipulated entrees per 100 g
1
Food Fat Carbohydrate Protein Fiber
2
Energy
3
Energy density Moisture
g (% of energy) g (% of energy) g (% of energy) g kJ (kcal) kJ/g (kcal/g) %
Pasta salad with Italian dressing
4
Low energy density 1.7 (l9) 12.4 (61) 4.2 (20) 1.5 343 (82) 3.4 (0.8) 79.7
Medium energy density 1.7 (14) 17.8 (65) 5.6 (20) 1.4 456 (109) 4.6 (1.1) 72.9
High energy density 2.5 (16) 22.7 (64) 7.1 (20) 1.3 590 (141) 5.9 (1.4) 65.7
Pasta salad with yogurt dressing
4
Low energy density 2.2 (27) 10.6 (58) 2.9 (16) 1.5 306 (73) 3.1 (0.7) 82.6
Medium energy density 2.7 (26) 14.0 (59) 3.8 (16) 1.2 400 (96) 4 0 (1.0) 77.7
High energy density 2.8 (22) 17.8 (62) 4.6 (16) 1.1 482 (115) 4.8 (1.2) 73.0
Italian pasta bake
Low energy density 2.3 (22) 13.8 (58) 4.8 (20) 1.5 399 (95) 4.0 (1.0) 76.6
Medium energy density 2.4 (19) 18.3 (62) 5.7 (19) 1.5 491 (117) 4.9 (1.2) 71.1
High energy density 3.8 (25) 20.2 (57) 6.5 (18) 1.3 592 (141) 5.9 (1.4) 66.5
Chicken noodle casserole
Low energy density 2.5 (28) 10.8 (54) 3.7 (19) 1.4 338 (81) 3.4 (0.8) 81.4
Medium energy density 2.6 (24) 13.7 (56) 5.1 (21) 1.5 410 (98) 4.1 (1.0) 77.1
High energy density 3.5 (24) 19.6 (58) 6.1 (18) 1.5 564 (135) 5.6 (1.3) 68.7
1
Serving dishes were filled to similar volumes.
2
Calculated based on information from manufacturers and from Bowes and Church’s Food Values of Portions Commonly Consumed (23).
3
Atwater constants [38 kJ/g (9 kcal/g) for fat; 17 kJ/g (4 kcal/g) for carbohydrate; and 17 kJ/g (4 kcal/g) for protein] were used to determine energy con-
tent.
4
Lunch entrees were provided as the entree in the take-home evening snacks. Evening snack servings were <600 g.
by guest on July 12, 2011www.ajcn.orgDownloaded from
differences between conditions (a = 0.05; 12b = 0.80). Post hoc
analyses were conducted with data from the current study to deter-
mine whether we had adequate power to detect hypothetical dif-
ferences in the amount of food consumed if individuals had varied
their intake to adjust for changes in energy density.
RESULTS
Food and macronutrient intakes
Weight of food consumed
There was no effect of energy density on the amount of food
consumed. Cumulative intake (by meals) did not differ across
conditions (Figure 1). Mean (±SEM) daily intakes were 1350.1
± 39.6, 1302.2 ± 48.4, and 1334.1 ± 46.3 g/d for the low,
medium, and high-energy density conditions, respectively. Sub-
jects consumed a greater amount of food on day 1 (1368.6 ± 35.1
g) than on day 2 (1289.0 ± 37.3 g) (F
[1,17]
= 4.46, P < 0.05). This
effect was the same across conditions. Session (1, 2, or 3) was
found to be a significant covariate on average daily gram intake
(F
[1,33]
= 6.27, P < 0.02, b = 263.50), with subjects consuming
less food over time (mean daily intake ± SEM: session 1, 1384.7
± 48.4 g; session 2, 1344.1 ± 54.0 g; and session 3, 1257.7 ± 65.2
g). This did not alter the effect of energy density on the amount
of food consumed. The amount (in g) of water consumed in the
laboratory did not vary by day or across conditions.
Energy consumed
Subjects consumed significantly more energy in the high-energy
density condition than in the medium and low conditions. This effect
was seen after consumption of the first manipulated meal (day 1
lunch) (F
[2,34]
= 10.92, P < 0.0002) and continued throughout the
study (Figure 2). There was no effect of test day (ie, day 1 or 2) on
energy intake and no interaction between test day and condition.
