Original Article Obesity

OBESITY BIOLOGY AND INTEGRATED PHYSIOLOGY

Independent, Additive Effects of Five Dietary Variables

on Ad Libitum Energy Intake in a Residential Study
Lorien E. Urban1, Megan A. McCrory2, Helen Rasmussen1, Andrew S. Greenberg1, Paul J. Fuss1, Edward Saltzman1 and
Susan B. Roberts1

Objective: To examine the relationship between dietary characteristics of self-selected foods and energy
balance in a cafeteria-style dining hall.
Methods: Ad libitum dietary intake from a self-selection menu was measured over two days in 151 adults
(70% female, mean age 41 years, mean BMI 24.9 kg/m2). The associations of dietary variables with
energy balance (calculated as measured energy intake/predicted energy requirements, pER) were
assessed.
Results: Measured energy intake was significantly correlated with pER (R250.83, P < 0.001). In mixed
multiple regression models, percent energy from protein was negatively associated with energy balance
(R250.04, P 5 0.02), and percent energy from liquid sources (R2 5 0.02, P 5 0.05), total dietary variety in
females (R2 5 0.39, P < 0.001), and energy density (R2 5 0.57, P < 0.001) were positively associated with
energy balance. In addition, glycemic index was inversely associated with energy balance in normal-
weight individuals (R2 5 0.14, P < 0.001) but not in overweight or obese individuals.
Conclusions: There are independent associations of dietary protein, liquid calories, energy density, die-
tary variety, and glycemic index with energy balance, indicating additive effects of these dietary factors
on energy intake and energy balance. Intervention studies are needed to determine whether dietary pre-
scriptions combining these dietary factors facilitate long-term prevention of weight gain.
Obesity (2014) 00, 00-00. doi:10.1002/oby.20798

Introduction (3). In theory, a better approach to maintaining EBal would be accu-

Recommendations to balance energy intake (EI) with energy
requirements are a cornerstone of national dietary guidelines for pre-
venting weight gain and obesity (1). Although self-monitoring EI is
widely recommended for achieving the neutral energy balance
(EBal) that prevents an increase in body fat (1,2), in practice self-
monitoring is difficult to sustain because it requires burdensome
measuring and recording, and is hampered by the lack of compre-
hensive, accurate data on the dietary composition of consumed foods
rate dietary self-regulation, i.e. eating in ways that allow for an
automatic balancing of EI with energy requirements through the
body’s innate regulatory systems (4). However, the extent to which
self-regulation of EI is feasible in the modern food environment is
very uncertain. Specifically, there is much controversy over the
extent to which dietary macronutrient composition and other dietary
variables can influence dietary self-regulation for the prevention of
weight gain, a topic that is addressed in this study.

1
Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, Boston, Massachusetts, USA. Correspondence: Susan B. Roberts
(susan.roberts@tufts.edu) 2 Department of Foods and Nutrition, Department of Psychological Sciences and the Ingestive Behavior Research Center (IBRC),
Purdue University, West Lafayette, Indiana, USA.

Funding agencies: Funding for this project was provided by DK046124, DK73321 and by the US Department of Agriculture through agreement No. 58-1950-0-014 with
Tufts University.
Disclosures: (i) LEU, MAM, HR, ES, ASG, PJF, SBR - no disclosures (iii) the authors’ spouses, partners, or children have no financial relationships that may be relevant to
the submitted work; and (iv) authors have no nonfinancial interests that may be relevant to the submitted work.
Author contributions: Study concept and design: LEU, MAM, HR, SBR. Acquisition of data: HR, ES, ASG, PJF. Analysis and interpretation of data: LEU, MAM, HR,
SBR. Drafting of the manuscript: LEU, SBR. Critical revision of the manuscript for important intellectual content: LEU, MAM, HR, ES, ASG, PJF, SBR. Statistical analysis:
LEU. Administrative, technical, or material support: LEU, MAM, HR, SBR. Study supervision: LEU, MAM, HR, ES, ASG, PJF, SBR.
Role of the Sponsor: The sponsor did not have any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and
preparation, review, or approval of the manuscript.
Disclaimer: Any opinions, finding, conclusions, or recommendations expressed in this article are those of the authors and do not necessarily reflect the views of the US
Department of Agriculture.
Additional Supporting Information may be found in the online version of this article.
Received: 11 October 2013; Accepted: 10 May 2014; Published online 00 Month 2014. doi:10.1002/oby.20798

www.obesityjournal.org Obesity | VOLUME 00 | NUMBER 00 | MONTH 2014 1

Obesity
Current dietary recommendations for individuals emphasize dietary
patterns for general health (1). They are consistent with the flexible
Acceptable Dietary Macronutrient Ranges (5) and do not specifically
recommend on the newer, more controversial dietary factors such as
glycemic index (GI). The flexible macronutrient ranges (10-35% for
protein, 20-35% for fat, and 45-65% for carbohydrate) contrast with
the narrower nutrient patterns identified as beneficial for weight con-
trol in several short-term and clinical studies. For example, results
from most single-meal and single-day studies that used precisely con-
trolled meals of specific energy and nutrient content, typically indicate
that low energy density (6), low GI (7), low dietary variety (8), low
percentage of energy from liquid calorie sources (9), high protein (10),
and high fiber (11) meals all reduce EI acutely via beneficial changes
in hunger or satiety. However, some of these findings have not been
incorporated into national dietary recommendations for several reasons.
One reason may be that long-term studies looking at the effectiveness
of different dietary recommendations have typically failed to show sig-
nificant effects of dietary composition on weight change, and in many
cases one potential explanation was that the dietary recommendations
were not adopted when individuals self-selected their own food (12-
15). Another important reason for why the consistent short-term data
may have not been translated into dietary recommendations is that
there is almost no information on whether factors such as protein, GI,
and dietary variety have additive effects on EBal if combined in the
same dietary prescription, or whether they offer alternate approaches
to influencing EBal that are not additive.
Some limited data does suggest the potential for different dietary fac-
tors to have additive effects on EI and EBal. For example, the Dioge-
nes study found that recommendations to consume diets that were
both high in protein and low in GI prevented weight regain following
weight loss more effectively than recommendations to change one or
other of the parameters, and also demonstrated that, at least for pro-
tein, participants did successfully adopt the recommended dietary pat-
terns (16). However, that study concerned protein and GI specifically,
and targeted only prevention of weight regain. There is very limited
information on whether dietary factors considered more broadly can
have additive effects on EI during ad libitum eating.
We describe here an analysis examining potential additive influences
of different dietary composition parameters on EI and EBal in adult
men and women. The protocol was a 2-day inpatient study in a meta-
bolic ward that examined the effects of different dietary choices from
a self-selection menu on ad libitum EI and EBal in a cafeteria setting.
Methods
This study is an analysis of data from the Tufts Twin Study
(17,18), a cross-sectional investigation of genetic and environmen-
tal factors associated with energy regulation in monozygotic male
and female twins raised apart or raised together and their biologi-
cal parents. Twin pairs, and available parents of twins raised
together, visited the Jean Mayer U.S. Department of Agriculture
Human Nutrition Research Center on Aging at Tufts University to
participate in the study. Subjects lived at the research center for 2-
4 days. This analysis focuses on the first 2 days of their stay, dur-
ing which time subjects ate all meals at the research center and
completed several examinations and questionnaires for assessment
of demographics, nutrient intakes, energy expenditure, body com-
2 Obesity | VOLUME 00 | NUMBER 00 | MONTH 2014
Independent, Additive Effects of Five Dietary Variables Urban et al.
position, physical activity, eating behavior, and collection of blood
samples.
Subjects
The study population was 253 adult men and women, and a subset
of 151 subjects were included in this analysis (105 twins and 46
parents from 67 families) who were <65 years of age at the time of
the study. Although some parents of enrolled twins were >65 years,
age was restricted to <65 years for the present analysis because
older adults have a diminished ability to regulate EBal (19) which
would have been a confounding factor in the present analysis. Sub-
jects were not enrolled in the study if they had any medical condi-
tion known to influence dietary intake including diabetes, active
cancer, heart disease, cachexia, eating disorders, AIDS, and preg-
nancy. In addition, individuals who reported treatment for a psychi-
atric disorder, or who had gained or lost >10 pounds of weight
within the previous 12 months or >5 pounds within the previous 6
months were ineligible. Monozygotic twins raised apart were
recruited through the Minnesota Study of Twins Reared Apart (20)
and lived in North America, Europe, South Africa, or Australia.
These subjects spent a week in Minnesota prior to travel to Boston,
which allowed for adjustment to time zone changes. Families of
twins raised together and their biological parents were recruited
through advertisement in the New England area. The reason for
including parents of twins reared together was to improve the statis-
tical power of heritability analyses (17,18). Subjects were reim-
bursed for travel expenses and received a stipend for study participa-
tion. All subjects provided informed consent and the study was
approved by the Institutional Review Board at Tufts University.
Measured dietary intake
Subjects were offered food ad libitum throughout their stay in the
research center. There were three meal times and one snack daily,
and subjects were told they could have more or fewer meals and
snacks, and larger or smaller portions as they wished. They self-
selected meals and snacks from a menu of 97 unique foods (180
total food choices, since some side dishes and desserts were dupli-
cated across eating occasions) designed to represent typical items.
Selection of meals occurred independently; however, meals may
have been consumed in small groups with other research volunteers.
The menu was the same on each study day and typical single-unit
portions sizes are shown in Table 1. The composition of individual
menu items by meal and food type is described in the Supporting
Information online. Subjects were allowed to write in items not on
the menu, and those items were provided if available in the research
kitchen. These requests were rare and were usually for small items
such as condiments. Each food was weighed to 0.1 g prior to serv-
ing, and leftovers were returned to the kitchen where they were
reweighed, so that exact consumption of each item could be calcu-
lated. The metabolizable energy, macronutrients, total fiber, and GI
of consumed foods were calculated using the Nutrient Data System
for Research (NDSR) (Nutrition Coordinating Center, University of
Minnesota, Minneapolis, MN). Dietary variety was determined as
the count of unique foods consumed over two days, and energy den-
sity was calculated as kcal per total gram of all foods and liquids
including noncaloric beverages that were provided by the kitchen.
The percentage of energy from nondairy liquid calories was also cal-
culated [dairy was excluded from this calculation on the grounds
that milk solidifies in the stomach (21)].
www.obesityjournal.org
 TABLE 1 Means and ranges for nutrient composition of single portions of menu items
Original Article