Because there was no effect of test day, this factor was dropped from
the model and further analyses were conducted by using mean daily
energy intakes. Subjects consumed significantly more energy daily
in the condition of high energy density (7532 ± 363.2 kJ, or 1800 ±
86.8 kcal) than in the medium (6356 ±280.7 kJ, or 1519 ± 67.1 kcal)
and low (5756 ± 178.2 kJ, or 1376 ± 42.6 kcal) conditions (F
[2,34]
=
20.08, P < 0.0001). The difference in total energy intake (entrees and
side dishes) across conditions was due to a difference in energy
intake from the main entrees (F
[2,34]
= 20.57, P < 0.0001).
Energy density and macronutrient composition of diets and
fiber intake
Energy density and macronutrient composition of the diets
(main entrees and side dishes) and fiber intake were calculated on
the basis of intakes of manipulated entrees and compulsory foods.
Mean energy density values were 5.61 kJ/g (1.34 kcal/g), 4.90 kJ/g
(1.17 kcal/g), and 4.27 kJ/g (1.02 kcal/g) for the high, medium, and
low conditions, respectively. The percentages of energy from car-
bohydrate, fat, and protein, respectively, for the conditions were
65.7 ± 0.4%, 16.0 ± 0.2%, and 20.4 ± 0.3% for low energy density;
66.1 ± 0.5%, 14.9 ± 0.3%, and 21.2 ± 0.3% for medium energy
density; and 65.4 ± 0.5%, 15.8 ± 0.3%, and 20.8 ± 0.3% for high
energy density. Average daily fiber intakes were 17.4 ± 0.6, 16.1 ±
0.7, and 15.08 ± 0.6 g for the low-, medium-, and high-energy den-
sity conditions, respectively. The small differences in macronutri-
ent composition and fiber intake across conditions were, arguably,
too small to affect human food intake.
VAS ratings
Hedonic ratings
There were no differences in ratings of pleasantness of taste or
texture, prospective consumption, or perceived energy content
for the three versions of each of the manipulated entrees before
meals (Table 3). Likewise, there were no differences in these rat-
ings after meals. These data confirm that entrees were equally
well-liked and did not vary systematically in sensory properties
with changes in energy density.
Hunger ratings
There were no main effects of energy density on ratings of
hunger, fullness, thirst, prospective consumption, or nausea
416 ENERGY DENSITY AFFECTS ENERGY INTAKE
FIGURE 1. Cumulative food consumption by condition. Energy densities of the diets (main entrees and side dishes): low, 4.3 kJ/g (1.02 kcal/g);
medium, 4.9 kJ/g (1.17 kcal/g); and high, 5.61 kJ/g (1.34 kcal/g). B, breakfast; L, lunch; D, dinner; S, evening snack. x
–
± SEM.
by guest on July 12, 2011www.ajcn.orgDownloaded from
before or after breakfast, lunch, and dinner on either day of the
study. Before breakfast, subjects reported a greater degree of
fullness on day 2 than on day 1 of the study (F
[1,17]
= 7.79, P <
0.01). After breakfast, subjects reported greater feelings of thirst
on day 2 (F
[1,17]
= 6.12, P < 0.02) than on day 1. Ratings of
prospective consumption before lunch were greater on day 1
than on day 2 (F
[1,17]
= 15.11, P < 0.001). Results from the ques-
tionnaire completed on day 3 showed that, retrospectively, sub-
jects did not experience differences in either hunger between
meals (low: 49.2 ± 5.3 mm; medium: 48.8 ± 5.1 mm; and high
energy density: 54.0 ± 4.4 mm) or in fullness over the 2 d across
conditions (low: 71.3 ± 5.8 mm; medium: 74.3 ± 4.8 mm; and
high energy density: 68.1 ± 6.6 mm).
Discharge questionnaire
No subjects correctly reported the purpose of the study or
indicated that there were any differences in energy content of the
manipulated entrees. Four subjects indicated that they believed
we were studying their perceptions of foods before and after eat-
ing. Three subjects reported that they believed that the study was
related to low-fat diets whereas three other subjects indicated
that they believed that the study’s purpose was to compare rat-
ings of hunger with amounts of food eaten.
Power analysis
Because the average amount of food consumed daily varied by
<50 g between any two conditions, it is not surprising that we were
not able to detect differences in intake (by weight). We did have
power (a = 0.01; 12b = 0.80) to detect hypothetical differences in
the amount of food consumed of 13.6% (181 g), 13.9 % (188 g),and
18.5% (247 g) between the low and high, low and medium, and
medium and high conditions of energy density.
DISCUSSION
The present study is the first to examine the effects of energy
density on food and energy intakes while specifically controlling
for the potential effects of both macronutrient composition and
palatability. Because individuals did not adjust the amount of
food consumed in relation to the energy density of foods, energy
intake was driven by the energy density of the meals provided.
The results of this study clearly show that energy density can
have a significant effect on energy intake independent of either
macronutrient composition or palatability.
We successfully controlled for palatability of the manipulated
entrees. Subjects rated pleasantness of taste, texture, perceived
energy content, and their prospective consumption of three ver-
sions of each main entree similarly. Other results from this study
also indicate that palatability was held constant. Specifically,
subjects consumed a similar amount of each main entree across
conditions and ate to similar states of fullness as shown by the
lack of difference in ratings of fullness after either lunch or din-
ner across conditions.
Formulation of foods and diets varying in energy density, but
not macronutrient composition (as a percentage of energy), is a
challenging task. Results from proximate analysis indicated that
we were successful in manipulating energy density of the main
entrees. As planned, energy density of the main entrees differed
by 20% between conditions. This difference was reduced to
<15% between diets with the inclusion of compulsory foods and
by individual variations in the amount consumed of manipulated
entrees. Thus, the diet of low energy density contained <30%
less energy per gram than the diet of high energy density.
Although the energy density of the diets varied with the inclu-
sion of compulsory food and differences in intake of manipu-
lated entrees, the macronutrient content of the diets was held
constant. It is unlikely that the slight variations in macronutrient
content were large enough to affect human intake.
One way in which the energy density of foods can be reduced,
without changing the macronutrient composition, is by adding
foods with a high water content. This is an effective manipula-
tion because water adds weight but not energy to foods. In the
present experiment, variations in the energy density of the main
entrees were accomplished primarily by altering the proportion
BELL ET AL 417
FIGURE 2. Cumulative energy intake by condition. Means with different letters are significantly different, including the meal indicated, at each
time point (P < 0.05). Energy densities of the diets (main entrees and side dishes): low, 4.3 kJ/g (1.02 kcal/g); medium, 4.9 kJ/g (1.17 kcal/g); and high,
5.6 kJ/g (1.34 kcal/g). B, breakfast; L, lunch; D, dinner; S, evening snack. x
–
± SEM.
by guest on July 12, 2011www.ajcn.orgDownloaded from
of low-fiber vegetables to pasta. Entrees of low energy density
contained the greatest amount of vegetables, whereas entrees of
high energy density contained the greatest amount of pasta. This
manipulation was successful because both of these foods derive
most of their energy from carbohydrate while differing primarily
in water content. Entrees with low energy density had a higher
percentage of moisture than their counterparts with greater
energy densities. The small differences in fiber content of the
entrees and likewise in fiber intake across conditions reflect the
dietary manipulation (ie, proportion of vegetables and pasta) and
are most likely not large enough to affect food intake (25).
In the current study, daily intakes of food varied by no more
than 50 g between any two conditions. Because the amount of
food consumed remained constant, energy intake was driven by
the energy density of the diets. These results are consistent with
findings from previous studies reporting changes in energy
intake, but not in the amount of food consumed when both energy
density and fat content of diets varied. For example, in one of the
earliest studies on energy density, Duncan et al (14) provided sub-
jects with a low-fat diet or a high-fat diet of twice the energy den-
sity (2.9 kJ/g or 0.7 kcal/g compared with 6.3 kJ/g or 1.5 kcal/g)
for 5 d. Individuals consumed nearly twice as much energy with
the diet of high energy density. Although not reported by the
authors, calculations based on the energy consumed and the
energy density of the diets show that subjects ate approximately
the same amount of food during the two diets. In another study,
in which men were fed each of three diets varying in energy den-
sity (4.8, 5.6, and 7.0 kJ/g) and fat content (20%, 40%, and 60%
of energy) over 14 d, subjects consumed a constant weight of
food across dietary treatments with energy intake greatest with
the high-fat, high-energy-density diet and lowest on the low-fat,
low-energy-density diet (18). A recent study by Rolls and Miller
(26) provides further evidence that individuals tend to consume
similar amounts of food, regardless of changes in fat content and
energy density, even during discrete eating periods such as snack-
ing. In this study, men and women ate similar amounts of no-fat
and regular potato chips and energy intake varied accordingly.