Glycemic

www.obesityjournal.org
% Energy % Energy Energy index
Food types by Energy from % Energy from Fiber density (glucose
meal (n) Portion (g) (kcal) carbohydratea from fata proteina (g/1000 kcal) (kcal/g) reference)

Breakfast
Entrees (14) 90 (18-227) 173 (70-353) 66 (3-92) 20 (1-62) 14 (3-44) 10 (0-55) 2.56 (0.50-4.18) 60 (0-82)
Sides (3) 100 (100-100) 56 (33-89) 92 (91-92) 3 (2-3) 6 (4-7) 35 (27-50) 0.56 (0.33-0.89) 42 (25-62)
Nondairy beverages (7) 189 (124-237) 26 (0-68) 70 (0-100) 5 (0-29) 11 (0-71) 1 (0-7) 0.21 (0.00-0.54) 22 (0-68)
Dairy beverages (3) 244 (242-246) 111 (82-148) 44 (30-56) 24 (2-49) 33 (21-42) 0 (0-0) 0.45 (0.34-0.60) 33 (32-34)
OBESITY BIOLOGY AND INTEGRATED PHYSIOLOGY

Condiments (7) 11 (1-30) 36 (4-105) 45 (0-100) 52 (0-99) 3 (0-10) 1 (0-4) 4.20 (1.30-7.19) 41 (0-104)
Lunch
Entrees (7) 195 (105-441) 319 (176-391) 42 (5-76) 26 (8-45) 31 (12-66) 8 (1-23) 2.00 (0.81-3.25) 65 (57-72)
Vegetable sides (9) 93 (73-110) 34 (12-78) 77 (68-89) 7 (3-15) 17 (4-27) 81 (30-142) 0.37 (0.17-0.81) 51 (41-75)
Nonvegetable sides (10) 92 (12-206) 118 (52-203) 67 (40-92) 25 (1-57) 7 (3-14) 12 (3-27) 2.30 (0.38-5.46) 74 (61-88)
Desserts (11) 87 (34-227) 167 (44-578) 63 (30-96) 28 (1-62) 9 (2-44) 15 (0-50) 2.39 (0.44-4.73) 48 (32-64)
Nondairy beverages (9) 239 (124-474) 26 (0-97) 55 (0-100) 4 (0-29) 8 (0-71) 0 (0-2) 0.16 (0.00-0.54) 19 (0-68)
Dairy beverages (3) 244 (242-246) 111 (82-148) 44 (30-56) 24 (2-49) 33 (21-42) 0 (0-0) 0.45 (0.34-0.60) 33 (32-34)
Condiments (15) 15 (1-30) 42 (3-115) 33 (0-100) 61 (0-99) 6 (0-38) 5 (0-49) 3.35 (0.67-7.19) 51 (0-104)
Snack
Food (12) 79 (29-227) 196 (34-391) 58 (15-96) 30 (1-71) 12 (2-44) 15 (0-50) 3.30 (0.47-5.85) 58 (13-86)
Nondairy beverages (3) 319 (240-474) 33 (0-98) 33 (0-98) 0 (0-0) 0 (0-0) 0 (0-0) 0.13 (0.00-0.40) 23 (0-68)
Dairy beverages (3) 244 (242-246) 111 (82-148) 44 (30-56) 24 (2-49) 33 (21-42) 0 (0-0) 0.45 (0.34-0.60) 33 (32-34)
Dinner
Entrees (7) 189 (70-399) 249 (121-550) 16 (0-60) 31 (8-51) 53 (17-92) 10 (0-40) 1.44 (0.76-2.04) 26 (0-57)
Vegetable sides (9) 93 (73-110) 34 (12-78) 77 (68-89) 7 (3-15) 17 (4-27) 81 (30-142) 0.37 (0.17-0.81) 51 (41-75)
Nonvegetable sides (10) 92 (12-206) 116 (52-203) 61 (33-92) 31 (1-60) 9 (3-21) 9 (3-24) 2.29 (0.32-5.46) 70 (40-88)
Desserts (11) 87 (34-227) 167 (44-578) 63 (30-96) 28 (1-62) 9 (2-44) 15 (0-50) 2.39 (0.44-4.73) 48 (32-64)
Nondairy beverages (9) 239 (124-474) 26 (0-97) 55 (0-100) 4 (0-29) 8 (0-71) 0 (0-2) 0.16 (0.00-0.54) 19 (0-68)
Dairy beverages (3) 244 (242-246) 111 (82-148) 44 (30-56) 24 (2-49) 33 (21-42) 0 (0-0) 0.45 (0.34-0.60) 33 (32-34)
Condiments (15) 15 (1-30) 42 (3-115) 33 (0-100) 61 (0-99) 6 (0-38) 5 (0-49) 3.35 (0.67-7.19) 51 (0-104)

a
Percent energy from the macronutrients may not sum to 100% because of rounding and minimal alcohol content not displayed.

Obesity | VOLUME 00 | NUMBER 00 | MONTH 2014

Obesity

3
Obesity Independent, Additive Effects of Five Dietary Variables Urban et al.

TABLE 2 Means and ranges for predicted energy requirement

Results
(pER) measured energy intake (mEI) and dietary composition There were 151 subjects. The sample was 70% female, the mean
of measured intake age was 41 years (range 18-64), and the mean BMI was 24.9 kg/m2
(range 17.6-43.0). Table 1 provides a summary of the mean dietary
Dietary component Mean (range) composition of different menu choices by food type and eating occa-
sion, and Table 2 provides the pER data and mean values for self-
pER (kcal)a 2325 (1757, 3980)
selected dietary composition. Mean mEI was 2305 kcal (range 1078-
mEI (kcal) 2305 (1078, 4711) 4711) and mean pER was 2325 (range 1757-3980), and mEI was
% Energy from carbohydrate 54 (40, 73) strongly correlated with pER (R2 5 0.83, P < 0.001). On average,
% Energy from fat 30 (12, 45) subjects consumed almost identical amounts of EI relative to pER,
% Energy from protein 16 (8, 27) but variability between individuals was wide (range 255% to 69%)
% Energy from nondairy beverages 5.31 (0.00, 18.95) (P 5 0.57) as depicted in Figure 1.
Fiber (g/1000 kcal) 10 (5, 22)
Energy density (kcal/g) 0.87 (0.46, 1.40) Table 3 provides a summary of the model resulting from backward
Glycemic Index (glucose reference) 63 (56, 71) selection of covariates of interest. The final model adjusted for age,
Total dietary varietyb 14 (8, 20) gender, and BMI included GI, energy density, percent energy from
protein and from nondairy beverages, and total variety score. There
Percent difference (mEI vs. pER)c 20.86 (255, 69)
were significant interactions for both GI and BMI and dietary vari-
a
ety and gender. No other interactions were significant, making the
pER was calculated from the Institute of Medicine equations (5) as described in
the Methods. effects of all remaining terms in the model independent and addi-
b
Total dietary variety is the count of unique foods consumed daily (mean for 2 tive. Tests of interactions between the demographic variables and
days). dietary factors were also not significant, indicating that the demo-
c
Percent error was calculated as [(mEI2pER)/pER]100%.
graphic variables were not significant predictors of EBal in this
study design. Fiber (g/1000 kcal) was also not predictive of the per-
Calculation of predicted energy requirements cent difference between EI and pER. During backward selection, the
percentages of energy from carbohydrates and fat were removed
for individuals because of their relationship to GI and energy density, respectively.
Values for measured EI (mEI) were compared to individuals’ pre-
Removing these parameters also resulted in a slightly lower Akaike
dicted energy requirements (pER) to give an index of EBal over the
Information Criterion (2314 without vs. 2311 with), indicating a
2-day study. pER values were calculated using the Estimated Energy
model with better fit.
Requirement equations of the Institute of Medicine (5) specific for
body mass index category, which use the individuals’ weight,
As depicted by the partial correlation plots in Figure 2, there was a
height, sex, age, and one of the four defined activity levels (i.e. sed-
negative relationship between EBal and percent energy from protein
entary, low active, active, very active) to calculate total energy
(R2 5 0.04, P 5 0.02) and positive relationships between EBal and
expenditure, which is equal to energy requirements in weight stable
percent energy from nondairy beverages (R2 5 0.02, P 5 0.05), and
individuals. Each individual’s physical activity category was
energy density (R2 5 0.57, P < 0.001). In other words subjects who
assigned as low active as this activity level is typical (22) and par-
selected a lower percent energy from protein, a higher percent
ticipants could not be physically active during their stay at the
research center.

Statistical methods
The outcome of interest was EBal, which was the difference
between mEI and pER expressed as a percent of pER to normalize
the outcome across individuals. The log of the ratio mEI:pER was
used in all statistical analyses to approximate the normal distribu-
tion. The signed rank test was used to determine whether the median
EBal was significantly different from zero. Mixed linear regression
models were fit to explore the effects of dietary variables on the
outcome of interest. Predictors of interest were calculated to be
independent of total EI, specifically percent of energy from carbohy-
drate, fat, protein, and nondairy beverages, and fiber (g/1000 kcal),
energy density (kcal/g), and total dietary variety. Backward selection
was used to identify significant predictors of EBal, and age, sex,
and BMI were included in all models. Interactions of BMI and sex
with each predictor were tested. To control for the interdependence
of the data from this study that included twin pairs and parents, fam-
ily unit was included as a random factor and family member type
(twin, mother, father) was included as a fixed effect. Statistical anal- Figure 1 Predicted energy requirement (pER) versus measured energy intake (mEI)
in males (•) and females (䊊). Predicted energy requirement was calculated as a
yses were performed by using SAS version 9.3 (2010, SAS Institute, percent difference: [(mEI - pER)/ pER]*100%.
Cary, NC) and significance was set at a 5 0.05.