Because palatability and macronutrient content were controlled
for, our present study confirms that individuals fail to adjust their
intakes in response to changes in energy density. Thus, energy
density, rather than fat content, might have accounted for differ-
ences in energy intake that occurred in previous studies when
diets varied in both fat content and energy density.
To consume equal amounts of energy, subjects would had to
have eaten <30% (400 g) more of the diet of low energy density
and 12% (160 g) more food with the diet of medium energy den-
sity than with the diet of high energy density. Calculations based
on results from the present study revealed that we had power to
detect differences in the amount of food consumed of 13.6%
(181 g) and 18.5% (247 g) between high and low and high and
medium conditions of energy density, respectively. Thus, we had
adequate power to detect a difference in the amount of food con-
sumed if individuals had adjusted the amount of food consumed
in response to changes in energy density.
Average daily energy intakes with the high, medium, and low
conditions of energy density were <7500 kJ (1800 kcal), 6400
kJ (1520 kcal), and 5800 kJ (1375 kcal), respectively. All of
these are below the recommended dietary allowance (27) for
energy of 9204.8 kJ (2200 kcal) for women 19–50 y of age.
Energy needs were also calculated for each subject by using the
Harris-Benedict equation (28) with an activity factor of 1.4. The
average energy requirement from this computation was 8318 kJ
418 ENERGY DENSITY AFFECTS ENERGY INTAKE
TABLE 3
Lunch and dinner entree visual-analogue-scale ratings
1
Meals and entrees Low energy density Medium energy density High energy density
Lunch
Pasta salad with Italian dressing
“How pleasant is the taste of this food right now?”
2
69 ± 4.3 69 ± 4.3 72 ± 3.8
“How pleasant is the texture of this food right now?”
2
58 ± 4.9 69 ± 4.6 66 ± 4.9
“How much of this food do you think you could eat right now?”
3
61 ± 4.4 53 ± 4.8 56 ± 5.4
“How many calories do you think this food has?”
4
40 ± 4.2 38 ± 4.1 42 ± 4.4
Pasta salad with yogurt dressing
“How pleasant is the taste of this food right now?” 66 ± 4.2 68 ± 4.9 64 ± 5.3
“How pleasant is the texture of this food right now?” 70 ± 4.3 72 ± 4.6 66 ± 5.2
“How much of this food do you think you could eat right now?” 60 ± 4.6 49 ± 4.6 60 ± 4.9
“How many calories do you think this food has?” 39 ± 3.5 39 ± 4.3 41 ± 4.2
Dinner
Italian pasta bake
“How pleasant is the taste of this food right now?” 73 ± 3.2 79 ± 3.3 78 ± 3.3
“How pleasant is the texture of this food right now?” 68 ± 3.9 71 ± 4.7 75 ± 3.0
“How much of this food do you think you could eat right now?” 65 ± 2.8 55 ± 4.9 68 ± 3.5
“How many calories do you think this food has?” 57 ± 3.7 58 ± 3.3 61 ± 3.7
Chicken noodle casserole
“How pleasant is the taste of this food right now?” 66 ± 4.5 61 ± 5.8 58 ± 4.9
“How pleasant is the texture of this food right now?” 61 ± 4.5 59 ± 5.1 60 ± 4.2
“How much of this food do you think you could eat right now?” 57 ± 4.6 52 ± 5.2 58 ± 2.5
“How many calories do you think this food has?” 48 ± 3.9 51 ± 4.5 51 ± 3.9
1
x
–
± SEM. Values are before-meal ratings. There were no differences in responses to questions across conditions.
2
Anchors were “not at all pleasant” and “extremely pleasant”.