4 Obesity | VOLUME 00 | NUMBER 00 | MONTH 2014 www.obesityjournal.org

Original Article Obesity
OBESITY BIOLOGY AND INTEGRATED PHYSIOLOGY

TABLE 3 Estimated associations of glycemic index, energy density, percent energy from protein and nondairy beverages, and
total variety score on the percent difference between measured energy intake (mEI) and predicted energy requirement
(pER)a

Covariate ß (SEE)b t value P value Model AIC

Intercept 31.48 (3.65) 2.67 <0.01 2313.8

Within-family subtype
Twins (compared to mothers) 1.00 (1.04) 20.04 0.97
Fathers (compared to mothers) 0.99 (1.06) 20.27 0.79
Age 1.00 (1.00) 2.75 <0.01
Gender (males compared to females) 1.37 (1.17) 1.99 0.05
Body mass index (kg/m2) 0.86 (1.05) 22.92 <0.01
Glycemic index (glucose reference) 0.93 (1.02) 23.36 0.001
% Energy from protein 0.99 (1.00) 22.32 0.02
% Energy from nondairy beverages 1.01 (1.00) 2.03 0.05
Energy density (kcal/g) 1.76 (1.07) 8.09 <0.001
Total dietary varietyc 1.05 (1.01) 8.17 <0.001
Total dietary variety x Gender 0.97 (1.01) 22.49 0.02
Glycemic index x Body mass index 1.00 (1.00) 2.69 <0.01

AIC, Akaike Information Criterion.

a
Percent difference between mEI and pER was calculated as [(mEI2pER)/pER]*100%. pER was calculated from the Institute of Medicine equations (5) as described in the
Methods.
b
ß values were obtained in the log10 scale of the ratio of mEI to pER which more closely approximated a normal distribution and were then transformed back to percent
difference for ease of interpretation; thus, b values less than 1 indicate negative slopes.
c
Total dietary variety is the count of unique foods consumed daily (mean for 2 days).

energy from nondairy beverages, and more energy dense foods had a
higher mEI relative to pER. In addition, selecting a diet with higher
variety was associated with a higher mEI relative to pER in females
(R2 5 0.39, P < 0.001). There was a similar trend for males that
almost achieved statistical significance (R2 5 0.08, P 5 0.06). Finally,
there was also a significant interaction effect of GI with BMI on
EBal. To depict this interaction of two continuous variables, BMI was
categorized by cut points for normal weight (<25 kg/m2), overweight
(25-29 kg/m2), and obese (30 kg/m2), and then partial correlations
of BMI with mEI relative to pER were calculated within each BMI
category. The association between GI and EBal was positive but not
significant in obese individuals (R2 5 0.06, P 5 0.25), and there was a
significant negative relationship between GI and EBal in normal
weight individuals (R2 5 0.14, P < 0.001), and this trend was also
observed in overweight individuals (R2 5 0.06, P 5 0.08). Energy den-
sity accounted for the majority of the variance in EBal in both gen-
ders (R2 5 0.57) followed by dietary variety in females (R2 5 0.39).
Discussion
There remains uncertainty over the extent to which dietary composi-
tion and other dietary variables influence EI, and can therefore help
prevent weight gain leading to obesity (6,11,23,24). This study
found that, in a cafeteria-style research setting in which ad libitum
food intake was measured over 2 days in adults self-selecting their
meals and snacks from a large menu of traditional American foods,
there were several dietary composition parameters that had signifi-
cant, independent associations with EI and EBal. In particular,
energy density, percentage of energy from nondairy beverages, and
dietary variety (in females), were all positively associated with
EBal, while percent energy from protein, and GI in normal weight
individuals, were negatively associated with EBal. Energy density
and dietary variety (in females) accounted for most of the variance
in EBal, indicating they had the strongest influence on EBal in this
study. The finding that multiple dietary variables were, as a group,
all significantly and independently associated with EI is consistent
with a model of additive influences of dietary variables on EI and
EBal, which we have proposed previously (25).
To our knowledge, this is the first study of dietary predictors of
short-term EBal and EI that mEI precisely when foods of different
composition could be self-selected and consumed ad libitum. The
results thus reflect the increasingly common practice of consuming
foods that are self-selected and ready-prepared (e.g. from restaurants
or supermarkets) (26), and are consistent with those previous pre-
load, short-term, and epidemiological studies identifying significant
positive relationships between increased EI and a higher proportion
of total daily energy from liquid calories (27,28), higher dietary
variety (24), and higher energy density (6,29). We also identified an
effect of gender on the relationships between dietary variety and EI,
which is consistent with previous reports (30). Specifically, the EI
of females was significantly influenced by variety while male EI
may not have been since, although a similar trend was observed in
males, it was not quite statistically significant. Further studies are
needed to examine the effects of gender on responsiveness to dietary
variety, which has implications for gender-specific dietary recom-
mendations for weight control.
The central purpose of this study was to examine whether different
dietary properties may exert additive influences on EI and, hence,
EBal. For example, it examined whether a diet that is both higher in
protein and lower in variety spontaneously results in more negative
EBal than a diet that is higher in protein or lower in variety. Most

www.obesityjournal.org Obesity | VOLUME 00 | NUMBER 00 | MONTH 2014 5

Obesity Independent, Additive Effects of Five Dietary Variables Urban et al.

Figure 2 Partial correlation plots of significant predictors of measured energy intake (mEI) with regard to predicted energy requirement
(pER). • 5 Male, 䊊 5 Female, BMI < 25 (•), 25-29 (䊊), >30 (䊊). Plots are in the log10 scale of the ratio of mEI to pER to approximate the
normal distribution. Plots are adjusted for age, gender and body mass index and for the other covariates in the model (Table 3). Total die-
tary variety is the count of unique foods consumed daily (averaged from two days).

6 Obesity | VOLUME 00 | NUMBER 00 | MONTH 2014 www.obesityjournal.org

Original Article
OBESITY BIOLOGY AND INTEGRATED PHYSIOLOGY
studies on the relationship between dietary factors and EI have
examined either the effects of single nutrients or have studied die-
tary patterns without distinguishing the effects of different nutrients.
Specifically, the mixed model used in this analysis, achieved by
backward stepping, indicated additive effects of dietary variety,
energy density, protein, liquid nondairy calories, and GI on EI rela-
tive to pER, with the strongest contributions from energy density
and dietary variety. In other words, subjects self-selecting diets with
favorable profiles for all of these parameters (i.e. higher protein,
lower liquid calories, lower energy density, lower variety, and lower
GI) had a lower EI than subjects who self-selected diets with favor-
able profiles for only one or two parameters.
To our knowledge this study provides the first evidence to support a
model of additive effects of dietary composition factors on EI and
EBal (25). Some epidemiological studies have reported data that
potentially relates to this issue, but such studies are recognized to be
confounded by widespread selective under-reporting of different
food groups and greater under-reporting among individuals with
high BMI (3), and have previously reported both significant (31-33)
and nonsignificant (34-36) additive effects of different dietary com-
position factors. Moreover, none of those studies reported the
broader range of dietary factors in the same model that we used
here, including dietary variety and GI along with more traditional
macronutrient parameters. In this study, EI was measured precisely
in subjects able to self-select foods and portions, and the results
indicate that self-selection of meals and snacks that are higher in
protein and lower in energy-containing beverages, lower in dietary
variety, and lower in energy density and GI may be particularly ben-
eficial for weight regulation. For example, at the maximum values
for percent energy from protein, energy containing beverages, die-
tary variety, energy density, and GI observed in the current study, a
41-year-old female with a BMI 5 30 kg/m2 would consume 55%
more energy than pER, whereas at the minimum values for these
dietary factors observed in the current study, the same female would
consume 40% less energy than pER. These comparisons likely over-
estimate the potential contribution of additive dietary factors to
energy regulation, especially with regard to dietary composition in
weight loss as other factors (e.g. hunger) may become important.
Nevertheless, even a decrease of just 5% from the highest levels in
each of the above categories would produce a 10% decrease in EI in
a 41-year-old female with BMI of 30 kg/m2 which, if sustained,
would translate into 200 kcal per day in a person typically consum-
ing 2000 kcal/day, which could potentially have a substantial effect
on body weight over time. The additive effects of dietary energy
density, variety, GI and percent of daily EI from protein and liquid
energy sources are, thus, very substantial within the context of levels
of change in EI that impact energy regulation.
The influence of GI on EI has been an area of considerable contro-
versy, and this study found a significant interaction between GI and
BMI such that individuals with a normal BMI (<25 kg/m2) had a
negative association between GI and BMI and obese individuals had
a positive but nonsignificant association between GI and EI. These
findings suggest that higher GI diets may have different and perhaps
even opposite effects on EI depending on body habitus, specifically
to facilitate under-eating in normal weight people in comparison
with no effect or overeating in obese people. The finding of a nega-
tive association of GI with EI in people with a normal BMI conflicts
with some (37) but not all (7) previous studies of GI and energy
regulation. Aside from the fact that this study focused on the effects
www.obesityjournal.org
Obesity
of dietary variables on ad libitum energy intake, whereas most stud-
ies of GI are concerned with its efficacy for facilitating intentional
weight loss, one methodological factor that may have influenced the
results obtained here is that very low GI carbohydrate sources (such
as legumes and intact wheat kernels) were not emphasized on the