3
Anchors were “nothing at all” and “a large amount”.
4
Anchors were “no calories at all” and “extremely high in calories”.
by guest on July 12, 2011www.ajcn.orgDownloaded from
(1988 kcal). Therefore, in all conditions, subjects failed to con-
sume enough food to maintain appropriate energy intake. This
may be partially explained by the low energy density of each of
the diets, which in turn is a function of their low fat content.
With low-fat, low-energy-density diets, as compared with diets
of higher energy density, subjects need to consume greater
amounts of food to reach similar energy intakes. For example, in
this study, subjects would have needed to eat <1500, 1700, and
1960 g of food for the diets of high, medium, and low energy
density, respectively, to maintain an energy intake of 8370 kJ
(2000 kcal). In comparison, subjects consumed <1330 g across
the three conditions.
Although we cannot be certain of the effect of the laboratory
environment or menus on food intake in this study, our results
suggest that it may be difficult to maintain adequate food intakes
with diets of low energy density. Specifically, gastric distention
and other satiety cues may have limited the amount of food able
to be consumed. Likewise, knowledge of eating an amount of
food that constitutes a culturally acceptable meal (29), or knowl-
edge of portion sizes appropriate for the satisfaction of hunger
(30), may have contributed to the termination of eating before
consumption of enough food to meet energy needs. It is also pos-
sible that food intake might have been limited by sensory-spe-
cific satiety (the decline in the pleasantness of a food as it is con-
sumed, whereas other foods remain pleasant). Usually,
sensory-specific satiety leads to the termination of intake of a
particular food while promoting the selection and consumption
of other foods (31). However, in this study, the intake of other
foods was limited because side dishes were controlled and the
manipulated main entrees were the only foods offered to be con-
sumed ad libitum. Therefore, the large amount of food that must
be consumed with diets of low energy density may prevent or
hinder adequate energy intakes, thus promoting a reduction in
energy intake.
This study provides preliminary evidence that diets of low
energy density may be helpful tools for weight loss and control.
Across the conditions of energy density, subjects reported no dif-
ferences in subjective sensations of hunger or fullness before and
after meals and no differences in feelings of hunger between
meals or fullness over the 2 study days. This is surprising
because <30% less energy was consumed from the diet of low
energy density than from the diet of high energy density. Fur-
thermore, as mentioned above, in no diet condition did subjects
consume enough food to meet their energy requirements. Thus,
diets of low energy density may reduce energy intakes while
allowing individuals to consume adequate amounts of food.
Also, with diets of low energy density, individuals may not expe-
rience adverse feelings, frequently associated with dieting, such
as hunger or food deprivation.
Several years ago, the message that fat restriction was the key
to weight loss was promoted widely (15) and some researchers
concluded that restriction of fat intake would be an effective diet
therapy for weight loss (32–34). Nonetheless, when two low-fat
diets were compared directly, the diet emphasizing both fat and
energy reduction was more effective in reducing body weight
than the diet recommending only the reduction of dietary fat
(35). Results from the third National Health and Nutrition Exam-
ination Survey (2) showing that the proportion of fat in the
American diet has decreased while both total energy intake and
body weight have increased, imply that reducing fat intake alone
may not be effective for weight control. An explanation for these
apparently contradictory findings, as shown by the present study,
could be that the overconsumption of energy is related to the
energy density and not to just the fat content of foods.
Before the advent of modern food technology, the substitution
of low-fat foods for high-fat foods often would have led to reduc-
tions in not only fat content, but also in energy density. This is
because individuals lowered the fat content of their diets primar-
ily through the incorporation of fruits and vegetables. However,
the food supply is changing and many commercially available
low-fat or fat-free foods are not lower in energy density than
their full-fat counterparts. Consumers must be made aware that
fat restriction alone will not result in a reduction in energy intake
because energy intake is dependent on the energy density, and
not just the fat content, of foods.
Although our study provides valuable insight into the poten-
tial effectiveness of diets of low energy density for reducing
energy intake and weight loss, the results must be interpreted and
applied with caution. This study was conducted under a con-
trolled experimental setting. Subjects were not allowed to con-
sume foods in addition to those provided by the laboratory, were
not given nutrition information about the foods, and were not
aware of the dietary manipulations. The diets, as a function of
their low fat content, were all low in energy density. We cannot
determine how individuals would respond to higher levels of
energy density or if the energy density of more than one food per
meal was varied, so that there was a wider range of both foods
and energy density levels. Also, it is not known whether subjects
would have adjusted their food intakes to compensate for the
energy deficits, observed across all conditions, if the same study
were conducted over a greater length of time or if subjects were
given free access to other foods varying in macronutrient com-
position and energy density. Furthermore, this study was con-
ducted in normal-weight women and the results cannot be
extrapolated to normal-weight men and obese individuals.
In conclusion, this study indicates that energy density influ-
ences energy intake independently of macronutrient (fat) content
and palatability. Individuals were not sensitive to variations in
energy density because they consumed similar amounts of food
across conditions. Energy intake was significantly greater from
the diets of high energy density than from the diets of lower
energy density. Thus, these results imply that the overconsump-
tion of high-fat foods may be due to their high energy density
rather than to their fat content, per se. Although energy intake
was greatly reduced in the diet of low energy density as com-
pared with the diet of high energy density, individuals reported
no differences in feelings of hunger or fullness. This study pro-
vides important preliminary information suggesting that diets
low in energy density may be helpful for weight loss.
REFERENCES
1. Pi-Sunyer X. Medical hazards of obesity. Ann Intern Med
1993;119:655–60.
2. Life Sciences Research Office. Federation of American Societies for
Experimental Biology. Third report on nutrition monitoring in the
United States. Washington, DC: US Government Printing Office,
1995.
3. Rolls BJ. Carbohydrates, fats, and satiety. Am J Clin Nutr
1995;61(suppl):960S–7S.
4. Blundell JE, Rogers PJ. Hunger, hedonics, and the control of satia-
tion and satiety. In: Friedman MI, Tordoff MG, Kare MR, eds.
BELL ET AL 419
by guest on July 12, 2011www.ajcn.orgDownloaded from
Chemical senses. Vol 4. Appetite and nutrition. New York: Marcel
Dekker, Inc, 1991:127–48.
5. Blundell JE, Burley VJ, Cotton JR, Lawton CL. Dietary fat and the
control of energy intake: evaluating the effects of fat on meal size
and postmeal satiety. Am J Clin Nutr 1993;57(suppl):772S–8S.
6. Schutz Y, Flatt JP, Jéquier E. Failure of dietary fat intake to promote
fat oxidation: a factor favoring the development of obesity. Am J
Clin Nutr 1989;50:307–14.
7. Poppitt SD, Prentice AM. Energy density and its role in the control
of food intake: evidence from metabolic and community studies.
Appetite 1996;26:153–74.
8. Castellanos VH, Rolls BJ. Diet composition and the regulation of
food intake and body weight. In: Dalton S, ed. Overweight and
weight management. Gaithersburg, MD: Aspen Publishers, Inc,
1997:254–75.
9. Drewnowski A. Fats and food acceptance: sensory, hedonic and atti-
tudinal aspects. In: Solms J, Booth DA, Pangborn RM, et al, eds.
Food acceptance and nutrition. New York: Academic Press,
1988:189–204.
10. Rolls BJ, Shide DJ. The influence of dietary fat on food intake and
body weight. Nutr Rev 1992;50:283–90.
11. van Stratum P, Lussenburg RN, van Wezel LA, Vergroesen AJ, Cre-
mer HD. The effect of dietary carbohydrate:fat ratio on energy
intake by adult women. Am J Clin Nutr 1978;31:206–12.
12. Stubbs RJ, Harbron CG, Prentice AM. Covert manipulation of the
dietary fat to carbohydrate ratio of isoenergetically dense diets:
effect on food intake in feeding men ad libitum. Int J Obes
1996;20:651–60.
13. Poppitt SD. Energy density of diets and obesity. Int J Obes
1995;19(suppl):S20–6.
14. Duncan KH, Bacon JA, Weinsier RL. The effects of high and low
energy density diets on satiety, energy intake, and eating time of
obese and nonobese subjects. Am J Clin Nutr 1983;37:763–7.