study menu because they are not considered traditional American
foods, and thus our range of GI was relatively narrow. Nevertheless,
our general finding of a different effect of GI on EBal depending on
obesity status is consistent with some previous work from our own
laboratory which demonstrated a trend towards lower EI with high
GI diets in individuals with low insulin secretion whereas in the
same study we found that high GI diets increased EI in individuals
with high insulin secretion (38). The reasons for this apparent
dichotomy in EBal responses to dietary GI are not understood and
further work in this area is clearly needed.
Limitations in this study included the lack of a direct measurement
of EBal during the study period. Nevertheless, the mean value for
EI was very close to pER, indicating that subjects consumed levels
of energy that were close to their expected requirement for weight
maintenance during the study period. Efforts were made to reduce
awareness of investigator scrutiny of dietary intake during this
study, but it is possible that this did influence food choices and food
intake. Additionally, although foods with a variety of fiber contents
were available on the menu, the lower fiber foods were routinely
selected by subjects so that data were only available for a narrow
range of fiber intakes [below national fiber recommendations of 21-
38 g/day (5)]. This prevented an assessment of the effects of dietary
fiber values spanning the recommended range on EI.
In summary, there appears to be substantial additive effects of
energy density, dietary variety, percent energy from protein and
non-dairy liquid sources, and GI on short-term EI and EBal. These
findings suggest that dietary recommendations which combine multi-
ple favorable dietary factors may be more effective for protecting
against overeating than recommendations focusing on individual die-
tary targets or general dietary patterns, and intervention studies are
needed to test this hypothesis and determine how such recommenda-
tions might be implicated on a practical level. O
Acknowledgments
The authors would like to acknowledge the help of Gerard Dallal
for his help with statistics, Thomas Bouchard for his help recruiting
twins raised apart and our dedicated participants, without whom this
study would not have been possible.
C 2014 The Obesity Society
V
References
1. U.S. Department of Agriculture and U.S. Department of Health and Human
Services. Dietary Guidelines for Americans, 2010, 7th ed. Washington, DC: U.S.
Government Printing Office; December 2010.
2. Burke LE, Wang J, Sevick MA. Self-monitoring in weight loss: A systematic
review of the literature. J Am Diet Assoc 2011;111:92-102.
3. Trabulsi J, Schoeller DA. Evaluation of dietary assessment instruments against
doubly labeled water, a biomarker of habitual energy intake. Am J Physiol
Endocrinol Metab 2001;281:E891-E899.
4. Suzuki K, Jayasena CN, Bloom SR. Obesity and appetite control. Exp Diabetes Res
2012;2012(824305).
5. Institute of Medicine. Dietary Reference Intakes for Energy, Carbohydrate, Fiber,
Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids (Macronutrients).
Washington DC: National Academy of Sciences; 2005.
Obesity | VOLUME 00 | NUMBER 00 | MONTH 2014 7
Obesity
6. Rolls BJ. The relationship between dietary energy density and energy intake.
Physiol Behav 2009;97:609-615.
7. Flint A, Gregersen NT, Gluud LL, et al. Associations between postprandial insulin
and blood glucose responses, appetite sensations and energy intake in normal
weight and overweight individuals: A meta-analysis of test meal studies. Br J Nutr.
2007;98:17-25.
8. McCrory MA, Burke A, Roberts SB. Dietary (sensory) variety and energy balance.
Physiol Behav 2012;107:576-583.
9. Cassady BA, Considine RV, Mattes RD. Beverage consumption, appetite, and
energy intake: What did you expect? Am J Clin Nutr 2012;95:587-593.
10. Martens EA, Lemmens SG, Westerterp-Plantenga MS. Protein leverage affects
energy intake of high-protein diets in humans. Am J Clin Nutr 2013;97:86-93.
11. Slavin JL. Dietary fiber and body weight. Nutrition 2005;21:411-418.
12. Melanson KJ, Summers A, Nguyen V, et al. Body composition, dietary
composition, and components of metabolic syndrome in overweight and obese
adults after a 12-week trial on dietary treatments focused on portion control, energy
density, or glycemic index. Nutr J 2012;11:57.
13. Sacks FM, Bray GA, Carey VJ, et al. Comparison of weight-loss diets with
different compositions of fat, protein, and carbohydrates. N Engl J Med 2009;360:
859-873.
14. Alhassan S, Kim S, Bersamin A, King AC, Gardner CD. Dietary adherence and
weight loss success among overweight women: Results from the A TO Z weight
loss study. Int J Obes. 2008;32:985-991.
15. Das SK, Gilhooly CH, Golden JK, Pittas AG, Fuss PJ, Cheatham RA, Tyler S,