15. Kendall A, Levitsky DA, Strupp BJ, Lissner L. Weight loss on a
low-fat diet: consequence of the imprecision of the control of food
intake in humans. Am J Clin Nutr 1991;53:1124–9.
16. Lissner L, Levitsky DA, Strupp BJ, Kalkwarf HJ, Roe DA. Dietary
fat and the regulation of energy intake in human subjects. Am J Clin
Nutr 1987;46:886–92.
17. Stubbs RJ, Harbron CG, Murgatroyd PR, Prentice AM. Covert
manipulation of dietary fat and energy density: effect on substrate
flux and food intake in men eating ad libitum. Am J Clin Nutr
1995;62:316–29.
18. Stubbs RJ, Ritz P, Coward WA, Prentice AM. Covert manipulation
of the ratio of dietary fat to carbohydrate and energy density: effect
on food intake and energy balance in free-living men eating ad libi-
tum. Am J Clin Nutr 1995;62:330–7.
19. Tremblay A, Lavallée N, Alméras N, Allard L, Després J-P,
Bouchard C. Nutritional determinants of the increase in energy
intake associated with a high-fat diet. Am J Clin Nutr 1991;
53:1134–7.
20. Stunkard AJ, Messick S. The three-factor eating questionnaire to
measure dietary restraint, disinhibition, and hunger. J Psychosom
Res 1985;29:71–83.
21. Garner DM, Garfinkel PE. The Eating Attitudes Test: an index of the
symptoms of anorexia nervosa. Psychol Med 1979;9:273–80.
22. Beck AT, Beamesdorfer A. Assessment of depression: the Depres-
sion Inventory. In: Pinchot P, ed. Psychological measurements in
psychopharmacology, modern problems in pharmacopsychiatry. Vol
7. Basel, Switzerland: Karger, 1974:151–69.
23. Pennington JT. Bowes and Church’s food values of portions com-
monly used. 16th ed. Philadelphia: JB Lippincott Company, 1994.
24. Cunniff P, ed, Official methods of analysis of AOAC International.
Arlington, VA: AOAC International, 1995.
25. Blundell JE, Burley VJ. Satiation, satiety and the action of fibre on
food intake. Int J Obes 1987;11:9–25.
26. Rolls BJ, Miller DL. Is the low-fat message giving people a license
to eat more? J Am Coll Nutr 1997;16:535–43.
27. National Research Council. Recommended dietary allowances. 10th
ed. Washington, DC: National Academy Press, 1989.
28. Harris JA, Benedict FG. Standard basal metabolism constants for
physiologists and clinicians: a biometric study or basal metabolism
in man. Philadelphia: Lippincott, 1919.
29. Rozin P. Sociocultural influences on human food selection. In:
Capaldi ED, ed. Why we eat what we eat. Washington, DC: Ameri-
can Psychological Association, 1996:233–63.
30. Rolls BJ, Hammer VA. Fat, carbohydrate, and the regulation of
energy intake. Am J Clin Nutr 1995;62(suppl):1086S–95S.
31. Rolls BJ, Hetherington M, Burley VJ. The specificity of satiety: the
influence of foods of different macronutrient content on develop-
ment of satiety. Physiol Behav 1988;43:145–53.
32. Shah M, McGovern P, French S, Baxter J. Comparison of a low-fat,
ad libitum complex-carbohydrate diet with a low-energy diet in
moderately obese women. Am J Clin Nutr 1994;59:980–4.
33. Raben A, Jensen ND, Marckmann P, Sandstrom B, Astrup A. Spon-
taneous weight loss during 11 weeks ad libitum intake of a low
fat/high fiber diet in young, normal weight subjects. Int J Obes
1995;19:916–23.
34. Siggard R, Raben A, Astrup A. Weight-loss during 12 weeks’ ad
libitum carbohydrate-rich diet in overweight and normal-weight
subjects at a Danish work site. Obes Res 1996;4:347–56.
35. Schlundt DG, Hill JO, Pope-Cordle J, Arnold D, Virts KL, Katahn
M. Randomized evaluation of a low fat ad libitum carbohydrate diet
for weight reduction. Int J Obes 1993;17:623–9.
420 ENERGY DENSITY AFFECTS ENERGY INTAKE
by guest on July 12, 2011www.ajcn.orgDownloaded from