Tsay M, McCrory MA, Lichtenstein AH, Dallal GE, Dutta C, Bhapkar MV, Delany
JP, Saltzman E, Roberts SB. Long-term effects of 2 energy-restricted diets differing
in glycemic load on dietary adherence, body composition, and metabolism in
CALERIE: A 1-y randomized controlled trial. Am J Clin Nutr. 2007;85:1023-1030.
16. Larsen TM, Stine-Mathilde D, van Baak M, Jebb SA, Papadaki A, Pfeiffer AFH,
Martinez JA, Handjieva-Darlenska T, Kunesova M, Pihlsgard M, Stender S, Holst
C, Saris WHM, Astrup A. Diets with high or low protein content and glycemic
index for weight-loss maintenance. N Engl J Med. 2010;363:2101-2113.
17. Elder SJ, Neale MC, Fuss PJ, Lichtenstein AH, Greenberg AS, McCrory MA,
Bouchard TJ, Saltzman E, Roberts SB. Genetic and environmental influences on
eating behavior - A study of twin pairs reared apart or reared together. Open Nutr
J. 2012;6:59-70.
18. Elder SJ, Roberts SB, McCrory MA, Das SK, Fuss PJ, Pittas AG, Greenberg AS,
Heymsfield SB, Dawson-Hughes B, Bouchard TJJ, Saltzman E, Neale MC. Effect
of body composition methodology on heritability estimation of body fatness. Open
Nutr J. 2012;6:48-58.
19. Rolls BJ, Dimeo KA, Shide DJ. Age-related impairments in the regulation of food
intake. Am J Clin Nutr. 1995;62:923-931.
20. Bouchard TJ, Lykken DT, McGue M, Segal NL, Tellegen A. Sources of human
psychological differences: The Minnesota study of twins reared apart. Science.
1990;250:223-228.
21. Ledikwe JH, Blanck HM, Khan LK, Serdula MK, Seymour JD, Tohill BC, Rolls
BJ. Dietary energy density determined by eight calculation methods in a nationally
representative United States population. J Nutr. 2005;135:273-278.
8 Obesity | VOLUME 00 | NUMBER 00 | MONTH 2014
Independent, Additive Effects of Five Dietary Variables Urban et al.
22. Haskell WL, Lee I, Pate RR, Powell KE, Blair SN, Franklins BA, Macera CA,
Heath GW, Thompson PD, Bauman A. Physical activity and public health: Updated
recommendation for adults from the American college of sports medicine and the
American heart association. Med Sci Sport Exerc. 2007;39(8):1423-1434.
23. Abete I, Astrup A, Martinez JA, Thorsdottir I, Zulet MA. Obesity and the
metabolic syndrome: Role of different dietary macronutrient distribution patterns
and specific nutritional components on weight loss and maintenance. Nutr Rev.
2010;68:214-231.
24. Raynor HA. Can limiting dietary variety assist with reducing energy intake and
weight loss? Physiol Behav. 2012;106:356-361.
25. Roberts SB, Urban LE, Das SK. Effects of dietary factors on energy regulation:
Consideration of multiple- versus single-dietary-factor models [published online
ahead of print April 17 2014]. Physiol Behav 2014. doi: 10.1016/
j.physbeh.2014.04.024.
26. .US Department of Agriculture. Away From Home: Percentages of Selected
Nutrients Contributed by Foods Eaten Away from Home, by Gender and Age, in
the United States, 2009-2010. NHANES; 2009-2010. Available at: www.ars.usda.
gov/ba/bhnrc/fsrg. Accessed December 17, 2012.
27. Kant AK, Graubard BI, Mattes R. Association of food form with self-reported 24-h
energy intake and meal patterns in US adults: NHANES 2003-2008. Am J Clin
Nutr 2012;96:1369-1378.
28. Dennis EA, Flack KD, Davy BM. Beverage consumption and adult weight
management: A review. Eating Behav 2009;10:237-246.
29. de Castro JM. Macronutrient and dietary energy density influences on the intake of
free-living humans. Appetite 2006;46:1-5.
30. Rolls BJ, Fedoroff IC, Guthrie JF. Gender differences in eating behavior and body
weight regulation. Health Psychology 1991;10:133-142.
31. Austin GL, Ogden LG, Hill JO. Trends in carbohydrate, fat, and protein intakes and
association with energy intake in normal-weight, overweight, and obese individuals:
1971-2006. Am J Clin Nutr 2011;93:836-843.
32. Howarth NC, Huang TT, Roberts SB, McCrory MA. Dietary fiber and fat are
associated with excess weight in young and middle-aged US adults. J Am Diet
Assoc 2005;105:1365-1372.
33. de Castro JM. Dietary energy density is associated with increased intake in free-
living humans. J Nutr 2004;134:335-341.
34. Ma Y, Olendzki B, Chiriboga D, et al. Association between dietary carbohydrates
and body weight. Am J Epidemiol 2005;161:359-367.
35. Iqbal SI, Helge JW, Heitmann BL. Do energy density and dietary fiber influence
subsequent 5-year weight changes in adult men and women? Obesity 2006;14:106-
114.
36. Lyles TE3, Desmond R, Faulk LE, et al. Diet variety based on macronutrient intake
and its relationship with body mass index. Med Gen Med 2006;8:39.
37. Roberts SB, Pittas AG. The role of glycemic index in type 2 diabetes. Nutr Clin
Care 2003;6:73-78.
38. Pittas AG, Das SK, Hajduk CL, et al. A low-glycemic load diet facilitates greater
weight loss in overweight adults with high insulin secretion but not in overweight
adults with low insulin secretion in the CALERIE trial. Diabetes Care 2005;28:
2939-2941.
www.obesityjournal.org